Introduction

A diverse array of transposable elements (TEs) have been identified in fungi, most of which are categorized as class I (transposing via RNA intermediate) or class II (DNA intermediate) (reviewed by Kempken and Kuck 1998; Wostemeyer and Kreibich 2002; Daboussi and Capy 2003). Transposition is generally thought to influence genome organization and/or gene expression, although there is little experimental support for this view particularly within the major phylum, Basidiomycota.

In the widely studied lignin-degrading basidiomycete Phanerochaete chrysosporium (Cullen and Kersten 2004), a remnant of a class II element, pce1, is inserted within lignin peroxidase allele lipI2. The 1,747-nt insertion transcriptionally inactivates lipI2 and is inherited in a strict Mendelian fashion (Gaskell et al. 1995). Southern blots of pulsed field gels localized four pce-like sequences to a single chromosome band, and segregation analysis detected distant linkage among three copies designated pce1, pce2, and pce3 (Gaskell et al. 1995; Stewart et al. 2000). Class II elements of Homobasidiomycetes also include Abr1 identified in Agaricus bisporus (Sonnenberg et al. 1999) and Scooter, which has been shown to disrupt two genes regulating signal transduction in Schizophyllum commune (Fowler and Mitton 2000).

A high quality draft assembly of the P. chrysosporium genome has been generated (http://www.genome.jgi-psf.org/whiterot1/whiterot1.home.html). Initial analysis revealed families of structurally related genes some of which are thought to be involved in lignocellulose degradation, e.g., cytochrome P450s, glycosyl hydrolases, copper radical oxidases, and multicopper oxidases (Martinez et al. 2004). The latter family occurs as a cluster of four genes designated mco1 through mco4 (Larrondo et al. 2003, 2004). Recently, Yadav and coworkers have examined the complex structure and organization of the cytochrome P450 families (Doddapaneni et al. 2005).

A pure whole genome shotgun sequencing strategy was used for the P. chrysosporium genome, and this approach tends to exclude highly repetitive sequences such as rDNA and TEs. Nevertheless, non-coding repetitive elements as well as class I and II transposons were identified (Martinez et al. 2004). The class I retroelements include copia-like, gypsy-like, and LINE sequences (reviewed by Kempken and Kuck 1998; Wostemeyer and Kreibich 2002; Daboussi and Capy 2003). Many of these elements appeared as truncated remnants and/or rearranged, and their long terminal repeats often lie apart as “solo LTRs” (Kim et al. 1998; Goodwin and Poulter 2000).

Most recently, a substantially improved assembly (v2.0) and revised gene models (v2.1) have become available (http://www.genome.jgi-psf.org/Phchr1/Phchr1.home.html). We have systematically examined the insertional context of repetitive elements within the latest assembly. pce-like insertional mutations were detected in a cytochrome P450 and a glycosyltransferase gene. Further, gypsy- and copia-like retroelements have insertionally mutated members of multicopper oxidase and cytochrome P450 gene families, respectively. Results are consistent with an important role for transposons in generating genetic variation in P. chrysosporium.

Materials and methods

Organism and culture conditions

Phanerochaete chrysosporium strains BKM-F-1767 and RP-78 were used throughout. Typical of basidiomycetes, the vegetative mycelium of BKM-F-1767 harbors two distinct haploid nuclei in constant association. This dikaryotic strain has been extensively used for decades in studies of P. chrysosporium genetics and physiology. Homokaryotic derivative RP-78 was isolated by regenerating protoplasts derived from BKM-F-1767 mycelium. The nuclear condition and physiological characteristics of RP-78 were established (Stewart et al. 2000) and the strain used for whole genome shotgun sequencing (Martinez et al. 2004). All alleles from the RP-78 haplotype have been arbitrarily assigned an “A” suffix. Both strains are available from several culture collections (e.g., Forest Mycology Center, Forest Products Laboratory, Madison, WI, USA).

For RNA, 200 ml defined medium (Eriksson and Hamp 1978) amended with 0.4% Avicel PH-101 (Fluka Chemie, Buchs, CH) was inoculated with 1×107 spores in 2-l flasks. Incubation was at 37°C, 250 rpm in a shaking incubator. The cultures were harvested after 6 days by filtration through Miracloth (Calbiochem, La Jolla, CA, USA). The mycelium was snap frozen in liquid N2 and stored at −90°C.

