Introduction

Claviceps paspali is a hypocrealean fungus that has been used in the pharmaceutical industry for decades to produce ergot alkaloids. These alkaloids serve as precursors for the manufacture of drugs that treat Parkinson’s disease and migraine (Arcamone et al. 1960; Tudzynski et al. 2001). In its natural environment, C. paspali forms an association with dallis grasses (Paspalum spp.) and produces not only ergot alkaloids but also indole diterpenes (IDTs) such as paspalitrems (Cole et al. 1977; Uhlig et al. 2014). Ingestion of grasses and grains contaminated with paspalitrem IDTs causes an array of symptoms in large animals often referred to as “Paspalum stagger,” characterized by tremor, ataxia, and convulsions. Although Paspalum stagger is rarely lethal for the intoxicated livestock, reduced body mass gain and culling due to accidents suffered by the animals as a result of uncoordinated movement lead to large losses in agriculture (Cole et al. 1977; Moyano et al. 2010; Cawdell-Smith et al. 2010). Importantly for the pharmaceutical industry, C. paspali produces IDTs not only when it forms associations with host plants but also in axenic cultures. The presence of IDTs complicates the isolation and purification of ergot alkaloids during downstream processing in industrial fermentations. Thus, safety concerns and process economics both led to a demand in the pharmaceutical industries for paspalitrem non-producing mutant C. paspali strains (Kozák et al. 2018).

Paspalitrems are derived from paspaline (1, Fig. 1), the simplest cyclic IDT that contains a tetracyclic diterpene moiety fused with an indole group (Kozák et al. 2019). The cyclic diterpene of paspaline is derived from geranylgeranyl diphosphate, while the indole originates from tryptophan via indole-3-glycerol phosphate (Liu et al. 2015). Tailoring of the common paspaline core by various enzymes (e.g., P450 monooxygenases, prenyltransferases, and FAD-dependent monooxygenases) yields the considerable chemical diversity within the IDT group of fungal secondary metabolites (Kozák et al. 2019).

Fig. 1
figure 1

Proposed biosynthesis of paspalitrem B in C. paspali. Structures of reverse prenylated paspaline derivatives obtained in vitro using purified AtmD of A. flavus or PaxD of P. paxilli are also shown

Full size image

Biosynthetic gene clusters for most of the known paspaline-derived IDTs have been identified, and the individual biochemical steps leading to the main IDT groups are well characterized (Young et al. 2005; Young et al. 2006; Nicholson et al. 2009; Tarui et al. 2014; Nicholson et al. 2015). However, IDT biosynthesis in the industrially important IDT producer C. paspali remained little studied. Previously, we provided functional proof for the involvement of a putative IDT biosynthetic locus of C. paspali DSM833, identified by genome sequencing, in the production of paspalitrems. Deletion of the idtCBGF genes of the paspalitrem cluster of C. paspali resulted in the complete abrogation of all IDT-related metabolites, while ergot alkaloid production continued undisturbed in the mutant. We also proposed a biosynthetic scheme for paspalitrem B (2, Fig. 1) in this fungus using bioinformatic analysis of the biosynthetic genes and by detecting IDT intermediates in fermentation extracts of the wild-type strain (Kozák et al. 2018).

During paxilline biosynthesis in Penicillium paxilli, the cytochrome P450 monooxygenase (PaxP) catalyzes the oxidative elimination of the pendant methyl group at C12 of the common intermediate paspaline (1, Fig. 1) and generates the C10 ketone to yield 13-desoxypaxilline (McMillan et al. 2003). In P. paxilli, deletion of paxP results in the accumulation of paspaline (Nicholson et al. 2015). In contrast, deletion of janP (encoding the corresponding PaxP orthologue) during the biosynthesis of shearinines in P. janthinellum results in the elimination of the production of all IDTs, including paspaline (Nicholson et al. 2015). Finally, although the PaxP orthologue TerP of Tolypocladium album is able to convert paspaline to 13-desoxypaxilline, during terpendole biosynthesis 13-desoxypaxilline is only a shunt metabolite, with paspaline serving as the substrate for TerQ which hydroxylates the C11 carbon, giving rise to terpendole E (Motoyama et al. 2012). The biosynthesis of lolitrems follows a similar logic in Neotyphodium lolii/Epichloë festucaë (Saikia et al. 2012). In C. paspali, the candidate enzyme catalyzing the conversion of paspaline to 13-desoxypaxilline is IdtP that shows 41% amino acid sequence identity to PaxP. Considering the above precedents, we were interested to establish whether deletion of idtP in C. paspali would lead to abrogation of IDT biosynthesis; accumulation of paspaline; or perhaps the biosynthesis of shunt paspaline derivatives modified by enzymes with broad specificity, such as a prenyltransferase that may accept paspaline for prenylation (Liu et al. 2013).

