Introduction

Ascomycetes such as Aspergilli have become a powerful platform to heterologously biosynthesize secondary metabolites (SM) from various fungi [1]. Several classes of SM enzymes including polyketide synthases (PKSs), non-ribosomal peptide synthetases (NRPSs), PKS-NRPS hybrids, and terpene cyclases (TC) were successfully produced [2, 3]. However, fungal natural product research mainly focused on SM genes of manageable and cloneable size [4,5,6,7]. In contrast, longer genes (> 12 kb) have been preferably studied by either knock-out or promoter replacement strategies (or combination of both) [8], but have hardly been investigated by production in heterologous hosts. Both strategies require cultivability and transformability of the investigated fungi. These requirement are frequently met by Ascomycota, but are scarcely applicable to EDF such as Mucoromycota or higher fungi, i.e. Basidiomycota [9]. Hence, current knowledge on fungal secondary metabolite genes and their natural products is mostly based on work on ascomycetes and is thus biased. This is also founded on the elevated number of SM biosynthetic genes that have been identified in ascomycetes (10–30 NRPS and NRPS-like, 10–30 PKS, and > 4 TC genes per genome) versus basidiomycetes (1–6 NRPS and NRPS-like, 1–6 PKS, and 6–34 TC genes per genome) or EDF (0–15 NRPS and NRPS-likes, 0–15 PKS, 0–10 TC genes per genome) [9,10,11,12].

Several tools for the production of SM genes have been developed in ascomycetes. The heterologous expression of genes of interest (GOI) in Aspergilli require either a strong and constitutive promoter, such as the glycerinaldehyde-3-phosphate dehydrogenase promoter (PgpdA) [13] and the α-amylase promoter (PamyB) [14], or an inducible promoter system, such as the alcohol-inducible PalcA/alcR system [15], the tetracycline-dependent TetON system [16], or the ATNT system, i.e. a combination of the TetON system with the regulatory terrein biosynthetic promoter PterA [17, 18]. To date, both the PalcA/alcR system in Aspergillus nidulans and the ATNT system in Aspergillus niger are frequently used fungal expression platforms for genes encoding SM biosynthetic enzymes including PKSs [19, 20], NRPSs [21], and NRPS-like enzymes [22, 23]. However, alternative fungal systems have been developed, among them (i) the HEx platform using an engineered Saccharomyces cerevisiae strain producing additional SM auxiliary enzymes [24], (ii) the multiauxotrophic Aspergillus oryzae strains M2-3 and NSAR1 [25, 26], and various systems developed for A. nidulans including (iii) polyauxotrophic strains sensitive to various antibiotics [27], (iv) strains with strongly reduced intrinsic metabolic background [28, 29], (v) the nitrate-inducible aflR/S CoIN system [30], and (vi) an AfoA-inducible platform [31]. Traditionally, single SM biosynthetic genes were expressed in phylogenetically distantly related hosts such as Escherichia coli and S. cerevisiae [4, 32]. However, the latter two platforms require adaptations and adjustments such as (i) the co-production of SM specific phosphopantetheinyl transferases to produce holoenzymes [33], or (ii) the expression of additional tRNAs to ensure active enzymes of sufficient yield [34] and, moreover, may fail due to limitations of intron splicing [24]. In any case and prior to transformation of the heterologous host, the GOI must be amplified from DNA and cloned in plasmids that are usually assembled and propagated in E. coli or S. cerevisiae [35, 36].

Fungal genes encoding highly reducing PKSs (8–9 kb), PKS-NRPS hybrids (12 kb) and multimodule NRPSs (10–26 kb) are of extraordinary size and can be even longer if interspaced by introns. Hence, a full-length amplification by PCR using genomic DNA as template is not recommendable. An amplification from cDNA is the gold standard, but is impaired as: (i) reverse transcriptases possess low processivities at long templates [37], and (ii) most fungal SM genes remain silent under standard laboratory conditions—a fact, which reduces the availability of a full-length RNA template [38]. Hence, an amplification and cloning in small fragments is a conceivable way to manage long GOI. Since 1990 artificial chromosomes (FACs, YACs and BACs) has been frequently used [39, 40], and are suitable to express entire gene clusters of 50–150 kb size. [40] However, identification and metabolic screening of correct clones is laborious and time-consuming. Moreover, FACs and BACs rely on an intrinsic activation of the transgenic promoters in the expression host which is hardly achieved for genes from distantly related species.

