Global survey of secondary metabolism in Aspergillus niger via activation of specific transcription factors

Abstract

Genomics analysis confirmed the status of filamentous fungi as a rich source of novel secondary metabolites; however, the discovery of these compounds is hampered by the cryptic nature of their biosynthetic pathways under laboratory conditions. Consequently, despite substantial research effort over the past decades, much of the secondary metabolome remains uncharacterized in fungal organisms. Our manual curation of biosynthetic gene clusters (BGCs) in the Aspergillus niger NRRL3 genome revealed that only 13 of 86 BGCs have had their cognate secondary metabolite products confirmed or reliably inferred. We also identified 60 transcription factors (TFs) associated with cryptic BGCs. To further characterize A. niger secondary metabolism, we created a collection of strains each overexpressing a single BGC-associated TF. We analyzed the strain collection using a standardized pipeline where we monitored phenotypic changes and compound production using mass spectrometry. Strains showing evidence of secondary metabolism activation were selected for gene expression analysis. Our approach resulted in the production of multiple potentially novel secondary metabolites and linked a specific BGC to tensidol production in A. niger. More broadly, this study found evidence counter to the existing paradigm of BGC expression controlled by colocalized TFs, lending credence to the emerging picture of a complex regulatory network governing fungal secondary metabolism.

Significance Statement.

Fungi produce an array of chemically diverse compounds that are routinely found to harbor valuable bioactivity. The products of secondary metabolism, these compounds have been a source of antimicrobials, anticancer agents, and other biopharmaceutical compounds termed natural products. Despite their demonstrated economic value, much is still unknown about the biosynthesis, regulation, and identities of these compounds. This study adopted a genome-wide approach to improve our understanding of the regulatory mechanisms that control fungal secondary metabolism, improving our ability to investigate the pathways responsible for natural product production.

Introduction

Fungal secondary metabolites are organic compounds that are rich in chemical diversity and include many molecules of clinical and industrial significance (1)). Examples include antimicrobials (e.g. penicillin, echinocandins), anticancer compounds (e.g. aurantiamine, wortmannin), immunosuppressants (e.g. cyclosporin, mycophenolic acid), and anticholesterolemic compounds (e.g. statins) (2–4). Filamentous fungi have traditionally been a rich source of secondary metabolites and display a capacity to produce compounds that are both beneficial through exploitation (e.g. gibberellic acid) and harmful (e.g. aflatoxins, ochratoxin) to humans (5, 6). The availability of fungal genome sequences, combined with improved bioinformatic tools, has further highlighted the wealth of untapped biosynthetic potential of filamentous fungi for the production of uncharacterized secondary metabolites (7). Despite the clinical, industrial, and economic potential associated with fungal secondary metabolism, the majority of the secondary metabolome of these organisms remain poorly characterized. As a result, substantial effort has been made towards improving our understanding of the genetics and processes that control expression of these small molecules with an eye towards combatting spoilage due to mycotoxin production, discovering new bioactive molecules, and linking known metabolites to their corresponding biosynthetic enzymes.

The biosynthetic genes responsible for production of secondary metabolites are typically colocalized into discrete biosynthetic gene clusters (BGCs) within fungal genomes (8). BGCs contain two or more genes that contribute to the biosynthesis of a secondary metabolite and can be defined by the enzyme class of the corresponding backbone enzyme(s) (9). Backbone enzymes catalyze extension of the polymer chain and include polyketide synthases (PKS), nonribosomal peptide synthetases (NRPS), hybrid PKS/NRPS (HPN), terpene synthases, fatty acid synthases (FAS), and dimethylallyl tryptophan synthases (DMAT). BGCs may also encode for proteins involved in decorating the backbone through various enzymatic activities (e.g. redox, methylation), transporters or efflux pumps, and in some cases transcription factors (TFs) that can play a role in regulating the overall biosynthetic pathway (8). Based on the class of backbone enzyme present within a BGC, general predictions can made about the chemistry of the cognate secondary metabolite(s); however, predicting the final chemical structure and corresponding biological function of the metabolites produced by BGCs remains elusive necessitating empirical characterization. Further hampering the characterization and exploitation of the fungal secondary metabolome is the fact that the majority of BGCs found in typical ascomycete species are transcriptionally silent under laboratory conditions (10).

Spurred by advances in molecular biology, considerable effort has been invested into awakening silent BGCs to elicit production of their corresponding secondary metabolites. Approaches include modification of culture conditions (pH, temperature), coculturing with other organisms (11, 12), deleting and/or overexpressing master or colocalized regulators (13), heterologous expression of BGCs (14), or in vitro reconstitution of entire biosynthetic pathways (15). While each of these techniques has been successful for identifying a small number of novel secondary metabolites, they are primarily used to parse only a subset of BGCs in a given organism. In cases where novel metabolites are discovered, they are typically produced in quantities insufficient for structural characterization or downstream functional characterization, thus necessitating further optimization.

Aspergillus niger is a haploid filamentous fungus that has been used extensively for the production of organic acids (e.g. citric acid and gluconic acid) and carbohydrate-degrading enzymes (e.g. glucoamylase) (16). It is also used as a model organism for the study of fungal systems, with comprehensive genetic tools available including a recently established protocol for gene editing using CRISPR–Cas9 (17, 18). Publicly available genome sequences of A. niger, including the manually curated NRRL3 genome, confirmed it to be a rich source of cryptic BGCs (19).

In this study, we report the manual curation of all BGCs within the A. niger NRRL3 genome, which included the identification of 60 cryptic BGCs featuring colocalized Zn₂Cys₆ TFs. As a tool to advance the study of A. niger secondary metabolism, we generated an overexpression strain collection comprised of 58 A. niger NRRL3-derived strains, each overexpressing a single Zn₂Cys₆ TF. We then analyzed each strain using a standardized workflow that involved phenotypic profiling, untargeted metabolomics and transcriptomics (Fig. 1). Previously characterized secondary metabolites from A. niger were well represented within our strain library, and their production could be reliably attributed to overexpression of specific TFs. Analysis of our strain library also found a number of unique metabolomic features associated with activation of cryptic BGCs, leading to correlation of a specific BGC to tensidol B production. Gene expression analysis of the strain library revealed multiple instances of cross-reactivity between BGCs and underscored the complex regulatory network that governs fungal secondary metabolism.

Fig. 1.

Workflow for investigating secondary metabolism using overexpression strain collection. 1) The NRRL3 genome was manually curated and all genes localized to BGCs were cataloged. Fifty-eight BGC-associated TF genes were identified and selected for overexpression. 2) Using CRISPR–Cas9, each TF was inserted into the glucoamylase locus by replacing the coding region of the glaA gene resulting in the generation of a 58-member overexpression strain library. 3) The strain collection was plated on induction (maltose) media for phenotypic profiling. 4) Strains were cultured in 96-well format on induction media. 5) Metabolite extraction and untargeted LC–MS analysis of samples were conducted. The results were compared to existing A. niger metabolite databases to positively identify metabolites where possible. 6) RNA-seq was performed 2 h postinduction on a subset of strains showing altered metabolite profiles to measure changes in gene expression resulting from overexpression of TF.

