Fungal strains used in this study are summarized in Supporting Information Table S1. C. purpurea Cp-1 strain and C. paspali No. 24 strain were cultured and fermented following the methods described previously^36,39.

Escherichia coli Trans1-T1 and BL21(DE3) cells were grown in LB medium (5 g/L yeast extract, 10 g/L tryptone, and 10 g/L NaCl). 50 mg/mL ampicillin was supplemented for the cultivation of recombinant E. coli strains.

Saccharomyces cerevisiae BJ5464-NpgA was used for in vivo yeast DNA recombination cloning. The cells were grown in the YPD medium (10 g/L yeast extract, 20 g/L peptone, and 20 g/L glucose). The SD medium lacking uracil was used for selection. Aspergillus nidulans A1145 was kindly provided by Prof. Yi Tang in the Department of Chemical and Biomolecular Engineering and Department of Chemistry and Biochemistry, University of California, California, United States. The strain was grown at 37 ℃ on GMM agar medium (1.0% glucose, 50 mL/L nitrate salt solution, 1 mL/L trace element solution) for sporulation and transformation with appropriate nutrition as required. The nitrate salt solution and trace element solution were described in previous references⁴⁰.

C. purpurea Cp-1 strain was cultured and preserved on T25D agar medium as described previously³⁶. For extraction of genomic DNA of C. purpurea Cp-1 strain, the mycelium was first grown on the T25D plate for 7 days, then inoculated in NL720 liquid medium (200 g/L sucrose, 15 g/L ammonium citrate, 0.25 g/L KH₂PO₄, 0.3 g/L MgSO₄, 1 mg/L FeSO₄, 30 mg/L ZnSO₄, at pH 5.2) and cultivated for 3 days, the mycelium was collected and ground in liquid nitrogen. The genomic DNA was extracted from the resulting cell powder. Polymerase chain reactions for cloning were performed using Q5 high-fidelity DNA polymerase (New England Biolabs, MA, USA) or FastPfu Fly DNA polymerase (TransGen Biotech, Beijing, China) as recommended by the manufacturer. The functional genes of the EAS gene cluster in the Cp-1 strain were amplified using the primers listed in Supporting Information Table S2. For obtaining the full sequences of Cp-lpsA1 and Cp-lpsA2 from the Cp-1 strain, the gene fragments were amplified using the primers Cp-lpsA1/A2-P1 and Cp-lpsA1/A2-P3, Cp-lpsA1/A2-P4 and Cp-lpsA1/A2-P6, Cp-lpsA1/A2-P2 and Cp-lpsA1/A2-P5, Cp-lpsA1/A2-S4F and Cp-lpsA1/A2-S4R, and Cp-lpsA1/A2-S5F and Cp-lpsA1/A2-S5R respectively. Then the chromosome walking technology was applied to obtain an unknown sequence at the 3′ terminal of Cp-lpsA1 and Cp-lpsA2 using the genome walking kit (Takara, Tokyo, Japan). Specifically, the three specific primers Cp-lpsA1/A2-SP1, Cp-lpsA1/A2-SP2, Cp-lpsA1/A2-SP3 with higher annealing temperature were designed, and thermal asymmetric PCR reactions were performed with the specific primers and degenerate primer (AP Primer, provided in the kit). Then the amplified fragment was cloned into the pEASY-simple blunt vector (TransGen) and sequenced. Finally, the segmented fragments were spliced together through overlapping regions to obtain the full sequences of Cp-lpsA1 and Cp-lpsA2.

To construct plasmids for heterologous expression in A. nidulans, the yeast homologous recombination method was adopted. Briefly, the gene of lpsB carrying 200 bp terminator was amplified from the genomic DNA of C. purpurea Cp-1 strain using primers (lpsB-F-pANP and lpsB-R-pANP) containing 30 bp overlapping regions with the A. nidulans vectors. The gene of lpsB and PacI-digested pANP were co-transformed into S. cerevisiae BJ5464-NpgA using PEG/LiAc mediated transformation and selected on uracil-dropout semisynthetic agar medium. The plasmids from the correct yeast colonies were extracted using Zymoprep Yeast Miniprep Kit (Zymo Research, CA, USA) and transformed to E. coli Trans2-Blue (TransGen) for propagation and sequencing. The resulting recombinant plasmid was designated as pANP-lpsB. The genes of Cp-lpsA1 and Cp-lpsA2 were amplified in two segments using primers Cp-lpsA1/A2-1F and Cp-lpsA1/A2-AR, Cp-lpsA1/A2-BF and Cp-lpsA1/A2-3R. There were 100–300 bp homologous regions between the amplified fragments. The amplified DNA fragments of Cp-lpsA1 and Cp-lpsA2 were inserted into PacI-digested pANU, respectively, using the same procedures. The resulting recombinant plasmids were designated as pANU-Cp-lpsA1 and pANU-Cp-lpsA2, respectively.

A. nidulans A1145 was used as the recipient host. The protoplast preparation and transformation were performed as follows. Firstly, 1 × 10⁹ spores of A. nidulans A1145 were inoculated and grown in 20 mL GMM liquid medium with 10 g/L yeast extract at 37 ℃ and 180 rpm for 5 h. The germinated spores were collected by centrifuging at 4 ℃ and 5500 rpm (Eppendorf, Hamburg, Germany) and washed twice with mycelium wash buffer (10 mmol/L sodium phosphate buffer, 0.6 mol/L MgSO₄, pH 7.0). Then 36 mg lysing enzymes from Trichoderma harzianum (Sigma–Aldrich, MO, USA) and 24 mg yatalase (Takara) were dissolved into 12 mL of osmotic buffer (10 mmol/L sodium phosphate buffer, 1.2 mol/L MgSO₄, pH 5.8) and added into the germlings. After digestion at 80 rpm and 30 ℃ for 6–8 h, the enzyme solution was poured into a 50-mL tube and 12 mL of trapping buffer (0.6 mol/L sorbitol, 0.1 mol/L Tris–HCl, pH 7.0) was gently added. The mixture was centrifuged at 4 ℃ and 3750 rpm for 15 min. The protoplasts at the middle layer were pipetted to a new tube, and added with an equal volume of STC buffer (10 mmol/L Tris–HCl, 1.2 mol/L sorbitol, 10 mmol/L CaCl₂, pH 7.5). The mixture was centrifuged at 4 ℃ and 5500 rpm for 15 min to obtain the protoplasts. The protoplasts were further resuspended with STC buffer and brought to a final concentration of 10⁸ protoplasts. The constructed expression plasmid pANP-lpsB in combination with pANU-Cp-lpsA1 or pANU-Cp-lpsA2 was transformed into 100 μL of protoplasts and incubated on ice for 60 min, and then the 600 μL PEG solution (10 mmol/L Tris–HCl, 60% PEG4000, 50 mmol/L CaCl₂, pH 7.5) was added to the mixture, followed by additional incubation at room temperature for 20 min. Then the mixture was plated on GMM-sorbitol medium (GMM solid medium with 1.2 mol/L of sorbitol) and cultured at 37 ℃ for 2–3 days. The positive transformants were designated as An01 and An02, respectively. The empty plasmids pANU and pANP were introduced to A. nidulans to obtain the strain An-control, which was used as the control in this study. The recombinant expression strains were transferred to the GMM plates containing riboflavin and cultured for 3 days at 37 ℃ to obtain spores.