During paspalitrem biosynthesis in C. paspali, 13-desoxypaxilline is proposed to be converted to paspalinine (3, Fig. 1) by the IdtQ cytochrome P450 monooxygenase via oxidations at the C13 and C7 positions. Paspalitrems A and C (4 and 5, Fig. 1) are formed by the prenylation of the C21 or C20 positions of paspalinine (3), respectively, by a monoprenyl transferase, suggested to be IdtF (Kozák et al. 2018). In A. flavus, prenylation of paspalinine is catalyzed by the AtmD prenyltransferase to yield aflatrem or β-aflatrem, the structural isomers of paspalitrems A and C, respectively. Surprisingly, IdtF shows only very low similarity to AtmD (21.2% identity over 74% coverage at the amino acid level). In fact, AtmD shows higher similarity to the DmaW prenyltransferase of the ergot alkaloid biosynthetic gene cluster of C. paspali (27.4% identity over 92% coverage at the amino acid level). Indeed, crosstalk between different fungal biosynthetic gene clusters and localization of some secondary metabolite biosynthetic genes outside of the main cluster is often encountered. Thus, considering that IdtF is not an AtmD orthologue, experimental verification of its deduced role in paspalitrem biosynthesis in C. paspali is necessary.

In the current work, we set out to provide proof for the function of IdtP and IdtF by knocking out the idtP and idtF genes of C. paspali and comparing the IDT profiles of the mutant strains to that of the wild type. At the same time, we hoped that we may be able to block IDT biosynthesis at the paspaline or paspalinine stage, thus opening the way for the establishment of fermentation technologies for the large-scale production of the IDT nucleus for subsequent chemical or biosynthetic derivatization and structure—activity relationship studies for pharmaceutical drug discovery.

Materials and methods

C. paspali DSM833 was used throughout this work. The maintenance of the fungus, fermentation conditions, sample preparation for IDT analysis, and genomic DNA isolation was carried out as describer earlier (Kozák et al. 2018). PCR reactions were carried out in 50 μL total volumes, containing 20 ng genomic DNA, or 1 ng plasmid DNA as the template, respectively; 0.2 mmol/L of each dNTP; 1 pmol/L of each primer; 1 μL Phusion® High-Fidelity DNA Polymerase; and 10 μL HF buffer (New England Biolabs, Ipswich, MA). Reaction conditions were as follows: 98 °C for 180 s for the initial denaturation, followed by 31 cycles of amplification (98 °C for 10 s, 55 °C for 15 s, 72 °C for 30 s/kbp), and a final extension step of 60 s/kbp at 72 °C.

Constructions of the pAg-idtF-KO and pAg-idtP-KO vectors for the disruption of the idtF and idtP genes, respectively, were carried out from overlapping PCR fragments using the Gibson Assembly Master Mix (New England Biolabs, Ipswich, MA) (Fig. 2). The hygromycin phosphotransferase gene (hph) (Gritz and Davies 1983) of the pAg-H3 vector (Zhang et al. 2003) and the rest of the vector were amplified by PCR in two separate reactions (primers are listed in Supplementary Table S1 of the Supplementary Information). The left and right targeting arms for the idtF and the idtP genes were also amplified by PCR using C. paspali genomic DNA as the template and appropriate primers (SI Supplementary Table S1). The four PCR amplicons (the hph gene, the rest of the pAg-H3 vector, and the appropriate left and right targeting sequences) were fused using the Gibson Assembly Master Mix, utilizing overlapping sequences between the adjacent DNA fragments at the 5′ ends of the primers. The resulting Gibson reaction products were transformed into E. coli XL1-Blue chemical competent cells (New England Biolabs, Ipswich, MA), and the transformed cells were grown on LB agar plates supplemented with 25 μg/mL kanamycin. Plasmids from kanamycin-resistant colonies were isolated using the EZ-10 Spin Column Plasmid DNA Minipreps Kit (Bio Basic Inc., Toronto, Canada) and verified by diagnostic PCR and DNA sequencing. The correctly assembled pAg-idtF-KO and pAg-idtP-KO plasmids were separately transformed into A. tumefaciens LBA4404 electrocompetent cells (Takara Bio Inc., Kusatsu, Japan), and the transformants were selected on LB agar plates supplemented with kanamycin (25 μg/mL) and streptomycin (50 μg/mL). A. tumefaciens-mediated transformation of C. paspali, and the selection and homogenotization of initial transformants were carried out as described previously (Kozák et al. 2018). The genotypes of the C. paspali transformants were validated for the expected gene knockout alleles using diagnostic PCR reactions as described in the “Results” section, using appropriate primers (SI Supplementary Table S1).

