Transcriptome Insights into the Effect of Sterigmatocystin on Growth and Aflatoxin B1 Production by Aspergillus Flavus


 Aspergillus flavus (A. flavus) is an important fungus that produces aflatoxins, of which aflatoxin B1 (AFB1) is the most toxic and prone to contaminating food. This leads to economic losses in agricultural products and human health risks. Sterigmatocystin (STC) is a precursor in the metabolic biosynthesis of AFB1. In this study growth of A. flavus was not affected by the presence of STC, but AFB1 production was inhibited at certain concentrations. To uncover the mechanism, transcriptomic profiles of A. flavus in the presence or absence of STC were evaluated in PDB. A total of 3377 differentially-expressed genes were identified; 1182 were up-regulated and 2195 were down-regulated. GO function and KEGG pathway enrichment analysis indicated that these genes were mainly involved in the organization and biosynthesis of cellular components, organelle part, organelle division, macromolecular compound functions; the main pathway was valine, leucine and isoleucine synthesis and aflatoxin synthesis. The clustered genes responsible for AFB1 biosynthesis were down-regulated to different degrees; norB expression was completely suppressed in the experimental group. This study provides new insights into control of A. flavus and the mechanisms regulating mycotoxin production.


Introduction
A atoxin B 1 (AFB 1 ) is one of the most important mycotoxins in the world due to its frequent detection at high concentrations as a contaminant in human food and animal feed [1][2][3] . AFB 1 is a polyketide compound synthesized by secondary metabolic pathways in Aspergillus. AFB 1 can cause pathophysiological changes in some organisms (e.g. reduced growth rate, disturbed gastrointestinal tract, silenced immune responses and malnutrition); while also inducing a variety of histopathological manifestations in hepatocytes (e.g. proliferation of the bile duct and fatty degeneration of the hepatocytes) [4][5][6][7] . AFB 1 is known to induce hepatocellular carcinoma in various animals including rodents, poultry, non-human primates and shes [8][9] . For these reasons, AFB 1 is classi ed by the International Agency for Research on Cancer (IARC) as a group 1 compound 10 .
Another important mycotoxin is sterigmatocystin (STC) which is produced mainly by A. nidulans and A. versicolor, but also some other species belonging to the genera Aschersonia, Botryotrichum, Emericella and Chaetomium [11][12] . STC is a potential carcinogen, mutagen and teratogen and was categorized by IARC as a group 2 compound [13]. Interestingly, STC shares its biosynthetic pathway with AFB 1 , and the relative concentrations of AFB 1 and STC in contaminated foods depends on the invasive species.
Aspergillus versicolor and A. nidulans are not able to metabolize STC into O-methylsterigmatocystin, which is the the direct precursor of AFB 1 . As a result, food and feed infested by these fungi often contain high concentrations of STC. In contrast, in food and feed invaded by A. avus and A. parasiticus, which are able to metabolize STC, concentrations of STC are generally low [14][15][16]11 .
The simultaneous occurrence of multiple mycotoxins in an individual product is a common phenomenon, and co-occurence of AFB 1 and STC has been reported in food commodities (De Saeger, personal communication, 2012) [17][18][19][20] . Exposure to multiple mycotoxins may lead to synergistic, additive or antagonistic toxic effects and the toxicity of mycotoxin combinations cannot always be predicted based on their individual toxicities [21][22][23][24] . Mycotoxins with similar modes of action would generally be expected to have the least additive effects 25 . The mechanisms of AFB 1 toxicity is through DNA modi cation, metabolic activation, cell death/transformation and cell deregulation [26][27][28][29] . In contrast, the toxicity of STC is through DNA adducts, inhibition of the cell cycle and mitosis, increases in ROS formation and lipid peroxidation [30][31][32][33] . Thus, there are similarities and differences in the mode of action of AFB 1 and STC.
Methods to remove mycotoxins from contaminated food, or degrade them into less toxic or non-toxic compounds are required.
