A Complex Ergovaline Gene Cluster in Epichloë Endophytes of Grasses

Abstract

Clavicipitaceous fungal endophytes of the genera Epichloë and Neotyphodium form symbioses with grasses of the subfamily Pooideae, in which they can synthesize an array of bioprotective alkaloids. Some strains produce the ergopeptine alkaloid ergovaline, which is implicated in livestock toxicoses caused by ingestion of endophyte-infected grasses. Cloning and analysis of a nonribosomal peptide synthetase (NRPS) gene from Neotyphodium lolii revealed a putative gene cluster for ergovaline biosynthesis containing a single-module NRPS gene, lpsB, and other genes orthologous to genes in the ergopeptine gene cluster of Claviceps purpurea and the clavine cluster of Aspergillus fumigatus. Despite conservation of gene sequence, gene order is substantially different between the N. lolii, C. purpurea, and A. fumigatus ergot alkaloid gene clusters. Southern analysis indicated that the N. lolii cluster was linked with previously identified ergovaline biosynthetic genes dmaW and lpsA. The ergovaline genes are closely associated with transposon relics, including retrotransposons and autonomous and nonautonomous DNA transposons. All genes in the cluster were highly expressed in planta, but expression was very low or undetectable in mycelia from axenic culture. This work provides a genetic foundation for elucidating biochemical steps in the ergovaline pathway, the ecological role of individual ergot alkaloid compounds, and the regulation of their synthesis in planta.

Fungal endophytes of the genera Epichloë and anamorphic Neotyphodium (Clavicipitaceae, Ascomycota) are obligate biotrophs that form symbioses with grasses of the subfamily Pooideae (45). Phylogenetic analysis suggests that the asexual Neotyphodium species are generally hybrids of the sexual Epichloë species (35). Some species, however, are direct asexual derivatives of sexual species; Neotyphodium lolii, for example, is a derivative of Epichloë festucae (35). The fungi colonize the intercellular spaces of all the aboveground parts of the plant, including the reproductive tissues, but do not infect the roots (14). Endophyte infection confers several ecological benefits to infected plants, including resistance to invertebrate and vertebrate herbivory as well as enhanced growth, mineral uptake, and resistance to drought (17, 32). Resistance to herbivory is conferred to host plants by fungus-produced secondary metabolite alkaloids (10). Four classes of epichloë alkaloids have been characterized: indole diterpenes, lolines, peramine, and the ergot alkaloids (10).

Ingestion of ergot alkaloids by livestock grazing on endophyte-infected pastures can cause toxic effects, including poor weight gain, hyperthermia, convulsions, reduced fertility, gangrene of the extremities, and death (2, 50). Losses as a result of ergot alkaloid poisoning are significant and a major cost to the global agricultural industry (29). The effects of ergot alkaloid poisoning are attributed mainly to the ergopeptine end product, ergovaline (2). However, transport across ruminant gastric membranes is much higher for intermediate lysergyl compounds than for ergopeptines, suggesting that intermediate ergot alkaloids may have a significant role (28).

Ergot alkaloids are produced by ascomycetous fungi from discontinuous taxonomic groupings, including plant-associated fungal genera from the family Clavicipitaceae and some members of the order Eurotiales, including the human pathogen Aspergillus fumigatus (46). Much of the chemistry for ergot alkaloid synthesis has been elucidated from the ergot fungus, Claviceps purpurea. The first committed step is the formation of dimethylallyl tryptophan (DMAT) catalyzed by DMAT synthase using primary metabolites tryptophan and dimethylallyl diphosphate (23). The gene dmaW, encoding DMAT synthase, has been characterized from Claviceps fusiformis (52), C. purpurea (53), A. fumigatus (19), and the endophyte Neotyphodium sp. strain Lp1 (Epichloë typhina × N. lolii) (57). Serial redox reactions form clavine intermediate compounds chanoclavine, agroclavine, and elymoclavine (22), with the latter converted to d-lysergic acid by a cytochrome P450 encoded by the cloA gene, which was recently characterized for C. purpurea (26). d-Lysergic acid can then be converted into several lysergyl amides and the ergopeptines (22).

Ergopeptines are formed by the nonribosomal peptide synthetase (NRPS)-catalyzed addition of a tripeptide to activated lysergic acid (56). NRPSs are large multimodular enzymes where, generally, one module activates and tethers one specific amino acid or carboxylic acid substrate and catalyzes peptide bond formation with a substrate activated and bound to an upstream module, and so on, resulting in small peptides with defined sequences (34). In ergopeptine synthesis, as elucidated for C. purpurea, lysergic acid is activated and tethered to the LpsB (LPS2) enzyme, a single-module NRPS (18, 42). The lysergyl group is then added to a specific amino acid tethered to the first module of LpsA (LPS1), a three-module NRPS (42). Two further specific amino acids are sequentially added, and the tripeptide is heterocyclized and released from LpsA as lysergyl peptide lactam (56). One further heterocyclization step occurs to yield the final ergopeptine product (40). The tripeptide moiety in ergovaline, the most abundant epichloë-produced ergopeptine, consists of alanine, valine, and proline (8). The genes encoding the LPS1 and LPS2 enzymes, lpsA and lpsB, have been cloned from C. purpurea (18, 25, 53), and the lpsA gene has been characterized from Neotyphodium sp. strain Lp1 (20, 38).

