Functional Operons in Secondary Metabolic Gene Clusters in Glarea lozoyensis (Fungi, Ascomycota, Leotiomycetes)

The glpks3 and glnrps7 genes are transcribed as an operon.

The glpks3-glnrps7 locus, annotated as a PKS-NRPS hybrid gene, is part of a typical secondary metabolic gene cluster (Fig. 1). To investigate the transcription of glpks3-glnrps7, we used 5′- rapid amplification of cDNA ends (5′-RACE), 3′-RACE, and reverse transcription (RT)-PCR to amplify seven overlapping cDNA fragments to obtain the whole mRNA transcript. Each of the seven overlapping cDNA fragments spans at least one intron in the corresponding genomic DNA sequence to ensure that they were amplified from cDNA templates. By comparing the genomic DNA and the cDNA sequences, we confirmed that glpks3-glnrps7 is transcribed as a single 12,562-nucleotide (nt) transcript (Fig. 2A and B). We assessed the cDNA sequence for potential open reading frames (ORFs) and confirmed two distinct ORFs separated by a 26-bp intergenic sequence (Fig. 2C and D). glpks3 contains a 9,021-bp ORF with six introns, and glnrps7 contains a 2,664-bp intronless ORF. The 26-bp intergenic sequence between glpks3 and glnrps7 lacks a polyadenylation signal, which is required for 3′-end cleavage and polyadenylation of pre-messenger RNA in eukaryotic protein-coding genes (11). In addition, neither a transcriptional termination site for glpks3 nor a transcription initiation site for glnrps7 was detected. Taken together, these results confirm that glpks3 and glnrps7 are cotranscribed into one dicistronic mRNA.

Since the glpks3-glnrps7 locus contains two ORFs, two distinct proteins are expected to be translated from the single 12,562-nt transcript. We carried out two experiments to confirm the translation of two proteins from the 12,562-nt transcript. First, a genomic DNA fragment spanning the promoter sequence and the region annotated to encode the ketosynthase domain (KS) of the glpks3 gene was fused downstream with a green fluorescent protein (GFP) reporter gene (Fig. 2E). The reporter plasmid was introduced into the G. lozoyensis wild-type strain by Agrobacterium tumefaciens-mediated transformation. A GFP signal was detected in the transformants (Fig. 2G), indicating that the transcript was translated. Second, we attempted to detect the predicted GLNRPS7 protein, containing 887 amino acids and with a molecular mass of about 98 kDa, by Western blotting. A rabbit polyclonal antibody (CAS2) was generated against a peptide containing residues 361 to 369 of GLNRPS7 (Fig. 2F). A protein band with a molecular mass close to that of the predicted GLNRPS7 protein of 98 kDa was detected in the mycelial homogenate of the G. lozoyensis wild-type strain by Western blotting using the CAS2 antibody (Fig. 2H). These results demonstrated that the single glpks3-glnrps7 transcript is translated into two distinct proteins.

To confirm that transcription of the glpks3-glnrps7 operon is controlled by a single promoter, the glpks3-glnrps7 locus without its endogenous promoter was cloned and introduced by protoplast transformation into the genetic model fungus Aspergillus nidulans under the control of the A. nidulans trpC promoter (PtrpC) (Fig. 3A). We mapped the PtrpC-glpks3-glnrps7 transcript in the A. nidulans transformant by amplifying five overlapping cDNA fragments, and the results showed that the PtrpC-glpks3-glnrps7 gene construct was transcribed into a single transcript which was identical to the transcript of glpks3-glnrps7 in G. lozoyensis (Fig. 3B). The transcription of glpks3 and glnrps7 in the heterologous host A. nidulans was further confirmed by RT-PCR using two primer sets (Fig. 3A). One pair (5′-Y/3′-Z) spanned the first intron of glpks3, and the other pair (5′-I/3′-999′) spanned the last intron of glpks3 and the intergenic region plus part of glnrps7. The results showed that the glpks3 and glnrps7 genes were transcribed in the A. nidulans strain transformed with PtrpC-glpks3-glnrps7 and the G. lozoyensis wild-type strain but not in the A. nidulans parent strain or the A. nidulans strain transformed with the promoterless glpks3-glnrps7 construct (Fig. 3C). These results confirmed that transcription of the glpks3 and glnrps7 operon is controlled by a single promoter. We also observed a protein of approximately 120 kDa only in the PtrpC-glpks3-glnrps7 A. nidulans transformant by Western blotting using antibody CAS2 (Fig. 3D). The larger molecular mass of the GLNRPS7 protein in A. nidulans may be the result of heavier glycosylation in the heterologous host. Based on these results, we conclude that the glpks3 and glnrps7 genes are cotranscribed into one dicistronic mRNA under the control of a single promoter and the resultant transcript is then translated into two individual proteins, GLPKS3 and GLNRPS7, proving that the glpks3-glnrps7 locus belongs to the second type of eukaryotic operon in which the dicistronic mRNAs are translated as in prokaryotic operons.