Results

Manual curation of BGCs in A. niger NRRL3

The publicly available A. niger NRRL3 genome sequence contains coverage of all eight chromosomes including 16 telomeric regions and manually verified gene models (17). We used the bioinformatic tools SMURF (20) and antiSMASH (21, 22) to identify BGCs that may be involved in secondary metabolite synthesis, as well as InterProScan to identify protein domains that correspond to proteins involved in backbone production of secondary metabolites. The results of this curation effort identified a total of 86 BGCs (Table 1), distributed across all eight chromosomes (Fig. 2). Notably, this number of predicted BGC is greater than previously published estimates based on similar analyses of the A. niger CBS513.88 genome (18, 19). Within these 86 BGCs in A. niger NRRL3, we identified 99 genes encoding backbone-type enzymes, the majority of which were classified as PKS (n = 40), NRPS (n = 38), or HPN (n = 9). We also identified a small number of terpene synthases (n = 6) and FAS (n = 5) encoding genes in BGCs. In addition, the 86 BGCs featured 69 genes encoding for Zn₂Cys₆ TFs, 53 genes for predicted transporters, and more than 300 additional genes encoding predicted tailoring enzymes as well as 77 genes encoding proteins of unknown function. Cross-referencing these BGCs against gene expression data from publicly available gene expression data for A. niger CBS513.88 (FungiDB) indicated that most of these BGCs lacked a gene expression profile, consistent with previous results showing >70% of A. niger BGCs appear to be transcriptionally silent under laboratory growth conditions (20). Among BGCs identified in A. niger NRRL3, nine have been previously characterized experimentally in terms of production of specific secondary metabolites and two have had their secondary metabolite products inferred on the basis of sequence similarity with characterized BGCs from other organisms (Table 1). Specifically, the NRRL3_02178–NRRL3_02189 BGC has been proposed to support the production of fumonisin based on similarity to corresponding BGCs in Fusarium verticillioides and Gibberella moniliformis (26, 27, 31), while the expression of NRRL3_07873–NRRL3_07888 BGC has been predicted to produce malformin A2 and C, based on similarity to a characterized BGC is Aspergillus brasiliensis (37). We also identified genes homologous to the ones responsible for production of naptho-γ-pyrone and dihydroxynaphthalene (DHN)–melanin in A. fumigatus; however, these genes are not colocalized into a specific BGC in the A. niger NRRL3 genome. In total, our manual curation efforts revealed that only 13 out of 86 BGCs have had their secondary metabolite products confirmed or reliably inferred, suggesting a systematic undertaking to explore the endogenous biosynthetic potential of the remaining 73 clusters is warranted.

Table 1.

BGC in A. niger NRRL3.