Genes for natural product biosynthesis are often found clustered in fungal genomes, and gene clusters for ergot alkaloid biosynthesis have been identified for C. purpurea (25, 53) and A. fumigatus (19, 54); they contain 13 and 14 genes, respectively. Putative functions have been proposed for the uncharacterized genes in the clusters based on bioinformatics; however, their conclusive roles in ergot alkaloid synthesis await functional analyses. Genes shared between the two clusters are presumed to be responsible for the early steps common to the two organisms, while genes unique to C. purpurea are proposed to be required for lysergic acid and ergopeptine biosynthesis and genes found only in the A. fumigatus cluster presumably are involved in the clavine decorations unique to that organism (19). Although the dmaW and lpsA genes have been cloned from epichloë species (20, 38, 57), additional biosynthetic genes assumed to be clustered were not isolated.

This study arose from the isolation, by degenerate PCR, of an N. lolii NRPS fragment, ps12, which is preferentially expressed during biotrophic growth (48). In this work, we show that the full-length gene associated with ps12 is the lpsB gene involved in ergovaline biosynthesis. Having isolated lpsB, our further objectives were (i) to determine whether ergot alkaloid biosynthetic genes were clustered in N. lolii by using chromosome walking and Southern blot analysis, (ii) to confirm the role of lpsB in ergot alkaloid biosynthesis by using targeted deletion and complementation, and (iii) to determine whether the genes in the cluster were preferentially expressed in planta, concurrent with ergot alkaloid production.

MATERIALS AND METHODS

Bacterial strains.

The Escherichia coli strains used in this study, XL1-Blue (9), KW251 (Promega Corp. Madison, WI), SOLR (Stratagene, La Jolla, CA), and Top 10 (Invitrogen Corp., Carlsbad, CA), were grown on Luria-Bertani agar plates supplemented, where necessary, with either ampicillin (100 μg/ml) or kanamycin (50 μg/ml).

Fungal strains and growth conditions.

Experiments were performed with N. lolii strain Lp19 (15) and E. festucae strain Fl1 (61). Fungal cultures were grown and maintained as described previously and supplemented where necessary with hygromycin (150 μg/ml) or Geneticin (220 μg/ml). Cultures of E. festucae Fl1 used for expression analysis were grown on either potato dextrose (PD) medium or one of two defined media, Mantle A (33), with high (2.0 g/liter) or low (0.2 g/liter) KH₂PO₄, or Czapek-Dox salts plus nitrogen (100 mM NH₄Cl) and/or a carbon source (100 mM glucose). Cultures incubated in plant extract were first grown in PD broth for three days and then filtered through sterile Whatman 3MM paper and added to the extract for 30 min. The plant extract was prepared essentially using the method of Lev et al. (31), and endophyte-free perennial ryegrass tillers were ground to a fine powder under liquid nitrogen and mixed 1:1 (wt/vol) with sterile distilled H₂O. The suspension was centrifuged at 13,000 rpm for 5 min, and the supernatant was removed for use.

Chemicals.

All solvents used for extractions and high-performance liquid chromatography (HPLC) were HPLC grade. Ergotamine tartrate was obtained from Sigma-Aldrich Co. (St. Louis, MO). Authentic ergovaline was provided by F. Smith, Auburn University. Authentic samples of lysergic acid and ergine were provided by M. Flieger (Czech Academy of Sciences).

Construction and screening of libraries.

A λZAPII library (Stratagene) was constructed by partially digesting N. lolii Lp19 genomic DNA with Tsp509I and purifying from an agarose gel in the 2- to 7-kb size range. This DNA was then ligated into an EcoRI-digested λZAPII insertion vector and packaged using Gigapack III packaging extract (Stratagene) per the manufacturer's instructions. Clones were also obtained from a previously described λGEM-12 library to N. lolii Lp19 (61). Libraries were screened by plaque hybridization using standard methods. Phagemid clones were recovered from λZAPII clones per the manufacturer's instructions.

Plant growth conditions and inoculation.

Inoculation of endophyte-free seedlings of perennial ryegrass (Lolium perenne cv. Nui) was carried out using the method of Latch and Christensen (30) with mycelium from 7- to 14-day-old cultures. Plants were tested for endophyte infection by tissue print immunoblotting (24) as described in reference 61.

Molecular biology.

Fungal genomic DNA was isolated from freeze-dried or frozen mycelium by using the methods described in references 1 and 11. Plasmid DNA was isolated and purified using a QIAGEN (Hilden, Germany) plasmid mini kit. Lambda DNA was isolated using a standard method (43).

PCR products amplified with Taq DNA polymerase (Invitrogen) were routinely cloned into pCR2.1-TOPO (Invitrogen) and transformed into E. coli Top 10.

For Southern analysis, genomic digests were transferred to positively charged nylon membranes (Hybond-N+; Amersham) by capillary transfer. Filters were probed with [α-³²P]dCTP (3,000 Ci/mmol; Amersham)-labeled probes. DNA was labeled by primed synthesis with the Klenow fragment and a RadPrime Kit (Invitrogen). Hybridizations were carried out at 65°C overnight in Church and Gilbert buffer (43) followed by three washes with 2× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate)-0.1% sodium dodecyl sulfate, the first at room temperature for 5 min and the second and third at 50°C for a total of 1 h. Blots were subsequently exposed to X-ray film.