For cultivation and the products analysis of recombinant A. nidulans strains, approximately 1 × 10⁷ fresh spores from positive transformants were used to inoculate 20 mL of CD-ST medium (GMM liquid medium containing 20 g/L starch without glucose) and cultured for 2 days at 25 ℃ and 180 rpm. 2 mg d-lysergic acid dissolved in DMSO was added to the broth with a final concentration of 0.1 mg/mL. The culture was shaken at 180 rpm for an additional 2 days at 25 ℃ and 180 rpm. The broth was adjusted to pH 8.0 and extracted with ethyl acetate. The extracts were evaporated under reduced pressure, resolved into methanol, and subjected to LC‒MS analysis on a C18 column (5 μm, 2.1 mm × 100 mm), at the flow rate of 0.3 mL/min and the UV wavelength of 314 nm. Notably, the solvents used were (A) aqueous solution with 0.01 mol/L ammonium formate and (B) acetonitrile. The solvent gradient was % B, initial, 10%; 10 min, 15%; 13 min, 42%; 25 min, 42%; 30 min, 100%; 31 min, 10%; 36 min, 10%, with the run time of 36 min.

For investigating the function of Cp-LPSA2, the three functional modules of Cp-LPSA2 were respectively utilized to replace their counterparts in Cp-LPSA1. To construct the recombinant Cp-lpsAs with module swapping between Cp-lpsA1 and Cp-lpsA2, the DNA fragments from each module of Cp-lpsA1 and Cp-lpsA2 were amplified, and these fragments and PacI-digested pANU backbone were mixed and transformed into yeast to generate plasmids pANU-Cp-lpsA2^M1Cp-lpsA1^M2Cp-lpsA1^M3, pANU-Cp-lpsA1^M1Cp-lpsA2^M2Cp-lpsA1^M3, pANU-Cp-lpsA1^M1Cp-lpsA1^M2Cp-lpsA2^M3, pANU-Cp-lpsA2^M1Cp-lpsA2^M2Cp-lpsA1^M3, and pANU-Cp-lpsA2^M2Cp-lpsA2^M1Cp-lpsA1^M3, respectively. Then these recombinant plasmids in combination with pANP-lpsB were transformed into the protoplasts of A. nidulans, resulting in the recombinant strains An03–An07 (Table S1), and the products of the recombinant strains were analyzed as mentioned above.

To generate the replacement fragments for Cp-lpsA1 and Cp-lpsA2, the fusion PCR method was used as previously described⁴¹. Primers used to design constructs are listed in Table S2. Firstly, the DNA fragments at 5′ flank and 3′ flank of Cp-lpsA1 and Cp-lpsA2 were amplified using the primers Cp-lpsA1/A2-P1 and Cp-lpsA1/A2-P3, Cp-lpsA1/A2-P4 and Cp-lpsA1/A2-P6, respectively, from genomic DNA of the C. purpurea Cp-1 strain. As a selectable marker, the hph fragment was amplified from the plasmid pAN7-1 using the primers hph-F and hph-R. The three PCR products were combined and used as the template to generate the gene deletion fragments using the primers Cp-lpsA1/A2-P2 and Cp-lpsA1/A2-P5, which were used to transform Cp-1 strain. To prepare the protoplasts, the mycelium of the Cp-1 strain grown on the T25D plate was inoculated in NL720 liquid medium and cultivated at 220 rpm and 24 ℃ for 3 days, and then transferred to fresh NL720 medium again at 10% inoculum volume, and continued to culture for 24 h. The hyphae were harvested by centrifugation at 10,000 rpm and 4 ℃ for 10 min and washed with distilled H₂O twice. The mycelia were then transferred into a 100-mL flask with 5 mL of KCl (0.7 mol/L) containing 50 mg lywallzyme (Guangdong Microbial Culture Collection Center, Guangzhou, China). After shaking at 100 rpm and 25 ℃ for 1–2 h, the mixture was filtered through sterile cotton. The protoplasts were further pelleted by centrifugation for 10 min at 2200 rpm and resuspended with STC buffer (0.85 mol/L sorbitol, 50 mmol/L CaCl₂ in 0.1 mol/L Tris-HCI, pH 7.5) and brought to a final concentration of 10⁸ protoplasts. Cp-lpsA1 or Cp-lpsA2 gene deletion fragments (∼10 μg) were mixed with 100 μL of the protoplasts, and the final volume was adjusted to 200 μL with STC. 50 μL of PEG solution (25% polyethylene glycol 4000, 50 mmol/L CaCl₂, 10 mmol/L Tris–HCl, pH 7.5) was added, and the mixture was incubated for 20 min on ice. Then, 2 mL STC buffer was added to the mixture and the mixture was further incubated at room temperature for 20 min. The process was stopped by dilution with 4 mL STC buffer. Finally, the suspension was added to 100 mL regeneration agar medium (10 g/L glucose, 4 g/L tryptone, 1 g/L yeast extract, 5 g/L beef extract, 2.5 g/L NaCl, 125 g/L sucrose, 5.1 MgCl₂, 2.78 g/L CaCl₂, 0.5 g/L K₂HPO₄, 25 mg/L sodium deoxycholate, 1.5 g/L agar, at pH 7.2) containing 1.2 mg/mL hygromycin B and spread on 5 Petri dishes. 7–10 days later, the transformants were transferred onto fresh T25D plates containing 1.5 mg/mL hygromycin B for isolation of genomic DNA to verify the gene deletion via PCR amplification. The diagnose primers were Cp-lpsA1/A2-P1 and hph-R, Cp-lpsA1/A2-P6 and hph-R, and Cp-lpsA1/A2-self-F and Cp-lpsA1/A2-self-R, respectively (Table S2). To screen the gene deletion homokaryons, a protoplasting process was performed due to the inability of the Cp-1 strain to sporulate. The potential gene-deleted C. purpurea strains and the parent Cp-1 strain were first cultivated in 50 mL seed medium at 240 rpm and 24 ℃ for 3 days in the dark. Then, 5 mL of this culture was transferred into the 80 mL fermentation medium and the cultivation was continued under the same condition for another 14 days. The ergopeptines were extracted and detected as previously described³⁶.