Functional analysis of the glpks3-glnrps7 gene cluster.

The glpks3-glnrps7 locus is clustered with two putative transporter genes, one putative oxidoreductase gene, and one putative transcriptional regulatory gene (Fig. 1), which constitutes a typical organization for a gene cluster for the biosynthesis of a secondary metabolite. Based on the catalytic domain structures of GLPKS3-GLNRPS7, it is predicted that the gene cluster is responsible for the biosynthesis of a polyketide chain linked to 1 amino acid residue. In order to detect this compound, we generated G. lozoyensis mutant strains in which the DNA fragment from 8,512 to 9,946 bp downstream from the translation start codon of glpks3 (including the 26-bp gap and partial region annotated as an adenylation domain [A]) was deleted by Agrobacterium tumefaciens-mediated transformation to inactivate GLPKS3-GLNRPS7 (Δglpks3-glnrps7) (see Fig. S1 in the supplemental material). By comparing the metabolite profiles of the wild-type and Δglpks3-glnrps7 strains, we hoped to identify the compound encoded by the glpks3-glnrps7 gene cluster. Repeated attempts under different fermentation conditions did not reveal any differences between the high-performance liquid chromatography-mass spectrometry (HPLC-MS) profiles of the wild-type strain and the Δglpks3-glnrps7 deletion strain. As an alternative strategy, we compared the HPLC-MS profiles of the A. nidulans parent strain and the A. nidulans strain transformed with the PtrpC-glpks3-glnrps7 gene construct. One new and unique peak was observed in the fermentation broth of the PtrpC-glpks3-glnrps7 transformant of A. nidulans but not in the fermentation broth of the A. nidulans parent strain or the A. nidulans strain transformed with the glpks3-glnrps7 construct without a promoter (Fig. 4A). We purified the target compound, referred to as compound 1, and elucidated its structure with the high-resolution electrospray ionization-mass spectrometry (HRESIMS) and one-dimensional and two-dimensional nuclear magnetic resonance (NMR) spectra (Fig. 4B; see also Fig. S2). It is a novel tetramic acid-containing metabolite and is given the trivial name xenolozoyenone. The prefix “xeno” refers to the biosynthetic gene’s expression in a foreign host. Xenolozoyenone is a polyketide chain linked to an amino acid residue, which is consistent with the expected biosynthetic product of the glpks3-glnrps7 gene cluster. Collectively, these results suggest that the glpks3-glnrps7 operon is functional at the transcriptional, translational, and chemical levels.

Evolutionary origin of GLPKS3-GLNRPS7 operon.