Large DNA fragments were separated using contour-clamped homogeneous electric field (CHEF) electrophoresis (16) in an agarose gel (Bio-Rad; pulsed-field grade) using a CHEF-DR III electrophoresis unit at 14°C in 0.5× Tris-borate-EDTA buffer (43) at 6 V/cm using switch times of 50 to 90 s for 22 h. Chromosomal DNA was embedded in agarose and digested as described in reference 62.

Semiquantitative PCR analysis.

RNA was isolated from frozen fungal mycelia or pseudostem tissue of infected perennial ryegrass by using TRIzol reagent (Invitrogen). DNA was removed by incubating 1-μg aliquots of total RNA with 10 U of RNase-free DNase I (Roche) for 30 min at 37°C, followed by 10 min at 75°C. RNA was reverse transcribed at 60°C for 60 min in a reaction volume of 20 μl containing 1× reaction buffer, 10 mM dithiothreitol, 1 mM (each) deoxynucleoside triphosphates, 2.5 μM oligo(dT)₂₀, 2 U of RNase inhibitor, and 0.75 U of ThermoScript reverse transcriptase (Invitrogen). Gene-specific amplification from cDNA was carried out in a 20-μl volume containing 1× Taq polymerase buffer, 50 μM (each) deoxynucleoside triphosphates, 200 nM (each) primers, and 0.5 U of Taq polymerase (Invitrogen).

The thermocycling conditions used for semiquantitative PCR were one cycle at 94°C for 2 min; 30 cycles at 94°C for 20 s, 60°C for 20 s, and 72°C for 30 s; and a final incubation at 72°C for 5 min. Primer pairs used were lpsBq-F/lpsBq-R (lpsB), DamP81/DamP82 (easH), DamP83/DamP84 (easA), DamP85/DamP86 (easG), DamP87/DamP88 (easF), and DamP89/DamP90 (easE); for descriptions, see Table S1 in the supplemental material.

Preparation of deletion construct.

The targeted deletion construct, pDF6, was constructed using a MultiSite Gateway (Invitrogen) system. Flanking sequences of 3.1 kb, 5′ and 3′ of lpsB, were amplified by PCR with attB1- and attB4-tailed primers (KOLEnd-F/KOLEnd-R) and attB3- and attB2-tailed primers (KOREnd-F/KOREnd-R), respectively, from E. festucae Fl1 genomic DNA. A 4.1-kb fragment containing P_gpdA hph was amplified from pAN7-1 (39) with attB1- and attB2-tailed primers (Hyg-F/Hyg-R). These fragments were recombined into pDONR P4-P1R, pDONR P2R-P3, and pDONR 221, respectively, using BP Clonase (Invitrogen). The three fragments were recombined into pDEST R4-R3 in a three-way recombination reaction using LR Clonase Plus (Invitrogen), resulting in pDF6. The linear product of pDF6 used for transformation was amplified using TripleMaster polymerase (Eppendorf) using conditions recommended by the manufacturer.

Transformation of E. festucae protoplasts and molecular analysis of transformants.

Protoplasts of E. festucae Fl1 were prepared using the method of Young et al. (60) except that 10 mg/ml Glucanex (Chemcolour Industry) was used to digest the cell walls. Protoplasts were either transformed with 5 μg of a linear PCR product of pDF6 for targeted gene replacement or cotransformed with 5 μg of pDF1 and 2.5 μg of pII99 (36) for complementation. Transformants were selected on RG medium (PD plus 0.8 M sucrose) containing either hygromycin (150 μg/ml) for replacement constructs or Geneticin (220 μg/ml) for complementation. To obtain clonal isolates, the resulting transformants were purified by subculturing three times as described by Young et al. (61).

Transformants were screened by PCR for homologous recombination of the replacement construct with primers lpsBKOs-F and lpsBKOs-R, which annealed to sequences in the replacement construct on either side of the hph cassette, amplifying a 400-bp product in wild-type E. festucae and a 4-kb product in the mutant. The mutant strain chosen for analysis was further screened by Southern blot analysis as described above.

Liquid chromatography-tandem mass spectrometry (LC-MS/MS) analyses.