Backbone enzyme(s)	Enzyme class	Enzyme domains	TF	Range	SM products; homologs; methodology
NRRL3_00030	NRPS-like	A-T-R	NRRL3_00028	00022–00032
NRRL3_00036	NRPS	T-C-A-T-C	NRRL3_00034	00034–00043	Unnamed yellow metabolite. Experimentally linked to cluster in (23)
			NRRL3_00042
NRRL3_00102	HR-PKS	KS-AT-DH-MT-ER-KR-ACP	NRRL3_00104	00098–00111
NRRL3_00135	NRPS	A-T-C-A-T-C	N/A	00134–00136
NRRL3_00147	NR-PKS	KS-AT-ACP-MT-Te	NRRL3_00148	00140–00153	Azanigerones A–F. Experimentally validated in A. niger ATCC 1015 (24)
NRRL3_00153	HR-PKS	KS-AT-DH-MT-ER-KR-ACP
NRRL3_00166	PKS-like	ACP-MT-Te	N/A	00166–00168
NRRL3_00204	Terpene synthase		NRRL3_00205	00202–00209
NRRL3_00410	NRPS-like	A-A-T-Te	NRRL3_00406	00406–00414
NRRL3_00430	HR-PKS	KS-AT-DH-ER-KR-ACP	NRRL3_00447	00424–00447
NRRL3_00462	NR-PKS	KS-AT-ACP-Te	N/A	Unclustered	NRRL3_00462 (albA) produces precursor of naptho-γ-pyrone and DHN–melanin. Experimentally validated in A. niger ATCC 11414 derivative KB1001 (25)
NRRL3_00492	Terpene synthase		N/A	00492–00493
NRRL3_00755	HPN	KS-AT-DH-KR-ACP–C-A-T	NRRL3_00759	00754–00759
NRRL3_00998	HPN	A-T–KS-AT-ACP	NRRL3_00988	00988–00998
NRRL3_01334	NRPS	C-A-T-C	NRRL3_01335	01332–01340
NRRL3_01339	HPN	KS-AT-DH-KR-ACP–C-A-T
NRRL3_01367	NRPS-like	A-T-R	NRRL3_01372	01366–01374
NRRL3_01369	HR-PKS	KS-AT-DH-ER-KR-ACP
NRRL3_01804	HR-PKS	KS-AT-DH-MT-ER-KR	NRRL3_01808	01804–01809
NRRL3_02185	NRPS-like	T-C	NRRL3_02186	02178–02189	Fumonisin B2. Cluster identified through homology with fum gene cluster characterized in F. verticillioides (26–28)
NRRL3_02189	HR-PKS	KS-AT-DH-ER-KR-ACP
NRRL3_02191	HR-PKS	KS-AT-DH-KR	N/A	02190–02191
NRRL3_02593	NRPS-like	A-T	N/A	02591–02597
NRRL3_02596	NRPS-like	A-T-Te
NRRL3_02621	PKS-like	KR-ACP	NRRL3_02628	02618–02629
NRRL3_02852	NRPS-like	A-T-R-Ki	NRRL3_02857	02851–02858
NRRL3_02887	NRPS-like	A-T-Te	N/A	02886–02888
NRRL3_02961	NRPS-like	A-T-Te	NRRL3_02959	02956–02962
NRRL3_03167	NRPS	A-T-C-A-T-C-A-T-Te	NRRL3_03177a	03167–03185
NRRL3_03184	HR-PKS	KS-AT-DH-ER-KR-ACP
NRRL3_03756	HR-PKS	KS-AT-DH-MT-ER-KR-ACP	NRRL3_03751	03750–03757
NRRL3_03891	HR-PKS	KS-AT-DH-MT-ER-KR-ACP	NRRL3_03898a	03886–03898
NRRL3_03977	HR-PKS	KS-AT-DH-MT-ER-KR-ACP	NRRL3_03978	03971–03978
NRRL3_04180	NRPS	A-T-C-A-T-C-A-T-C-A-T-C-A-T-C	N/A	04173–04189
NRRL3_04207	PKS-like		N/A	—	Remnant of ochratoxin A synthase, a cluster intact in A. niger CBS 513.88 but absent in A. niger ATCC 1015 (26)
NRRL3_04226	HPN	KS-AT-DH-KR-ACP–C-A-T	NRRL3_04231	04223–04236	Putative involvement in tensidol B production, this study
NRRL3_04305	PR-PKS	KS-AT-DH-ACP	N/A	04299–04313
NRRL3_04420	DMAT		N/A	04219–04220
NRRL3_05440	NRPS-like	A-T-Te	NRRL3_05457	05437–05457
NRRL3_05484	HR-PKS	KS-AT-DH-MT-ER-KR-ACP	NRRL3_05469	05469–05485
NRRL3_05588	HPN	KS-AT-DH-KR-ACP–C-A-T-E	NRRL3_05582a	05580–05588
NRRL3_05848	NRPS	A-T-C-T-C	N/A	05847–05850
NRRL3_06189	NRPS-like	A-T-Te	N/A	06185–06189
NRRL3_06217	HR-PKS	KS-AT-DH-MT-ER-KR-ACP	N/A	06217–06223
NRRL3_06237	NRPS-like	A-T-Te-DH	NRRL3_06232	06232–06239
NRRL3_06291	HR-PKS	KS-AT-DH-KR-ACP	NRRL3_06287 (yanR)	06287–06296	Yanuthones (D, F, G, H, I, J, X1). Experimentally validated in the A. niger ATCC 1015 derivative KB1001 (29)
NRRL3_06340	FAS		N/A	06337–06342
NRRL3_06341	FAS
NRRL3_06801	NRPS	A-C-T-A-T-C-A-T-C-A-T-C-A-T-C	NRRL3_06795a	06793–06801
NRRL3_07380	NRPS-like	A-T-R	N/A	07376–07381
NRRL3_07399	Terpene synthase		NRRL3_07394a	07394–07405
NRRL3_07443	HR-PKS	KS-AT-DH-MT-ER-KR-ACP	NRRL3_07461	07440–7463
			NRRL3_07463
NRRL3_07486	NR-PKS	KS-AT-ACP-Te	N/A	07485–07487	Kotanin, 7-demethylkotanin, orlandin. Experimentally validated in A. niger FGSC A1180 (30)
NRRL3_07739	NRPS	A-T-C-A-T-C	NRRL3_07736	07735–07741
			NRRL3_07741
NRRL3_07881	NRPS-like	A-T-Te	NRRL3_07873	07873–07888	Malformin A2, C. Based on homology to mlfA in A. brasiliensis (31)
NRRL3_07884	PR-PKS	KS-AT-DH-KR-ACP	NRRL3_07888
NRRL3_08167	NRPS-like	A-T-Te	NRRL3_08195	08167–08195
NRRL3_08318	HR-PKS	KS-AT-DH-MT-ER-KR	NRRL3_08316	08310–08319
			NRRL3_08319
NRRL3_08341	NRPS	A-T-C-A-T-Te	NRRL3_08344	08341–08344
NRRL3_08369	NRPS-like	A-T-R	NRRL3_08350	08350–08369
NRRL3_08388	NR-PKS	KS-AT-ACP-Te	NRRL3_08381	08381–08389
NRRL3_08411	HR-PKS	KS-AT-DH-MT-ER-KR-ACP	N/A	08410–08411
NRRL3_08436	Terpene synthase		N/A	08435–08438
NRRL3_08538	NRPS	A-T-C-A-T-C	NRRL3_08550a	08537–08553
NRRL3_08647	NR-PKS	KS-AT-ACP-Te	NRRL3_08635	08635–08652
			NRRL3_08652
NRRL3_08729	NRPS	A-T-C-A-T-C	NRRL3_08721a	08721–08730
NRRL3_08732	Terpene synthase		N/A	08731–08733
NRRL3_08775	NR-PKS	KS-AT-DH-ACP-ACP	NRRL3_08770	08770–08781
			NRRL3_08781
NRRL3_08790	NRPS	T-C-A-T-C-A-T-C-A-T-C-A-T-C	N/A	08788–08790
NRRL3_08891	NRPS	A-T-C-A-T-C-A-T-C-A-T-C	NRRL3_08890a	08885–08914
			NRRL3_08914
NRRL3_08969	NRPS	A-T-C-A-T-C-A-T-C-A-T-C	NRRL3_08965	08965–08969
NRRL3_08978	NRPS-like	A-T-C	NRRL3_08989	08974–08989
NRRL3_08980	HR-PKS	KS-AT-DH-ER-KR-ACP
NRRL3_08984	HR-PKS	KS-AT-DH-MT-ER-KR-ACP-Te
NRRL3_09034	NR-PKS	KS-AT-ACP-MT-Te	NRRL3_09035	09034–09041
NRRL3_09038	FAS
NRRL3_09039	FAS
NRRL3_09351	PR-PKS	KS-AT-DH-MT-KR-ACP-Te	NRRL3_09365	09349–09365
NRRL3_09549	NR-PKS	KS-AT-ACP	NRRL3_09545 (adaR)	09545–09550	TAN-1612/BMS-192548, asperthecin. Experimentally validated in A. niger ATCC 1015 and in vitro (32)
NRRL3_09616	HPN	KS-AT-DH-KR-ACP–C-A-T	N/A	09613–09620
NRRL3_09628	Terpene synthase		NRRL3_09625	09625–09629
NRRL3_09686	HR-PKS	KS-AT-DH-MT-ER-KR-ACP	NRRL3_09689#	09686–09691
NRRL3_09693	NRPS-like	A-T-R	NRRL3_09696a	09692–09696
NRRL3_09789	HR-PKS	KS-AT-DH-KR-ACP	NRRL3_09799	09787–09799
NRRL3_09792	NRPS-like	A-T-NADR
NRRL3_09827	HR-PKS	KS-AT-DH-MT-ER-KR-ACP	NRRL3_09831	09827–09833
NRRL3_09848	HR-PKS	KS-AT-DH-ER-KR-ACP	NRRL3_09846	09846–09849
NRRL3_10128	HPN	KS-AT-DH-KR-ACP–C-A-T	NRRL3_10124 (pynR)	10120–10130	Pyranonigrin E. Experimentally validated in A. niger ATCC 1015 (33)
NRRL3_10148	NRPS	A-T-T-C-A-T-C-A-M-T	N/A	10146–10157
NRRL3_10912	NRPS	A-T-C-A-C-A-T-C-A-T-C-T-C-T	N/A	10911–10913
NRRL3_10209	HPN	KS-AT-DH-ER-KR-ACP–C-A-T-Te	N/A	10204–10213
NRRL3_10375	PR-PKS	KS-AT-DH-ER-KR-ACP	NRRL3_10370	10368–10375
NRRL3_11031	HPN	KS-AT-DH-KR-ACP–C-A-T	NRRL3_11029 (caaR)	11025–11031	Carlosic acid, carlosic acid methyl ester, agglomerin F. Experimentally validated in A. niger ATCC 1015 (34)
NRRL3_11121	NRPS-like	A-T-R	N/A	11121–11125
NRRL3_11458	NRPS-like	C-T	NRRL3_11460	11456–11461
NRRL3_11645	NRPS	A-T-C-A-T-C-A-T-C-T-C-T-C	N/A	11643–11646
NRRL3_11726	Type III PKS	KS	N/A	11723–11728	Protocatechuic acid. Experimentally validated in A. niger ATCC 9029 derivative MA169.4 (35)
NRRL3_11763	FAS		NRRL3_11765 (akcR)	11756–11767	Biosynthesis of alkylcitric acids with unlinked genes. Experimentally validated in A. niger NRRL3 (36)
NRRL3_11767	FAS

^aCluster genes are transcriptionally active under standard laboratory conditions.

AT, acyltransferase; ACP, acyl carrier protein; DH, dehydratase; KR, keto reductase; KS, keto synthase; ER, enoylreductase; MT, methyltransferase; Te, thioesterase; A, adenylation; C, condensation; R, reductase; T, peptidyl carrier protein; NADR, terminal reductase.

Fig. 2.

Genomic locations of BGC in A. niger NRRL3. Location of all 86 BGCs shown on chromosome maps, colored by backbone enzyme type. TF locations are demarcated by asterisk above colocalized backbone enzymes.