Here, an alternative strategy to clone and express long GOI is presented, which is based on the well-established ATNT expression system in A. niger and does not require cloning of the full-length gene. Instead, the GOI is amplified in up to five fragments and is assembled with high fidelity by homologous recombination within the host in one single step. On-site integration into the fwnA locus responsible for conidial pigmentation allowed a simple visual screening for the correct transformants. The system’s versatility was proven by the integration of two GOI from phylogenetically divergent fungi: First, the lpaA gene (8.2 kb) from the basidiomycete Laetiporus sulphureus, known to encode a PKS responsible for the chromophoric laetiporic acids [41], was precisely integrated in five DNA fragments. Furthermore, an unknown NRPS gene (calA, 20.0 kb) from the EDF Mortierella alpina was successfully expressed and assigned as a synthetase for calpinactam, an anti-mycobacterial peptide. Hence, the system sets the basis to study long biosynthetic genes from various biological sources including EDF.

Results

Deletion of the akuB gene in ATNT favors homologous recombination

Aspergilli randomly integrate transgenes into their genomes and multiple integration events are frequently observed [42]. It has been demonstrated, that Aspergillus or Neurospora deletion mutants defective in the non-homologous end-joining (NHEJ) repair mechanism such as ∆akuAKU70,∆akuBKU80, or ∆ligA, facilitate the targeted gene deletion by homologous recombination [43,44,45]. Moreover, these mutants can integrate transgenes in two overlapping fragments [29]. This “split-marker” recombination is broadly used for efficient targeted gene deletion in fungi [46]. To investigate the applicability of a deletion of the NHEJ system for a targeted multi-fragment gene integration, the akuB gene from the A. niger expression strain ATNT∆pyrG was replaced by the hygromycin B resistance cassette and the akuB deletion in the obtained mutant A. niger ATNT∆akuB (tLK01) was confirmed by Southern Blot (Additional file 5: Fig. S1). To test the frequency of transgene integration, the fwnAalbA/pksP gene encoding the polyketide synthase responsible for conidial pigmentation [47] was deleted in both A. niger ATNT and mutant tLK01 (Fig. 1A). To this end, the overexpression plasmid pPS01 [41] containing the inducible promoter PterA, the terminator TtrpC, and an URA blaster was expanded by the upstream and downstream flanks of fwnA (5′ fwnAup and 3′ fwnAdn) resulting in expression plasmid pLK04 (Additional file 6: Fig. S2). When both ATNT and tLK01 were transformed with the overlapping fragments of the construct, the frequency of targeted recombination was significantly higher for tLK01, i.e. the relative number of non-pigmented transformants was increased from 15 to 80% when compared to the parental strain (Fig. 1B, Additional file 2: Table S1, Additional file 7: Fig. S3). A successful genomic integration was additionally confirmed by PCR (Additional file 8: Fig. S4). One of these A. niger tLK01∆fwnA transformants (tLK07) served as null mutant (empty vector control) in subsequent metabolite analyses.

Fig. 1
figure 1

Principle of the recombination and ∆fwnA selection system. A Construction of the fwnA deletion mutants in the parental strains A. niger ATNT and tLK01 (ATNT∆akuB). Homologous recombination of the two PCR amplicons into the fwnA locus in the parental strains led to the deletion of the fwnA gene resulting in A. niger strains tLK06 (ATNT∆fwnA) and tLK07 (tLK01∆fwnA). B Representative agar plate of a transformation of the A. niger tLK01 parental strain with the split-marker fwnA deletion constructs. Non-pigmented, fawn colonies are resulting from the fwnA deletion suggesting a successful recombination event whilst pigmented, black colonies indicate false-positive transformants. The frequency of homologous recombination is significantly increased in the A. niger ∆akuB deletion strain tLK01

Full size image

Aspergillus niger tLK01 is competent to integrate five-fragment expression constructs

As a proof of concept to integrate long SM genes in the fwnA locus of A. niger tLK01 by application of multiple DNA fragments, we used the 8196 bp PKS gene lpaA from the “chicken of the woods” mushroom L. sulphureus [41]. Recently, this gene has been successfully expressed in Aspergilli and was assigned to the biosynthesis of polyenes of various chain lengths (C26–C32), i.e. laetiporic acids A1-D1 (LA A1–D1) and their 7-trans-isomers A2 -D2 (LA A2–D2) [41], which both are currently discussed to be natural alternative colorants in cosmetics and for textiles [48]. Due to their absorption maxima between 450 and 470 nm, the mycelium of producing cultures exhibit an orange hue allowing an easy read-out of positive transformants. We fused a 1 kb fragment of either end of lpaA and ligated the fusion fragment into pLK04. Five fragments with a 1 kb overlap were amplified by PCR (Fig. 2A), mixed equimolarly and used to transform A. niger tLK01 using uridine prototrophy as selection marker. Five non-pigmented transformants were randomly picked and checked by PCR for the full-length integration of the expression construct (Fig. 2B) and by Southern Blot for homologous integration into the fwnA locus (Additional file 9: Fig. S5). Indeed, four out of five fawn tLK04 transformants (4/5, 80%) successfully assembled and integrated the lpaA expression construct into the fwnA locus. Finally, the transformant tLK04 and the respective empty vector control strain tLK07 were fermented under non-inducing and inducing conditions (by addition of doxycycline). Metabolites were extracted and analyzed by UHPLC-MS (Fig. 3A) and by high resolution MS fragmentation (Additional file 10: Fig. S6). During cultivation the color of the fungal mycelium turned orange and the expected polyene products LA A1–D2 were primarily detectable from mycelium of the doxycycline-induced cultures of tLK04 suggesting that the five DNA fragments have been accurately assembled. Overall, seven out of seven PCR-positive transformants (Fig. 2B) produced the expected compounds (7/7, 100%). Minor signals of LAs with 100-fold less abundances were detectable from non-induced cultures, suggesting that the promoter is not entirely silent (not shown). In addition, no laetiporic acids were obtained from the empty vector control tLK07 under inducing conditions (Fig. 3A).

