Chromosome Number Reduction in Eremothecium coryli by Two Telomere-to-Telomere Fusions

Abstract

The genus Eremothecium belongs to the Saccharomyces complex of pre-whole-genome duplication (WGD) yeasts and contains both dimorphic and filamentous species. We established the 9.1-Mb draft genome of Eremothecium coryli, which encodes 4,682 genes, 186 tRNA genes, and harbors several Ty3 transposons as well as more than 60 remnants of transposition events (LTRs). The initial de novo assembly resulted in 19 scaffolds, which were assembled based on synteny to other Eremothecium genomes into six chromosomes. Interestingly, we identified eight E. coryli loci that bear centromeres in the closely related species E. cymbalariae. Two of these E. coryli loci, CEN1 and CEN8, however, lack conserved DNA elements and did not convey centromere function in a plasmid stability assay. Correspondingly, using a comparative genomics approach we identified two telomere-to-telomere fusion events in E. coryli as the cause of chromosome number reduction from eight to six chromosomes. Finally, with the genome sequences of E. coryli, E. cymbalariae, and Ashbya gossypii a reconstruction of three complete chromosomes of an Eremothecium ancestor revealed that E. coryli is more syntenic to this ancestor than the other Eremothecium species.

Introduction

Comparative genomics is most powerful when comparing essentially complete draft genomes. This can yield insight into the evolution of species and compiling several genomes of closely related species may allow the reconstruction of ancestral genomes. The precision of such a paleogenomic reconstruction depends on the degree of synteny, that is, conserved gene order in the studied species and on the number of sequenced genomes (Bhutkar et al. 2007; Muffato and Roest Crollius 2008; El-Mabrouk and Sankoff 2012).

Yeast species of the Saccharomyces complex have been of considerable interest based on their fermentative properties and their large evolutionary timescale spanning at least 100 Ma from an ancient whole-genome duplication (WGD) event (Wolfe and Shields 1997). Compiling the data of 11 sequenced yeast species a pre-WGD ancestor was reconstructed harboring 4,700 genes distributed on eight chromosomes (Gordon et al. 2009). Due to a WGD modern Saccharomyces sensu stricto species contain 16 chromosomes per haploid genome. From an ancestral genome, the evolutionary paths in terms of duplications, inversions, and reciprocal translocations can be inferred. Interestingly, a comparison of the protoploid Lachancea kluyveri, which contains eight chromosomes, with this pre-WGD ancestor allowed the reconstruction of the complete evolutionary genome rearrangement history of L. kluyveri (Gordon et al. 2011). Chromosome number, however, is not static and several protoploid, that is, “pre-WGD” and post-WGD species of the Saccharomyces complex have undergone chromosome number reductions.

There are basically two mechanisms for a reduction in chromosome number without loss of coding information: 1) By telomere-to-telomere fusion and inactivation of one of the two centromeres of such a newly formed chromosome or 2) by breakage of a chromosome at a centromere and fusion of the two chromosomal arms to two telomeres of other chromosomes. The first seems to be more widespread than the latter as breakage of a chromosome at a centromere was so far only observed in Eremothecium/Ashbya gossypii (Gordon et al. 2011).

The genus Eremothecium constitutes clade 12 of the Saccharomyces complex (Kurtzman and Robnett 2003). The type of strain of this genus, Eremothecium cymbalariae, was first isolated and described in 1888 by Borzi and recently its genome sequence has been determined (Borzi 1888; Wendland and Walther 2011). Eremothecium species are known to cause fruit rotting, for example, on cotton or tomato (Miyao et al. 2000). Insect vectors are required for dispersal of the fungi, particularly milkweed bugs, boxelder bugs, or other stink bugs (Dietrich et al. 2013). The disease caused is referred to as stigmatomycosis or “yeast spot disease” (Ashby and Nowell 1926).

Major interest in Eremothecium species was attracted by A. gossypii as a potent overproducer of riboflavin/vitamin B₂ (Kato and Park 2012). Based on its molecular genetic tractability, Ashbya soon became a model for studies of fungal cell biology and filamentous growth (Wendland and Walther 2005). Comparisons of the complete genomes of the filamentous fungi A. gossypii and E. cymbalariae revealed that E. cymbalariae harbors greater similarity to the pre-WGD ancestor than A. gossypii (Dietrich et al. 2004; Wendland and Walther 2011). This includes 1) eight chromosomes in E. cymbalariae compared with only seven in A. gossypii, 2) a low GC content of 40.3% in E. cymbalariae (as found in other yeast species) versus the remarkably high GC content of 51.8% in A. gossypii, 3) larger blocks of synteny, 4) a similar gene density between E. cymbalariae and the yeast ancestor, and 5) the presence of a Ty3 transposon in E. cymbalariae, which is absent in A. gossypii (Wendland and Walther 2011). Ashbya gossypii is thus characterized by a more divergent, more rearranged, and much more compact genome—largely due to size reductions in intergenic regions—compared with the E. cymbalariae genome.