Construction of the A. niger NRRL3 TF overexpression strain collection

Previous studies have shown that overexpression of colocalized TF can activate expression of the corresponding BGC in a fungal genome resulting in the production of secondary metabolites (37). This specific strategy has led to characterization of BGCs in A. niger involved in production of azanigerone A–F, agglomerin F and carlosic acid, and TAN-1612 and BMS-192548 (24, 32, 34). However, to our knowledge, this strategy has never been applied to investigate a wider spectrum of the A. niger NRRL3 secondary metabolome. Accordingly, we decided to pursue the genome-wide overexpression of TFs associated with cryptic BGCs.

As mentioned above, our analysis revealed 69 BGC-associated TF genes, of which 60 appeared to be transcriptionally silent under previously tested growth conditions. The nine TF genes showing expression under previously tested growth conditions were not included in our pipeline for generating overexpression strains, and two of the TFs, NRRL3_09689 and NRRL3_11460, were recalcitrant to cloning and after several attempts were omitted as well. The remaining 58 TFs were selected for overexpression, and we undertook construction of the A. niger NRRL2270 (ATCC11414) (a spontaneous derivative of ATCC1015 (38)) overexpression strain collection where each of these 58 TF was individually targeted (Fig. 2). To generate the TF overexpression (TF^OE) strains, we used Streptococcus pyogenes CRISPR–Cas9-mediated integration (17) of each of the 58 TF genes into the glucoamylase A locus, thereby replacing the coding region of the glucoamylase A region with the gene encoding a TF. The integration site selected was downstream of the strong inducible promoter (P_glaA) that has been shown to drive high levels of gene expression during growth on maltose (39, 40). Using this approach, we successfully generated 58 TF^OE strains, with each TF gene integration verified by PCR amplification and restriction enzyme digestion (Fig. S1).

Phenotypic analysis of overexpression strains

To estimate the potential impact of individual TF gene overexpression on metabolite production, we first monitored our A. niger NRRL2270 strain collection for potential phenotypic changes. Distinct phenotypic features such as pigmentation (24), delayed or reduced growth (36), or changes in sporulation (41) are often early indicators of changes in fungal secondary metabolism. All 58 TF^OE strains were grown on transformation media and induction media (2% maltose as a carbon source) and monitored over 7 days. In at least 14 strains, we observed unique phenotypic markers including pigmentation, delayed/slow growth, and impaired sporulation (Figs. S2 and S3).

Pigmentation during growth on transformation or induction media was observed in eight of the 14 strains (Fig. S2). This group included six strains where the overexpressed TF is associated with previously characterized or functionally predicted BGC, including NRRL3_00042, NRRL3_00148, NRRL3_06287, NRRL3_09545, NRRL3_10124, and NRRL3_11029. We had previously described the NRRL3_00042 overexpression strain as being involved in production of a yellow pigment (23). Overexpression of NRRL3_00148 resulted in production of a distinct yellow–orange pigment on induction media. This TF is associated with a BGC shown to produce several azaphilone compounds including the yellow-colored azanigerone A compound (24). Yellow pigmentation as well as irregular branching was observed in the NRRL3_06287 overexpression strain, which colocalizes with the BGC previously linked to production of yanuthone compounds (29). For this strain, the phenotypic changes were most evident when the strain was grown on transformation media using sucrose as a carbon source but could also be seen when the strain was plated on induction media. These differences could be due to the lower expression levels of the TF during growth on sucrose-containing media (42). The concomitant effect of this may lead to greater accumulation of yellow intermediates, such as yanuthone A and E, due to lower expression levels of the BGC (43). Growth of this strain on induction media resulted in high levels of yanuthone D production (discussed below), which has been proposed as the terminal compound of yanuthone biosynthesis and described as a pale-colored compound (29, 43).

One of the most prominent instances of pigment production was in response to overexpression of NRRL3_09545. The BGC associated with this TF produces multiple compounds including TAN-1612, a yellow aromatic polyketide (32). NRRL3_10124 overexpression also produced noticeable yellow pigmentation, reduced growth, and irregular branching compared with the parental strain when grown on induction media. This TF has been linked to a BGC that produced the yellow compound pyranonigrin E (33). The NRRL3_11029 overexpression strain displayed pink pigmentation during growth on transformation media and at later stages of growth on induction media (34). Finally, the NRRL3_11765 overexpression strain showed irregular branching and impaired sporulation. This TF was previously shown to regulate a BGC linked to the production of alkylcitric acids (36).

In six of the strains where we observed unique phenotypic traits (Fig. 3), the corresponding TFs were not associated with any previously characterized BGC. Overexpression of NRRL3_00205 resulted in a phenotype displaying an irregular branching shape and strong black/brown pigmentation (Fig. S2). NRRL3_01335 overexpression resulted in distinct changes in sporulation. The NRRL3_04231 overexpression was marked by delayed growth and an irregular branching phenotype. Strong yellow pigmentation was observed in response to overexpression of NRRL3_09035 (Fig. S2). Finally, the TF^OE strains overexpressing NRRL3_07741 and NRRL3_10370 displayed an accelerated growth rate compared with the parental strain (NRRL2270), which was accentuated in the later stages of culture (Fig. S2).

Fig. 3.

Hierarchical clustering of the secreted secondary metabolites of the TF-overexpressing strains. Supernatants from a 5-day standing liquid culture were analyzed by mass spectrometry. X-axis displays the 58 strains; Y-axis displays the compounds identified by FT–MS. The compound abundance detected by MS is an arbitrary unit calculated as the area under an individual peak. Values with a ratio >5 and a P-value <0.05 were included. The hierarchical clustering was generated using the statistical analysis software R (1) and the d3heatmap package (2). Intensities are centered and scaled by row (across samples).

In summary, 14 of the 58 strains (∼26%) included within the overexpression strain library displayed phenotypic changes when grown on transformation or induction media. The distinct phenotypes associated with overexpression of BGC-associated TF provided initial support for our strategy and for detailed analysis of the metabolomic repertoire in our strain collection.

Surveying the secondary metabolism of TF^OE strains

To investigate whether overexpression of a given TF resulted in changes in the profile of secondary metabolites produced in A. niger, we analyzed the extracellular metabolite fraction using liquid chromatography–mass spectrometry (LC–MS). We chose to focus on extracellular metabolite production due to the propensity for fungal secondary metabolites to be secreted and to simplify the parallel analysis of metabolite production by all 58 TF^OE strains. All TF^OE strains were cultured in liquid medium under inducing conditions (1 or 15% maltose as a carbon source) for 5 or 7 days. The culture supernatant was then collected, and secreted metabolites were extracted and analyzed via LC–MS. The extracellular metabolite profile of each TF^OE strain was then compared with that of the parental strain grown under the same conditions. The MS features corresponding to compounds that displayed a minimum 5-fold increase in relative abundance across two biological replicates were selected for further analysis. We also analyzed the LC–MS profiles of the TF^OE strains for production of specific metabolites using a manually curated A. niger metabolite database (Fig. 3).

Overall, our metabolomics analysis detected the production of unique metabolites enriched more than 500-fold compared with the parental strain in 15 of 58 overexpression strains. In an additional 27 strains, we identified unique metabolomic features enriched 100-fold compared with the parental strain. In the remaining 15 strains, no significant enrichment of specific metabolites was observed, or the enrichment was inconsistent across biological replicates (Fig. 3). Of the 43 TF^OE strains identified as producing unique metabolite profiles, only eight are associated with previously characterized BGCs.