Pseudostems (two to three per plant, approximately 35 mm in length) were collected from perennial ryegrass plants that were either endophyte-free or infected with E. festucae Fl1, an lpsB mutant, or complemented strains, after 8 weeks of growth in the greenhouse. Samples were also collected from perennial ryegrass cv. Rosalin infected with Neotyphodium sp. strain Lp1 (37) as a reference association. Samples (approximately 50 mg) were extracted by maceration with a ceramic bead in a tissue disrupter (FastPrep FP120; Savant) for 60 s (setting 4) in 500 μl methanol-1% aqueous acetic acid (1:1) containing 0.112 μg/ml ergotamine tartrate as the internal standard. Subsamples (15 μl) of the supernatant were eluted through a C₁₈ Luna column (Phenomenex) (150 by 2 mm; column particle size, 5 μm) at a flow rate of 200 μl/min using a Surveyor HPLC (Thermo Finnigan) with a solvent gradient, starting with 5% MeCN:95% H₂O (containing 5 mM ammonium acetate) for 5 min and then increasing to 50% MeCN over 38 min followed by a column wash at 100% MeCN. To limit carry-over between samples, four wash cycles were carried out between each injection. Cross-contamination between samples was very low but detectable (ca. 1/1,000), and endophyte-free samples were interspersed between endophyte-infected samples to avoid ambiguity. Mass spectrometry was carried out by electrospray ionization in the positive mode with a linear ion trap mass spectrometer (LTQ; Thermo Finnigan). The spray voltage was 5.0 kV, and the capillary temperature was 275°C. The flow rates of sheath gas and auxiliary gas were set to 20 and 10 (arbitrary units), respectively. Metabolites were identified by comparisons of elution times and product ion spectra from collision-induced dissociations of selected precursor ions (35% collision energy) with authentic standards or, in the case of lysergyl-alanine and its isomer and 6,7-secolysergine, with previously characterized metabolites of Neotyphodium sp. strain Lp1 (37). Measurements were carried out by selective reaction monitoring of the transitions from m/z 269 to 223 (lysergic acid and isolysergic acid), m/z 340 to 223 (lysergyl-alanine and isolysergyl-alanine), m/z 268 to 223 (ergine and erginine), m/z 241 to 210 (6,7-secolysergine and putative isomers), m/z 534 to 268 (ergovaline and ergovalinine), and m/z 582 to 268 (ergotamine and ergotaminine).

DNA sequencing and bioinformatics analyses.

DNA fragments were sequenced by the dideoxynucleotide chain termination method using Big-Dye 3.1 chemistry with oligonucleotide primers (Invitrogen), either M13 forward and reverse or specific to N. lolii or E. festucae genomic sequences. Products were separated with an ABI Prism 3730 sequencer (PerkinElmer).

Sequence data were assembled into contigs and analyzed using Vector NTI suite 9 (Invitrogen). Sequence comparisons with public databases were performed via the Internet at the National Center for Biotechnology Information site (http://www.ncbi.nlm.nih.gov/), by employing BLASTN, BLASTX, and BLASTP algorithms.

Repeat sequences were identified and characterized using MEME (Multiple EM for Motif Elicitation) (3) and the etandem and einverted algorithms of EMBOSS (41).

Nucleotide sequence accession number.

The EAS cluster sequence isolated in this study has been submitted to the DDBJ/EMBL/GenBank databases (accession number EF125025).

RESULTS

Isolation and analysis of a putative gene cluster for ergovaline biosynthesis.

PCR amplification of N. lolii NRPS adenylation domains in a previous study identified a clone, ps12, that was preferentially expressed in planta (48). To clone the full-length gene, the ps12 PCR product was used to probe genomic libraries. This region of the genome was found to be underrepresented in all libraries screened. Just two clones (pDF1 and pDF2), from ∼80,000 recombinant phages plated, were isolated from an N. lolii Lp19 small-insert λZAPII library, and no clones were identified for N. lolii Lp19 bacterial artificial chromosome or E. festucae Fl1 cosmid and fosmid libraries. The sequence of pDF1 and pDF2 covered 9.8 kb (Fig. 1), and bioinformatics analysis revealed two open reading frames (ORFs) encoding predicted proteins with 49% and 48% identity, respectively, to genes for a single-module NRPS, lpsB (formerly cpps2), and an oxidoreductase, easE (formerly cpox1), found in a gene cluster for ergopeptine biosynthesis in C. purpurea (18). These results suggested that the two putative genes, designated lpsB and easE following the convention of Schardl et al. (46), encode genes for the biosynthesis of ergovaline, the predominant ergopeptine product in N. lolii.

FIG. 1.

Physical map and clones covering the N. lolii EAS locus. Putative genes are shown as arrows indicating the direction of transcription. Introns are shown as gaps. The solid line above lpsB denotes the ps12 sequence used to probe genomic libraries. “Retro” denotes two highly degenerate nested long terminal repeat retrotransposons. Regions containing putative nonautonomous transposable elements are expanded. SSR, simple sequence repeat (eight times 10 bp).

An analysis of the predicted amino acid sequence of LpsB revealed the single module to contain an adenylation domain, a thiolation domain, and a condensation domain. As the LpsB NRPS is predicted to interact with a second NRPS, LpsA, in order to catalyze the formation of lysergyl peptide lactam, we aligned the N. lolii and C. purpurea LpsB carboxy-terminal sequences with those of bacterial NRPSs involved in multienzyme complexes (data not shown). No alignment was found with COM domains required for protein-protein interaction in bacterial systems.

To extend the cluster, additional clones were isolated by screening λZAPII and λGEM-12 libraries. No clones could be isolated from the λZAPII library using a probe to the right end of the cluster amplified with primer set DamP10/DamP11, and just one clone, pDF3, was isolated using a probe to the left end of the cluster amplified with primers DamP21/DamP22 that extended the sequence only 791 bp. Left-end sequence was present, albeit underrepresented, in the λGEM-12 library with four clones, λDF1-4, isolated from the 110,000 recombinant phages plated. The four clones spanned 19.4 kb and extended the previously isolated sequence by 4.1 kb at the left end and 5.5 kb at the right end (Fig. 1).