The Eremothecium genus is not only composed of true filamentous fungi but it contains also dimorphic yeasts, for example, Nematospora/Holleya sinecauda and Nematospora/Eremothecium coryli. Although E. cymbalariae and A. gossypii grow only in the filamentous form, dimorphic fungi generate yeast cells, pseudohyphal cells, or filaments. Emil Christian Hansen, who worked at the Carlsberg Laboratory, first described the genus Nematospora in 1904 (Hansen 1904). Later Ashbya, Nematospora, Holleya, and Eremothecium were placed in a single genus that was seeded within the Saccharomycetaceae (Kurtzman 1995; Prillinger et al. 1997). This grouping suggested that filamentous growth may have been gained in the Eremothecium genus whereas the yeast ancestor was unicellular/dimorphic (Schmitz and Philippsen 2011). To further elucidate genome evolution in Eremothecium, we established the draft genome of the dimorphic species E. coryli. Using comparative genomics and functional analysis tools, we identified the mechanism of chromosome number reduction from 8 to 6 chromosomes in E. coryli. Furthermore, based on conserved synteny, three chromosomes of an Eremothecium ancestor (ERA) could be reconstructed. Comparisons of the recent Eremothecium genomes with ERA indicate that E. coryli is most syntenic to ERA supporting the hypothesis that the lineage ancestor was a unicellular/dimorphic yeast and true filamentous growth may be an apomorphy in the Eremothecium lineage.

Materials and Methods

Strains and Media

Eremothecium coryli strain CBS 5749 was sequenced. For plasmid stability assays H. sinecauda (CBS 8199) served as a host. Strains were grown using complete media (1% yeast extract, 1% peptone, and 2% dextrose) supplemented with G418/geneticin (200 μg/ml) for the selection of antibiotic-resistant plasmid transformants or minimal media with either asparagine or ammonium sulfate as nitrogen source. For plasmid propagation, Escherichia coli DH5α was used.

Transformation of H. sinecauda

Transformation and plasmid stability assays in H. sinecauda were done as described previously (Schade et al. 2003).

Plasmid Constructs

Episomal plasmids were generated for testing of plasmid stability and centromere activity. To this end centromere DNA fragments of the E. coryli centromere loci of chromosome 1 (734 bp), 2 (1,075 bp), 3 (785 bp), 4 (821 bp), 7 (772 bp), and 8 (445 bp) were amplified by polymerase chain reaction and cloned into the high copy (autonomously replicating sequence [ARS]-containing) shuttle vector pHC shuttle (#310; Schade et al. 2003) using XbaI and XhoI restriction sites provided with the primers. This generated plasmids C875-C880. A low copy pLC shuttle (#268) containing A. gossypii ARS and centromere DNA sequences was used as a control.

Sequencing Strategy

The E. coryli genome was sequenced using Illumina HiSeq2000 next-generation sequencing with 100-bp paired-end reads and an 8-kb mate-pair library (LGC Genomics, Berlin, Germany). Sequencing generated approximately 40 million reads corresponding to more than 100× coverage of the E. coryli genome. Assembly of the genome sequencing data produced 19 scaffolds/supercontigs.

Annotation of the E. coryli Genome

The 19 scaffolds of the E. coryli draft genome were submitted to GenBank with a BioProject number (PRJNA229863) and have been deposited under accession number AZAH00000000. The mitochondrial genome has not been assembled.

The E. coryli genes were compared with the A. gossypii, E. cymbalariae, and Saccharomyces cerevisiae genomes available from Ashbya Genome Database (http://agd.vital-it.ch/index.html, last accessed May 15, 2014) and Saccharomyces Genome Database (http://www.yeastgenome.org, last accessed May 15, 2014) and GenBank using local blast tools (available at http://blast.ncbi.nlm.nih.gov, last accessed May 15, 2014). LTR sequences were identified using BLASTN. Fine annotation of the E. coryli genome used syntenic relationships to A. gossypii, E. cymbalariae, and S. cerevisiae. Unidentified E. coryli ORFs were also searched against the nonredundant data set of National Center for Biotechnology Information. The assembly of the E. coryli genome into six chromosomes was based on syntenic gene order and the prediction of reciprocal translocations. A systematic nomenclature based on this chromosome assembly was generated. As species identifier for E. coryli “Eco_” was used followed by the chromosome number (1.–6.) and the feature number (1–n starting from the first ORF at the left telomere running continuously to the last ORF [n] at the right telomere of the chromosome, e.g., Eco_1.001 for the first ORF at the left end of chromosome 1). For the identification of tRNA genes, tRNAscan (http://lowelab.ucsc.edu/tRNAscan-SE/, last accessed May 15, 2014) was used (Schattner et al. 2005).