Specific examples of metabolite production include overexpression of the TF NRRL3_00148. In a previous study, this TF (annotated as azaR) was overexpressed using a plasmid-encoded gene under control of the glyceraldehyde-3-phosphate dehydrogenase A promoter (P_gpdA) (24). This led to overexpression of the cognate BGC resulting in production of several azaphilone compounds. Analysis of the strain overexpressing NRRL3_00148 in our strain collection showed a 1,200-fold increase in relative abundance of a specific peak corresponding to azanigerone D (mw = 361.4) and a 500-fold increase in a peak corresponding to azanigerone E (mw = 250.2). In another example, we observed similarly significant increases in relative abundance of specific metabolites produced by the strain overexpressing the NRRL3_09545 TF. This TF has been previously shown to regulate the adjacent BGC linked to the production of the TAN-1612 and BMS-192548 compounds (32). The NRRL3_09545 overexpression strain produced a 1,000-fold increase in relative abundance of a compound with a m/z value expected of BMS-192548 (m/z = 414.1) (Fig. 3). Taken together, these specific examples demonstrate that stable integration of TF genes is an effective approach for eliciting the overproduction of specific secondary metabolites. By utilizing genomic integration, the strains within our collection do not require nutritional or antimicrobial selection during inductive growth, allowing for increased flexibility in potential downstream applications.

Hierarchical clustering of LC–MS data showed multifold increases in relative abundance of metabolites that could be linked to overexpression of a specific TF gene in 43 of the strains within our collection. Examples included fumonisin B₂ that was linked to overexpression of NRRL3_02186, yanuthone D linked to overexpression of NRRL3_06287, carlosic acid and agglomerin linked to NRRL3_11029 overexpression, and hexylitaconic acid that was found in increased relative abundance in response to NRRL3_11765 overexpression. In other examples, the specific metabolites overproduced lacked matches within existing LC–MS compound databases, suggesting the production of uncharacterized compounds.

Transcriptomic analysis of metabolite-producing TF^OE strains

The metabolomics-guided survey of overexpression strains within our collection identified significant changes in metabolite profiles in response to TF^OE in 43 out of 58 strains. Accordingly, we further investigated the changes in transcriptome profile in this subset of strains using RNA sequencing (RNA-seq). RNA-seq was performed on each sample and the data was processed into transcript-per-million (TPM) format.

The results from the RNA-seq analysis also showed strong up-regulation of the specific TF genes in response to induction with 1% maltose compared with their expression in the parental strain grown under the same conditions. However, we also observed significant variation in absolute expression levels of different TF genes and fold modulation compared with the parental strain despite their identical site of insertion (glucoamylase gene locus). For example, the NRRL3_00447 TF^OE strain exhibited a 56-fold increase in expression of the corresponding TF gene, while in the NRRL3_08350 TF^OE strain the expression level of the TF gene increased 6,000-fold relative to the parental NRRL2270 strain (Table S2, Fig. S4). This large disparity in the magnitude of the changes in expression level occurred despite NRRL3_00447 and NRRL3_08350 showing similar expression levels, less than a 2-fold difference, in the parental strain (3.35 and 1.84 TPM, respectively).

Next, we investigated the expression level of genes residing within the cognate BGCs of the TFs targeted for overexpression. This analysis revealed three distinct scenarios. In the first scenario, we observed up-regulation of expression of all genes predicted to belong to a BGC in response to TF^OE. This occurred in nine of the overexpression strains tested. Within this set of nine strains, six represent BGCs that have been previously experimentally characterized in A. niger. The remaining three strains are as-yet uncharacterized experimentally and the specific chemical structures of the secondary metabolites produced in response to overexpression of these BGCs (NRRL3_04231, NRRL3_07741, and NRRL3_10370) are unknown. In the second scenario, we observed partial up-regulation of the associated BGC, where only a subset of genes was up-regulated. This occurred in six of the strains analyzed via RNA-seq (Table S2). The specific pattern of colocalized gene up-regulation differed among the six strains within this category, with two distinct trends emerging: (i) backbone genes up-regulated and (ii) backbone genes not up-regulated. More specifically, overexpression of NRRL3_02186 resulted in up-regulation of 10/12 genes within the BGC. The two genes that were not up-regulated include NRRL3_02187 encoding for a predicted cytochrome P450-type enzyme and NRRL3_02188 encoding for a predicted ABC transporter. In contrast, while overexpression of NRRL3_07461 resulted in multiple colocalized genes being up-regulated, the PKS-encoding gene within this cluster (NRRL3_07443) was not up-regulated.

The final scenario observed was where there was no evidence of associated BGC up-regulation in response to TF^OE. We observed this trend in the remaining 27 strains we analyzed via RNA-seq. Taken together, our transcriptional analysis suggested that overexpression of individual TFs resulted in full or partial up-regulation of only 26% of associated BGCs, with the remaining TFs failing to directly activate their adjacent BGC under the conditions tested. As a result, activation of these BGCs appears to be a more complex process and may involve additional layers of regulation governing BGC expression.

Up-regulation of BGC correlates with overproduction of secondary metabolites

Next, we compared the LC–MS metabolomics data and the transcriptomic results for the nine strains where we observed up-regulation of the complete BGC in response to TF^OE. Unsurprisingly, all nine strains showed clear overproduction of secondary metabolites. Notably, this group included six strains overexpressing TFs associated with previously experimentally characterized BGCs. Overexpression of the TF (NRRL3_06287) colocalized with the BGC involved in yanuthone biosynthesis resulted in strong up-regulation of the entire BGC, and we observed a 53-fold enrichment in a compound with a mass corresponding to yanuthone D (mw = 502.25). The NRRL3_11765 overexpression strain also showed complete up-regulation of its cognate BGC and produced a metabolite profile with multiple compounds significantly enriched compared with the parental strain. Two of the most prominent features within the metabolite profile of this strain corresponded to hexylcitrate (mw = 276.12) and hexylitatonic acid (mw = 230.11), both of which represent expected metabolic products of this BCG (Fig. S5) (36). Additional overexpression strains linked to previously characterized clusters where we observed up-regulation of the complete BGC and strong changes in metabolite production include those discussed above, specifically NRRL3_00148, NRRL3_09545, NRRL3_10124, and NRRL3_11029, all of which displayed distinct phenotypic markers in addition to robust changes in metabolite profile and complete BGC up-regulation. Taken together, this result supports phenotypic changes as a strong, though imperfect (discussed below), proxy for alterations in fungal secondary metabolism.