Amplification of genomic and cDNA targets

PolyA RNA from frozen pellets was isolated by magnetic capture as described (Vanden Wymelenberg et al. 2002, 2005). First strand cDNA was synthesized by preparing a 500-μl reverse transcription master mix containing 1× PCR buffer (Promega Inc., Madison, WI, USA), 5 mM MgCl2, 4 mM dNTPs, 500 units RNAsin (Promega), 105 pmol oligo dT15, 1,250 units MMLV-RT (Invitrogen Corp., Carlsbad, CA, USA), and 100 μl mRNA. Fifty microliter aliquots were divided among 0.5-ml Eppendorf tubes and placed in a Perkin Elmer DNA Thermal Cycler 480 (Applied Biosystems, Foster City, CA, USA). Cycling conditions were 23°C 10 min, 42°C 45 min, 95°C 5 min. All tubes were combined and stored at −20°C.

Genomic and cDNA targets were amplified by PCR in a Perkin Elmer DNA Thermal Cycler 480. Each PCR reaction contained 1× PCR buffer, 1 mM MgCl2, 1.25 units Taq DNA polymerase (Promega Inc., Madison, WI, USA), 10% DMSO (Sigma-Aldrich, St Louis, MO, USA), 11 pmol each primer and 10 ng DNA template. Cycling conditions were 94°C 6 min, 54°C 2 min, 72°C 40 min (1 cycle), followed by 94°C 1 min, 54°C 2 min, 72°C 5 min (35 cycles), 72°C 15 min (1 cycle). Using primer D (Table 1) and primer I (5′-CCGTTCCATCTGCACGGACA-3′), elongase (Invitrogen Corp., Carlsbad, CA, USA) was used for long-range PCR amplification of pcret1. Elongation time was 10 min and the annealing temperature was 58°C. Buffer A and B were used in a 2:3 proportion. Amplicons to be sequenced were first cloned using pGEM-T Easy (Promega). Primer sequences are listed in Tables 1 and 2.

Table 1 PCR primers and predicted product lengths (nt)
Full size table
Table 2 Primers extending transcripts in RP-78 homokaryon
Full size table

Computational analysis

Overlapping approaches were used to detect transposon-interrupted coding regions in assemblies v1.0 and v2.0. Blastp and tblastn queries with well-known fungal TEs (Daboussi and Capy 2003), tyrosine recombinase retrotransposons (Goodwin and Poulter 2004), and helitrons (Poulter et al. 2003) identified numerous elements as did word or phrase searches (http://www.genome.jgi-psf.org:8080/annotator/servlet/jgi.annotation.-Annotation?pDb=Phchr1). Significant hits (E<10−4) were rarely intact, and subsequent analysis of the adjacent regions for direct and inverted repeats were performed using GeneQuest (DNASTAR, Madison, WI, USA). In an iterative fashion, putative P. chrysosporium nucleotide repeats and TE-related proteins were used to re-scan the genome by blastn and blastp/tblastn, respectively. For each hit, the completeness of surrounding gene models was assessed. In the absence of adjacent gene models, the flanking regions were examined for extended ORFs and subject to blastx searches of NCBI. Finally, the 24-longest scaffolds (representing 95% of assembly v2.0) were examined for partial or rearranged TEs by manually scanning the browser’s repeat track. In assembly v2.0, this track encompasses well-known fungal elements (e.g., maggy, skippy, etc.) and short non-coding repeats. Because many transposon-related domains (integrases, reverse transcriptases) are not captured in final v2.1 “Best models” (Vanden Wymelenberg et al. 2006), manual examination included Fgenesh ab initio models, particularly those with unusually long intervening sequences.

Unless otherwise specified, our analysis emphasized assembly v2.0. This assembly was generated by a pure shotgun approach, and the homology-based gene predictions were recently described (Vanden Wymelenberg et al. 2006). The v2.0 and archived v1.0 assemblies, predicted genes, supporting evidence, annotations, and analysis are accessible through an interactive browser and tools at the Joint Genome Institute’s web portal (www.jgi.doe.gov/whiterot). The protein information and related links for the most recent v2.1 genes can be rapidly accessed by appending model numbers to the end of the following URL: http://www.genome.jgi-psf.org/cgi-bin/dispGeneModel?db=Phchr1&id=. Throughout, similarity scores are based on the Smith–Waterman algorithm (Smith and Waterman 1981) using the BLOSUM62 matrix. ClustalW (Thompson et al. 1994) generated multiple alignments using DNASTAR software.