The mycelia of C. purpurea Cp-1 strain under the different fermentation stages were collected and lyophilized, and the total RNA was isolated using Trizol (Invitrogen, CA, USA) following the manufacturer’s instructions. 100 mg of mycelia per sample was used as the starting material for the determination of total RNA. The reverse transcription polymerase chain reaction (RT-PCR) was carried out using TransScript All-in-One First-Strand cDNA Synthesis SuperMix (TransGen), and the cDNA was used for the real-time analysis. For real-time reverse transcription quantitative PCR (RT-qPCR), independent assays were conducted using the Ultra-SYBR One Step RT-qPCR kit (CWBIO, Beijing, China) with three biological replicates, and the expression levels were normalized to the mRNA level of tubulin. The 2^‒ΔΔCT method was used to determine the change in expression. The primers used for RT-qPCR are listed in Table S2.

The boundaries of A domains in Cp-LPSA1 and Cp-LPSA2 were predicted by CD-search in NCBI, and then the intron-free sequences of A domains were amplified by using the cDNA of C. purpurea Cp-1 strain as template and the primers Cp-lpsA1/A2^Ade1-F and Cp-lpsA1/A2^Ade1-R, Cp-lpsA1/A2^Ade2-F and Cp-lpsA1/A2^Ade2-R, respectively (Table S2). The PCR products were digested with EcoR I and Xho I and ligated into pCold TF digested with the same restriction enzymes to generate the plasmids pCold TF-Cp-lpsA1^Ade1, pCold TF-Cp-lpsA1^Ade2, pCold TF-Cp-lpsA2^Ade1 and pCold TF-Cp-lpsA2^Ade2, respectively (Table S1). All the correct constructs were verified by DNA sequencing. These constructed plasmids were transformed to chemically competent E. coli BL21 (DE3) by heat shock and grown overnight at 37 ℃ on an LB agar plate supplemented with 50 μg/mL ampicillin. For protein induction and purification, the single colony was inoculated into LB media overnight as seed culture at 37 ℃. After inoculated into fresh LB medium at 1% (v/v), cells were grown at 37 ℃ and 220 rpm supplemented with the 50 μg/mL ampicillin to a final OD₆₀₀ around 0.6. The culture was quickly cooled to 15 ℃ in ice water, and let stand for 30 min. Then the 0.1 mmol/L IPTG was added to induce protein expression and the cells were incubated with shaking at 15 ℃ for an additional 24 h. Then, the cells were harvested by centrifugation for 10 min at 10,000×g, and washed and resuspended in buffer A (150 mmol/L NaCl, 20 mmol/L Tris–HCl, pH 8.0). The cells were lysed by sonication (60 × 5 s cycles), and the cellular debris was removed by centrifugation for 30 min at 12,000×g two times. After filtration through a 0.45 μm filter, the supernatant was subjected to a 5-mL HisTrap HP column (GE Pharmacia, NY, USA) using an AKTA purifier, and then eluted with gradient from 0 to 100% buffer B (150 mmol/L NaCl, 300 mmol/L imidazole, 20 mmol/L Tris–HCl, pH 8.0) for 20 column volumes at a flow rate of 2 mL/min. Then the elution was concentrated and further applied to a HiTrap desalting column (GE Pharmacia) washed with storage buffer (50 mmol/L Tris–HCl, 50 mmol/L NaCl, 5% glycerol, 0.5 mmol/L EDTA, pH 8.0). The purified protein was confirmed by SDS-PAGE and stored at −80 ℃ with 5% glycerol. The activities of the A domains of Cp-LPSA1 and Cp-LPSA2 on 19 types of standard essential amino acids as substrates were measured by monitoring the release of PPi at 360 nm using the EnzChek pyrophosphate assay kit according to the manufacturer’s recommendation (Life Technology, NJ, USA). In detail, 10 μL of suitably diluted A domain proteins was added to each 100 μL reaction system, which contained 5 μL 20 × reaction buffer, 20 μL MESG, 1 μL of PNP (1 U), 5 mmol/L ATP, 5 mmol/L substrate. Reactions were carried out at 30 ℃ on a multimode plate reader Enspire (PerkinElmer, MO, USA) using a 96-well plate, measured by monitoring PPi release at 360 nm continuously for 30 min. The standard curve of the pyrophosphate assay was generated using standard pyrophosphate. Reactions for each A domain protein were conducted in triplicate with boiled A domain proteins as control.