Results

Eremothecium Genome Comparisons

Eremothecium coryli is a dimorphic fungus that lacks dichotomous tip branching characteristic for hyphal tip growth in its filamentous relatives A. gossypii and E. cymbalariae (Gastmann et al. 2007). The E. coryli strain CBS 5749 was sequenced using Illumina HiSeq2000 with 8 kb mate-pair libraries and paired-end sequencing with more than 100× genome coverage. The draft genome was assembled into 19 scaffolds (table 1). The genome size is approximately 9.1 Mb and thus of intermediate size compared with E. cymbalariae (9.7 Mb) and A. gossypii (8.7 Mb). We identified 4,682 genes, which is close to the slightly over the 4,700 genes for the other Eremothecium species indicating that our assembly is basically complete. The E. coryli genome consists of 73.6% encoding DNA with a GC content of 41.5% very similar to E. cymbalariae (73.6% coding with 40.3% GC) and in contrast to A. gossypii (79.5% coding and 51.8% GC). The apparently higher similarity between the E. coryli and E. cymbalariae genomes is also reflected by the amount of synteny blocks: Longer stretches of conserved gene order between these two species result in fewer synteny blocks (139) compared with E. coryli and A. gossypii (198) (see table 1). Interestingly, we also identified several Ty3 transposons and 83 remnants of transposition marked by LTRs (supplementary table S1, Supplementary Material online). Of these LTRs 73, that is 88%, are adjacent to tRNA genes in E. coryli (supplementary table S4, Supplementary Material online). The paired-end sequencing and scaffold assembly indicate that there are at least six full-length Ty3 transposons present in the E. coryli genome. Sequence analysis of the E. cymbalariae genome indicated only one Ty3 transposon that—based on the orientation of the LTRs—may, however, have lost its ability to transpose. We also found several LTRs positioned at the end of scaffolds in E. coryli. In three cases, we inferred reciprocal translocations at these positions for the assembly of the E. coryli genome (see below).

Table 1.

Eremothecium coryli Genome Summary

Scaffolda	Number of Genes	Scaffold Length (bp)	% Encoding	GC Content (%)	tRNAs	LTRsb	Blocks to Eremothecium cymbalariaec	Blocks to Ashbya gossypii
0	1,012	1,827,054	76.0	41.59	27	5 (1)	26	47
1	567	1,105,492	75.7	41.34	33	13 (1)	19	25
2	536	1,035,239	73.9	41.39	8	3	12	20
3	521	1,037,803	73.2	41.04	30	11 (2)	19	26
4	277	590,229	72.6	41.11	9	6	6	7
5	275	544,843	75.7	41.59	10	5	14	16
6	254	521,725	72.9	41.37	8	2	2	9
7	251	503,936	70.0	41.10	9	1	13	13
8	235	488,308	72.9	40.74	7	3	6	7
9	217	415,668	71.7	41.69	6	2	6	10
10	143	262,922	75.7	41.30	13	4	3	5
11	113	222,772	69.8	40.98	7	2	6	5
12	98	173,376	71.6	42.47	2	0	2	3
14	44	85,504	68.2	39.71	2	0	2	2
15+13	69	142,227	69.4	40.29	8	5 (1)	1	1
16	30	53,536	60.6	40.82	1	0	1	1
17+18	40	84,386	59.1	42.13	7	3 (1)	1	1
	4.682	9.095.020	73.6	41.57	187	65	139	198

^aScaffolds 15+13 and 17+18 were combined based on synteny.

^bLTRs were identified based on the direct repeat sequences flanking full-length Ty3 transposons (number in brackets).

^cBlock synteny based on conserved gene order.

Morphological differences between the filamentous Eremothecium species E. cymbalariae and A. gossypii compared with the dimorphic species including H. sinecauda and E. coryli are not necessarily also manifested in the average similarity of the protein-coding genes. Comparison of the proteomes between the three sequenced species shows an average identity of approximately 60% between these species, which is slightly higher between E. coryli and E. cymbalariae (63.2%) compared with E. coryli and A. gossypii (62.3%) (fig. 1A). Overall the three Eremothecium species share about 95% of their genes. Furthermore, E. coryli shares an additional 1% of its genes with E. cymbalariae but not with A. gossypii and a similar number with A. gossypii but not with E. cymbalariae (fig. 1B).

Fig. 1.—

Proteome and genome comparisons. (A) Pairwise proteome comparisons between Eremothecium coryli, E. cymbalariae, and Ashbya gossypii using all protein-coding genes of these Eremothecium species. (B) Diagram showing the distribution of homologous genes within Eremothecium species. Central genes (4,461 of ∼4,700) are shared by all three species. Genes in intersections are shared by only two species.