Results

Structure and organization of non-autonomous pce1-like repeats

Five sequences >40% identical to pce1 (Gaskell et al. 1995) were identified by blastn searches of assembly v2.0. Elements pce2 (AF134289), pce3 (AF134290), and pce4 (AF134291) are nearly identical to pce1, as previously shown (Stewart et al. 2000). Designated pce5, a 1,967-nt paralog was identified on scaffold 9 (coordinates 438,871–440,837). The pce5 nucleotide sequence is 44.5% identical to pce1 and, like pce1–pce4, contains no extended ORFs. A truncated element on scaffold 33 (coordinates 5,333–6,260), pce6, extends only 928 nt but is otherwise identical to pce1. Blastn and blastx searches of NCBI or the v2.0 database failed to identify any additional pce1-like sequences.

Excluding the incomplete pce6 element, the sequences show remarkable sequence conservation at their termini. All but pce3 are flanked by dinucleotide repeats, and all feature imperfect inverted terminal repeats (ITRs, Fig. 1). But more surprisingly, the positions of these seven “imperfections” are consistently retained in pce1, pce2, pce3, and pce4. The more divergent element, pce5, retains five of these nucleotides and features two additional imperfect bases. In addition to the ITRs, a 16-bp direct repeat (GTTTGTGCATGTCTGT) was also conserved in pce1, pce2, pce3, and pce4 (positions 1014 and 1602 in pce1).

Fig. 1
figure 1

Inverted terminal repeats (capital letters) of elements pce1 through pce5. Imperfect bases are boxed. Ten base pairs of flanking region (lower case letters) are shown with dinucleotide repeats underlined. Numbering on margins indicate scaffold no.: position on scaffold

Full size image

Corroborating the long-range structure of assembly v2.0 and consistent with earlier segregation analysis (Stewart et al. 2000), pce1, pce2, and pce3 were located on a single scaffold (number 19). Element pce4 resides on a separate scaffold (number 8), an observation consistent with cosegregation analysis showing no detectable linkage (Stewart et al. 2000). However, the v2.0 assembly and segregation data are at variance with pulsed field gel blots which suggest pce4 resides on the same 3.7-Mb chromosome as pce1, pce2, and pce3 (Gaskell et al. 1995). Possibly, pce4 is distantly linked, and scaffolds 8 and 19 may lie on the same chromosome. Elements pce5 and pce6 are located on scaffolds 9 and 33, respectively. Regions surrounding the elements were examined for the presence of incomplete gene models and/or extended ORFs. None were detected adjacent to pce2 and pce3.

Integration context and transcriptional impact of pce4

Gene model 4905 (v2.1), encoding a putative UDP-glucosyltransferase (ugt1A), lies 184 nt to the left of pce4 on scaffold 8 (coordinates 291,078–292,799). Blastp and blastx analysis, together with manual inspection of the region, suggested an incomplete N-terminus. Automated gene predictions failed to assign any models to the right of pce4, but an ORF potentially continuing ugtA was observed. PCR primers were designed within this ORF (Fig. 2a, primer D) and within the putative glucosyltransferase coding region (Fig. 2a, primer A). Using genomic DNA templates, the outermost primer pair (A/D) efficiently PCR amplified the 3,605-nt ugt1A allele from the homokaryon (Fig. 2b, lane 6) and the uninterrupted 1,858-nt ugt1B allele from the parental dikaryon (Fig. 2b, lane 12). Using dikaryon-derived RNA and the same three primer pairs, cDNAs corresponding to ugt1B were RT-PCR amplified (Fig. 2b, lanes 7–9) and the longest amplicon (1,386 nt) sequenced (GenBank DQ400694). (The 244-nt N-terminal ugt1B cDNA is only faintly visible in Fig. 2, lane 8.) Comparisons of ugt1A genomic sequence and the ugt1B cDNA pinpointed pce4 insertion within ugt1A coding region. Using RP-78-derived RNA as template, RT-PCR amplifications of ugt1A (Fig. 2b, lanes 1–3) generated a 421-nt C-terminal cDNA. This transcript was subsequently shown to extend through the coding region (Fig. 2c, primer pairs A/B, A/E, A/F) but primers located >45 nt inside pce4 (primer G, H) failed to amplify a ugt1A cDNA. Additional primers tested within this region yielded non-specific products (data not shown), and the precise transcriptional start point remains uncertain. Irrespective of the transcript start point, these results clearly indicate that pce4 interferes with ugt1A transcription.