Fig. 2
figure 2

Construction of the pAg-idtF-KO and pAg-idtP-KO vectors. The left and right targeting arms (LTA and RTA, respectively) for the idtF and idtP genes, respectively, were PCR amplified from the genomic DNA of the C. paspali DSM833 strain. The backbone of the pAg-H3 plasmid (without the hph gene) and the hph gene alone were amplified separately and fused with the appropriate LTA and RTA amplicons using Gibson assembly. Gene symbols: hph, hygromycin phosphotransferase; nptII: neomycin phosphotransferase II (kanamycin resistance); trfA, plasmid replication initiation gene; oriV, vegetative origin of replication; LB and RB, left and right T-DNA border, respectively

Full size image

Three validated isolates were collected for each knockout mutant strain, and these were evaluated in shake flask fermentations for IDT congener production. Two isolates of each mutant strain and the wild-type strain were subsequently analyzed in detail by determining their IDT profiles in the fermentation extracts, using liquid chromatography–high resolution tandem mass spectrometry (LC-HRMSn) analysis using an Agilent (Santa Clara, CA) 6550 iFunnel Q-TOF mass spectrometer connected to a 1290 Infinity LC System. Experimental conditions for the fermentation, HPLC separation, and mass spectrometry were described earlier (Kozák et al. 2018). In the current work, the target compounds (paspaline 1, paxilline, paspalinine 3, paspalitrem A 4, paspalitrem C 5, and paspalitrem B 2) were identified by comparison of their retention times and exact masses to our previous data (Kozák et al. 2018). The presence of additional paspalitrem-related IDTs (SI Supplementary Table S2) has been investigated using extracted ion chromatograms (EICs) with narrow mass windows (0.02 Da) for the analytes, with specificity ensured by the acquired high resolution MS data. The fragmentation profiles of the main IDTs isolated from C. paspali were described in our previous work (Kozák et al. 2018). Data shown are representative of at least two fermentation experiments with two independent isolates each per strain, in two technical replicates.

Results

To verify the proposed functions of the idtP and idtF genes in paspalitrem biosynthesis in C. paspali, we used A. tumefaciens-mediated transformation (Kozák et al. 2018) to separately replace these target genes in their entireties with the hygromycin resistance (hph) selectable marker gene. Genomic DNA of 15 hygromycin-resistant transformants each from the idtF- and the idtP-targeted mutant strains, respectively, was isolated and used as templates for a set of PCR reactions to validate the gene knockout events. Four transformants (CPIDTF2, CPIDTF3, CPIDTF7, and CPIDTF9) were found to lack the idtF gene, and three transformants (CPIDTP1, CPIDTP8, and CPIDTP11) were validated to miss the idtP gene. The wild-type C. paspali DSM833 and a strain transformed with the pAg-H3 plasmid were included as positive controls, and these displayed the intact idtF or idtP alleles (Fig. 3d). Further PCR reactions with primers specific for the hygromycin resistance gene and those bracketing the mutant alleles showed that the selected transformants lacking the wild-type alleles have undergone double homologous recombination replacing the idtF or the idtP gene, respectively, with the hyg hygromycin resistance gene (Fig. 3e and f). Genomic DNA from the wild-type strain and that transformed with the pAg-H3 vector did not yield PCR amplicons, as expected. Taken together, these PCR experiments proved that the CPIDTF2, CPIDTF3, and CPIDTF7 isolates are homokaryotic for the ΔidtF allele, while the CPIDTP1, CPIDTP8, and CPIDTP11 isolates are homokaryotic for the ΔidtP allele. These independent isolates representing the same strain (∆idtP, or separately ∆idtF) were verified to be indistinguishable during subsequent fermentations.