Fig. 2
figure 2

Determination of the full-length integration of the PterA:lpaA:TtrpC construct into the fwnA genomic locus of the A. niger recipient strain tLK01. A Schematic representation of the integration event of the five lpaA fragments into the genomic fwnA locus of A. niger tLK01. To mutate the SAM binding site in the C-methyltransferase domain of LpaA, the triplet GAC (pos. 4245–4247) encoding Asp1415 (probably binding the ribose moiety of SAM) was site-mutated into the triplet GCC encoding Ala1415. B An agarose gel of a PCR targeting the PterA promoter and the TtrpC terminator has been performed (oligonucleotides oMG370/oMG116). The expected amplicon size is indicated. Full length integration was evident for seven tLK04 (producing native LpaA) and three tLK05 transformants (producing LpaAD1415A). The genomic DNA of tLK07 (transformed with an empty vector) and the lpaA-encoding plasmid pPS03 [41] served as negative and positive controls, respectively

Full size image
Fig. 3
figure 3

Laetiporic acid production in the A. niger ∆fwnA mutants tLK04, tLK05 and tLK07. A Cultures photographs, representative domain structure of LpaA mutant proteins and HPLC profiles of metabolite extracts. Photographs depict 3-days cultures at 30 °C, 150 rpm. HPLC chromatograms of extracts of mycelia from tLK04, tLK05 and tLK07 were monitored at a wavelength of λ = 450 nm. Note, that tLK04 expressing the native lpaA gene produces laetiporic acids A1–D2, whereas no metabolites are detectable in tLK05 (expressing the mutant lpaAD1415A gene) or in tLK07 (empty vector control). Authentic standards of laetiporic acids (LA-A1/2, LA-B1/2 and LA-C1/2) served as controls. An authentic standard for LA-D1/2 was not available. For MS and MS/MS spectra please refer to Additional file 10: Fig. S6. The pearls on a string represent the domain structure of the PKS LpaA and include: KS, β-ketoacyl synthase; AT, acyl transferase; DH, dehydratase; MT, C-methyltransferase; ER0, (inactive) enoyl reductase; KR, keto reductase; ACP, acyl carrier protein; TE, thioesterase. B Chemical structure of laetiporic acids A1–D2. Note the common 1-methyl-2-oxo-propylidene group, which requires the action of the intrinsic C-MT domain

Full size image

The expression system is suitable for point mutation analysis

Next, the suitability of system to study point mutations of long GOI was investigated. Structurally, linear polyenes including the antifungal laetiporic acids from L. sulphureus [41], piptoporic acid from curry punk fungus Piptoporus australiensis [49], and the two antilarval polyenes from the undescribed stereaceous basidiomycete BY1 [50, 51], feature a 1-methyl-2-oxo-propylidene head as plausible pharmacophore (Fig. 3B). We speculated on the intrinsic activity of the C-methyl transferase (C-MT) domain in LpaA to catalyze this S-adenosyl-methionine (SAM) dependent C-transfer after two cycles of acyl condensation. Hence, we aimed at the inactivation of the C-MT domain. In a previous study, an aspartate (D2019) in the active site of citrinin polyketide synthase PksCT was shown to be essential for the methylation of the polyketide citrinin by coordination of 2′-OH and 3′-OH of the ribose moiety of the cofactor SAM [52]. Hence, to inactive the C-MT activity of LpaA, the GAC triplett encoding the homolog aspartate (D1415) was mutated into GCC (A1415). To this end, five DNA fragments were used to transform A. niger tLK01, of which fragment 3 contained the mutated sequence encoding the potential dysfunctional MTD1415A (Fig. 2A). Full-length integration of lpaAD1415A was confirmed by PCR (Fig. 2B) and the point mutation was additionally confirmed by Sanger sequencing. Surprisingly, whilst cultures of the lpaA-expressing strain tLK04 turned orange after induction, the induced transformants expressing the lpaAD1415A gene (tLK05) neither changed mycelial color nor produced any additional compound when compared to the empty vector control (tLK07) (Fig. 3A). This observation is not based on an altered transcription because lpaA in tLK04 and lpaAD1415A in tLK05 showed identical expression levels (Additional file 11: Fig. S7). Hence, the C-MT seemed to be essential for polyketide production. This finding in turn, implies that the 1-methyl-2-oxo-propylidene head in laetiporic acids is required for a successful chain elongation. In sum, the strain tLK01 is convenient to assemble DNA constructs from up to five fragments and directly express the GOI in one step. Moreover, the system is suitable to induce point mutations in long GOIs to study modified enzymes.