For examining the functions of the P1-LPSA1 and P1-LPSA2 from C. purpurea P1 strain, the P1-lpsA1 and P1-lpsA2 genes were synthesized by BGI (Beijing, China) based on the sequences registered in GenBank. The 3′ end of P1-lpsA2 was found to have additional 16 bases of discontinuous insertions and one base of deletion compared to the other reported lpsAs, which may potentially cause the downstream mutation, thus a modified version of P1-lpsA2 (P1-lpsA2^∗) was also synthesized, in which the inserted or missing bases are corrected. Each of the synthesized P1-lpsA1, P1-lpsA2, and P1-lpsA2^∗ was mixed with PacI-digested pANU and transformed into yeast to generate plasmids pANU-P1-lpsA1, pANU-P1-lpsA2 and pANU-P1-lpsA2∗, respectively. Then the recombinant plasmids in combination with pANP-lpsB were transformed into the protoplasts of A. nidulans, resulting in the recombinant strains An12–An14 (Table S1). To construct chimeric enzymes containing functional modules of LPSAs from the Cp-1 strain and the P1 strain, the DNA fragments of different LPSAs modules were amplified, and the recombinant plasmids were constructed, named pANU-P1-lpsA1^M1Cp-lpsA2^M2Cp-lpsA1^M3, pANU-P1-lpsA1^M1_mCp-lpsA2^M2Cp-lpsA1^M3(where P1-lpsA1^M1_m includes the restored missing 767 bp at the 5′ end using the homologous sequence of Cp-lpsA2), pANU-P1-lpsA2^M1Cp-lpsA2^M2Cp-lpsA1^M3, pANU-Cp-lpsA1^M11-lpsA1^M2Cp-lpsA1^M3, pANU-Cp-lpsA2^M1P1-lpsA1^M2Cp-lpsA1^M3, pANU-Cp-lpsA1^M1P1-lpsA2^M2Cp-lpsA1^M3, pANU-Cp-lpsA2^M1P1-lpsA2^M2Cp-lpsA1^M3, pANU-Cp-lpsA1^M1Cp-lpsA2^M2P1-lpsA1^M3, pANU-Cp-lpsA1^M1Cp-lpsA2^M2P1-lpsA1^M3# (where P1-lpsA1^M3# includes the restored missing 363 bp at the 3′ end using the homologous sequence of Cp-lpsA1), pANU-Cp-lpsA2^M1Cp-lpsA2^M2P1-lpsA2^M3∗(where P1-lpsA2^M3∗ refers to the correction of the inserted or missing bases in its module 3), pANU-Cp-lpsA1^M1Cp-lpsA2^M2P1-lpsA2^M3∗, and pANU-Cp-lpsA1^M1Cp-lpsA2^M2P1-lpsA2^M3# (where P1-lpsA1^M3# refers to the compensation for the missing 118 bp at the 3′ end using the homologous sequence based on P1-lpsA2^M3∗), respectively (Table S1). Then the recombinant plasmids in combination with pANP-lpsB were transformed into the protoplasts of A. nidulans, resulting in the generation of recombinant strains An15–An21 and An24–An28 (Table S1). The products of the recombinant strains were analyzed as previously mentioned.

The DNA fragment of easH1 was amplified from the genomic DNA of Cp-1 strain using primers (easH1-F and easH1-R) and cloned into pANR to generate pANR-easH1. Then the pANR-easH1 was introduced into the corresponding ergopeptam-producing strains, resulting in the generation of recombinant strains An08–An11, and An22–An23 (Table S1). The products of the recombinant strains were analyzed as previously mentioned. To isolate compounds 3, 6, 8, and 9, the fresh spores from the recombinant strains An09, An11, An22, and An23 were respectively used to inoculate CD-ST liquid medium, and cultured for 2 days (25 ℃, 180 rpm). d-lysergic acid dissolved in DMSO was added to the broth. The culture was sequentially shaken at 180 rpm for an additional 2 days (25 ℃, 180 rpm). The products were extracted as mentioned previously. The targeted compound was further isolated and purified using semi-preparative HPLC. The isolated products were dissolved in DMSO-d₆ or CDCl₃. NMR spectra of the samples were recorded at room temperature on an INOVA 600 NMR spectrometer (Varian, CA, USA) to obtain ¹H NMR, ¹³C NMR, HSQC, and HMBC spectra.

According to the method mentioned above, the recombinant plasmids containing multiple or single point mutations were constructed, named as pANU-Cp-lpsA1^M1Cp-lpsA1^M2−pentaCp-lpsA1^M3, pANU-Cp-lpsA1^M1P1-lpsA1^M2−pentaCp-lpsA1^M3, pANU-Cp-lpsA1-A196V, pANU-Cp-lpsA1-V254M, pANU-Cp-lpsA1-G256A, pANU-Cp-lpsA1-I285V, pANU-Cp-lpsA1-I286G, and pANU-Cp-lpsA1-I286A, respectively. The recombinant plasmids in combination with pANP-lpsB were transformed into the protoplasts of A. nidulans, resulting in the generation of recombinant strains An29–An36 (Table S1). The products of the recombinant strains were analyzed by LC‒MS as previously mentioned.

The three-dimensional structures of the second A domain in Cp-LPSA1 and P1-LPSA1 (referred as Cp-LPSA1^Ade2 and P1-LPSA1^Ade2) were predicted from Swiss model http://swissmodel.expasy.org/)⁴² using the available structure of A domain of gramicidin synthetase 1 (Protein Data Bank code 11AMU), which has 28% sequence identity and 34% sequence similarity to Cp1-LPSA1^Ade2, and 29% sequence identity and 34% sequence similarity to P1-LPSA1^Ade2 as a template. The predicted 3D structure and substrate aminoacyl-AMP were selected for the docking experiments with AutoDockTools Proteins, preprocessed by deleting the water molecules, adding hydrogen, and computing Gasteiger charges. The ligands were preprocessed by adding polar hydrogen and detecting roots. The grid size was set to cover the ten key positions that are relevant for substrate specificity which are collectively referred to as the nonribosomal code. The conformations that the ligand bound to the cavity inside the protein and had the lowest binding free energies were selected as the optimal docking poses. PyMOL was used to visualize molecular docking results.

The C. paspali No. 24 strain was first cultured for 10 days as previously reported³⁹. On the 8th day of C. paspali No. 24 cultivation, the recombinant A. nidulans was cultured in a liquid CD-ST medium as mentioned above. Subsequently, C. paspali No. 24 and recombinant A. nidulans were co-cultivated for 2 days at different ratios of inoculation volumes, or different volumes of C. paspali No. 24 fermentation supernatant were added to the culture of recombinant A. nidulans, followed by continued cultivation for another 2 days. The production of specific ergopeptines was analyzed as previously described³⁶.