Eremothecium species are pre-WGD and thus contain unduplicated protoploid genomes. Yet, these species are not completely devoid of gene duplications. Some of them occur dispersed throughout the genome but others are present as tandem duplications. These give rise to evolutionary diversification and subfunctionalization as has been demonstrated for RHO1 paralogs in A. gossypii (Walther and Wendland 2005; Köhli et al. 2008). Out of 21 tandem duplications found in A. gossypii, E. coryli shares 13 and E. cymbalariae 9 (supplementary table S2, Supplementary Material online). The remaining A. gossypii duplications are either telomeric in A. gossypii or may hint to species-specific functions, for example, A. gossypii MCH4, which is currently under investigation. In addition to these shared duplications, there are seven tandem duplications that are specific for E. coryli. Interestingly, ABR156W/YJL212C occurs in four tandem copies. YJL212C encodes the oligopeptide transporter OPT1 in S. cerevisiae, which also transports phytochelatin (Osawa et al. 2006). This multiplication may be functionally relevant for metal homeostasis. Furthermore, there is a tandem duplication of the E. coryli paralogs of AER22W/YBR139W, which encodes a serine carboxypeptidase that is required for phytochelatin synthesis in yeast (Wünschmann et al. 2007). This suggests a functional linkage of these duplications that is specific for E. coryli.

Synteny Relationships within Eremothecium Species

Synteny describes the conservation of gene order and transcriptional orientation of homologous genes between two-related species. Comparisons of the E. coryli genome with those of E. cymbalariae and A. gossypii revealed four types of synteny relationships (fig. 2). First, by far the largest parts of all three Eremothecium genomes show synteny between all Eremothecium species. A long stretch of conserved synteny encompassing, for example, 108 genes or 230 kb of DNA, is found at the centromere locus of E. coryli chromosome 6 (fig. 2A). Second, there are regions of single block synteny between E. coryli and A. gossypii that are fragmented into multiple blocks in the E. cymbalariae genome. One example of 44 genes distributed over 85 kb on E. coryli chromosome 3 is shown in figure 2B (see below for chromosome assignments). The syntenic A. gossypii locus harbors the genes from AAL174C to AAL131C. Homologs of these genes are found in five blocks on four different chromosomes in E. cymbalariae (fig. 2B). Conversely, there are regions of single block synteny between E. coryli and E. cymbalariae that are dispersed to multiple regions in the A. gossypii genome (fig. 2C). In the example shown, also derived from E. coryli chromosome 3, 78 genes found on 138 kb in E. coryli are syntenic to E. cymbalariae Ecym_5.451 to Ecym_5.528. Finally, there are positions in the E. coryli assembly in which both A. gossypii and E. cymbalariae genomes show synteny breaks. However, we found several locations in which the E. coryli gene order is syntenic with that of the pre-WGD ancestor (fig. 2D). The region of synteny shown harbors 106 genes on 205 kb dispersed on three to four chromosomes in E. cymbalariae and A. gossypii, respectively. An analysis of the E. coryli genome for positions of such conserved ancient synteny between E. coryli and the yeast ancestor that are not conserved in either A. gossypii or E. cymbalariae identified 20 such cases (supplementary table S3, Supplementary Material online). Eleven of these were found to be associated with tRNA genes that often occur at breakpoints of synteny. All tRNAs and their scaffold positions are listed in supplementary table S4, Supplementary Material online. Due to the efficient homologous recombination machinery in Eremothecium, short homology regions provided, for example, by tRNA genes can readily serve as templates for reciprocal translocations (Steiner et al. 1995). The examples presented in figure 2B–D indicate species-specific genome evolution events. Of course, they are by far outnumbered by syntenic gene organization. Yet, these regions could be drivers of species-specific evolution and thus of interest for targeted functional analyses.

Fig. 2.—

Synteny relationships in Eremothecium genomes. (A) Single block synteny among Ashbya gossypii (Ag), Eremothecium coryli (Eco), and E. cymbalariae (Ecym). See text and supplementary material, Supplementary Material online, for the E. coryli chromosome assignments and the E. coryli systematic gene nomenclature. (B) Single block synteny between E. coryli and A. gossypii but not between E. coryli and E. cymbalariae. (C) Single block synteny between E. coryli and E. cymbalariae but not between E. coryli and A. gossypii. (D) Conserved ancient synteny between E. coryli and the reconstructed pre-WGD ancestor (Anc) not found in A. gossypii and E. cymbalariae. Such cases not only support our scaffold assembly but are also instrumental in generating an ancestral gene order. Red connectors were used to link each homologous gene pair between Eremothecium species; green connectors in (D) were used to link homologs between E. coryli and the pre-WGD ancestor. Graphs were generated using Strudel software (http://bioinf.hutton.ac.uk/strudel/, last accessed May 15, 2014).