Operons are common in prokaryotes. In order to clarify the evolutionary origin of the glpks3-glnrps7 operon, we constructed a phylogenetic tree based on the KS domain of GLPKS3, 30 fungal PKSs, and 9 bacterial PKSs (Fig. 5; see also Fig. S3 and S4 and Table S1 in the supplemental material). The 30 fungal PKSs and 9 bacterial PKSs have been linked to products with known chemical structures (Table S1) and, thus, were proven to be functional PKSs. GLPKS3 and the 30 fungal PKSs were grouped together (bootstrap value of 92%), and the clade was distantly related to and distinct from the bacterial PKSs. Thus, the glpks3-glnrps7 locus and its fungal orthologs appear to have a monophyletic origin among fungal PKSs, and the glpks3-glnrps7 locus did not result from horizontal gene transfer from prokaryotes. Among the fungal PKSs, GLPKS3 and its fungal orthologs were grouped in a monophyletic clade with a high bootstrap value (100%). In this clade, 26 proteins were from 13 genera belonging to four classes of Pezizomycotina, which suggests that these kinds of enzymes are widespread and possibly serve a basic function in these fungi.

Two other G. lozoyensis gene loci transcribed as operons.

In addition to glpks3-glnrps7, the G. lozoyensis genome revealed four more genes (glpks26-nrps, glpks27-nrps, glpks28-nrps, and glpks29-nrps) that are predicted to be responsible for the biosynthesis of polyketide linked to 1 amino acid residue (10). To find out if the occurrence of operons is common among these secondary metabolic genes in G. lozoyensis, we determined the cDNAs of the other four PKS-NRPS hybrid genes. Whole transcripts assembled with overlapping cDNA fragments obtained by RACE-PCR and RT-PCR revealed that glpks28-nrps and glpks29-nrps possess gene organization patterns similar to that of the glpks3-glnrps7 locus (Fig. 6). Because not every cDNA fragment spans an intron, one primer set, PF5-1 and Ourer5 (located in the third intron of glpks29-nrps), was used as a genomic DNA contamination control. No PCR product was amplified by this primer set with the template cDNA, indicating there was no genomic DNA contamination in the RNA samples used for cDNA amplification. The glpks28-nrps locus was transcribed as a dicistronic transcript with an intergenic sequence of 443 nt, while the glpks29-nrps locus also was transcribed as a dicistronic transcript with an intergenic sequence of 179 nt. No polyadenylation signal sequences or promoter elements were detected in the intergenic region. Moreover, no transcriptional termination site for the upstream gene or transcription initiation site for the downstream gene were detected in the intergenic regions. These results demonstrated that the glpks28-nrps and glpks29-nrps loci are organized as operons. Based on these results, we renamed these two loci as glpks28-glnrps8 and glpks29-glnrps9. glpks28 contained a 966-bp ORF, and glnrps8 contained a 10,822-bp ORF. glpks29 had a 9,212-bp ORF, and glnrps9 had a 3,439-bp ORF. Among the four genes, glpks28 contains no introns, while glnrps8, glpks29, and glnrps9 contain multiple introns. These two loci are organized as typical secondary metabolic gene clusters which are predicted to be responsible for the biosynthesis of compounds consisting of a polyketide chain linked to an amino acid residue.

DNA and RNA procedures.

Fungal genomic DNA was extracted as described previously (10). Plasmids and PCR products were purified using the E.Z.N.A. gel extraction kit (Omega Bio-Tek, Norcross, GA). Phusion high-fidelity DNA polymerase (New England Biolabs, Ipswich, Massachusetts) and EasyTaq DNA polymerase (TransGen Biotech, Beijing, China) were used for PCRs. Sequences were determined by SinoGenoMax Co., Ltd. (Beijing, China). Restriction endonucleases and DNA-modifying enzymes were from New England Biolabs (Ipswich, Massachusetts).

RNA was extracted with the Qiagen RNeasy minikit, and carryover DNA was removed by DNase I digestion (Qiagen, Valencia, CA). cDNA fragments were synthesized using the TransScript II first-strand cDNA synthesis supermix (TransGen Biotech, Beijing, China). Introns were identified by comparing genomic and cDNA sequences. The 5′ ends of the mRNA and the poly(A) attachment sites were mapped by 5′- and 3′-RACE–PCR (FirstChoice RLM-RACE kit; Ambion, Austin, Texas).

All primers used in the study are listed in Table S3 in the supplemental material.