The NRPS gene calA encodes the calpinactam synthetase in Mortierella alpina

In contrast to Dikarya, early diverging fungi (EDF) are a comparably new resource of natural products. Although the genomes of Mortierella sp. (Phylum Mucoromycota) and Basidiobolus sp. (Phylum Zoopagomycota) encode numerous of long NRPS genes (up to 26 kb), only a few have been assigned to specific products [10]. Calpinactam, produced by M. alpina FKI-4905, is a hexapeptide with an unusual C-terminal ε-caprolactam moiety (Fig. 4A) [53]. Calpinactam—but not its chemically closely related derivatives [54, 55]—is a proven anti-mycobacterial agent with MICs of 0.78 µg mL−1 and 12.5 µg mL−1 against Mycobacterium smegmatis or Mycobacterium tuberculosis, respectively [53]. Although of high pharmaceutical interest, the corresponding gene for calpinactam biosynthesis has never been identified. We examined the metabolome and genome of the M. alpina sequencing reference strain ATCC32222 [56]. The strain produces and secretes calpinactam into the supernatant as confirmed by UHPLC-MS and comparative MS–MS fragmentation patterns versus a commercially available authentic standard (Fig. 4C and D). Three hexamodular candidate NRPS genes (nps5, nps6 and nps7) were identified in the M. alpina genome [57]. Since Mortierella NRPS genes are likely of bacterial origin [58], we predicted the substrate specificity codes of the candidate enzymes versus the bacterial gramicidin S synthetase GrsA from Aneurinibacillus migulanus as reference (Table 1). While at least three of the adenylation (A) domains encoded by nps6 or nps7 are expected to use identical amino acids, the six A domains encoded by nps5 (calA) accept six dissimilar substrates as required for calpinactam (Fig. 4A, Table 1). Moreover, the arrangement of condensation (C) and dual epimerization/condensation (E/C) domains in CalA directly reflects the sequence of l- and d-amino acids in the final peptide chain of calpinactam according to the NRPS collinearity principle [59] (Table 2). Hence, the postulated domain structure and the predicted A domain substrate specificities of CalA perfectly matched an NRPS assembly line for calpinactam.

Fig. 4
figure 4

Calpinactam production in M. alpina and transgenic A. niger ∆fwnA mutants tJMW06 and tLK07. A Chemical structure and calculated MS/MS fragments of calpinactam. B Schematic representation of the integration event of the five calA fragments into the genomic fwnA locus of A. niger tLK01. The pearls on a string represent the domain structure of the 6-module NRPS CalA and include: C, condensation domain; A, adenylation domain; T, thiolation domain; E/C dual epimerization/condensation domain; TE, thioesterase. Note that the 1st (starter) C domain is truncated and inactive. C Extracted ion chromatograms of the synthesized calpinactam standard and metabolic extracts from various fungal mycelia. M. alpina ATCC32222 was cultivated on MEP, whilst transgenic A. niger tJMW06 (calA expressing) and tLK07 (empty vector control) was cultivated in Aspergillus Minimal Medium (AMM) under inducing (+ doxycycline) or non-inducing (− doxycycline) conditions. MS data was monitored in positive ionization mode at m/z 768.6844 [M + H]+. D MS/MS spectra highlighting the specific daughter ion fragments of calpinactam of the authentic standard and the metabolic extracts from M. alpina ATCC32222 and transgenic A. niger tJMW06

Full size image
Table 1 Substrate specificity code for adenylation domains of hexamodular NRPSs of Mortierella alpina
Full size table
Table 2 Domain architecture of predicted hexamodular NRPSs in M.alpina
Full size table

Full-length calA gene expression in A. niger enables a heterologous calpinactam production

Given its length of 20,015 bp, we decided to heterologously express calA from M. alpina in A. niger tLK01 in a similar fashion as lpaA using collectively five calA DNA fragments of up to 7000 bp (Fig. 4B). 30 transformants were obtained of which three randomly picked transformants (tJMW06.3, tJMW06.13, tJMW06.26) showed a full-length integration of calA into the genome of A. niger as confirmed by multiple PCRs (Additional file 12: Fig. S8) and by Southern Blot (Additional file 13: Fig. S9). Profiling the metabolome of the tJMW06 transformants, an additional metabolite was detected under inducing conditions, whose ion mass (m/z 768.4411 [M + H]+) and retention time (2.1 min) was identical to that of the commercial calpinactam standard (MW 768.4330 Da) (Fig. 4C). The identical MS/MS-fragmentation pattern confirmed the heterologous production of calpinactam (Fig. 4D). In contrast, calpinactam was not detectable in the empty vector control strain tLK07 (Fig. 4C). Hence, we established calA as the calpinactam biosynthetic gene.