Fig. 2
figure 2

Context and transcriptional consequence of pce4 insertion within ugt1A. Position and orientation of all primers (short arrows) are shown in panel a relative to exons (thick black boxes), introns (thin black lines), and pce4 (gray line). In panel b, genomic DNA (gDNA) and cDNA derived from homokaryotic (RP-78) and dikaryotic (BKM-F-1767) strains were amplified with primer pairs A/B, C/D, and A/D. Based on these results, the full-length of C-terminal transcript was further examined using RP-78 RNA (panel c, c lanes). RP-78 genomic DNA (panel c, g lanes) served as positive controls. Primer sequences and predicted product lengths are in Tables 1 and 2. Amplicon lengths in nt are shown to the right of the ethidium bromide-stained gels. Low molecular weight extraneous bands barely visible in panel c, lanes 5 (∼550 nt) and 9 (∼700 nt, 440 nt) were sequenced and identified as non-specific amplicons

Full size image

Blast analysis using the ugt1B cDNA and the corrected 4905 gene model showed greatest similarity to higher plant glucosyltransferases including Oryza sativa hydroquinone glucosyltransferase (gi 50252251, bit score 104) and a multifunctional triterpene/flavanoid glycosyltransferase from Medicago truncatula (gi 83753974, bit score 102). No significant fungal homologues were detected by blastn, blastx, or blastp searches of NCBI. In contrast, P. chrysosporium v1.0 gene models gx.22.38.1, pc.22.87.1, and pc.22.101.1 are 76, 66, and 53% identical to the deduced protein of ugt1B, respectively. (No corresponding v2.1 models were generated.) Sequences corresponding to these paralogs reside within a 16-kb region on v2.0 scaffold 6 (coordinates 869,664–885,611).

Integration context and transcriptional impact of pce5

pce5 lies on scaffold 9 between gene models 5526 and 5527 (v2.1), which are predicted to encode family CYP614/534 cytochrome P450 proteins. Close inspection of these automated model predictions suggested a single gene with the pce5 insertion (Fig. 3a). Employing a strategy similar to that used for pce4, primers were designed to assess the transcriptional consequence of insertion (Fig. 3a). Using RNA derived from the parental dikaryon, an 1,133-nt cDNA corresponding to cyp614/534B was RT-PCR amplified, cloned, and sequenced (Fig. 3b, lane 9, GenBank DQ444269). Comparisons of cDNA and genomic sequences pinpointed the insertion to the 3′ end of an intron suggesting the possibility of pce5 splicing. However, no flanking cDNA product was obtained from strain RP-78 (Fig. 3b, lane 3), indicating the insert is not spliced from the transcript, at least under the culture conditions examined.

Fig. 3
figure 3

Context and transcriptional consequence of pce5 insertion within cyp614/534A. Using a strategy similar to pce4 (Fig. 2a), primer position and orientation relative to exons (thick black boxes), introns (thin black lines), and pce5 (gray line) are shown in panel a. Genomic DNA (gDNA) and cDNA derived from homokaryotic (RP-78) and dikaryotic (BKM-F-1767) strains were amplified with primer pairs A/B, C/D, and A/D (panel b). With increased brightness/contrast control, faint bands corresponding to genomic cyp614/534A (3,687 nt) were visible (panel b, lanes 6 and 12; data not shown). Amplification of cDNAs corresponding to the N- and C-termini of cyp614/534A (panel b, lanes 1 and 2), prompted further investigation of these transcripts using RP-78 cDNA (panels c and d). RP-78 genomic DNA template (g lanes) served as positive controls for cDNA amplifications (c lanes). Primer sequences and predicted product lengths are in Tables 1 and 2. Faint bands slightly larger than the target cDNAs (panel c, lane 5; panel d, lane 5) were sequenced and identified as splice variants with retained introns

Full size image

Transcripts corresponding to the N- and C-termini of cyp614/534A were detected (Fig. 3b, lanes 1 and 2), and further examined by RT-PCR amplification of RP-78 RNA (Fig. 3c, d). The N-terminal (Fig. 3c, primer pairs A/E, A/F) and C-terminal transcripts (Fig. 3d, primer pair J/D) extended at least 512 and 68 nt, respectively, into pce5. Longer cDNAs were not detected (Fig. 3c, primers A/G, A/H; Fig. 3d, primer pairs K/D, L/D) even though these primer pairs efficiently amplified the corresponding genomic regions (Fig. 3c, lanes 8 and 10; Fig. 3d, lanes 8 and 10). RT-PCR amplification with primer pair A/F (Fig. 3c, lane 5) yielded a cDNA of 779 nt (GenBank DQ677350), 121 nt less than the predicted amplicon. Comparison with genomic DNA revealed the presence of an intron within this region of pce5. Although considerably larger than most P. chrysosporium introns (mode = 54 nt), canonical 5′ and 3′ splice sites (Martinez et al. 2004) were clearly present. Truncated chimeric transcripts have also been detected in Fot1 mutated nitrate reductase (niaD) alleles of Fusarium oxysporum (Deschamps et al. 1999).