Genetic manipulation of glpks3-glnrps7 in G. lozoyensis.

To detect the expression of GLPKS3, a genomic fragment extending from the promoter to the region annotated as the KS domain of glpks3, the GFP reporter gene, and the A. nidulans trpC terminator were amplified and inserted into pAg1-H3 to give the expression vector pAg1-H3L-KS2-TtrpC (see Text S1 and Fig. S6 in the supplemental material).

To disrupt glpks3, the 3′ regions of glpks3 and glnrps7 were amplified and inserted into pAg1-H3 to give pAg1-H3-glpks3-glnrps7, containing the gene replacement cassette (see Fig. S1 in the supplemental material).

Conidia of G. lozoyensis were prepared and transformed as previously described (10). After isolation of single spores from the transformants, the positive transformants of Δglpks3-glnrps7 were determined by PCR using two primer pairs (GLPKS3-F2/999′ and GLPKS3-F1/T). Transformants positive for KS::GFP were determined by PCR using two primer pairs (GFP-F/GFP-R and GFP-F/N).

The wild-type strain of G. lozoyensis, the KS::GFP transformant, and the Δglpks3-glnrps7 mutant were grown in LYCP-5 medium to prepare seed cultures (10). The seed cultures were then inoculated into H medium and grown for 14 days (21). For the wild-type strain of G. lozoyensis, the culture was divided into two parts. One part was examined by RT-PCR using two primer pairs (Y/Z and I/999′) and Western blotting using the CAS2 antibody. The other part and the culture of the Δglpks3-glnrps7 strain were analyzed by HPLC-MS following the same procedures used for A. nidulans. For the KS::GFP transformant, mycelia were collected and observed by fluorescence microscopy.

Heterologous expression of glpks3-glnrps7 in A. nidulans.

The glpks3-glnrps7 gene was amplified from genomic DNA of G. lozoyensis, and the constitutive A. nidulans trpC promoter was amplified from pAg1-H3 (22). After restriction enzyme digestion and ligation, glpks3-glnrps7 under the control of the constitutive A. nidulans trpC promoter was constructed and inserted into pAg1-H3 to give the expression vector pAg1-PtrpC-glpks3-glnrps7 and control vector pAg1-glpks3-glnrps7 (see Text S1 and Fig. S6 in the supplemental material). Protoplast preparation and transformation were performed as previously described (23). One microgram each of pRG3-AMA1 and SbfI-digested pAg1-PtrpC-glpks3-glnrps7 or pAg1-glpks3-glnrps7 were cotransformed into A. nidulans A773 (pyrG89 wA3 pyroA4 veA1). Positive glpks3-glnrps7-containing transformants were identified by PCR using five primer sets (Pc800F/P22′, A/B, U/F, and M/D for PtrpC-glpks3-glnrps7 transformants and K/D for glpks3-glnrps7 transformants) and DNA sequencing.

The positive transformants and parent strain of A. nidulans were incubated on MAG with appropriate supplements (23). Conidia were harvested and adjusted to 10⁸ conidia ml⁻¹. Amounts of about 2 ml of spore suspension were inoculated into MMV (1% glucose, 0.001% thiamine, nitrate salts, trace elements, and vitamins, pH 6.5) with appropriate supplements (23). The cultures were shaken at 200 rpm in the dark for 7 days. RNA preparations from the cultures were examined for GLNRPS7 transcription and translation by RT-PCR using two primer pairs (Y/Z, and I/999′) for all strains of A. nidulans, five primer pairs (Y/P30R-2, GLPKS3-F7/U-R, U/GLPKS3-F2-R, GLPKS3-F2/GLNRPS7-F-R, and I/ GLNRPS7-R1) for PtrpC-glpks3-glnrps7 transformants, and protein preparations were examined by Western blotting using the polyclonal antibody CAS2.

Protein analysis.