Bioinformatics analysis of the 4.1-kb left-end sequence identified four putative genes encoding an oxidoreductase (easA), two conserved ORFs with unknown functions (easF and easG), and a dioxygenase gene (easH) (Table 1 and Fig. 1). The easH gene was truncated at the 5′ end in a chimeric lambda clone, λDF1. Each of the additional putative genes in the EAS cluster had homologues in the C. purpurea and A. fumigatus ergot alkaloid gene clusters (Table 1). Sequence analyses of each of the three gene clusters (Fig. 2) showed substantially different gene orders among the three clusters.

TABLE 1.

Bioinformatics analysis of genes within the N. lolii EAS cluster^a

Gene	Introns (bp)	Size (aa)	BLASTP E value		Function or putative function
			C. purpurea	A. fumigatus
lpsB	75	1,352	0.0	—	Single-module NRPS
easE	75, 60	605	e-123	e-113	Reductase/dehydrogenase
easF	69	344	e-118	e-115	Methyltransferase
easG	78, 53	309	4e-78	1e-64	Reductase/dehydrogenase
easA	—	380	e-152	57e-124	Reductase/dehydrogenase
easH	—	188	ND	0.53	Oxygenase/hydroxylase

^aeasH is truncated at the 5′ end in λDF1; therefore, only a partial sequence was studied. ND, not determined as the C. purpurea easH gene sequence is not available in public databases; —, absent; aa, amino acids.

FIG. 2.

Comparative eas gene order among N. lolii, C. purpurea, and A. fumigatus. Cp, C. purpurea; Nl, N. lolii; Af, A. fumigatus. Open arrows, genes shared between each organism; black arrows, genes proposed to be required for ergot alkaloid synthesis but not yet identified for N. lolii; light gray arrows, genes found in N. lolii and C. purpurea but not A. fumigatus; dark gray arrows, genes found only in A. fumigatus. N. lolii dmaW, cloA, and lpsA are not shown, as their locations relative to the EAS cluster are not known.

The 5.5-kb right-end sequence was extremely AT rich (76%) compared with the eas genes (50%). A BLASTX analysis of this region indicated that this sequence is derived from two nested retrotransposons but is highly degenerate, with no evidence of intact ORFs. This degeneracy is probably a result of repeat-induced point mutation (RIP), which results in C:G-to-A:T transitions (12) and was previously observed in two epichloë retroelements, Tahi and Rua, associated with the LTM gene cluster (61).

Linkage of the EAS gene cluster with dmaW and lpsA and identification of cloA.

Two genes for ergovaline biosynthesis have been previously characterized: dmaW, a DMAT synthase-encoding gene required for the first committed step in ergot alkaloid biosynthesis (52), and lpsA, a gene encoding a trimodular NRPS that, along with a second NRPS, predicted to be LpsB, is required for formation of lysergyl peptide lactam (38). To test the hypothesis that these genes were physically linked with the gene cluster isolated in this study, we screened the λZAPII library for clones containing these genes. No clone could be isolated for dmaW; however, a BLASTX analysis of the Neotyphodium coenophialum dmaW-2 flanking sequence in GenBank (accession number AY259839.1) revealed the presence of a P450 monooxygenase gene orthologous with the C. purpurea cloA-encoded monooxygenase required for conversion of elymoclavine to d-lysergic acid (26) (Fig. 3). PCR analysis showed cloA to be present in N. lolii Lp19 and E. festucae Fl1 but directly upstream of dmaW only in E. festucae (data not shown).

FIG. 3.

Sequence diversity at dmaW loci. Nc1, N. coenophialum dmaW locus 1 (accession number AY259838); Nc2, N. coenophialum dmaW locus 2 (accession number AY259839); Lp1, Neotyphodium sp. strain Lp1 dmaW locus (accession number AY259837). LTR, long terminal repeat.

One clone that was isolated for lpsA contained 1,204 bp of unique sequence 5′ of the lpsA sequence deposited in GenBank (accession AF368420). This sequence contained 655 bp of a Tahi long terminal repeat retrotransposon relic at the 5′ end. The AT richness (76%) of this element precluded further chromosome walking. We therefore performed Southern analyses of NotI digests of N. lolii Lp19 and E. festucae Fl1 genomic DNA (Fig. 4). Probes for lpsB, lpsA, and dmaW each hybridized to a 114-kb fragment in E. festucae and a 340-kb fragment in N. lolii, indicating that dmaW and lpsA were linked with the EAS cluster in each species.

FIG. 4.

Southern blot analysis indicating linkage of lpsB, lpsA, and dmaW. N. lolii Lp19 (Nl) and E. festucae Fl1 (Ef) genomic DNA was digested with NotI, separated by CHEF pulsed-field gel electrophoresis, and transferred to a nylon membrane. The blot was hybridized with ³²P-labeled probes to lpsB (pDF2), lpsA (amplified with primers lpsA-F and lpsA-R), and dmaW (amplified with primers dmaWq-F and dmaWq-R). The lpsA probe contains a NotI site and hence hybridizes to two bands.

Transposable elements associated with the EAS cluster.