Blast analysis using the cDNA and corrected models showed the cyp614/534A sequence was most closely related to a hypothetical protein from Gibberella zeae (gi 42546901, bit score 164). P. chrysosporium models 4556, 3368, 4557, and 6351 were 47, 55, 58, and 60% identical, respectively, to the protein sequence deduced from cDNA. Of these, only models 4556 and 4557 are linked, occurring within a 5.3-kb region of scaffold 7. However, blastx analysis of a region 6.2 kb to the right of model 5527 (coordinates 448,750–452,500), reveals the presence of a more distantly related CYP614/514-like sequence. (No v2.1 models were generated in this region.)

Identification of pcret1 in a mco3 allele

On scaffold 9 of assembly v2.0, clustered multicopper oxidase genes mco1A, mco2A, and mco4A appear intact (Larrondo et al. 2004; Fig. 4a), but mco3A, located furthest toward the scaffold’s left end, contains an extended intervening sequence within the 12th intron (http://www.genome.jgi-psf.org/cgi-bin/dispGeneModel?db=Phchr1&id=132237). Sequence analysis revealed the presence of an 8.14-kb insertion with features common to gypsy-class retroelements. Designated pcret1, the element contains identical LTRs of 336 bp flanking a 7.46-kb region with two extended ORFs (Fig. 4e). A 5-bp target site duplication (TSD) with the sequence AGTCT flanks the 5′ and 3′ LTRs (Fig. 4d). The first ORF present in pcret1 extends 1,417 bp, and corresponds to a predicted gag protein of 475 residues. The second ORF of 5,222 bp encodes a putative pol protein, with domains corresponding to protease, reverse transcriptase, RNAseH, and integrase (Fig. 4e). Blastx searches of the NCBI database show the pol protein most closely related (bit score 786) to Tricholoma matsutake marY1 (gi 5002510), and the full-length sequence is most closely related (bit score 709) to an unnamed Cryptococcus neoformans retrotransposon (gi 57227612).

Fig. 4
figure 4

The gypsy-like element pcret1 is inserted within mco3A, a member of the multicopper oxidase family located on scaffold 9 (panel a). Scaffold coordinates are shown below mco3A and mco4A. Contiguity of the insertion was confirmed by long-range PCR using primers D and I (panel b). Positions of exons (thick black lines) and introns (thin black lines) were established by comparisons of cDNA and genomic sequences (panel c). Insertion occurred within the 12th intron of mco3A at which position a 5-nt duplication is observed (italicized lettering, panel d). Panel e shows the structural organization of pcret1 with long terminal repeats (LTRs) and pol protein coding regions. Genomic DNA (gDNA) and cDNA derived from homokaryotic (RP-78) and dikaryotic (BKM-F-1767) strains were amplified with primer pairs A/B, C/D, and A/E (panel f). The length of the N-terminal mco3A transcript (panel f, lane 1) was further examined using RP-78 RNA (panel g, c lanes), and RP-78 genomic DNA (g lanes) served as positive controls. Primer sequences and predicted product lengths are in Tables 1 and 2

Full size image

As in the case of pce, pcret1 insertion altered transcript patterns. cDNAs corresponding to the N-terminal, C-terminal, and flanking regions were detected in the parental dikaryon (Fig. 4f, lanes 7–9, respectively). In contrast, RT-PCR amplification of RNA derived from the homokaryon only yielded a 1,008-nt cDNA corresponding to the N-terminal region (Fig. 4f, lane 1). Additional RT-PCR amplifications of RP-78 RNA slightly extended this transcript (Fig. 4g, lane 3), but none was obtained using primers lying within the adjacent gag region (Fig. 4g, lanes 5 and 7). The absence of transcript flanking the insertion point and transcript 3′ to pcret1 (Fig. 4f, lanes 2 and 3) strongly suggest that pcret1 is not spliced, at least under the conditions tested.

pcret1-related elements

Blastn analysis showed a minimum of 36 sequences with >95% sequence identity to the pcret1-LTR. Often unpaired and without nearby retroelement domains (e.g., reverse transcriptase, integrase), the LTRs were distributed on 25 separate scaffolds. Examination of automated models in the surrounding regions and blastx submissions (NCBI) of flanking sequences failed to identify insertions within detectable genes. pcret1-like coding regions flanked by the highly conserved LTRs were detected on scaffold 2 (coordinates 2,656,822–2,673,069 and 2,464,083–2,471,738) and on scaffold 7 (coordinates 1,655,590–1,665,209 and 1,978,909–1,987,005). In these instances, the pol coding regions are interrupted by sequence ambiguities of varying lengths.