Fig. 3
figure 3

PCR validation of C. paspali transformants. a The IDT biosynthetic loci of C. paspali DSM833. Arrow pairs indicate the locations and the sizes (in base pairs) of the PCR amplicons that were designed to detect the intact copies of the idtF (primer pair: idtF-F–idtF-R) and the idtP genes (primer pair: idtP-F–idtP-R), respectively. b and c The ∆idtF (b) or the ∆idtP (c) mutant alleles, respectively, whereby the hph gene replaces the target gene. Arrow pairs indicate the locations and sizes (in base pairs) of the amplicons that were designed to validate the replacement. Primer pairs for the ∆idtF allele: idtF-out-F–hph-in-R; and hph-in-F–idtF-out-R. Primer pairs for the ∆idtP allele: idtP-out-F–hph-in-R and hph-in-F–idtP-out-R. d PCR analysis to detect the presence of the intact idtF (top row) or idtP gene (bottom row). DNA from the wild-type C. paspali DSM833 and a strain transformed with the pAg-H3 plasmid (CP-hph-1) are included as positive controls (intact idtF or idtP alleles). Transformants CPIDTF2, CPIDTF3, CPIDTF7, and CPIDTF9 lack the idtF gene. Transformants CPIDTP1, CPIDTP8, and CPIDTP11 miss the idtP gene. e and f PCR analyses to detect the ΔidtF (e) or the ΔidtP (f) mutant alleles, respectively. DNA from the wild-type C. paspali DSM833 and a strain transformed with the pAg-H3 plasmid (CP-hph-1) are included as negative controls (intact idtF or idtP alleles). In transformants CPIDTF2, CPIDTF3, and CPIDTF7, the idtF gene is replaced by the hph gene. In transformants CPIDTP1, CPIDTP8, and CPIDTP9, the idtP gene is replaced by the hph gene

Full size image

Next, the IDT congener profiles were determined in fermentations with the wild type, the ∆idtP (represented by isolates CPIDTP1 and CPIDTP8) and the ∆idtF C. paspali strains (represented by isolates CPIDTF2 and CPIDTF7). We searched extracts of 12 days old fermentation cultures of these strains using LC-HRMSn for the presence of expected biosynthetic intermediates such as paspaline (1), β-PC-M6, 13-desoxypaxilline, paspalicine, paxilline, and paspalinine (3); and for main fermentation products such as paspalitrem A (4), paspalitrem C (5), and paspalitrem B (2) (Fig. 4, SI Supplementary Tables S2 and S3). As expected, the wild-type C. paspali DSM833 strain produced paspalitrem B (2, m/z 518.2881 for the [M+H]+ ion, calculated 518.2906 for C32H40NO5). The wild-type strain also produced two IDTs with different retention times that correspond to the structural isomers paspalitrems A and C (4 and 5, m/z 502.2946 and 502.2936 for the [M+H]+ ions, calculated 502.2957 for C32H40NO4). Intermediates of paspalitrem biosynthesis were not detected in the extracts of the wild-type strain in this set of experiments, although trace amounts of paspaline, paxilline, and paspalinine had previously been observed in similar fermentations (Kozák et al. 2018).

Fig. 4
figure 4

IDT profiles of the wild-type and mutant C. paspali strains. Extracted ion chromatograms are shown for the LC-HRMSn analysis of the selected IDTs in the fermentation extracts of the wild-type C. paspali DSM833, the ∆idtP mutant CPIDTP1 and the ∆idtF mutant CPIDTF2 strains. Compounds were identified based on their exact mass with a narrow mass window (0.02 Da), and their fragmentation patterns were compared to those described in previous literature examples (Uhlig et al. 2014; Kozák et al. 2018). Peak intensities (Y axis) and chromatographic retention times (X axis) are shown with the same scale for each analyte to facilitate comparison. The representative ion chromatograms (EICs) are arranged in a matrix: the studied IDTs vary within the columns and the different strains vary within the rows (see labels on the left side of the rows and above the columns)