BLASTN and MEME analyses of the EAS cluster revealed the presence of putative nonautonomous transposons between easG and easA and between easE and lpsB (Fig. 1). Two elements in the intergenic region between easG and easA have the characteristics of a miniature inverted repeat transposable element (MITE) (21). They are 82% identical, are 139 and 141 bp in size, respectively, and have 61-bp imperfect terminal inverted repeats (TIRs) and a putative AT target site duplication (TSD). The 24 bp between the TIRs is AT rich and has no similarity with any sequence in the public databases. We named the element Toru (Maori for three) following the convention adopted for naming the two retrotransposons in the LTM cluster called Tahi and Rua (one and two) (61).

A BLASTN analysis showed Toru elements to be present upstream of epichloë dmaW (Fig. 3) and also in other epichloë biosynthetic gene clusters (see Table S2 in the supplemental material). Analysis of the “extra” sequence between N. coenophialum dmaW-1 and the Toru MITEs, compared with dmaW-2 and Neotyphodium sp. strain Lp1, revealed a putative type 2 Mutator-like element relic, Wha (Maori for four) (Fig. 3). Wha has 95-bp TIRs and a 10-bp putative TSD. The 2,413 bp between the putative TIRs was highly degenerate (75% AT), probably due to RIP, and shared 32% identity with a 312-amino-acid putative protein (CHGG_06856) from Chaetomium globosum, with several closely related proteins in the C. globosum genome. When submitted for BLASTX analysis itself, this sequence had 31% identity with a transposase from the Hop Mutator-like element in Fusarium oxysporum (13).

A second MITE-like element, Rima (Maori for five), is present between the lpsB and easE genes. This element is 295 bp in size and has two 63-bp imperfect TIRs, a putative TA TSD, and AT-rich noncoding sequence between the TIRs. No other copies of Rima were found in GenBank by BLASTN analysis. Southern analysis showed the element to be single copy in the E. festucae Fl1 and N. lolii Lp19 genomes (data not shown); however, this may be due to degeneracy of other Rima elements, as is observed with Toru elements.

A third putative nonautonomous repeat element is present 97 bp from Rima. This element, termed Ono (Maori for six), is 236 bp long and is 82% and 66% identical to sequences upstream of ltmM and ltmE, respectively, in the LTM gene cluster. Of the three Ono elements, only the ltmM-associated element has a putative TSD of 7 bp. This element had been identified previously as a 115-bp sequence shared between the ltmM and ltmE promoter regions and was described as possibly being a relic of a short interspersed nuclear element (62).

Functional analysis of lpsB.

To confirm that lpsB was required for ergovaline biosynthesis, deletion of a 276-bp region of the adenylation domain was performed by the introduction of a targeting construct containing 6 kb of homologous sequence and a hygromycin resistance cassette (Fig. 5A). A PCR-generated linear construct was transformed into E. festucae Fl1 protoplasts. E. festucae, the sexual progenitor of N. lolii (35), was used rather than N. lolii because of its genetic tractability (44, 61). Transformants that grew in the presence of hygromycin were screened by PCR, and 15% of these (7 of 48 screened) showed a replacement event (data not shown). An lpsB mutant, DFM3, was confirmed to be a replacement by Southern analysis (Fig. 5B). This mutant showed no difference in morphology or growth rate in culture compared with the parent isolate and showed a wild-type plant-interaction phenotype, demonstrating that disruption of lpsB has no effect on the symbiosis under normal growth conditions.

FIG. 5.

Targeted gene replacement of lpsB. (A) Diagrammatic representation of homologous recombination at lpsB. The wild-type (WT) gene locus and targeted replacement construct (KO) are shown. The solid line above lpsB represents the PCR probe amplified with primers KOprobe-F and KOprobe-R, used for the results shown in panel B. E, EcoRI site. (B) Southern blot analysis. Wild-type and lpsB mutant DFM3 (KO) genomic DNA was digested with EcoRI and transferred to a nylon membrane, which was hybridized with the probe shown in panel A.

LC-MS/MS with selected reaction monitoring was performed to determine the effect of the lpsB mutation on ergot alkaloid production (Fig. 6) (see Table S3 in the supplemental material). The structures of lysergic acid, lysergyl alanine, and ergovaline are shown in Fig. 6. Ergovaline and its stereoisomer ergovalinine were not detected in samples from symbiota containing the lpsB mutant compared with the wild type, confirming the expected requirement of lpsB for ergovaline biosynthesis. Lysergic acid and its C-8 stereoisomer were observed to accumulate to elevated levels in lpsB mutant symbiota compared to wild-type associations. Lysergyl alanine, an amide of lysergic acid previously identified by Panaccione et al. (37) as a missing compound in an lpsA mutant of Neotyphodium sp. strain Lp1, was also identified together with its C-8 stereoisomer in symbiota containing E. festucae strain Fl1 but was absent in lpsB mutant symbiota. Synthesis of ergine, another lysergic acid amide, was also shown to be blocked by lpsA mutation in Neotyphodium sp. strain Lp1 (37). While no samples of E. festucae lpsB mutant-infected symbiota contained ergine, production of ergine was not always detected in wild-type symbiota. The novel clavine compound 6,7-secolysergine, observed to accumulate in the Neotyphodium sp. strain Lp1 lpsA mutant (37), was not observed at significant levels in any E. festucae association. However, an unknown peak detected by LC-MS/MS monitoring of the selective reaction characteristic of 6,7-secolysergine (m/z 241 to 210) did appear to increase in concentration in the E. festucae lpsB mutant symbiota (see Table S3 in the supplemental material). Analyses of other clavine intermediate compounds did not show any reduction or accumulation resulting from lpsB mutation (data not shown).