The production of calpinactam in A. niger tJMW06 was highest (20 µg g−1 fungal dry weight), when the transgenic fungus was cultivated at 25 °C (Additional file 14: Fig. S10), i.e. the temperature optimum for M. alpina enzymes [60]. However, titers of calpinactam remained low compared to the original producer strain M. alpina (1438 µg g−1). The > 20 kb calA gene includes six potential introns of 90 to 112 bp length (Fig. 5A). Hence, we checked for the correct splicing pattern of the calA transcript by comparative PCR on the M. alpina and A. niger tJMW06 cDNA (Fig. 5B). The complete set of introns were spliced in its original producer strain M. alpina as predicted, although introns 3 and 4 showed only partial splicing. Similarly, the splicing of introns 1, 2, 3, 5 and 6 appeared at the correct sites in the transgenic A. niger tJMW06. However, only marginal pre-mRNA maturation was detectable for intron 4, as genomic DNA and cDNA show the same PCR fragment sizes (Fig. 5B). The unspliced intron 4 would result in a premature termination at a UAG stop codon at position 11,148. Hence, this incomplete splicing event results in a loss-of-function version of CalA and may explain the low calpinactam titers in A. niger tJMW06 (Fig. 5C). To bypass this limitation, we removed the intron 4 by amplification and PCR-fusion of the adjacent exons of fragment 3 and transformed the recipient strain A. niger tLK01 with the altered version of calA fragments. However, this manipulation did not change the titers (data not shown), suggesting that additional adjustments such as codon-optimization and balanced supply of precursors might be required for increased calpinactam titers. In summary, in this pilot study we established for the first time a successful five-fragment assembly and heterologous expression of a > 20 kb natural product biosynthetic gene from an early diverging fungus.

Fig. 5
figure 5

Splicing events in the calA gene in M. alpina and transgenic A. niger tJMW06. A Structure of the full-length 20 kb calA gene. The gene is disrupted by six introns. B Splicing pattern of natively expressed calA in M. alpina and heterologously expressed calA in A. niger tJMW06. Both strains were cultivated under inducing conditions (M. alpina in MEP medium, A. niger tJMW06 in AMM with doxycycline). Genomic (gDNA) and cDNA served as templates for diagnostic PCRs spanning the intron in a housekeeping gene (actB or gpdA, respectively) or individually spanning the six introns of calA. Expected fragment sizes are indicated. C MS profile at m/z 768.44 [M + H]+ of the induced cultures from M. alpina (upper lane) and A. niger tJMW06 (lower lane). The indicated titer of calpinactam is normalized to the fungal dry weight (calculated from three replicates)

Full size image

Discussion

Over the past four decades, 60% of all new antibiotic lead structures were based on natural products [61]. Although new antibiotics are urgently needed especially to treat nosocomial infections, natural product research slowed down as mainly bacteria and ascomycetes were investigated concerning their metabolic potential and rediscovery of known compounds is frequently observed [62]. In contrast basidiomycetes and especially EDF are an attractive resource of bioactive agents but are rather genetically uninvestigated which is due to their challenging cultivability and the lack of genetic accessibility [9, 63]. Here, we present a molecular tool to express entire long SM genes to unravel the metabolic treasure chest of these fungi.

Multiple gene expression tools have been developed for ascomycetes (genera Aspergillus, Fusarium, Penicillium, Pichia and Saccharomyces) and partially for basidiomycetes (genera Ustilago, Coprinus). No system for EDF has been established yet. Although anectodical evidence for their transformability exists using the more complex molecular tools such as CRISPR/Cas9 or TALEN/exonuclease [64,65,66], no common strategy for a targeted gene deletion exists since the ku70/80 homologs are generally not expressed in these fungi [64]. Hence, unrevealing the function of the SM biosynthetic genes from Basidiomycota and EDF still rely on heterologous expression. However, main issue is the size of the SM genes and its interruption with introns which impedes an expression in traditional bacterial hosts [9]. Several molecular tools to clone and express entire bacterial gene clusters of > 40 kb were developed in Streptomyces species [67]. However, the current fungal expression systems mainly vary in their type of promoter activation, but most of them require are a pre-cloning of the GOI into a specific vector, which limits their applicability to long SM biosynthetic genes. Effort was done with the yeast/E.coli shuttle vectors in yeast, allowing a multi-gene assembly and co-expression of smaller biosynthetic genes in yeast and Aspergillus [68,69,70]. Alternatively, reiterative recombination systems assemble multigene constructs in S. cerevisiae, but this endonuclease-induced recombination requires multiple rounds of transformation to recycle selectable markers and is hence a robust, but more laborious technique [71]. Moreover, the genes of S. cerevisiae rarely possess introns (0.04 introns per gene) in comparison to Aspergillus species (0.96) making yeast to a less suitable host for multiple spliced transgenes [72].