Many methods for degradation of AFB 1 have been reported, including adsorption, physical and chemical reduction [34][35][36][37] . In addition, AFB 1 is degraded by microorganisms including bacteria, yeasts and moulds; this has been studied extensively in recent years 38-42 .
To our knowledge, there is no information in the literature about interactions between the presence of STC in foods and AFB 1 production. In the present work the effect of STC on A. avus growth and AFB 1 biosynthesis were studied in potato dextrose broth

Results
Effect of STC on mycelial growth and AFB 1 production.
The effect of STC, at ve concentrations (0.02, 0.1, 2.0, 10.0, 50.0 µg/mL), on mycelial growth of A. avus and AFB 1 production in PDB medium over 11 days are shown in Table 1. The dry weight of mycelia increasing with increasing time in culture. In the control A. avus grew well in PDB medium with sustained growth over the 11 days. When STC was added to the medium, mycelial dry weights at the highest concentration of STC were signi cantly different to the control and the other treatments (p>0.05) after 3 and 4 days growth (except for 10.0 µg/mL in 3 days); From the 5th day, there was no signi cant difference in A. avus dry weights amongst any of the STC concentrations and the control except in the 10 µg/mL treatment (p<0.05). This indicates that A. avus growth was not affected by the presence of STC, even at high concentrations. µg/L and 804.90 µg/L in the same four STC concentrations, and 1145.67 µg/L in the control. Hence, a certain concentration of STC inhibited AFB 1 production, and at an STC concentration of 0.1 µg/mL AFB 1 production was signi cantly different than the control (p>0.05). It is noteworthy that at an STC concentration of 50 mg/L AFB 1 production was not signi cantly different to the control (p 0.05) between days 4 and 7 of growth. Reference sequence alignment analysis.
Comparative analysis of the genome of A. avus was done using clean reads after ltering with TopHat 2.0 software. The results showed that over 85% of reads were uniquely mapped to the A. avus genome and multiple mapped reads or fragments coincided with the standard of half-point ratio below 10% in multiple localization-sequencing. This meant that the reference genome was appropriate and three was no pollution in the experiment.
Gene expression level analysis.
Gene expression level is re ected in transcription abundance; the lower the transcription abundance, the lower the level of gene expression. The overall transcriptional activity of genes in our data were quanti ed by calculating the number of reads per kilobase of exon per million mapped reads (FPKM); it is generally recognized that an FPKM 1 means that the gene is expressed. Gene expression levels are shown in Table 3. Table 3 Proportion of genes at different expression levels Identi cation and analysis of DEGs Differential expression analysis.
To identify differences in molecular responses (based on read count) between experimental groups and the control group, we identi ed 3377 differentially transcribed (FDR 0.05, |log 2 FC|≥1) genes, of which 1182 genes (35.00%) were up-regulated and 2195 (65.00%) genes were down-regulated.
GO and KEGG analysis of DEGs.
Functional assignments were de ned by Gene Ontology (GO) terms (http://www.genontology.org), which provided a broad functional classi cation of genes and gene products for various biological processes (BP), cellular components (CC) and molecular functions (MF). GO functional enrichment analysis revealed that: 13 MF terms were enriched, mainly involved with oxidoreductase activity, catalytic activity, structural constituent of ribosome and monooxygenase activity; 18 CC terms were enriched including intracellular organelle part, intracellular part, cell part and macromolecular complexes; 65 BP terms were enriched including cellular component organization or biogenesis, oxidation-reduction processes, single-organism metabolic processes, single-organism processes and cellular protein metabolic processes (Table 4). were involved in 110 pathways. The top ten down-and up-regulated genes enriched in KEGG is shown in Table 5. For downregulated genes, the most enriched were involved in a atoxin biosynthesis, followed by starch and sucrose metabolism and tyrosine metabolism; this explains why gene expression involved in a atoxins synthesis and glucose metabolic pathway was suppressed in the experiment group. For up-regulated genes, valine, leucine and isoleucine biosynthesis was the most enriched, followed by alanine, aspartate and glutamate metabolism; this accounts for why genes participating in amino acid metabolism in the experimental groups were highly expressed. KEGG metabolic pathway analysis showed that enrichment of valine, leucine and isoleucine biosynthesis and a atoxin biosynthesis was signi cant at p≤0.05. difference in expression of other genes in this cluster. Similarly, the genes AFLA_082220, AFLA_082230 and AFLA_082250 in the 27 # gene cluster and AFLA_101690 and AFLA_101710 in the 33 # gene cluster were all up-regulated to varying degrees. Amongst the up-regulated genes, AFLA_112890 and AFLA_066740 encode major facilitator superfamily transporters (MFS), which have an important role in transfer of some metabolites. Furthermore, genes involved in amino acid synthesis were differentially expressed to varying degrees; for example, AFLA_005510, AFLA_041550, AFLA_083270 and AFLA_126970, which encode amino acid permease, cysteine β-lyase, GABA permease and arginine permease, respectively, were up-regulated, but AFLA_038620 (encoding branched amino acid transferase) and AFLA_062910 (encoding speci c proline permease) were down-regulated.