Based on the published sequence information of the EAS gene cluster of C. purpurea 20.1 strain available on NCBI (GenBank accession no. JN186799.1), all the genes within the EAS gene cluster of C. purpurea Cp-1 strain were amplified. Specifically, the full sequences of Cp-lpsA1 (GenBank accession no. PQ097018.1) and Cp-lpsA2 (GenBank accession no. PQ097019.1), each with a length of 10.9 kb, were obtained through segmented amplification combined with chromosome walking technology. Conserved domain search analysis confirmed that both Cp-LPSA1 and Cp-LPSA2 in the Cp-1 strain belonged to trimodular NRPS, with each module containing the A, T, and C domains. Most genes of the EAS gene cluster between the Cp-1 and 20.1 strains exhibited 97%–100% sequence identity (Supporting Information Table S3). However, both Cp-lpsA1 and Cp-lpsA2 only maintained 84.6%–94.8% amino acid sequence identity with those from other C. purpurea strains (Supporting Information Table S4), confirming the previous opinion that the sequence heterogeneity in lpsAs resulted in different ergopeptines profiles among C. purpurea strains²⁰. Furthermore, the amino acid sequence identity between Cp-lpsA1 and Cp-lpsA2 was only 86.7%, suggesting that they may be responsible for synthesizing different ergopeptines in the Cp-1 strain.

Due to its clean genetic background, robust native gene-splicing system, and superior eukaryotic protein expression capability^43-45, the filamentous fungus A.nidulans A1145 was selected as a heterologous expression host to investigate the function of Cp-LPSAs. These advantages ensure accurate gene splicing and efficient expression of the giant fungal LPSAs. Consequently, either Cp-lpsA1 or Cp-lpsA2, along with lpsB, was introduced into the A. nidulans A1145 host to generate strains An01 and An02, respectively. The recombinant strains were cultured by feeding d-lysergic acid as one of the substrates, and the resulting products were analyzed by liquid chromatography-mass spectrometry (LC‒MS). Co-expression of Cp-lpsA1 and lpsB in A. nidulans host led to the formation of 2a (the [M + H]⁺ ion at m/z 546), 3a (the [M + H]⁺ ion at m/z 560), and 4a (the [M + H]⁺ ion at m/z 560) (Fig. 2Aii and Supporting Information Fig. S2). The oxygenase EasH1 was added to strain An01 to generate strain An08, where 2a and 3a were further converted into ergocornine (2) and α-ergocryptine (3), respectively (with 4 potentially being too low to be detected) (Fig. 3A and Bii). The structures of these products were identified through MS detection and NMR analysis (Supporting Information Figs. S2, S19 and S20, Table S6). Thus, 2a and 3a were identified as ergocornam and α-ergocryptam, respectively. The identification of 3a was further supported by the product analysis of the recombinant strain An09, which produced only 3 (Fig. 3A and Biii). Consequently, 4a was inferred to be β-ergocryptam. These ergopeptams were the precursors of 2, 3, and 4, which were the primary products of the C. purpurea Cp-1 strain (Fig. 2Aix). To our surprise, no ergopeptams were produced when Cp-lpsA2 and lpsB were co-expressed in the A. nidulans host (Fig. 2Aiii). The reason was later confirmed to be the inactivation of module 3 of Cp-LPSA2 (see below).

Meanwhile, we also tried to delete the Cp-lpsA1 and Cp-lpsA2 genes from the C. purpurea Cp-1 strain, respectively. Due to the non-sporulation of C. purpurea under laboratory conditions, the gene-deleted mutants had to be selected by the protoplasting process. However, it is challenging to acquire a complete gene knockout because of the multinucleate characteristic of the protoplasts. After screening hundreds of single colonies, the homokaryotic Cp-lpsA2 deletion mutant (ΔCp-lpsA2) was obtained, while the Cp-lpsA1 knockout mutant (ΔCp-lpsA1_hetero) remained heterokaryotic, meaning that the Cp-lpsA1 gene was deleted in some nuclei but still existed in other nuclei within the same cells (Supporting Information Fig. S3). High-performance liquid chromatography (HPLC) analysis of the ethyl acetate extracts from cultures showed that the contents of 2, 3, and 4 in the ΔCp-lpsA1_hetero strain were significantly reduced compared to the wild-type strain, even though the Cp-lpsA1 gene had not been completely deleted. In contrast, there was no noticeable difference in the ΔCp-lpsA2 strain (Fig. 2B). In addition, throughout the entire fermentation stages of the wild-type strain, the mRNA expression level of Cp-lpsA1 was distinctly higher than that of Cp-lpsA2 (Fig. 2C). These results demonstrated that only Cp-LPSA1 was responsible for the biosynthesis of ergopeptines (2, 3 and 4) in C. purpurea Cp-1 strain and the Cp-LPSA2 was inactive. These data were also consistent with the heterologous expression results mentioned above.

Based on the tripeptide chain composition of 2, 3, and 4, it can be concluded that module 1 of Cp-LPSA1 recognizes Val, while module 2 recognizes Val, Leu, or Ile. The first two A domains of Cp-LPSA1 were further expressed and purified, and they contained ten highly conserved regions (motif 1–motif 10) (Supporting Information Fig. S4). In vitro activity tests revealed that the first A domain of Cp-LPSA1 (Cp-LPSA1^Ade1) exhibited the highest catalytic activity towards Val, while the second A domain of Cp-LPSA1 (Cp-LPSA1^Ade2) showed the higher activities towards Leu, Val, and Ile (Supporting Information Fig. S5A), which largely confirmed the function of Cp-LPSA1. It was also noted that Cp-LPSA1^Ade2 possessed considerable activity towards Ala, but the product with the second amino acid of the tripeptide moiety being Ala was not detected. It implied that the individual structural domain alone could not reflect the enzyme’s function.

Collectively, it was confirmed that Cp-LPSA1 catalyzed the formation of 2, 3, and 4. Its first A domain accepted Val, its second A domain accepted any one substrate among Val, Leu, and Ile, while its third A domain accepted Pro (Fig. 2D).