The herein described system combines current knowledge of fungal biotechnology including the higher recombination frequency by deletion of the akuB gene [73], the easy transformability of Aspergilli [27], the discrimination of positive transformants by altered pigmentation [29] and the possibility to express > 20 kb genes in Aspergilli [16, 18]. The method is advantageous since (i) genomic DNA can be used as template, (ii) cloning steps are set to a minimum, (iii) assembly and expression of long genes is accomplished in a single step, and (iv) gene expression from different fungal divisions can be achieved. Similar in vivo recombination approaches have been conducted by heterologous expression of the entire aurofusarin and bikaverin biosynthetic gene clusters from Fusarium spec. in A. nidulans [74]. Moreover, along with our studies on A. niger, an AMA1-based vector was constructed for an in vivo recombination of various DNA fragments (encoding the uidA color enzyme or fluorescent proteins) into the conidial pigment gene wA of A. nidulans, i.e. homologous to the fwnA locus in A. niger, highlighting the general suitability of this biobricks approach [75]. A third example comprises the successful integration of the calbistrin cluster using six DNA fragments in an engineered Penicillium rubens platform strain resulting in the production of decumbenones A–C [76]. Although the calbistrin cluster is derived from the highly related species Penicillium decumbens, no production of calbistrins A–C was observed in the heterologous host, indicating that host specific effects can affect the metabolite production.

Indeed, inter-division gene expression is not a simple task because altered codon usage, intron recognition, protein folding or the lack of accessory proteins sometimes impedes successful heterologous protein production [9, 74]. Hence, it is the more astonishing as the NRPS gene calA from an EDF was successfully expressed in the ascomycete A. niger. However, the detected calpinactam titers are 71 times lower compared to the native producer, which may be caused by undesired splicing events or unstable transcripts in the host. The targeted recombination resulted in a single integration in the fwnA gene locus whilst multi-copy integrations may yield higher metabolite titers as shown in the biotechnological production of other natural compounds [17, 77]. To bypass this limitation, the introduced URA blaster might be excised by 5-fluoroorotic acid treatment of the expression strain to perform a second round of transformation into another locus [78].

The described system is suitable for expression of long genes of Basidiomycetes. We successfully assembled the 8 kb lpaA gene from five DNA fragments in A. niger and confirmed the functionality by the production of various methyl-branched polyenes of different chain length (C26–C32) and ∆7,8 cis/trans stereochemistry, which is in accordance with observations made by the heterologous expression of lpaA in Aspergilli in a previous report [41]. Increasing the metabolic diversity by premature chain truncation is a well-recognized phenomenon of fungal polyketide synthesis [20, 79, 80]. The C4-methyl group as part of the 1-methyl-2-oxo-propylidene head is a common feature of many fungal linear polyenes [49,50,51]. We showed that the inactivation of the C-MT of LpaA is critical for chain elongation and completely abolished laetiporic acid production. As demonstrated for the iterative type I PKS LovB in the lovastatin biosynthesis of A. terreus, the highly regioselective and comparably fast C-MT reactions are essential for the full extension of the lovastatin nonaketide chain [81, 82]. Deconstruction studies of single domains of the PKS PksCT producing the highly (C2,C4,C6)-trimethylated pentaketide citrinin showed that in absence of the C-MT domain, the minimal PKS solely produced an unmethylated triketide pyrone [52]. Similar to our results on the LpaA C-MT, single mutations in the catalytic dyad of the C-MT of the PKS MtaltA in the alternapyrone biosynthetic pathway completely abolished product formation [83]. Since no intermediates are detectable in the lpaA C-MT mutant, C4-methylation occurs most likely at the thioester α-carbon following the first extension cycle and not on the fully elongated hexacosaketide.

In a second approach, we successfully integrated the 27.7 kb calA expression construct (including the calA gene and the required regulatory elements) into the genome of A. niger. The calpinactam synthetase CalA (712 kDa) is approximately one and a half the size of the largest heterologously NRPSs produced in S. cerevisiae, among them the α-l-aminoadipyl-l-cysteinyl-d-valine synthetase PcbAB (426 kDa), the fumiquinazoline F synthetase Afu6g12080 (438 kDa), the tryptoquialanine synthetase TqaA (450 kDa) and the aspyridone synthetase ApdA (498 kDa) [84,85,86,87]. Our work enables both an optimized calpinactam production by the original producer strain M. alpina and the biotechnologically tractable A. niger strain tJMW06. Although, further improvement is required to increase the yield of the compound, e.g. by codon-optimization [88], the described tool provides the first evidence for a functional heterologous expression of a natural product gene in full-length from an early diverging fungus. Calpinactam represents a promising alternative to conventional, often less effective antibiotics to treat tuberculosis [89]. The compound possesses an unusual C-terminal ε-caprolactam ring and resembles the structure of mycobactin, i.e. the hydroxamate siderophore of M. tuberculosis, and may interfere with the iron uptake system [53]. Inhibition of iron metabolism is a prospective and highly specific target for antimycobacterial drug development [90].