AFB 1 synthesis genes were located on the 54 # gene cluster. In this study, 30 genes were all down-regulated to varying degrees; norB was completely repressed in the experimental group (Table 6). However, there was no difference in expression of the global regulatory genes of secondary metabolite, laeA and veA,. Also, brlA, which is involved with growth was down-regulated. Table 6 Analysis of genes expressed in the biosynthesis of a atoxins Aspergillus avus reproduces asexually depending on conidia production for growth and development, which is precisely controlled by multiple genes. The genes wetA, brlA and abaA are the main regulatory proteins involved in the sporulation process and play important roles in the different stages of conidia production. Activation of the brlA gene is a key step in conidia production by Aspergillus 44 ; knockout of brlA makes Aspergillus isolates form quill-like structures with slender stems that cannot produce conidia, while over-expression interrupts isolate growth 45 . Generally, a atoxin synthesis is always associated with slow growth 46 . In our study, brlA gene expression was down-regulated meaning that growth of A. avus remained active, leading to the synthesis of results of the phenotypic study, we speculate that STC cannot affect the growth of A. avus. Lin et al (2013) found that 5azacytidine induced up-regulation of brlA, which blocked sporulation, resulted in a uffy phenotype, and in uenced AFB 1 synthesis 47 .
Secondary metabolic pathways in fungi are relatively complex, involving polyketone synthase, epoxide hydrolase, methyl transferase, reductase, dehydrogenase, cytochrome P450 monooxygenase and fatty acid synthase. However, one type of enzyme can be produced as a result of more than one gene, and it is di cult to determine whether a gene is involved in secondary metabolism. Using metabolic analysis tools and SMURF software, Georgianna et al., (2010) found 55 secondary metabolic pathways in A. avus, including 27 NRPS, 22 PKS, ciproanilic acid synthesis, spore pigment synthesis and a atoxin synthesis metabolic pathways that perform speci c biological functions [48][49] . A R is a transcriptional regulatory gene for a atoxin synthesis and is responsible for activation of almost all structural genes. As the transcription factor auxiliary gene, A S cooperates with a R to prevent inhibition and guarantee a atoxin synthesis. LaeA and veA are global regulatory factors involved in regulation of a atoxin synthesis and were down-regulated in our study.
A atoxins are highly oxidative metabolic products and oxidative stress is a necessary condition for their formation 50 . Transcriptome analysis demonstrated that Cu-Zn superoxide dismutase expression was raised, which is consistent with previous research 43 . Thirty genes involved in a atoxin synthesis, were all down-regulated demonstrating that STC directly inhibits the genes encoding enzyme in this gene cluster which inhibits a atoxin production. Some antioxidant enzymes encoded by gene updated in the presence of STC can also inhibit synthesis of a atoxins.