Protein extraction, quantification, and Western blotting were performed as previously described (24). In order to detect GLNRPS7, we prepared a polyclonal antibody, CAS2, by immunizing a rabbit (Epitomics, Hangzhou, Zhejiang, China) with the peptide antigen PSSKSNEPA, which was predicted as an immune epitope against the predicted GLNRPS7 protein by the PeptideStructure program of the University of Wisconsin Genetics Computer Group Sequence Analysis Software Package. The IgG titer against the PSSKSNEPA peptide antigen reached 64,000. The polyclonal antibody CAS2 was purified by protein A column as described previously (25).

Fermentation and extraction of A. nidulans and transformants for HPLC-MS analysis.

The transformants and parent strain of A. nidulans were cultured on MAG with appropriate supplements (23). Fifteen pieces (0.5 by 0.5 by 0.5 cm³) of agar cultures were inoculated and incubated in 100 ml H medium with appropriate supplements at 30°C and 220 rpm for 7 days. The cultures were extracted with 100 ml of methyl ethyl ketone (MEK), and the organic phase was evaporated to dryness and redissolved in methanol (MeOH) to 30 mg·ml⁻¹.

NMR and MS methods.

All ¹H, ¹³C, and two-dimensional (¹H–¹H correlation spectroscopy [COSY], heteronuclear single quantum coherence [HSQC] spectroscopy, and heteronuclear multiple-bond correlation [HMBC] spectroscopy) NMR spectra were acquired on a Bruker 600- or 500-MHz spectrometer equipped with a 5-mm triple-resonance cryoprobe at 298 K. Residual solvent signals were used as references (acetone-d₆, δ_H 2.05/δ_C 29.8, 206.0, and CDCl₃, δ_H 7.26/δ_C 77.0). HPLC-MS spectra were obtained on an Agilent 6120 quadrupole mass spectrometer using a positive ESI source. Amounts of 10 µl of the test samples were injected for HPLC-MS analysis (Agilent Zorbax Eclipse plus C₁₈ reverse-phase column, 5 µm, 4.6 by 150 mm, 10% to 90% CH₃CN in H₂O with 01% formic acid for 30 min, 1 ml·min⁻¹). HRESIMS data were obtained using the Agilent 6520 quadrupole time of flight (Q-TOF) LC-MS instrument equipped with an electrospray ionization (ESI) source.

Isolation of xenolozoyenone.

The scaled-up fermentation culture (PtrpC-glpks3-glnrps7 transformant of A. nidulans in 2.0 liters of H medium) was extracted repeatedly with MEK (3 times per 2.0 liters), and the organic solvent was vacuum evaporated to dryness to obtain the crude extract (1.2 g), which was fractionated by using a Sephadex LH-20 column with 1:1 CH₂Cl₂-MeOH as eluents. The resulting subfractions were combined and further purified by semipreparative reverse-phase HPLC (Agilent Zorbax SB-C₁₈ column, 5 µm, 9.4 by 250 mm, 65% CH₃CN in H₂O with 01% formic acid for 25 min, 2 ml·min⁻¹) to obtain xenolozoyenone (compound 1) as follows: 4.5 mg, t_R 17.25 min (MeOH); UV (MeOH) λ_max (log ε) 217 (4.46), 256 (4.10), 390 (4.86) nm; ¹H NMR (acetone-d₆, 600 MHz) δ 7.73 (1H, brs, OH-1 or OH-6), 7.50 (1H, dd, J = 15.0, 11.4 Hz, H-8), 7.12 (1H, d, J = 15.0 Hz, H-7), 6.82 (1H, dd, J = 15.0, 10.8 Hz, H-10), 6.50 (1H, dd, J = 15.0, 11.4 Hz, H-9), 6.32 (1H, dd, J = 15.0, 10.8 Hz, H-11), 6.10 (1H, dt, J = 15.0, 7.2 Hz, H-12), 3.81 (2H, m, H₂-5), 2.19 (2H, m, H₂-13), 1.44 (2H, m, H₂-14), 1.31 (6H, m, H₂-15, H₂-16, and H₂-17), 0.89 (3H, t, J = 6.6, H₃-18); ¹³C NMR (CDCl₃, 125 MHz) δ 193.4 (C-4), 176.1 (C-2), 175.0 (C-6), 146.3 (C-8), 144.6 (C-10), 143.4 (C-12), 130.2 (C-11), 128.8 (C-9), 120.1 (C-7), 99.7 (C-3), 62.9 (C-5), 33.2 (C-13), 31.4 (C-15), 29.7 (C-17), 28.6 (C-14), 22.5 (C-16), 14.0 (C-18); HRESIMS m/z [M+H]⁺ 306.1708 (calculated for C₁₇H₂₄NO₄, 306.1700).