The TFs overexpressed in the remaining three strains in this category are all associated with previously uncharacterized BGCs. In most cases, we were unable to match the additional chemical species produced by these strains with known A. niger metabolites, raising the possibility that they represent novel secondary metabolites. Notably, we observed overproduction of two distinct compounds in the NRRL3_07741 overexpression strain. These compounds were enriched 80-fold and 160-fold compared with the parental strain and had molecular weights of 435.18 and 217.59, respectively. The backbone enzyme within the cluster associated with NRRL3_07741 is an NRPS, and the molecular weight of these chemical species puts them within the range (200 to 3,000 Da) of most known nonribosomal peptides (44)). Another uncharacterized BGC showing complete gene up-regulation in response to TF^OE was associated with the TF NRRL3_10370. This strain produced a metabolite profile showing multiple compounds enriched compared with the parental strain; however, like the situation described above we were unable to identify them upon comparison to an A. niger metabolite database. The final uncharacterized cluster showing complete BGC up-regulation also produced distinct metabolomic features. Upon overexpression of the TF NRRL3_04231, we observed 50-fold enrichment in production of tensidol B compared with the parental strain. Tensidol B is a potentiator of miconazole activity against the human pathogen Candida albicans (45) and had been previously isolated from the A. niger FKI-2342 strain; however, its biosynthetic origin is unknown. This prompted a further investigation into the possibility that this gene cluster is responsible for tensidol biosynthesis. Tensidol A and tensidol B have been previously identified using metabolomics within extracts produced by two other Aspergilli, namely A. brasiliensis (46) and Aspergillus tubingensis (47, 48). A BLASTP search of genes within the BGC associated with NRRL3_04231 identified clear orthologs in both of these species, indicating a high level of synteny between A. niger, A. brasiliensis, and A. tubingensis at the site of this BGC (Fig. 4, Table S3). A previous study that identified tensidol A and B in the metabolite extracts of A. niger and A. tubingensis found that these compounds were lacking in extracts produced by 15 other Aspergilli analyzed (48)). Correspondingly, we initiated a search for orthologs of the NRRL3_04231-associated BGC within these species (Aspergillus acidus, Aspergillus aculeatinus, Aspergillus aculeatus, Aspergillus carbonarius, Aspergillus costaricaensis, Aspergillus ellipticus, Aspergillus heteromorphus, Aspergillus homomorphus, Aspergillus ibericus, Aspergillus japonicus, Aspergillus sclerotiicarbonarius, Aspergillus sclerotiniger, Aspergillus uvarum, and Aspergillus vadensis) and found no matches in any of them. Taken together, this comparative genomic analysis provides corroborating support that the BGC containing NRRL3_04231 may be involved in tensidol biosynthesis.

Fig. 4.

Comparative genomics implicated NRRL3_04231 in tensidol B biosynthesis. a) Schematic representation of the NRRL3_04231 BGC. b) High degree of synteny exists between A. niger, A. brasiliensis, and A. tubingensis at the site of this BGC, which is absent in other Aspergilli identified as nonproducers of tensidols. c) Chemical structure of tensidol B. d) Extracted ion chromatograms (EIC) of tensidol B production found in culture extract from A. brasiliensis, NRRL3_04231 overexpression strain, and the parental strain NRRL2270. The peak intensity for each biological replicate is shown inset in the respective EIC panels. n.d., not detected. e) Base peak intensity of tensidol B found in culture extract from NRRL2270 (wild type), NRRL3_04231 OE, and NRRL3_04231 OE + ANEp8-4226. A Welch's one-way ANOVA was run. There was a significant difference between NRRL3_04231 and NRRL3_04231 OE + ANEp8-4226 P = 0.022. n.d., not detected.

To further assess the ability of the NRRL3_04231 OE strain to produce tensidol B, we obtained a known producer of tensidol B, A. brasiliensis strain SN26 (ATCC 9642). We grew A. brasiliensis SN26 and analyzed culture extract for SM production and observed tensidol B production (m/z = 342.0986, retention time [rt] = 7.98), which was consistent across three biological replicates. During this assessment, we also reanalyzed the NRRL3_04231 OE strain and the parental strain NRRL2270 by culturing them under the same conditions (1% maltose minimal medium) and analyzing their respective culture extracts for SM production. Our results showed clear evidence of tensidol B production by the NRRL3_04231 OE strain, with a peak produced consistently across three biological replicates matching the m/z value and rt observed for tensidol B in the A. brasiliensis culture extract. In contrast, we saw no evidence of tensidol B production within the culture extract obtained from the parental strain (Fig. 4).

The backbone enzyme (NRRL3_04226) within this cluster is a HPN consisting of a single NRPS module (C-A-T) and PKS domains KS-AT-DH-KR-ACP (Table 1). The chemical nature of tensidol A and B supports biosynthesis involving a HPN; the presence of a phenyl moiety and nitrogen-containing heterocycle implicates an NRPS adenylation domain, while the keto-acyl chain of tensidol B suggests a PKS origin. To further investigate the role of this cluster in tensidol biosynthesis, we attempted to generate a NRRL3_04226 knockout strain; however, this was unsuccessful possibly due to the proximity of this BGC to the telomeric region of chromosome IV (Fig. 2). As an alternative approach, we generated an extrachromosomal plasmid (ANEp8-4226) containing a codon-optimized gene encoding NRRL3_04226 downstream of the constitutive pyruvate kinase A promoter (PkiA). Transcriptomic analysis of the NRRL3_04231 OE strain showed that while each gene within the cluster was up-regulated, the degree to which the backbone enzyme was up-regulated (3.65 TPM in NRRL3_04231 OE; not detected in parental strain) was less than what we observed for other genes in this BGC (Table S2). As a result, we hypothesized that complementing the NRRL3_04231 OE strain with the ANEp8-4226 plasmid providing additional constitutive expression of the backbone enzyme would result in increased tensidol production. In line with our hypothesis, we observed a ∼10-fold increase in tensidol production by the NRRL3_04231 OE + ANEp8-4226 strain compared with the NRRL3_04231 OE strain, providing additional experimental support implicating this cluster in tensidol biosynthesis (Fig. 4E).

Investigating the effect of BGC partial up-regulation

Further analysis of the overexpression strains with only part of the BGC up-regulated by overexpression of an associated TF allowed us to investigate the specific role played by transporters encoded as part of the BGC. In two such strains, the genes encoding for predicted transporters showed little-to-no up-regulation compared with the parental strain, while backbone and decorating enzyme encoding genes were up-regulated (Table S2). In the NRRL3_02186 overexpression strain, the corresponding BGC was predicted to produce compounds from the fumonisin class of mycotoxins based on shared similarity with fumonisin biosynthetic genes from F. verticillioides (26). Within this BGC, the colocalized transporter NRRL3_02188 was not up-regulated in response to NRRL3_02186 overexpression as previously reported (49)). Our LC–MS analysis of the extracellular metabolites produced by this strain detected compounds corresponding to fumonisin B1/B6, B2, B4, and B5 based on matching m/z values and rt scores. We extracted and analyzed the intracellular metabolites produced by the NRRL3_02186 overexpression strain and compared the relative abundance of fumonisin B2 in the intracellular and extracellular metabolite fractions. Our results showed 2-fold enrichment of fumonisin B2 in the intracellular faction compared to the secreted fraction.

The second instance where we observed no up-regulation of a co-localized transporter was in the NRRL3_01335 overexpression strain. This BGC is uncharacterized, and its secondary metabolite products are unknown. Transcriptomic analysis revealed up-regulation of 8/9 genes within the BGC, with the lone exception being a putative ABC transporter encoded by NRRL3_01330. The extracellular metabolome produced by this strain displayed modest changes compared with the parental stain. The failure of TF^OE to elicit up-regulation of the clustered ABC transporter provides a strong rationale for why we did not observe more substantive changes in metabolite production, as secondary metabolite products of this BGC may have been retained intracellularly similar to what was observed with the fumonisin BGC.

Other instances of partial up-regulation were observed in the NRRL3_00104, NRRL3_06232, NRRL3_07461, and NRRL3_08989 overexpression strains. In each case, overexpression of the TF failed to up-regulate the corresponding backbone enzyme(s) within these BCGs. The extracellular metabolite profile of these strains lacked unambiguous signals of additional secondary metabolite production, an observation consistent with failure to activate biosynthetic enzyme expression.