To further investigate the function of Cp-LPSA2, the three functional modules of Cp-LPSA2 were respectively utilized to replace their counterparts in Cp-LPSA1. The recombinant Cp-LPSAs were heterologously expressed in A. nidulans, generating strains An03‒An05. Excitingly, replacing module 1 of Cp-LPSA1 with that of Cp-LPSA2 led to the accumulation of three additional products (5a, 6a, and 7a) in strain An03 compared to the control strain (Fig. 2Ai and 2Aiv). These products were identified as ergovalam (5a), α-ergosam (6a), and β-ergosam (7a) through LC‒MS detection, comparative MS/MS fragmentation analysis, and NMR analysis of the final ergopeptine products (Fig. 3C, Supporting Information Figs. S6-S8, S21-S24, Table S7). Especially, 5a and 6a were further converted into ergovaline (5) and α-ergosine (6), respectively, following the action of EasH1 in An10 (Fig. 3A and Cii). The identification of 6a was also confirmed by the product analysis of the recombinant strain An11, which produced only 6 (Fig. 3A and Ciii). Consequently, 7a was inferred to be β-ergosam. Since the first amino acid of the cyclic peptide moieties of 5a, 6a, and 7a was Ala, the first A domain of Cp-LPSA2 was demonstrated to specifically activate Ala. Similarly, replacing module 2 of Cp-LPSA1 with that of Cp-LPSA2 resulted in the production of 3a in strain An04 (Fig. 2Av), indicating that the second A domain of Cp-LPSA2 specifically activated Leu. The purified first two A domains of Cp-LPSA2 (Cp-LPSA2^Ade1 and Cp-LPSA2^Ade2) exhibited the highest activities towards Ala and Leu in vitro, respectively (Supporting Information Fig. S5B). As the first two modules of Cp-LPSA2 were specific to Ala and Leu, respectively, the normal Cp-LPSA2 should be able to catalyze the formation of 6a if module 3 could recognize substrate Pro.

Next, we focused on the function of its module 3 of Cp-LPSA2 and replaced the module 3 of Cp-LPSA1 with that of Cp-LPSA2. It was found that the activity of the recombinant enzyme in strain An05 was almost lost (Fig. 2Avi), and only a trace amount of products were detectable through EIC (Supporting Information Fig. S9), indicating that module 3 of Cp-LPSA2 was inefficient, which led to the functional loss of Cp-LPSA2. As expected, the activity of Cp-LPSA2 could be apparently restored by replacing its module 3 with that of Cp-LPSA1, as evidenced by the generation of the expected product 6a and 6 in the recombinant strains An06 and An11, respectively (Figure 2, Figure 3Ciii). This observation once again confirmed that the first two modules of Cp-LPSA2 were specific to Ala and Leu, respectively, endowing Cp-LPSA2 with the capacity to synthesize 6 when module 3 was normalized (Fig. 2D). Later, the sequence identity of module 3 between Cp-LPSA1 and Cp-LPSA2 was also compared, and it was found that the sequence identity between the two modules was 94.83%, with the ten specific sites responsible for recognizing Pro remaining unchanged (Supporting Information Fig. S10). The reason for the deficiency of module 3 in Cp-LPSA2 needs to be further investigated.

In addition, we also attempted to swap the positions of modules 1 and module 2 of Cp-LPSA2 to observe if the recombinant enzyme could be functional. Unfortunately, no product was detected in the recombinant strain An07 (Fig. 2Aviii), suggesting that the biosynthesis of ergopeptines was of directionality from module 1 through module 2 to module 3.

A previous study reported that P1-LPSA1 (GenBank accession no. AJ011964.1) from C. purpurea P1 strain was responsible for producing ergotamine (9) (the cyclopeptide: Ala-Phe-Pro), as evidenced by gene deletion³⁵. Meanwhile, P1-LPSA2 (GenBank accession no. AJ884678.1) was deduced to be involved in the biosynthesis of α-ergocryptine (3) (the cyclopeptide: Val-Leu-Pro)³⁵, but lacking substantive supporting evidence. The aforementioned platform was also used to examine the functions of the P1-LPSA1 and P1-LPSA2. Since the original strain was not available, we had to download the relevant sequences from the NCBI database and synthesize them. The P1-lpsA1 and P1-lpsA2, along with the lpsB of the Cp-1 strain, were individually expressed in A. nidulans to generate recombinant strains. Unfortunately, no product accumulation was observed in any of the recombinant strains (Supporting Information Fig. S11), indicating that the registered data in the NCBI database should be incorrect. Thus, the functions of individual modules of P1-LPSAs were examined by combining these modules with the validated functional modules of Cp-LPSAs to form the chimeric enzymes.

Firstly, we checked the module 1 functions of P1-LPSA1 and P1-LPSA2 by combining each of the modules 1 with modules 2 and 3 of Cp-LPSAs. Sequence analysis revealed that P1-lpsA1 lacked 767 bp at the 5′ end compared to other LPSA sequences (Supporting Information Fig. S12A), which could lead to its inactivity. We constructed two chimeric enzymes: P1-LPSA1^M1Cp-LPSA2^M2Cp-LPSA2^M3 (where P1-LPSA1^M1 lacks 767 bp) and P1-LPSA1^M1_mCp-LPSA2^M2Cp-LPSA2^M3 (where P1-LPSA1^M1_m includes the restored 767 bp derived from homologous sequence of Cp-lpsA2). Our results demonstrated that the latter enzyme could produce 6a (the cyclopeptide: Ala-Leu-Pro), confirming that this 767 bp sequence was crucial for the activity of module 1 in P1-LPSA1, specifically recognizing Ala (Fig. 4Ai‒iii). In addition, the chimeric enzyme P1-LPSA2^M1Cp-LPSA2^M2Cp-LPSA1^M3 could produce 3a (the cyclopeptide: Val-Leu-Pro), indicating that module 1 of P1-LPSA2 specifically recognized Val (Fig. 4Aiv). Ala and Val are also the first amino acids of ergotamine and α-ergocryptine cyclopeptides, respectively.