Full size image

Analyses of fermentation extracts of the ∆idtP strain (isolates CPIDTP1 and CPIDT1P8) revealed the absence of all paspalitrems (Fig. 4, SI Supplementary Tables S2 and S3). The only IDT congener which could be detected was paspaline (1). Just as with the ∆idtP transformants, the analyses of the fermentation extracts of the ∆idtF strain (isolates CPIDTF2 and CPIDTF9) also indicated the complete lack of prenylated IDTs. However, a new [M+H]+ ion at m/z 434.2304 was detected in the extracts (Fig. 4) at the retention time characteristic of paspalinine (3) and corresponding to the molecular formula of this intermediate (434.2331 for C27H32NO4). In addition to paspalinine (3), the production of small quantities of paspaline (1) was also observed (Fig. 4).

Besides the above mentioned biosynthetic intermediates and major products, a number of additional IDTs featuring the paspalitrem or the paspaline scaffold were also detected in Paspalum dilatatum samples infected with C. paspali (Uhlig et al. 2014). We carefully searched for the exact masses of the protonated ions of these metabolites, as well for prenylated paspaline analogues (Liu et al. 2013) in the fermentation extracts of the wild type, the ∆idtP and the ∆idtF strains, but none of these metabolites could be detected in the current experiments.

Discussion

C. paspali produces tremorgenic paspalitrem IDT congeners in both axenic cultures and in its natural association with Paspalum spp. (Uhlig et al. 2014; Kozák et al. 2018). Ingestion of grass contaminated with paspalitrem mycotoxins by livestock causes serious losses for agriculture, especially in the Southern Hemisphere (Moyano et al. 2010). In our previous work, we identified the paspalitrem gene cluster of C. paspali by the disruption of the idtCBGF gene locus using A. tumefaciens-mediated gene disruption. As the protein products of three out of the four deleted genes are responsible for the assembly of paspaline (1), a common intermediate of paspaline-type IDTs, deletion of this locus resulted in the complete elimination of the full spectrum of IDTs in C. paspali (Kozák et al. 2018).

In the current work, we set out to substantiate bioinformatic predictions for the function of the idtP and idtF genes using specific gene knockouts that are not expected to disturb the functions of any other genes in the IDT biosynthetic gene cluster of C. paspali. The encoded product of the idtP gene is a cytochrome P450 monooxygenase with 41% sequence identity to PaxP of P. paxilli. In paxilline biosynthesis, PaxP mediates the oxidative conversion of paspaline (1, Fig. 1) to 13-desoxypaxilline. Deletion of paxP in P. paxilli resulted in the elimination paxilline production and the accumulation of the intermediate paspaline (1) (Saikia et al. 2007). Just as with the ∆paxP mutation in P. paxilli, we found that the deletion of idtP in C. paspali leads to the accumulation of paspaline (1) and the complete elimination of any downstream biosynthetic products such as paspalinine (3) and paspalitrems A, B, and C (2, 4, and 5). Importantly, the inactivation of idtP orthologues in other IDT producer fungi yielded different results. Thus, disruption of janP in the shearinine producer P. janthinellum resulted in the elimination of the production of the full spectrum of IDTs, including the production of paspaline (1) that is the substrate of the JanP cytochrome P450 monooxygenase (Nicholson et al. 2015). This indicates that the overall regulation of the assembly of the IDT scaffold differs in P. janthinellum from that of C. paspali and other IDT producers such as P. paxilli. Even more surprisingly, inactivation of the idtP orthologue terP in the terpendole producer T. album leads to the accumulation of terpendole E, a C11-hydroxylated derivative of paspaline (1), and the appearance of a new shunt metabolite, 11-ketopaspaline. This indicates that during the biosynthesis of the terpendoles (and by extension, that of the lolitrems), oxidation of paspaline (1) catalyzed by TerQ (and LtmQ for lolitrems) precedes that of TerP (and LtmP), in contrast to the biosynthetic order of the orthologous enzymes in the rest of the paspaline-derived IDTs such as paspalitrems, aflatrems, shearinines, penitrems, sulpinines, and janthitrems (Saikia et al. 2012; Motoyama et al. 2012).