BGC TFs are involved in a complex regulatory network

We observed secondary metabolite production in 28 of 43 strains where TF^OE failed to promote gene expression of its colocalized BGC. The discordance between the results of metabolomics and transcriptomics was consistent across biological replicates and prompted additional analyses of the transcriptomic data. This analysis identified examples hinting at further complexity and dynamism in the regulation of secondary metabolism. In the NRRL3_09035 overexpression strain, we identified multiple metabolomic features enriched >500-fold compared with the parental strain (Fig. 3). This set of overproduced compounds included members of the previously characterized yanuthone class of metabolites, as well as compounds for which no match was found in our existing A. niger compound database. The relative complexity of the metabolite profile produced in response to overexpression of NRRL3_09035 led us to hypothesize that this TF may activate BGCs in trans. Analysis of the transcriptome-level changes in response to overexpression of NRRL3_09035 supported this hypothesis, as multiple genes within the yanuthone BGC showed significant up-regulation (Fig. 5). Specifically, overexpression of NRRL3_09035 resulted in an 81-fold up-regulation of the gene encoding the yanuthone BGC backbone enzyme YanA (NRRL3_06291) and 152-fold up-regulation of the YanG-encoding gene, NRRL3_06290. Other genes within the yanuthone BGC were similarly up-regulated, including the colocalized TF YanR (NRRL3_06287), suggesting that overall activation of the yanuthone BGC in response to NRRL3_09035 overexpression may be mediated through specific up-regulation of YanR (Fig. 5). In this example, it appears that YanR (NRRL3_06287) may itself be under regulation by the out-of-cluster TF, NRRL3_09035.

Fig. 5.

Trans-activation of yanuthone BGC by NRRL3_09035. a) Schematic representation of the yanuthone BGC located on chromosome V. The TF NRRL3_09035 is encoded on chromosome VII. b) LogTPM of genes in the yanuthone BGC. Most clustered genes are silent (not expressed) in the parental strain but are overexpressed in both the NRRL3_06287 and NRRL3_09035 overexpression strains.

Discussion

With A. niger serving as an industrial host for the production of proteins and organic acids, there exists a wealth of information surrounding its large-scale cultivation. This makes A. niger compelling for the study of secondary metabolism; however, limited understanding of control mechanisms governing the expression of fungal BGCs has long posed a challenge to uncovering, producing, and characterizing novel secondary metabolites, which has broad applications in pharma and bioindustry. As a result, the number of characterized BGCs and corresponding SM products is a small subset of the overall biosynthetic potential encoded by fungal genomes. This is readily apparent in A. niger, where more than 75% of its BGCs remain cryptic and uncharacterized (50). In this study, we generated a collection of A. niger TF^OE strains by targeting TF genes associated with cryptic BGCs for inducible overexpression. Our subsequent analysis of these strains for phenotypic markers and changes in secreted metabolomes indicated a strong correspondence between TF^OE and SM production; however, our transcriptomic analysis highlighted the complex and dynamic nature of BGC expression regulation.

The pipeline used for overexpression strain analysis (Fig. 1) showed that phenotypic markers were a strong indicator of metabolomics changes but an imperfect proxy for gene expression. This was most apparent in our NRRL3_00205 overexpression strain, where a clear phenotype of brown pigmentation was not associated with any BGC up-regulation in our transcriptomic data. We also observed metabolomics changes in many overexpression strains that did not display any obvious phenotype through the period of observation, highlighting clear practical limitations with using phenotype as an indicator of SM production.

Overall, we registered significant changes in the secreted metabolome in response to individual TF^OE in 43 out of 58 strains produced in this study. Most of the changes corresponded to peaks and compounds that could not be reliably matched to known entities populating our A. niger compound database. These compounds are candidates for follow-up experimentation and structural characterization. Additional assays, such as bioactivity screening, could be added to the end of our analysis pipeline to aid in prioritizing which of the candidate strains/compounds to pursue for structural characterization. Of note, we were able to increase production levels of tensidol B by combining constitutive overexpression of a backbone enzyme (NRRL3_04226) in a TF (NRRL3_04231) OE strain. This approach paves the way for future experiments that incorporate mutagenesis or alternative genetic backgrounds to gain mechanistic insight and further confirm the role of BGCs in the production of specific molecules. Additional OE strains in our collection, particularly those that displayed partial up-regulation of BGCs, would likely benefit from this combinatorial approach.

Our strategy of utilizing TF^OE to promote SM production was based on the canonical mechanism of BGC regulation in fungal species, whereby a colocalized TF is responsible for activating gene expression of clustered biosynthetic genes. Our results provided an abundance of data that appear counter to this paradigm and hint at regulatory complexity within A. niger. Our RNA-seq analysis suggested that in the majority of the overexpression strains, TF^OE did not trigger up-regulation of the adjacent BGC. This was despite them being selected for transcriptomic analysis due to the presence of unique metabolomic features identified in our LC–MS analysis. In some cases, we were able to reconcile these results by identifying other nonadjacent BGCs activated in response to overexpression of a TF. The most pronounced example we observed was in the NRRL3_09035 overexpression strain, which was selected for transcriptomic analysis on account of its strong yellow-pigmented phenotype and LC–MS profile. Overexpression of NRRL3_09035 failed to up-regulate any of its colocalized biosynthetic genes; however, it resulted in strong up-regulation of genes within the yanuthone BGC which appeared to be mediated through activation of the yanuthone TF, yanR. This result is a clear example of trans-activation of BGCs by a distant TF and lends further credence to emerging hypotheses that there has been an overinterpretation of the significance of physical clustering, suggesting a more complex relationship between TFs and associated BGCs (50). One explanation often provided for the observation of physical clustering of SM biosynthetic genes in fungal genomes is that it facilitates horizontal gene transfer of BGCs. Given this, and the multiple modalities of BGC expression (e.g. complete up-regulation, partial up-regulation) we observed in response to overexpression of specific TF, a possibility emerges that BGCs that adopt the more canonical mechanism of regulation (i.e. activated by a colocalized TF) may have been more recently acquired. By the same rationale, the BGCs that have had more time to coevolve with their fungal host may then come under control of the more complex regulatory mechanisms we observed in the majority of our TF^OE strains.

Additional and well-characterized parameters influencing secondary metabolism in A. niger are also likely to have played a role in BGC activation within our overexpression strains. Master regulatory proteins, such as McrA and LaeA, have been previously shown to influence gene expression over multiple BGCs. McrA functions as a repressor protein and has been linked to more than 10 BGCs (13). A comparison between LaeA overexpression and deletion strains revealed differential expression of more than half (34 of 61 in strain FGSC A1279) of BGC-associated TFs in A. niger, highlighting the interplay between this methyltransferase and cluster-specific regulatory elements (51). The role of epigenetics has also been well established as a determinant in the tractability of activating BGCs (52). Previous studies have shown that treatment with small molecules known to modulate the epigenetic state of chromosomes led to changes in secondary metabolism and facilitated the discovery of new compounds (53). In cases where strains within our collection failed to display changes in gene expression and metabolite profiles, the epigenetic state of the chromosomal location housing the BGC may have been inaccessible to the overexpressed TF. In the context of these additional layers of BGC regulation, it is possible that overexpression of individual TFs is required but insufficient to activate BGC gene expression. Consequently, it may be justified to incorporate adjunctive approaches in the future that, in addition to TF^OE, may lead to complimentary strategies for efficiently activating BGC gene expression and facilitating novel small molecule discovery.