Additionally, module 2 of P1-LPSA1 or P1-LPSA2 was used to replace their counterpart in Cp-lpsAs. The results showed that the chimeric enzymes Cp-LPSA1^M1P1-LPSA1^M2Cp-LPSA1^M3 in strain An18 and Cp-LPSA2^M1P1-LPSA1^M2Cp-LPSA1^M3 in strain An19 led to the formation of ergocristam (8a) with the [M + H]⁺ ion at m/z 594 and ergotamam (9a) with the [M + H]⁺ ion at m/z 566, respectively (Fig. 4Av‒vi, Supporting Information Fig. S13). These compounds were further converted into ergocristine (8) and ergotamine (9) by introducing oxygenase EasH1 into the corresponding recombinants (Fig. 4B and C). The structures of 8 and 9 were also confirmed by LC‒MS detection and NMR analysis (Supporting Information Figs. S13, S25-S30, Tables S8 and S9). These results confirmed that module 2 of P1-LPSA1 specifically recognized Phe. Similarly, the strains An20 and An21, which harbored chimeric enzymes Cp-LPSA1^M1P1-LPSA2^M2Cp-LPSA1^M3 and Cp-LPSA2^M1P1-LPSA2^M2-Cp-LPSA1^M3, respectively, produced 3a and 6a, indicating that module 2 of P1-LPSA2 specifically recognized Leu (Fig. 4Avii‒viii). Phe and Leu are also the second amino acids of ergotamine and α-ergocryptine cyclopeptides, respectively.

We further investigated the functions of module 3 of P1-LPSAs. Sequence alignment analysis revealed that P1-lpsA1 lacked 363 bp at its 3′ end compared to other lpsA sequences (Supporting Information Fig. S12B), which may result in the inactivity of module 3 (Fig. 4Di). Using the homologous sequence from the 3′ end of Cp-lpsA1 to fill the gap, we constructed the chimeric enzyme Cp-LPSA1^M1Cp-LPSA2^M2P1-LPSA1^M3# to successfully produce 3a, indicating that the corrected module 3 of P1-LPSA1 could recognize Pro (Fig. 4Dii). Additionally, sequence analysis showed that module 3 of P1-lpsA2 contained an additional 16 bases of discontinuous insertions and one base of deletion compared to the other reported lpsAs, which might potentially cause the downstream mutation (Supporting Information Fig. S14). Furthermore, P1-lpsA2 also lacked a length of 118 bp at its 3′ end compared to other lpsA sequences (Fig. S12B). To address these abnormal changes, we first synthesized a modified version of P1-lpsA2^M3∗, in which the additional 16 bases of discontinuous insertions and one base of deletion were corrected. Based on P1-lpsA2^M3∗, we further synthesized a modified version of P1-lpsA2^M3#, in which the missing 118 bp at the 3′ end was compensated by using the homologous sequence. The two modified third modules were used to construct the following chimeric enzymes: P1-LPSA2^M1P1-LPSA2^M2P1-LPSA2^M3∗, Cp-LPSA2^M1Cp-LPSA2^M2P1-LPSA2^M3∗, Cp-LPSA1^M1Cp-LPSA2^M2P1-LPSA2^M3∗, and Cp-LPSA1^M1Cp-LPSA2^M2P1-LPSA2^M3#. Unfortunately, none of these chimeric enzymes could synthesize the corresponding ergopeptams (Fig. 4Diii‒v, Supporting Information Fig. S11), indicating that the other errors still existed in the NCBI registered sequence. Briefly, except the registered module 3 of P1-LPSA2, all other modules of P1-LPSAs have been characterized in terms of their substrate specificities, in which modules 1, 2 and 3 of P1-LPSA1 recognized Ala, Phe and Pro [three amino acids of ergotamine (9) cyclopeptide], respectively, while the modules 1 and 2 of P1-LPSA2 recognized Val and Leu [first two amino acids of α-ergocryptine (3) cyclopeptide], respectively (Fig. 4E).

Ten key positions relevant to substrate specificity have been identified within A domains in NRPS⁴⁶, and the amino acid residues occupying these positions were collectively referred to as the non-ribosomal peptide code (NRP code)^47,48. These NRP codes within the LPSAs from Cp-1 and P1 strains were compared (Fig. 5A). The sequence identity of these key sites in the first two A domains was lower, whereas the corresponding sites in the third A domain were identical because the binding substrate was Pro.

In module 1, the NRP codes in Cp-LPSA1 and P1-LPSA2 were identical (D¹⁹²A¹⁹³I¹⁹⁶F²³³C²⁵⁴G²⁵⁶G²⁷⁷P²⁸⁵L²⁸⁶K⁵¹⁷), specifically recognizing Val. Similarly, the NRP codes in module 1 of Cp-LPSA2 and P1-LPSA1 were also the same (D¹⁹²L¹⁹³F¹⁹⁶F²³³C²⁵⁴G²⁵⁶G²⁷⁷P²⁸⁵L²⁸⁶K⁵¹⁷), and both modules activated Ala (Fig. 5A). The sequence identity of NRP codes in module 2 of LPSAs was lower, and this module was crucial in determining the diversity of ergopeptine profiles between different C. purpurea strains. Comparison of the NRP codes in module 2 of Cp-LPSA1 and P1-LPSA1 revealed differences in five residues (Fig. 5A), which might be the key sites leading to the recognition of different substrates. It was hypothesized that if these five differential residues in Cp-LPSA1 (A196, V254, G256, I285 and I286) were mutated into corresponding residues at the equivalent positions in P1-LPSA1 (V196, M254, A256, V285 and G286), the ergopeptine profiles might be switched from the product synthesized by Cp-LPSA1 to that of P1-LPSA1. Encouragingly, our result confirmed this prediction. In the recombinant strain An29 harboring the pentameter mutant Cp-LPSA1^M1Cp-LPSA1^M2−pentaCp-LPSA1^M3 (where Cp-LPSA1^M2−penta indicated that Cp-LPSA1^M2 contained five point mutations: A196V, V254M, G256A, I285V, and I286G), the production of 2a, 3a, and 4a was significantly reduced, while 8a was produced at a content of 71.45% (Fig. 5Bi‒iii, Supporting Information Table S5). Likewise, the five differential residues in P1-LPSA1 were mutated into the corresponding residues in Cp-LPSA1, generating the pentameter mutant Cp-LPSA1^M1P1-LPSA1^M2−pentaCp-LPSA1^M3 (where P1-LPSA1^M2−penta indicated that P1-LPSA1^M2 contained five point mutations: V196A, M254V, A256G, V285I, and G286I). The recombinant strain An30 harboring this chimeric enzyme almost lost the ability to synthesize 8a but produced 2a and 3a (Fig. 5Biv, and Table S5). These results confirmed that the combinatorial directed mutagenesis of these differential sites among NRP codes could change substrate specificity. The homology models of the second A domain of Cp-LPSA1 (Cp-LPSA1^Ade2) and P1-LPSA1(P1-LPSA1^Ade2) were constructed to reveal the molecular mechanism underlying their substrate specificity. The binding pockets of Cp-LPSA1^Ade2 and P1-LPSA1^Ade2 were surrounded mainly by the ten key sites, most of which were hydrophobic residues except D192 and K517. The binding pocket of P1-LPSA1^Ade2 was more spacious than that of Cp-LPSA1^Ade2 due to residues at positions 285 and 286 with relatively short side chains, which might provide additional flexibility to accommodate Phe with a bulky phenyl ring structure (Fig. 6A).