In our studies, no linear calpinactam peptides with hydrolyzed ε-caprolactam ring system were detected as side products—neither in the original nor in the transgenic producer. This suggests that the 7-membered azacycle is not spontaneously formed, but rely on the intrinsic activity of the releasing module of CalA. Caprolactam-containing metabolites—such as the nucleoside antibiotic capuramycins [91], bangamide A [92], the antifungal circinatin [93], the cytotoxin caprolactin A [94], terreazepine [39], the muscarinic acetylcholine receptor inhibitor nocardimicin [95] and the siderophore mycobactin [90]—are frequently isolated from microorganisms. The caprolactam structure in capuramycin originates from l-lysine and has been assigned to the action of the NRPS-like module CapU [91]. Future investigations on the 6th module of CalA might be of importance, as the current industrial production routes to ε-caprolactam, i.e. the monomeric precursor of the widely used synthetic fibers nylon-6 and PEBA2000 [96, 97], rely on expensive and energy-wasting classical chemical syntheses [98]. Alternative, sustainable production routes in mammalian cells [99], plants [98] or fungi are highly requested bio-based strategies.

Conclusions

The heterologous expression of genes encoding megasynth(et)ases for natural products has long time been a very challenging task. The presented expression system allows integration and expression of long genes of interest of > 20 kb in a single step. The tool is suitable for genes of various biological sources, including higher and early diverging fungi, and provide the basis to tap yet unexplored secondary metabolite producers such as non-cultivatable fungi.

Methods

Organisms and culture conditions

Aspergillus niger strains ATNT, tLK01 (ATNT∆akuB), the lpaA-expressing strains tLK04 and tLK05, the calA-producing strain tJMW06 and the empty vector controls tLK06 and tLK07 were cultivated on Aspergillus Minimal Medium (AMM) agar plates supplemented with 2 mm l-glutamine [100]. To compensate uracil auxotrophy, 10 mm uridine (Carl Roth) was added for cultivation of the strains ATNT, tLK01 and tLK06. For tLK01 cultivation, 140 µg mL−1 hygromycin B (Carl Roth) was added. Plates were incubated at 30 °C for 4 d and conidia were harvested as previously described [41]. 100 mL AMM with 200 mm glucose and 20 mm l-glutamine or YPD (10 g L−1 yeast extract, 20 g L−1 soy peptone, 20 g L−1 d-glucose) were inoculated with a titer of 1 × 106 A. niger conidia mL−1 to produce laetiporic acids at 30 °C for 72 h or to produce calpinactam at 25 °C for 72 h, respectively. Transgenic A. niger cultures were induced with 30 µg mL−1 doxycycline (Merck). Mortierella alpina ATCC32222 was cultivated on MEP agar plates (20 g L−1 malt extract, 3 g L−1 soy peptone, 20 g L−1 agar, pH 5.6) at 25 °C for 7 d. To produce calpinactam, 100 mL MEP liquid medium was inoculated by addition of three agar blocks (3 × 3 mm) and cultivated at 25 °C for 4 d. Escherichia coli XL1 blue were used for plasmid propagation and were cultivated in LB medium (5 g L−1 yeast extract, 10 g L−1 tryptone, 10 g L−1 NaCl) supplemented with 100 µg mL−1 carbenicillin (Roth), if required. Authentic standards for laetiporic acids were obtained from fruiting bodies of Laetiporus sulphureus JMRC:SF:012,599 as described [41]. All strains used in this study are listed in Additional file 3: Table S2.

Strain construction

Details on the cloning procedures, strain transformation, used strains and oligonucleotides are given in the Additional file 1: Experimental procedures, Additional file 3: Table S2, Additional file 4: Table S3. In brief, the akuB deletion mutant (tLK01) was generated by replacing the entire gene (2651 bp) with the hygromycin B resistance cassette (hph) [45] from plasmid pLK03. To replace the spore pigment polyketide synthase gene fwnA by overexpression constructs, the pSMX2-derived [18] expression vector pPS01 [41] was expanded by an 1 kb upstream and 1 kb downstream flank of fwnA to yield plasmid pLK04. The plasmid served either as vector to ligate DNA sequences or as template for subsequent fusion PCRs. For heterologous expression of lpaA (A. niger strains tLK04 and tLK05) and calA (A. niger strain tJMW06) in the recipient strain tLK01, the five gene fragments of the laetiporic acid synthase gene (from L. sulphureus) and the calpinactam synthetase gene (from M. alpina) were amplified from the plasmid pPS03 [41] or the genomic DNA of M. alpina strain ATCC32222, respectively. In either case, the orotidin-5′-phosphate-decarboxylase gene, pyrG, from Aspergillus nidulans was used as selectable marker. The generated transformants were checked by PCR (Fig. 2, Additional file 8: Fig. S4, Additional file 12: Fig. S8). Deletion of akuB in A. niger tLK01 and fwnA in A. niger tLK04 as well as the full-length integration of calA in tJMW06 were additionally confirmed by Southern blot analysis with digoxigenin-labeled probes as described (Figures S1, Additional file 9: Fig. S5, Additional file 13: Fig. S9) [45].