Identification of Centromere Loci in E. coryli Scaffolds

Previously, we identified eight centromere loci in E. cymbalariae providing evidence that an ERA, similarly to the yeast ancestor, also contained eight chromosomes (Wendland and Walther 2011). By searching for homologs of centromere-associated E. cymbalariae genes in E. coryli, we identified all eight syntenic loci (fig. 3). At these loci, some additions are present in E. coryli, for example, a YCR004C homolog of unknown function that is absent from both A. gossypii and E. cymbalariae. These loci provide clear direction for the search for centromere DNA in E. coryli. Centromere DNA in Eremothecium is very similar to that of S. cerevisiae in that there are conserved centromere DNA elements (CDEI, CDEII, and CDEIII) with the sole difference that the AT-rich CDEII is twice as long in Eremothecium as in S. cerevisiae (Dietrich et al. 2004). Alignment of the putative centromere regions allowed the identification of six bona fide centromeres in E. coryli. In the syntenic E. coryli region harboring CEN8 in E. cymbalariae, we could not locate any centromere DNA. For the syntenic region of CEN1 similarity to the core sequence of CDEIII was found, however, the surrounding sequence did not match the CDEIII consensus and, furthermore, CDEI was not present. Moreover, two of the centromere loci, CEN4 and CEN8, are located on scaffold 1 (fig. 4). This suggests that only six of these eight loci harbor functional centromeres. To test for centromere function of the E. coryli CEN1 and CEN8 loci in vivo, we used a plasmid stability assay that was originally developed for yeast (Murray and Szostak 1983). Holleya sinecauda/E. sinecaudum served as a host as previously described (Schade et al. 2003). In this assay, transformants harboring ARS-plasmids will form only small colonies compared with transformants carrying CEN-ARS-plasmids, which is based on the improved segregation properties of centromere-bearing plasmids. Because of the plasmid-encoded antibiotic resistance gene, daughter cells without plasmid are sensitive to the antibiotic and die. With this assay, we could demonstrate that the intergenic regions of, for example, CEN4 and CEN7 harbor functional centromeres whereas E. coryli CEN1 and CEN8 are nonfunctional (fig. 5).

Fig. 3.—

Centromere loci in Eremothecium. Identification of eight Eremothecium coryli loci harboring six functional centromeres was based on synteny to Ashbya gossypii and E. cymbalariae. Arrows indicate transcriptional orientation of genes. Arrows for centromeres indicate the orientation of centromere DNA elements (CDEI–CDEII–CDEIII). Special features are highlighted (YCR004C and TY3 absent from A. gossypii and E. cymbalariae; ABL004W absent from E. cymbalariae) and systematic gene nomenclature was used for each species. Eremothecium coryli CEN1 and CEN8 do not harbor conserved centromere DNA elements (see also fig. 4).

Fig. 4.—

Analysis of centromere DNA elements in Eremothecium coryli. CEN sequences were identified based on the highlighted CDEI (CAYCTG) and CDEIII (TCCGAA) consensus sequences. The CDEII spacers are AT rich (>70%) and about 165 bp in length. The intergenic region between the E. coryli homologs of AAL174C and ACR029C is only 291 bp lacking conserved sequences for CEN8 (marked as CEN8). EcoCEN1 sequence is without conserved CDEI and with only partially conserved CDEIII (CEN1). Positions of these loci on E. coryli scaffolds and assembled chromosomes (see below) are indicated.

Fig. 5.—

In vivo assay for centromere activity. Holleya sinecauda was transformed with ARS-plasmids additionally containing regions harboring Eremothecium coryli centromere loci as indicated. Control plasmids with only an ARS give rise to small and irregular colonies. The addition of centromere DNA (AgCEN5) leads to faithful plasmid segregation of plasmids and results in large colonies. Nonfunctional E. coryli centromere loci are marked by asterisks. Five initial transformants were repicked on selective plates and incubated at 30°C for 3 days prior to photography.

Chromosome Number Reduction in E. coryli

The previous section indicated that E. coryli has decommissioned two centromeres. As we identified eight syntenic centromere loci in E. coryli, this can be explained by two cases of telomere-to-telomere fusion of two chromosomes. Concomitant with each telomere-to-telomere fusion, loss of function mutations in one of the two centromeres of each new chromosome must have occurred. In total E. coryli should thus contain six chromosomes. We therefore analyzed the E. coryli genome data for traces of these telomere-to-telomere fusion events.