Identification of pcret2/3 in a putative cytochrome P450 allele

Initial analysis of the draft genome assembly (v1.0) suggested the presence of several copia-like retroelements, including one located within a putative cytochrome P450 gene (Martinez et al. 2004). Analyses of the more recent assembly (v2.0) reveal an extensive cluster of >14 family 64 cytochrome P450s on scaffold 1, 5 of which lie within a 43.5-kb region (coordinates 2,490,057–2,533,590) (Fig. 5a). Models 132401, 133482, 133291, and 132914 appear intact while model 791 is incomplete due to a 27,869-nt insertion (Fig. 5a). The assemblies generally agree in this region, except the insertion is longer and more complex in v2.0 relative to v1.0.

Fig. 5
figure 5

Family 64 cytochrome P450 gene with complex insertion. Schematic of five CYP64 gene models (black vertical lines represent predicted exons) and complex insertion (broken gray box) is illustrated in panel a. Primers A (Table 1) and E (5′-CTCTGCCTCTGGAAAACGCA-3′) are designed to the predicted coding region of model 791 and to a potential ORF, respectively. Positions of exons (thick black lines) and introns (thin black lines) were established by comparison of dikaryon-derived cDNA encoding allele B with the genomic sequence of the RP-78 A allele (panel b). Intron sequence surrounding insertion point with potential duplication in italics is shown in panel c. Structural elements identified within the insertion (panel d) include five extended ORFs (boxed arrows) and two pairs of long terminal repeats (LTRs, thin-lined arrows). The outermost 356 bp LTRs lie adjacent to copia-like element pcret3. The pcret3 ORFs of 1,086 and 3,606 bp are interrupted by another retroelement pcret2 (4,566 bp), ORF1 (6,027 bp), ORF2 (9,366 bp), and 984 bp LTRs (gray arrows). An expanded view of pcret3 showing domain organization typical of copia elements is shown in panel e. Genomic DNA (gDNA) and cDNA derived from homokaryotic (RP-78) and dikaryotic (BKM-F-1767) strains were amplified with primer pairs A/B, C/D, and A/D (panel f). cDNAs of 102 nt (panel f, lane 1) and 504 nt (panel f, lane 2) correspond to the N- and C-terminal P450 coding regions, respectively. Length of transcripts was further examined using RP-78 RNA (panel g, c lanes), and RP-78 genomic DNA (g lanes) served as positive controls. Primer sequences and predicted product lengths are in Tables 1 and 2. The non-target ∼320-nt band (panel g, lane 3) was sequenced and identified as a splice variant with retained intron

Full size image

Components of this complex insertion appear to include a disjoint copia-like element, pcret3, the pol domain of a second retroelement, pcret2, and a duplication of an extended ORF of unknown function (Fig. 5d). LTRs (356 bp) lie adjacent to ORFs corresponding to the carboxy (3′-pcret3) and N-terminal (5′-pcret3) regions of copia-like retrotransposons from Ipomoea batatas (gi 50838657, bit score 563), O. sativa (gi 77555860, bit score 516), and other higher plants. The approximate positions of gag, protease, integrase, reverse transcriptase/RNAseH domains were inferred from the presence of conserved motifs (Jordan and McDonald 1999) and alignment with the related Candida albicans copia element, tca5 (gi 68492301, bit score 361; Plant et al. 2000; Fig. 5e).

The pcret2 sequence (Fig. 5d) encodes a predicted protein of 1,511 residues with significant similarity to pol proteins from Yarrowia lipolytica (gi 50554773, bit score 344) and Schistosoma mansoni (gi 44829171, bit score 281, DeMarco et al. 2004). Conserved integrase core (pfam00665) and reverse transcriptase (pfam00078) domains align at amino acid positions 557–681 and 1164–1317, respectively. Immediately to the left of the pcret2 ORF lies a 984-nt direct repeat that is duplicated approximately 20 kb to the right. It seems likely that the minimal active progenitor of pcret2 is composed of these LTRs and the pol proteins contained within the 4,533-nt coding region. This view is supported by the structure of closely related “intact” elements within the P. chrysosporium genome (below).

In addition to the LTRs and retroelements within this large insertion, two extended ORFs are present (Fig. 5d). Overlapping by 165 nt, ORF1 and ORF2 encode nearly identical proteins of 2009 and 3122 amino acids, respectively. The duplicated nucleotide sequences (underlined in Fig. 5d) are 6,639 nt long and >99% identical. Blastp and blastx analysis of these ORFs shows no significant similarity to any proteins. However, they are distantly related (bit score 59) to a P. chrysosporium conceptual translation (gi 71148689) containing a putative retroviral aspartyl protease domain (RVP, pfam00077). Possibly, these ORFs represent duplicated fragments of divergent retroelements.