Among the various IDTs, paspalitrems are most similar to and constitute the structural isomers of the aflatrems that are produced by Aspergillus flavus (Cole et al. 1977; Gallagher and Wilson 1979). Thus, normal prenylation of the indole moiety of paspalinine (3) at the C21 and C20 positions yields the 2-methylbut-2-ene side chain in paspalitrems A and C, respectively (4 and 5, Fig. 1). For aflatrems, reverse prenylation of paspalinine at the same positions affords the 3-methylbut-1-ene side chains of β-aflatrem and aflatrem A, respectively. IdtF of C. paspali is a predicted prenyltransferase with 21% identity over 78% of the amino acid sequence to AtmP encoded in the aflatrem gene cluster of A. flavus that is responsible for the reverse prenylation of paspalinine (3) (Liu et al. 2013). IdtF shows similar levels of identities to PtmD and PenD (both 21% over 95% coverage at the amino acid level), normal prenyltransferases that modify β-paxitriol during penitrem production in Penicillium strains (Oikawa et al. 2016). We proposed that IdtF in C. paspali is responsible for the normal prenylation of paspalinine (3) at the C21 or C20 positions to afford paspalitrem A or C (4 or 5), respectively (Kozák et al. 2018). In agreement with this, the ∆idtF knockout strain of C. paspali was unable to produce prenyl-elaborated IDTs, and accumulated paspalinine (3) with minor amounts of paspaline (1) also detectable. This result clearly validates the predicted function of IdtF as a paspaline C20 or C21 normal prenyltransferase. This also indicates that normal prenylation of paspalinine (3) (as with IdtF) vs. reverse prenylation of the same substrate (as with AtmD) demanded the parallel evolution of non-orthologous prenyltransferases despite the identical substrate. On the other hand, normal prenyltransferases such as IdtF vs. PtmD or PenD are also highly divergent, reflecting the different substrates (paspalinine 3 vs. β-paxitriol).

Purified AtmD from A. flavus and PaxD from P. paxilli were reported to accept paspaline (1) as an alternative substrate in vitro and catalyze the formation of paspaline derivatives α-prenylated at the C21 or C22 positions (Fig. 1) (Liu et al. 2013). Since the C. paspali ∆idtP strain accumulates paspaline (1) and has an intact copy of the idtF gene, we considered that the prenyltransferase IdtF may also be able to produce prenylated paspaline derivatives. We also considered the possibility that the prenyltransferase DmaW from the ergot alkaloid biosynthetic gene cluster of C. paspali (27.4% identity over 92% coverage at the amino acid level to AtmD) may engage in metabolic crosstalk to produce prenylated paspaline derivatives. However, no metabolites could be detected with the calculated m/z of monoprenyl-paspaline or hydroxyprenyl-paspaline in the fermentation extracts of the wild type or the ∆idtP strains. This result indicates that at least in vivo, neither IdtF nor DmaW would accept paspaline (1) as a substrate, although just as with AtmD and PaxD, purified IdtF or DmaW may catalyze paspaline monoprenylation with a low efficiency in vitro.

Uhlig and coworkers (Uhlig et al. 2014) showed that in addition to the expected biosynthetic intermediates of paspalitrem biosynthesis, field samples of C. paspali (such as the C. paspali-P. dilatatum association) also produce a number of compounds with the paspaline or paspalitrem scaffold. Thus, we investigated the presence of such analogues in axenic cultures of C. paspali DSM833 and its ∆idtP and ∆idtF knockout strains. However, none of these analogues (SI Supplementary Table S2) was detectable. This discrepancy between the IDT profiles of the field samples vs. the axenic cultures may be explained by dissimilar regulation of gene expression and/or the presence of plant-derived enzymes and the extended cultivation period in the case of field-collected samples.

In this study, we demonstrated that targeted disruption of the idtP and idtF genes in C. paspali modulates the IDT product spectrum towards paspaline-type IDTs such as paspaline (1) and paspalinine (3). These results highlight the metabolic engineering potential of the A. tumefaciens-mediated C. paspali transformation system (Kozák et al. 2018) in creating efficient platforms for the production of the IDT nucleus for combinatorial biosynthesis and the large-scale production of complex IDTs for various biomedical applications in the future (Kozák et al. 2019).