The reconstructed pre-WGD ancestor provides 8 chromosomes with 16 ancient telomeres (Gordon et al. 2009). Remarkably, 15 of these loci are conserved at telomeres in E. cymbalariae and 9 out of those loci are also at telomeres in A. gossypii (fig. 6). We then went on to identify the scaffold positions of the respective telomere-linked genes in E. coryli. Ten of these were located at scaffold ends, six were internal. Interestingly, two scaffolds, S5 and S7, harbor homologs located at two different telomeres in the pre-WGD ancestor each (fig. 6). Strikingly, these telomeric loci are directly adjacent to each other on both scaffolds providing direct evidence for two telomere-to-telomere fusion events. According to the nomenclature of the yeast ancestor, these fusions involved the telomeres of Anc3R and Anc8R in one case and Anc6R and Anc7L in the other (fig. 7A and B). Interestingly, the telomere-to-telomere fusion located on scaffold 5 would not have been detected unambiguously without the reconstructed pre-WGD ancestral genome. The respective homologs in A. gossypii are found at internal positions in three different chromosomes. In E. cymbalariae, the telomere of Anc_3R is also telomeric at chromosome 6L, whereas the telomere of the ancestral chromosome 8R became internalized.

Fig. 6.—

Identification of telomere loci in Eremothecium coryli. The positions of Eremothecium homologs of telomere linked genes of the pre-WGD ancestor were identified. In E. cymbalariae 15/16 ancestral telomere loci are conserved telomeres, for example, genes located at the left end of chromosome 1 (Anc_1L) of the yeast ancestor are found at E. cymbalariae chromosome 1L, and genes at Anc_1R are found at E. cymbalariae chromosome 3R. Genes from Anc_5R were relocated between two telomeres in E. cymbalariae (5R+4L). Lack of conservation of telomere positioning is indicated as (—). In Ashbya gossypii, 9/15 telomere loci are conserved. For analysis of E. coryli, the assembled scaffolds were used. Here, telomere linked genes were found at the end of 10 scaffolds. The remaining six ancestral telomere positions were found within scaffolds (e.g., intS5). Note two scaffolds (S5 and S7) were identified twice—directing our search for telomere-to-telomere fusion events in E. coryli.

Fig. 7.—

Telomere-to-telomere fusion events in Eremothecium coryli. Two loci indicative of telomere-to telomere fusion in E. coryli were identified on scaffolds 5 and 7. The order of E. coryli genes of scaffold 5 on CHR6 (A) and scaffold 7 on CHR4 (B) is shown aligned with homologs from Ashbya gossypii, E. cymbalariae, and the pre-WGD ancestor. Telomere ends are drawn with round-shaped edges, internal regions are depicted as open bars. Positions of E. coryli genes on the assembled E. coryli chromosomes are shown. Numbers within the E. coryli chromosomes correspond to the contributing scaffolds (see also fig. 8).

Evidence of a telomere-to-telomere fusion found in E. coryli scaffold 7 is based both on conservation in Eremothecium and the pre-WGD ancestor. In A. gossypii, one telomeric end is conserved, whereas the location of ACR293C is telomeric both in E. cymbalariae and A. gossypii, but this gene has not been annotated in the yeast ancestor. The genes found linked in E. coryli are dispersed to two telomeres in E. cymbalariae indicating that this is a composite locus in E. coryli.

In the yeast ancestor Anc_7.1 encodes a glutamate dehydrogenase, the S. cerevisiae ortholog of YAL062W/GDH3. This gene is absent from both A. gossypii and E. cymbalariae. Interestingly, this gene has been conserved in E. coryli at the junction of the telomere fusion. The gene is functional and conveys growth to E. coryli using ammonium sulfate as sole nitrogen source. Minimal media for growing A. gossypii or E. cymbalariae are supplemented instead with asparagine as nitrogen source as they cannot grow in standard minimal medium without amino acids and with ammonium sulfate generally used for S. cerevisiae propagation (to be published elsewhere).

Next to E. coryli GDH3 two tRNAs are located. This suggests that the telomere-to-telomere fusion may have been brought about by homologous recombination involving these tRNAs rather than by head-to-head fusion of two telomeres (fig. 7B).