To ascertain the intron/exon context of the pcret2/3 insertion, cDNA corresponding to the P450 “B” allele was RT-PCR amplified from the parental dikaryon (GenBank DQ444268). Sequence analysis extended and corrected errors in the truncated model (Fig. 5a, model 791). Consistent with a recent insertion, allele comparisons showed 96% nucleotide and 99% amino acid identity. The insertion was pinpointed to an intron with a putative 7-nt TSD (Fig. 5b, c).

As in the case of pcret1, insertion within an intron and the absence of deleterious mutation within the P450 coding regions suggested the possibility of an intact transcript through splicing. However, transcripts flanking the insertion point were RT-PCR amplified from the dikaryon (Fig. 5f, lane 9) but not the homokaryon (Fig. 5f, lane 3). As above, the data strongly suggest that the allele is functionally inactive due to the insertion.

Transcripts corresponding to the N- and C-termini of the mutated P450 gene were detected (Fig. 5f, lanes 1 and 2), and RT-PCR amplification of RP-78 RNA extended their lengths (Fig. 5g). However, in contrast to pce5, cDNAs did not extend into the element on either the N-terminus (Fig. 5g, primer pairs A/G and A/H) or the C-terminus (Fig. 5g, primer pairs J/D and K/D).

pcret2-related elements

Blast analysis of the pcret2 LTRs and coding region yielded results similar to pcret1. Specifically, >10 highly conserved LTRs were identified some of which were adjacent to pcret2-like coding regions and others appeared to be “solo” LTRs. Tblastx searches of the P. chrysosporium genome revealed >15 highly similar (bit scores >200) pol proteins. On the left end of scaffold 15, pcret2 and pcret3 are again adjacent but the region is clearly truncated and the corresponding LTRs are absent.

pcret3-related elements

Using the reconstructed pcret3 protein (Fig. 5e, 1563 amino acids), tblastn analysis of assembly v2.0 translations identified 49 sequences with significant similarity (bit scores >100). In most instances, further examination revealed significant similarity only to the 3′ pol proteins (integrase, reverse transcriptase, RNAseH). As in the case of pcret1 paralogs, browser examination of nearby gene models, including ab initio FgenesH models, failed to identify additional mutations. These finding were confirmed by blastx queries (NCBI) using nucleotide sequences surrounding the retroelement paralogs.

Regions of extended pcret3 homology, including conserved LTRs, were detected on assembly v2.0 scaffold 2 (coordinates 2,731,968–2,739,855) and scaffold 23 (coordinates 150,316–155,943). Although 250 nt shorter and containing sequence ambiguities, the scaffold 23 element was remarkably conserved, showing >94.1% nucleotide and 98.2% amino acid identity with pcret3. In contrast to blastn searches with pcret1 LTRs, only four significant “hits” (bit scores >100) were obtained with the pcret3 LTRs. Viewed together with the tblastn results, we conclude that P. chrysosporium strain RP-78 harbors many diverse copia-like elements, but pcret3 copy numbers are relatively low. In this connection, it remains formally possible that additional copies may have been excluded from the shotgun assemblies.

Discussion

Transposable elements are widely distributed among the fungi and are presumed to play important roles in gene expression and genome organization. Most reports have focused on ascomycetes, although relatively few have emphasized TE location (Cambareri et al. 1998; Hua-Van et al. 2000; Thon et al. 2004) or identified spontaneous mutations in structural genes (Nishimura et al. 2000; Kang 2001; Kang et al. 2001; Fudal et al. 2005). Basidiomycetes have received comparatively less attention, but an increasingly diverse array of TEs have been characterized from P. chrysosporium, A. bisporus, S. commune, T. matsutake, C. neoformans, Laccaria bicolor, Pisolithus microcarpus, and Microbotryum violaceum. To our knowledge, pce1 (Gaskell et al. 1995) and scooter (Fowler and Mitton 2000) represent the only examples of insertional mutagenesis in a basidiomycete. The P. chrysosporium draft genome provides an opportunity to extend our understanding of TE structure, organization and effects on gene expression.

Our analysis identified insertions in a minimum of five P. chrysosporium genes. These repeats included non-autonomous class II elements (pce1, pce4, pce5), a gypsy-like retroelement, pcret1, and a complex insertion containing rearranged retroelements pcret2 and pcret3. Structurally, pcret1 appears intact and similar to many other gypsy-class elements. In contrast, the pcret2/pcret3 insertion features long, duplicated ORFs, and the LTRs are disjoint from their respective coding regions. Likely, this complex insertion represented nested and rearranged elements, a common feature of transposon organization in higher eukaryotes.