Materials and methods

Manual curation of BGCs

BGCs in A. niger NRRL3 were manually curated following the using the publicly available NRRL3 genome sequence (available at https://mycocosm.jgi.doe.gov/Aspni_NRRL3_1/Aspni_NRRL3_1.info.html) (54). In brief, backbone enzymes were initially identified via BLASTP searches using experimentally verified backbone enzyme sequences from other Aspergilli as queries (E-value ≥ 1 × 10⁻⁵). All hits were evaluated for acceptance into their respective categories by protein domain content using the online applications Pfam (http://pfam.xfam.org/), conserved domain database (CDD) (http://www.ncbi.nlm.nih.gov/cdd/), and InterProScan 5 (http://www.ebi.ac.uk/interpro/search/sequence-search). BGCs were defined using automated tools SMURF (20) and antiSMASH (21) following the methods described in Inglis et al. (55). These clusters were modified to incorporate our updated gene annotation data and took precedents over the original annotations including those defined by synteny.

Cloning of TF donor plasmids and ANEp8-4226

TF genes were PCR-amplified from A. niger NRRL2270 genomic DNA using Phusion DNA polymerase (New England Biolabs), and the resulting PCR products were purified via a PCR Clean and Concentrator kit (Zymo). Cloning of TF genes into the pJET, a plasmid containing 655 and 700 bp of the glucoamylase promoter and the terminator regions, respectively (23), was done via ligation-independent cloning (56). Plasmids were transformed into E. coli DH5α, purified using the Presto Mini Plasmid Kit (Geneaid), and verified via DNA sequencing.

For ANEp8-4226, the ANEp8 backbone was digested using FesI and PacI and a codon-optimized gene (TWIST DNA) was cloned into the vector using a Gibson Ultra Assembly Kit (Telesis Bio). A list of primers used in this study is given in Table S1.

Host transformation

All TF^OE strains are descendants of CSFG_7003 (NRRL2270 ΔpyrG ΔkusA), a uridine auxotroph, and NHEJ-deficient strain that allows for high-efficiency gene replacement (17). Protoplasts were prepared by incubating mycelium for 3 h at 37 °C in digestion solution (40 mg/mL VinoTaste Pro [Novozymes], 1.33 M sorbitol, 20 mM MES pH 5.8, 50 mM CaCl₂). PEG-mediated transformation was performed mostly as described in (25); briefly, protoplasts were incubated with 1 µg of ANEp8-Cas9-gRNAglaA and 10 µg of pJET/TF donor plasmid and plated on selective media. Two to three colonies from each transformation plate were selected and propagated on minimal media (MM). Genomic DNA was extracted, and the glaA locus of each candidate colony was assayed by PCR amplification using primers Fw_TF-insertion screen and Rv_TF-insertion screen. The PCR amplicon was then subjected to BglII digestion (Fig. S1) to determine whether gene replacement was successful based on the corresponding banding pattern.

Growth conditions

Validated TF^OE strains were cultured in 200 µL of either MM with 1 or 15% maltose in 96-well plates. Plates were kept stationary and incubated at 30 °C for 5 days.

Sample preparation for LC–MS analysis

From standing cultures, 75 µL of culture media was collected in 1.5-mL microfuge tubes and centrifuged at 16,000×g for 45 min to remove mycelia, spores and cellular debris. The supernatants were transferred to new tubes and an equal volume of cold methanol (−20 °C) was added for protein precipitation. Following incubation on ice for 10 min, samples were centrifuged at 16,000×g for 45 min to remove precipitated proteins. Supernatants were transferred to fresh tubes, and an equal volume of 0.1% formic acid was added. The soluble fraction was diluted 20-fold with cold 50% methanol and incubated on ice for 15 min with frequent mixing for protein precipitation and metabolite extraction. The samples were centrifuged for 45 min at 10,000×g, 4 °C to remove precipitated proteins. Methanol extracted metabolites were stored at −80 °C until LC–MS analysis was performed.

High-performance liquid chromatography–mass spectrometry analysis of metabolites

High-performance liquid chromatography–mass spectrometry (HPLC–MS) was performed as described in (23); briefly, 10 µL of sample was injected into a Kinetex C18 column coupled to an Agilent 1260 Infinity II HPLC system. Column flowthrough was delivered to an LTQ-FT mass spectrometer for electrospray ionization–MS. Ionization voltage was 4,900 V in a positive mode and 3,700 V in a negative mode, and the scan range was 100 to 1,400 m/z at a resolution of 200 m/z. Data analysis was performed using Compound Discoverer 3.1 software, and annotation was done using a custom database of 968 known A. niger metabolites.

RNA-sequencing samples and analysis of data

The whole-genome expression of A. niger parental and mutant strains was analyzed using a two-step culture, a primary culture generating fungal biomass and a second transfer-culture triggering gene induction. Fifty milliliters of Aspergillus complete medium, 2% fructose, were inoculated in 500-mL flasks with 2 × 10⁶ spores/mL and incubated 16–18 h at 30 °C, 250 rpm. From the primary cultures, wet mycelium was collected, filtered on a Buchner funnel with Miracloth (Calbiochem), and rinsed two times with Aspergillus minimum medium (57). Five milliliters (equivalent 2 g wet weight) of mycelia were transferred to 50 mL Aspergillus minimum medium, 1% maltose. The secondary cultures were incubated 2 h at 30 °C, 250 rpm. The mycelium was collected, filtered on a Buchner funnel with Miracloth (Calbiochem), press-dried with paper, flash-frozen in liquid nitrogen, and stored at −80 °C.

Frozen mycelium was ground with a mortar and a pestle into a fine powder. Total RNA was extracted as described (58) and quantified using a NanoDrop Spectrophotometer ND-1000 (NanoDrop Technologies, Inc.), and its integrity was assessed on a 2100 Bioanalyzer (Agilent Technologies). Samples were sent to the Genome Quebec Innovation Centre for analysis; briefly, samples were loaded on an Illumina cBot and the flowcell was ran on a HiSeq 4000 for 2 × 100 cycles (paired-end mode). A phiX library was used as a control and mixed with libraries at 1% level. The Illumina control software was HCS HD 3.4.0.38; the real-time analysis program was RTA v. 2.7.7. Program bcl2fastq2 v2.20 was then used to demultiplex samples and generate fastq reads.

Supplementary Material

Supplementary material is available at PNAS Nexus online.

Funding

This research was funded by the Industrial Biocatalysis Strategic Network (A.T. and A.S.) and the Discovery Grant (I.B.-G.) of the Natural Sciences and Engineering Research Council of Canada.

Author Contributions

Cameron Semper (Data curation, Formal Analysis, Investigation, Methodology, Writing—original draft, Writing—review & editing), Thi Thanh My Pham (Investigation), Shane Ram (Investigation), Sylvester Palys (Investigation), Gregory Evdokias (Investigation), Jean-Paul Ouedraogo (Investigation), Marie-Claude Moisan (Investigation), Nicholas Geoffrion (Data curation), Ian Reid (Data curation), Marcos Di Falco (Data curation), Zachary Bailey (Investigation), Adrian Tsang (Funding acquisition, Conceptualization, Supervision, Writing—review & editing), Isabelle Benoit-Gelber (Funding acquisition, Investigation, Methodology, Supervision, Validation, Writing—review & editing), and Alexei Savchenko (Conceptualization, Funding acquisition, Supervision, Validation, Writing—review & editing).

Data Availability

RNA-sequencing data have been deposited in the National Center for Biotechnology Information database (NCBI accession no. PRJNA1105038. https://www.ncbi.nlm.nih.gov/search/all/?term=PRJNA1105038).