Assembly of the E. coryli Genome

The initial assembly of the E. coryli genome provided 19 scaffolds. Using conserved/ancient synteny, we aligned these scaffolds into six chromosomes. This required linking of scaffolds at 13 positions. In seven cases, these assignments were based on synteny with the other Eremothecium species and the pre-WGD ancestor. One other case was Eremothecium specific regarding the duplication of FLO5 (AFL092C/AFL095C). Another one involved synteny at the rDNA-repeat locus. The remaining four cases involved reciprocal translocations. For chromosome 6, two single reciprocal translocations can be inferred. One involved the A. gossypii homologs AGL220W-AER272C and AGL219W-AER273C whereas the other occurred between AER168C-ABL066C and AER169C-ABL065W. More than one reciprocal translocation is required to generate chromosome 1. In this case, both tRNA sequences and LTRs can be found at the scaffold ends, which generated difficult regions for automated assembly and regions that were also not covered by the 8 kb library used for sequencing. We conclude that based on the low number of scaffolds and by using comparative genomics, the assembly of the E. coryli genome into six chromosomes can be done (fig. 8). We thus assigned systematic names to all identified E. coryli genes based on their position in this assembly, for example, Eco_1.001 for the first ORF at the left end of chromosome 1 counting up to the right end of chromosome 1 harboring Eco_1.514 (see supplementary material, Supplementary Material online).

Fig. 8.—

Assembly of Eremothecium coryli chromosomes. The 19 scaffolds from the original assembly left 13 gaps. Scaffolds were conceptually linked based on conserved synteny, which closed 10 gaps. Single reciprocal translocation closed one gap (between scaffolds 4 and 5). The remaining two gaps on chromosome 1 (scaffolds 11 and 10 and scaffolds 10 and 18) were linked by a set of reciprocal translocations. The size of each chromosome is according to scale. Scaffolds (also to scale) merged into chromosomes are indicated above the individual chromosomes. Key genome features are shown in the legend.

Based on this assembly, the E. coryli chromosomes are between 985 and 2,330 kb in size. We identified three mating type loci: A presumably active MATα and a telomeric HMLα on chromosome 2 and a telomeric HMRa on chromosome 4. The dispersal of mating type loci to different chromosomes has also been found in A. gossypii, whereas in E. cymbalariae all three mating type loci are located on chromosome 1 (Wendland and Walther 2005, 2011; Dietrich et al. 2013).

Assembly of an ERA

Eremothecium coryli now presents the third Eremothecium genome that has been sequenced next to completion. Due to the large degree of synteny and with the ability to compare gene order with the reconstructed pre-WGD ancestor, we aimed at reconstructing individual segments of an ERA. We used a manual parsimony approach based on block synteny. We started at the eight centromere loci and assembled synteny blocks in both directions toward the telomeres. At breakpoints of synteny in one Eremothecium species or the pre-WGD ancestor, the conserved gene order of at least two Eremothecium genome assemblies was relied on. This generated a telomere-to-telomere assembly of three ERA chromosomes, termed CHR3, CHR4, and CHR7 based on the founding centromeres (fig. 9). ERA_CHR3 contains 701 genes, ERA_CHR4 451 genes, and ERA_CHR7 732 genes in this assembly (see supplementary material, Supplementary Material online). At positions were all Eremothecium genomes differ among themselves and compared with the pre-WGD ancestor no conclusive progression could be called. Inclusion of further Eremothecium genomes will be required to improve this ERA assembly.

Fig. 9.—

Comparative view of genome rearrangements. The compiled ERA was compared with the pre-WGD ancestor and Ashbya gossypii (A) and to Eremothecium coryli and E. cymbalariae (B). Each pair of homologous genes is linked by one line between the genomes—consecutive blocks of homology show as bars. The more individual lines emanating from ERA toward one genome the more genomic rearrangements occurred. This identifies E. coryli with the least number of rearrangements and A. gossypii with most rearrangements (for full details, see supplementary material, Supplementary Material online). Strudel software (http://bioinf.hutton.ac.uk/strudel/, last accessed May 15, 2014) was used to generate the overviews.

However, the ERA chromosome assembly of at present three chromosomes allows a view on the series of rearrangements that led from the ERA to the present-day Eremothecium species. Interestingly, this shows that the E. coryli genome is more syntenic to ERA than either of the other Eremothecium species or the pre-WGD ancestor, whereas A. gossypii harbors the most rearranged genome of these Eremothecium species (fig. 9).

Discussion

Once the yeast genome project was finished the wealth of information that can be drawn from a genome project became immediately clear (Goffeau et al. 1996). One striking result was the discovery of duplicated groups of genes on chromosome XIV and, more comprehensively, the WGD (Philippsen et al. 1997; Wolfe and Shields 1997). The yeast genome sequence was instrumental in getting other genome sequencing efforts under way. Particularly the genomes of A. gossypii and Lachancea waltii, two protoploid, “pre-WGD,” species, reinforced the concept of genome evolution by a WGD in the Saccharomyces lineage (Dietrich et al. 2004; Kellis et al. 2004). With an increasing number of complete genomes and draft genome sequences available for the Saccharomyces lineage, it became possible to reconstruct a yeast ancestral genome as it may have existed just prior to the WGD based on syntenic gene order conservation (Gordon et al. 2009).