Unlike other reported modules with rigid substrate specificity, module 2 of Cp-LPSA1 accepted three structurally similar amino acids, exhibiting relatively broad substrate specificity. Further site-directed mutagenesis experiments were conducted to identify the residue determining ergopeptine formation in Cp-LPSA1. The five differential sites in module 2 of Cp-LPSA1 (template) were mutated towards their counterparts P1-LPSA1 or Cp-LPSA2, generating strains An31–An36. Among the mutants with single point mutations from Cp-LPSA1 to P1-LPSA1, only the I286G mutation in Cp-LPSA1 could produce a small amount of 8a (Fig. 5Bix). This might be due to the mutation of the residue at position 286 to Gly, which had the smallest side chain, allowing for the expansion of the active cavity (Supporting Information Fig. S15D), and providing conditions for accommodating Phe. Among the mutants with single-point mutations from Cp-LPSA1 to Cp-LPSA2, the V254M or I286A mutation in Cp-LPSA1 showed a significant increase in the levels of 3a (Fig. 5Bvi and 5Bx), particularly the latter. Specifically, the I286A mutation increased the content of 3a from 33.6% to 81.1% (Table S5), indicating enhanced affinity of this mutant for Leu. Regarding the change in substrate affinity, the following speculations were put forward. When Ile286 in Cp-LPSA1 was mutated into an amino acid with a smaller side chain, such as Ala, the binding pocket became larger, potentially allowing for easier entry of Leu (Supporting Information Fig. S15C). Moreover, the docking model of the enzyme-aminoacyl-AMP complex indicated that the mutant with I286A mutation might have a stronger interaction with Leu (three hydrogen bonds formed among Leu-AMP with D192 and I285) than with Val or Ile (Fig. 6B and C).

Furthermore, it was found that the G256A mutation in Cp-LPSA1 predominantly produced 2a, indicating that the mutated module 2 almost exclusively recognized Val, rather than the initial three substrates (Fig. 5Bvii and Table S5). The possible reason for the phenomenon was that the G256A mutation made the binding pocket smaller (Supporting Information Fig. S15B), which was not conducive to the entry of Leu/Ile with longer side chains but facilitated the entry of Val. In addition, the G256A mutant had a stronger interaction with Val (two hydrogen bonds formed among Val-AMP with A256 and I285) than with Leu or Ile (Fig. 6D). Further investigations, such as solving the crystal structure, would help elucidate the molecular mechanism.

Within the Claviceps genus, Claviceps paspali can not produce ergopeptines due to the absence of lpsAs. The C. paspali No. 24 strain, a mutant isolated and preserved in our lab, exclusively produced d-lysergic acid at 24.4 mg/L (Supporting Information Fig. S16). To omit the d-lysergic acid feeding and efficiently obtain 3, a mixed-culture platform of C. paspali−A. nidulans was designed (Fig. 7A), and the C. paspali No. 24 strain was utilized to supply the substrate d-lysergic acid, while the An09 (the recombinant A. nidulans strain overexpressing the chimeric enzyme Cp-LPSA1^M1Cp-LPSA2^M2Cp-LPSA1^M3, LPSB and EasH1) was dedicated to convert d-lysergic acid to 3, a precursor for clinical drugs such as α-dihydroergocryptine and bromocriptine. Initially, C. paspali No. 24 and the An09 strains were fermented separately for 10 days and 2 days, respectively. Subsequently, they were co-cultivated for 2 days at different ratios of inoculation volumes (V_C.paspali:V_{A. nidulans}). As shown in Fig. 7B, the production of 3a and 3 varied with the inoculation ratios. Co-cultivation at an incubation ratio of 1:1 (C2 group) produced 3a and 3 at 0.47 ± 0.09 and 0.84 ± 0.11 mg/L, respectively, which was 1.9 and 1.5 times higher than compared to the initial inoculation ratio of 1:2 (C1 group). Furthermore, the fermentation supernatant of C. paspali No. 24, either at half volume (C3 group) or equal volume (C4 group), was added to the 2-day culture of the An09 strain. It was found that this method significantly enhanced the production of 3a and 3. In the C4 group, the yields of 3a and 3 reached 0.65 ± 0.11 and 1.46 ± 0.11 mg/L, respectively, representing 2.7 and 2.6 times higher than the C1 group. Similarly, adding an equal volume of C. paspali No. 24 fermentation supernatant to recombinant A. nidulans An23 (which overexpressed the chimeric enzyme Cp-LPSA2^M1P1-LPSA1^M2Cp-LPSA1^M3, LPSB, and EasH1 and was responsible for producing the clinical drug 9), yielded 0.49 ± 0.13 mg/L 9a and 1.09 ± 0.14 mg/L 9, respectively (Fig. 7C).