Expression and splicing analyses

To determine the expression of lpaA in A. niger (lpaA-expressing), tLK05 (lpaAD1415A-expressing) and tLK07 (null mutant), fungi where cultivated in 200 mL AMM supplemented with 200 mm d-glucose and 20 mm l-glutamine at 30 °C for 48 h. To determine the expression of calA and its processed splicing products M. alpina ATCC32222 and A. niger tJMW06 was cultivated in 200 mL MEP for 72 h or in 200 mL YPD amended by 30 µg mL−1 doxycycline for 36 h, respectively. Mycelium was harvested, ground under liquid nitrogen, and RNA was isolated using the SV Total RNA Isolation System (Promega). Residual gDNA was digested using Baseline-Zero DNase (Biozym). cDNA was synthesized using the anchored oligo(dT)20 primers and the RevertAid First Strand cDNA Synthesis Kit (Thermo Scientific). Semi-quantitative PCR was carried out with DreamTaq Green Polymerase (Thermo Scientific) and oligonucleotides listed in Additional file 4: Table S3 according to the manufacturer’s instructions. As internal standard, the housekeeping genes encoding actin B (actB; for M. alpina) or the glycerinaldehyde-3-phosphate-dehydrogenase (gpdA; for A. niger) was used.

Extraction of laetiporic acids and calpinactam

Mycelia of A. niger tLK04 (lpaA-expressing), tLK05 (lpaAD1415A-expressing), tLK07 (empty vector control) and L. sulphureus fruiting bodies were used to extract laetiporic acids. Accordingly, mycelia of A. niger tJMW06 (calA expressing), tLK07 (empty vector control) and M. alpina were used to produce calpinactam. Mycelia were harvested by filtration, rinsed with water and finally lyophilized to entire dryness. Mycelia were ground to a fine powder using a mortar and pestle and subsequently extracted with 20 mL acetone per gram dry weight (laetiporic acids) or with 20 mL methanol per gram dry weight (calpinactam) at 150 rpm for 2 h. The solvent phase was decanted, filtered and evaporated under reduced pressure in a rotary evaporator. The residue was dissolved in 1 mL methanol and 5 µL thereof was subjected to UHPLC-MS analyses.

Chemical standards

Authentic standards of laetiporic acids A1, A2, B1, B2, C1, and C2 were isolated from L. sulphureus fruiting bodies as described previously [41]. A commercial standard for calpinactam was obtained from Santa Cruz Biotechnology, Inc. (CAS 1205538-83-5).

UHPLC-MS measurement

Extracts and standards were measured on an Agilent Infinity II 1290 system connected to an Agilent 6130 single quadrupole mass spectrometer. To detect laetiporic acids the following gradient was applied on a Zorbax Eclipse Plus C18 RRHD column (Agilent; 50 mm × 2.1 mm, 1.8 µm) at a flow rate of 1 mL min−1 at 30 °C using water + 0.1% formic acid (solvent A) and acetonitrile (solvent B): 0–3 min: 5–50% B; 3–7 min: 50–75% B; 7–8 min: 75–100% B. Signals were recorded at wavelengths λ = 210–600 nm by a diode array detector and UV profiles were extracted at λ = 450 nm. Additionally, extracted ion chromatograms were recorded by ESI–MS in positive ionization mode (m/z 421, 447, 473, 499 [M + H]+). Purified polyenes LA A1–LA C2 from L. sulphureus served as reference standard [41]. To detect calpinactam, the identical column and chromatograph were used and the following gradient were applied: 0–4 min: 5–72% B, 4–4.5 min: 72–95% B, 4.5–5 min: 95% B. Since calpinactam is hardly chromophore, mass signals were recorded in positive ionization mode and detection of calpinactam was based on selected ion monitoring (SIM) signals of m/z 768 [M + H]+. For quantifications, a calibration curve was recorded with a commercial calpinactam standard (Santa Cruz Biotechnology, Inc.) in concentrations ranging from 0.488 to 125 µg mL−1.

High resolution mass spectrometry

HR-MS and HR-MS/MS spectra of identified compounds and the respective standards were additionally recorded on a Q Exactive Plus mass spectrometer (Thermo Scientific) as described for polyenes [41] and peptides [57].