The Saccharomyces complex has been resolved into 14 clades with clade 12 representing the genus Eremothecium (Kurtzman and Robnett 2003). This genus harbors both dimorphic (E. coryli and H. sinecauda) but also true filamentous fungi (A. gossypii and E. cymbalariae). The genus is of 2-fold commercial interest. Ashbya gossypii has long been known as an overproducer of riboflavin but species of this genus cause yeast spot disease or stigmatomycosis (Stahmann et al. 2000; Dietrich et al. 2013). For dispersal plant-feeding insect vectors of the suborder Heteroptera are used. A very persuasive hypothesis on how Ashbya developed into a riboflavin overproducer has been put forward: Some insects may be enabled to feed on toxic alkaloid-producing plants such as oleander when harboring Ashbya, whose riboflavin detoxifies these alkaloids and thus opens this ecological niche for both fungal and insect species (Dietrich et al. 2013).

Here, we have sequenced the first dimorphic Eremothecium species. Based on synteny, we identified eight E. coryli loci homologous to E. cymbalariae centromere loci. Previously, the heterologous function of A. gossypii centromere DNA in H. sinecauda was shown (Schade et al. 2003). Using this assay, we could show that CEN1 and CEN8 were decommissioned in E. coryli. Concomitantly, we identified two sites of telomere-to-telomere fusion based on conserved sequences located to telomeres in E. cymbalariae and the pre-WGD ancestor (Gordon et al. 2011; Wendland and Walther 2011). Interestingly, CEN8 in A. gossypii has also been eliminated. However, the mechanism has been different. Instead of a telomere-to-telomere fusion in Ashbya a break (or nonreciprocal translocation) at the centromere and fusion of the two chromosome arms to two different telomeres occurred. The consequences of this restructuring of CEN8 are unclear. Yet, since E. coryli is a dimorphic fungus (lacking the characteristic Y-shaped dichotomous tip branching) and A. gossypii is a true filamentous fungus, we do not consider these events to be decisive for the evolution of hyphal growth—also given that the filamentous E. cymbalariae possesses a functional CEN8.

Eremothecium CEN8 has been assigned to chromosome 5 of the pre-WGD ancestor (Anc_CEN5), whereas CEN1 of Eremothecium corresponds to Anc_CEN1. Anc_CEN5 was also lost in Candida glabrata. Similarly, Anc_CEN1 was lost in C. glabrata and also in Vanderwaltozyma polyspora (Gordon et al. 2011).

The internalization of telomeres, for example, via telomere-to-telomere fusions may preserve genes by placing them in a genomic context that may constrain their further evolution or alteration of expression patterns compared with more rapidly evolving telomeric loci (Teixeira and Gilson 2005; Batada and Hurst 2007; Ottaviani et al. 2008). In the case of the Anc6R-Anc7L fusion in E. coryli, a homolog of glutamate dehydrogenase (ScGDH3) was retained that has been lost in A. gossypii and E. cymbalariae. EcoGDH3 enables E. coryli growth in media containing ammonium sulfate as sole nitrogen source. Similarly, via internalization of telomere Anc4L in E. coryli, a homolog of a Lachancea thermotolerans gene with similarity to a zinc-finger transcription factor (ScRDS1) has been retained.

With the currently available genome sequences of Eremothecium species and in combination with the pre-WGD ancestor, the reconstruction of an ERA was initiated and generated three of the eight chromosomes. This ancestral karyotype allows insight into chromosomal evolution that occurred within the Eremothecium lineage and also in comparison to other genera of the Saccharomyces complex. The E. coryli genome is more syntenic to ERA than the filamentous Eremothecium species. This may suggest that the ERA was a unicellular/dimorphic yeast whereas true hyphal growth is an apomorphy in the Eremothecium lineage. The independent evolution of hyphal growth in different ascomycetous lineages will fuel future comparative mechanistic studies to understand the molecular wiring of hyphal growth.

Paleogenomic studies of reconstructing ancestral karyotypes may provide hints of decisive evolutionary steps in a lineage (Yegorov and Good 2012). Comparison of lineage-specific ancestral genomes may provide insight into evolutionary steps at branch-points in phylogenetic trees. This directs future research to positions of synteny breaks, for example, between ERA and the pre-WGD ancestor for gene functions or changes in gene regulation that may have distinguished the Eremothecium clade from other Saccharomycetes in terms of filamentous growth, sporulation, or general metabolism.

Finally, by using build-a-genome methodologies, it has been demonstrated that synthetic DNA segments can be assembled (Dymond et al. 2009, 2011). With this technology even complete synthetic ancestral genomes could be generated and studied in the future.

Supplementary Material

Supplementary tables S1–S4 and files S1 and S2 are available at Genome Biology and Evolution online (http://www.gbe.oxfordjournals.org/).