FIG. 6.

Extracted ion chromatograms from LC-MS/MS of extracts of perennial ryegrass plants infected with wild-type E. festucae Fl1 (WT), lpsB mutant DFM3 (KO), or DFM3 transformed with pDF1 (Complement), showing accumulation of lysergic acid (shown above; peak a) and isolysergic acid (peak b) (m/z 268 to 223), lysergyl-alanine (peak c) and isolysergyl-alanine (peak d) (m/z 340 to 223), and ergovaline (peak e) and ergovalinine (peak f) (m/z 340 to 223). Relative abundance values are on the same (arbitrary) scale for each extract.

In order to complement the lpsB mutation, pDF1, containing the wild-type lpsB gene and 379 bp of upstream sequence (Fig. 1), was transformed into lpsB mutant protoplasts. An arbitrary selection of these transformants (DFM7, DFM8, and DFM9) was used to infect perennial ryegrass seedlings, and the ergot alkaloid phenotype of these associations was determined by LC-MS/MS (Fig. 6; see Table S3 in the supplemental material). These strains were able to synthesize ergovaline and lysergyl alanine and, on occasion, ergine.

Expression analysis.

Ergovaline is synthesized preferentially during biotrophic growth. To test whether each of the eas genes is preferentially expressed in planta, semiquantitative reverse transcription (RT)-PCR analysis was performed with RNA extracted from L. perenne Nui infected with E. festucae Fl1 and from E. festucae Fl1 grown axenically in PD broth (Fig. 7). To normalize the fungal RNA levels in each sample (48), cDNA derived from mycelia grown in PD broth was serially diluted to 1/100. At this dilution, a band with an intensity similar to that of the in planta cDNA was amplified by PCR using primers for E. festucae Fl1 tubB. Each gene in the EAS cluster was strongly expressed in planta but either weakly expressed, in the cases of easG and easH, or not at all expressed in axenic culture when grown in complex media.

FIG. 7.

Expression analysis of genes in the EAS cluster. RT-PCR was performed with RNA extracted from E. festucae Fl1 grown axenically in PD medium (A) and L. perenne infected with E. festucae Fl1 (B). cDNA prepared from cultured mycelium was diluted 1/100 to compensate for the biomass differences in planta.

To test the hypothesis that ergovaline genes may be subject to catabolite repression in culture, we isolated RNA from E. festucae Fl1 grown in carbon-, nitrogen-, and phosphate-limited minimal media, as well as from PD media and infected perennial ryegrass. This RNA was tested for lpsB expression by real-time RT-PCR, and no expression could be detected under any of the growth conditions. Transfer of E. festucae Fl1 grown in PD media to a perennial ryegrass extract for 30 min also failed to induce lpsB expression.

DISCUSSION

We describe here the isolation of a cluster of genes proposed to be required for ergot alkaloid biosynthesis in the grass endophyte N. lolii. Molecular cloning and functional analysis of a plant-regulated NRPS indicates that it is likely to be an orthologue of the C. purpurea lpsB gene, encoding the LpsB NRPS required for synthesis of ergovaline. Isolation and analysis of sequences flanking lpsB identified a cluster of genes homologous to ergot alkaloid biosynthetic gene clusters found in C. purpurea (18, 25, 53) and A. fumigatus (19, 54). With the exception of easH, which is likely to be a pseudogene in A. fumigatus (46), the genes common to all three clusters are proposed to encode enzymes for the early steps in the ergot alkaloid biosynthetic pathway. Genes that are shared by C. purpurea and N. lolii but absent from A. fumigatus—lpsA, lpsB, and cloA—have been shown to be required for steps leading to ergopeptines (18, 26, 38). Just two genes, easC, a predicted catalase-like gene, and easD, a predicted oxidoreductase, common to C. purpurea and A. fumigatus clusters remain to be identified for N. lolii. Sequencing of the E. festucae strain E2368 genome is currently under way (C. Schardl, personal communication), and the results will facilitate future cloning of these remaining genes.

While gene sequence is relatively highly conserved between each of the three ergot alkaloid gene clusters, there are several differences in gene order, and the N. lolii cluster is more complex in structure and organization. The transposon insertions in the EAS cluster have generated several repeat sequences that could be signatures of previous recombination events that led to these rearrangements. The presence of putative transposons in the intergenic regions of each of the fully sequenced genes in the N. lolii cluster that are rearranged compared with C. purpurea supports this hypothesis. The EAS cluster is linked with dmaW and lpsA in N. lolii and E. festucae on substantially differently sized NotI fragments, indicative of the presence of additional transposon blocks in N. lolii, as is the case at the LTM locus, which is associated with three large blocks of repeat sequences in N. lolii and just two in E. festucae (62). The abundance of transposon relic sequences adds to the evolutionary potential of the EAS cluster. In Magnaporthe oryzae, transposable element clusters correlate with increased recombination rate, loss of synteny, gene duplication, and sequence divergence from orthologous genes (51). Inter- and intrachromasomal recombination mediated by these repeat sequences in the N. lolii genome could be important in generating genetic diversity.