Independent pcret2 and pcret3 insertions are located elsewhere in the genome. Based on the relative position of reverse transcriptase and integrase domains, the overall structure of pcret2 is similar to gypsy-class elements. However, a gag domain could not be clearly assigned in the absence of a zinc finger motif (Jordan and McDonald 1999). In contrast, pcret3 organization is typical of numerous copia-class retroelements. Although relatively rare in fungi, an increasing number of copia elements have been identified in ascomycetes (reviewed by Daboussi and Capy 2003) and more recently in the basidiomycetes L. bicolor, P. microcarpus (Diez et al. 2003), and M. violaceum (Hood 2005).

In contrast to the retroelements, the pce sequences lack clear relationship to any known TE. Highly conserved, albeit imperfect, ITRs, together with putative TSDs are typical of non-autonomous class II elements. Such elements sometimes share ITR sequence identity with an autonomous transposase (Chomet et al. 1991; Rezsohazy et al. 1997; Hua-Van et al. 2000), but we were unable to identify any candidates in the RP-78 genome. Possibly, an active progenitor resides in other strains or in the parental dikaryon.

Undoubtedly, this analysis underestimates the frequency of TE-disrupted genes, in part because shotgun sequencing tends to exclude highly repetitive sequences. In addition, genome scanning relied on sequence similarity to known TEs and/or recognition of gene models with intervening sequences. Highly divergent genes, particularly those disrupted by long elements, might not be modeled by Fgenesh. Fortunately, in the case of pce4 (Fig. 3a) and pcret2/3 (Fig. 5a) insertions, at least one flanking model could be recognized. Still, our approach would likely overlook highly divergent genes including those that might have accumulated nonsense mutations following TE insertion. Tenuous examples of additional TE-disrupted genes include a putative family 3 glycoside hydrolase on scaffold 2 (coordinates 1,647,844–1,679,733). In other instances, TEs located adjacent to, but not within, gene models may influence gene expression. As a potential example, a solo pcret1 LTR lies 1,012 bp upstream of the translational start of a putative cytochrome P450 gene (model 40563) on scaffold 5.

Interestingly, insertions seem biased toward gene families, members of which tend to cluster. Assuming genes within such families have redundant or overlapping functions, the impact of such insertions would be lessened. Further, in each case the corresponding allelic variant was uninterrupted and transcriptionally active. As a consequence, the negative affects of insertion would be further diminished by stable heterozygosity within the dikaryon. Such sheltering by heterozygosity might be expected to extend to single copy genes or even to essential genes. However, the sequenced strain, RP-78, is homokaryotic, and deleterious mutations would be negatively selected in the absence of some mitigating mechanism.

One such mechanism, well established in higher plants (Weil and Wessler 1990), D. melanogaster (Fridell et al. 1990) and C. elegans (Rushforth and Anderson 1996), involves splicing to remove TE sequences from mRNAs. Intron insertion points were identified for pce5, pcret1, and pcret2/3, prompting an examination of this possibility in P. chrysosporium. As in the case of certain Fot1 insertion in niaD (Deschamps et al. 1999), we observed transcripts 5′ and/or 3′ to the insertion point (Figs. 3, 4, 5, lanes 1 and 2). In the case of pce5, the transcripts extended into the element (Fig. 3c, d). However, our RT-PCR analysis showed no splicing of the full-length elements (Figs. 3, 4, 5).

Conceivably, the truncated cDNAs might encode active protein, although this seems highly unlikely. In the case of mco3A, the cDNA lacks a Cys and related residues necessary for copper coordination (Canters and Gilardi 1993; Larrondo et al. 2003). Similarly, incomplete transcripts from pce5- and pcret2/3-interupted sequences cannot give rise to active cytochrome P450s because the cDNAs lack either the active site Cys or the β1 domain conserved in all P450s (Graham and Peterson 1999). Finally, the previously described pce1 insertion occurs at Ala135 within lipI (Gaskell et al. 1995), a position that divides residues (Arg43, His47, His176) essential for catalysis (Tien and Tu 1987). Thus, all five alleles are likely inactivated by the spontaneous insertions.

It remains unclear whether these loss-of-function mutations are generally targeted toward gene families. Future research may identify conditions favoring transposition. Ultimately, TEs might provide useful gene-tagging tools and thereby resolve fundamental questions about the role of gene multiplicity in P. chrysosporium.