Phylogenetic analysis of PKSs and PKS-NRPSs.

To test whether the glpks3-glnrps7 locus is a product of horizontal gene transfer from bacteria, we built a phylogenetic tree based on the KS domain of GLPKS3-GLNRPS7, its orthologs in other fungi (coverage, ≥95%, and identity, ≥64%), GLPKS28-GLNRPS8, GLPKS29-GLNRPS9, 30 characterized fungal PKSs, and 9 characterized bacterial PKSs (see Table S1 in the supplemental material). All domains were identified by using the program anti-SMASH (26) or manually in multiple alignments. The amino acid sequences of the KS domains were aligned with MUSCLE and analyzed phylogenetically with MEGA 6.0 by the maximum-likelihood (ML) algorithm using an LG+F+G+I model selected by Prot-test and a 1,000-replication bootstrap test (27, 28). The R/DKC domains were also aligned with MUSCLE. To test whether the glpks3-glnrps7 locus might be of single or multiple evolutionary origins, we constructed two phylogenetic trees, one based on the KS domain of GLPKS3, all orthologs in other fungi (coverage, ≥95%, and identity, ≥55%), and 3 characterized fungal PKS-NRPSs and the other based on the A domain of GLPKS3, all orthologs in other fungi (coverage, ≥95%, and identity, ≥40%), and 3 characterized fungal PKS-NRPSs. The KS and A domains were identified by the program anti-SMASH (26) or manually in multiple alignments. The amino acid sequences of these domains were aligned with MUSCLE and analyzed with RAxML BlackBox by the ML method under a WAG (Whelan And Goldman) model with or without a constraint from the reference fungal taxonomy (27, 29). The A domain tree was also analyzed with RAxML BlackBox by the ML method under the WAG model with a constraint from the KS domain tree.

Alternative hypotheses based on the tree topologies under the null hypothesis that all topologies are equally good explanations of the data were tested with the Shimodaira-Hasegawa test (30), weighted Shimodaira-Hasegawa test, and approximately unbiased test (31), as implemented in TREEFINDER (32).

Bioinformatics.

Putative promoter sequences in the intergenic region of glpks3-glnrps7 and 1 kb upstream from the genes in other gene pairs were investigated with Promoter 2.0 Prediction Server (33). The polyadenylation signals (AATAAA and ATTAAA) in the intergenic region of glpks3-glnrps7 and other gene pairs were identified by using the nucleotide sequence pattern search tool Fuzznuc in EMBOSS (http://emboss.bioinformatics.nl/cgi-bin/emboss/fuzznuc). To identify potential polycistrons, genome and transcriptome data of A. nidulans (GenBank accession numbers PRJNA13961 and PRJNA182228), G. graminicola (GenBank accession numbers PRJNA37879 and PRJNA151285), and D. stenobrocha (GenBank accession numbers PRJNA67941 and PRJNA236481) were downloaded from NCBI (http://www.ncbi.nlm.nih.gov/). The genomic loci were annotated by InterProScan, KEGG, FunCat, and UniProt-GO (34–37) to rule out the possibility of pseudogenes. The RNA-Seq data were assembled as described previously and then mapped and aligned to the genome (38, 39).

The mRNA sequences of glpks3-glnrps7, glpks28-glnrps8, and glpks29-glnrps9 have been deposited at GenBank under the accession numbers KM603664, KM603665, and KM603666, respectively.