The predicted N. lolii LpsB enzyme has a modular structure similar to that of the orthologous C. purpurea enzyme, including an adenylation domain, thiolation (peptidyl carrier protein) domain, and condensation domain. LpsB is unusual among fungal NRPSs in that it forms a multienzyme complex with LpsA (42). Multienzyme NRPSs are common in prokaryotes, but the LpsA/LpsB system is the only described example in the fungal kingdom, where NRPSs are otherwise found on a single large multimodular polypeptide (55). There is a condensation domain found at the carboxy end of the LpsB protein and also a partial condensation domain at the amino-proximal end of the predicted N. lolii, but not C. purpurea, LpsA enzyme (20). Whether the partial LpsA condensation domain plays a role in the condensation of the lysergyl-alanine peptide bond in the N. lolii enzyme system remains to be determined. A feature of multienzyme NRPSs in prokaryotic systems is the presence of recently described COM domains, which are required for specific protein-protein recognition (27). Alignment of the carboxy end of the predicted N. lolii and C. purpurea LpsB sequences with bacterial donor COM domains showed that the TPSD motif present at the junction between epimerization and COM domains is absent, a result not totally unexpected given that the C terminus of LpsB is preceded by a condensation domain. The amino-terminal LpsA sequence from N. lolii and the LpsA-1 and LpsA-2 sequences from C. purpurea are also dissimilar to the bacterial acceptor COM domain consensus sequence and, interestingly, to each other. With no other fungal multienzyme NRPS systems yet described, functional analysis is required to identify the residues important for LpsB-LpsA protein-protein interaction.

To confirm the role of lpsB in the formation of ergovaline, targeted disruption was performed with E. festucae, the more genetically tractable sexual progenitor of N. lolii. Plants infected with this mutant were unable to synthesize ergovaline, while complementation with a fragment containing the lpsB gene sequence restored this ability, confirming that lpsB is required for ergot alkaloid biosynthesis. While complementation did not restore ergovaline production to wild-type levels, possibly due to the length of upstream sequence (379 bp) or position effects of the ectopically transformed constructs, complemented strains were clearly able to synthesize ergovaline. A previous analysis of an lpsA mutant of Neotyphodium sp. strain Lp1 had shown a loss of lysergic acid amides lysergyl alanine and ergine along with ergovaline, suggesting that they are derived from either the lysergyl peptide lactam or ergopeptine (37). Our results, showing that synthesis of lysergyl alanine is also blocked by mutation of lpsB, support this idea. Although production of ergine by wild-type E. festucae symbiota was not always detected, its absence from all plants infected with the lpsB mutant strain and occurrence in some plants with the complemented strains suggest that lpsB is required for its synthesis. As with the lpsA mutant of Neotyphodium sp. strain Lp1, accumulation of the major identified clavine alkaloids did not appear to be affected by mutation of lpsB. An increased concentration of an unidentified minor clavine was noted but, interestingly, this appears to be an isomer of the clavine tentatively assigned the structure 6,7-secolysergine, which increased in concentration in the Neotyphodium sp. strain Lp1 lpsA mutant compared with the wild type (37).

Ergot alkaloids are specifically produced during biotrophic growth of epichloë endophytes. Attendant with this, the eas genes are all expressed in planta but not under any axenic culture condition tested. Ergot alkaloid biosynthetic genes in C. purpurea cultures are repressed by high phosphate levels (18, 25). However, our results suggest that in E. festucae, at least the lpsB gene is not derepressed in culture by phosphate, carbon, or nitrogen catabolite starvation. These results suggest that specific plant conditions may be required for induction. Several genes that are specifically expressed in response to plant signals have been identified for plant-associated fungi (4, 5, 49, 58); however, it was only recently, when Yang et al. (59) showed that the pelD gene of Nectria haematococca was expressed in response to homoserine and asparagine, that a specific plant factor responsible for expression of one of these genes was identified. The inability of E. festucae Fl1 to induce lpsB expression in the presence of a plant extract suggests that the inducing factor was not stable or, alternatively, that specific signaling between the epichloë endophyte and the grass host occurs.

Chromatin remodeling as a method of coordinate gene regulation is a possible factor causing selection pressure for secondary metabolite genes to be clustered. The presence of transposons at the EAS locus, particularly AT-rich, RIP-degenerated sequence, a potent inducer of DNA methylation in Neurospora crassa (47), suggests that regulation at the level of chromatin may be important for eas genes. The identification in Aspergillus nidulans of a putative histone methyltransferase, LaeA, that is a global secondary metabolite gene regulator (6), and the finding that transfer of a housekeeping gene to a location within the sterigmatocystin gene cluster partially silences expression (7), supports a gene silencing mechanism. Experiments to determine cis and trans factors required for transcription of eas genes and the effect of genomic location are currently under way. Identifying the regulatory mechanisms that activate the expression of the eas genes in planta and determining whether there is some commonality with mechanisms for activating other plant-induced endophyte genes will be of considerable future interest.