The role of amino acid metabolism in a atoxin synthesis is complex. Some amino acids can be used as carbon or nitrogen sources in growth and a atoxin synthesis. In A. avus, phenylalanine, tyrosine, tryptophan, proline and arginine can be used in this way for a atoxin synthesis 51 . Arginine is necessary for a atoxin synthesis in A. parasiticus, and can replace aspartate and alanine [52][53] . KEGG enrichment analysis found that, in 67 metabolic pathways, BCAAs metabolic activity was associated with biosynthesis of a atoxins, and other basic ammonia acids and acidic amino acid metabolism mainly related to the fungal growth.
Eighteen ribosome biosynthesis genes were up-regulated showing that STC may promote the growth of A. avus to a certain extent, but inhibited the synthesis of AFs. This result is consistent with previous studies 54 .
STC is a precursor of a atoxin synthesis and, theoretically, an increase in STC should enhance the ability of isolates to synthesize a atoxins. However, cell metabolism and microbial secondary metabolism are sophisticated processes. Although precursor substances can inhibit metabolic production, this has not been reported before for a atoxin synthesis; there is a large body of research on synthesis of antibiotics for example. In general, exogenous precursor added during synthesis of antibiotics can control synthetic direction and increase antibiotics production. Epothilone is a type of polyketide secondary metabolite, and when precursors are added they have an inhibitory effect on epothilone synthesis 55 . Similarly, FK506 output is signi cantly improved during fermentation in the presence of precursors, but only at low concentrations 56 . A atoxin synthesis in A. avus CGMCC declined in the presence of STC this study. However, under the same conditions, a atoxin production by other isolates increased. This con rms that a atoxin synthesis is isolate speci c, and can vary depending on the nutrient medium, external environment and the degree of tolerance to particular substances 57 .
Synthesis of secondary metabolites is closely related to primary metabolism because cell energy, precursors and co-factors can potentially limit secondary metabolism. Many enzymes involved in a atoxin biosynthesis and catalytic reactions are NADPH dependent. Glucose may control a atoxin biosynthesis by NADPH generated carbon metabolism repression and the tricarboxylic acid cycle. Therefore, NADPH/NADP + levels in cells may also affect a atoxin synthesis 58-59 . Lipid metabolism is also closely associated with a atoxin synthesis. NADPH/NADP determines whether acetyl-CoA enters fatty metabolism or a atoxin anabolism. It is possible that a high proportion of NADPH/NADP enables acetyl-CoA to enter fatty metabolic pathways, while low levels encourage its entry into other pathways 60 . The pentose phosphate pathway is the main source of intracellular NADPH in organisms, some important gene expressed signi cant difference may be the other possible in uencing factors in a atoxin synthesis.
In order to discuss the in uence of STC on secondary metabolite synthesis in A. avus, we used transcriptome sequencing analysis to identify 3377 differentially expressed genes of which 1182 genes were up-regulated and 2195 genes were down-regulated. GO function and KEGG pathway enrichment analysis indicated these genes mainly participated in cellular component organization and biosynthesis of components, organelle part, organelle division, macromolecular compound functions, and the main pathway was valine, leucine and isoleucine synthesis and a atoxin synthesis. We also showed that STC had least in uence on 55 secondary metabolic gene cluster, but that 30 genes on the a atoxin synthesis gene cluster were expressed to varying degrees; norB was completely suppressed in the experimental group, suggesting that STC probably increases oxidoreduction enzyme activity, facilitating growth of isolates, and improving branched chain amino acid biosynthesis, thus inhibiting synthesis of a atoxins.

Materials And Methods
Apparatus. Preparation of spore inoculum. Determinations of fungal dry weights and AFB 1 concentration.
Fungal dry weights of A. avus in each treatment and at each sampling time were determined by ltering out the mycelium from each ask using Whatman No 4 lter paper, washing in sterilized water, and then freeze-drying for 24 h at 65℃ before weighing.
The ltrate from each ask was used for AFB 1 analysis. Quanti cation of AFB 1 was done following the method described by Guo et al., with some modi cation 61 . For extraction of AFB 1 from media, 2 mL ltrate was transferred by pipette to a centrifuge tube before centrifugation at 11000 rpm for 10 min. Then 1.5 mL of the supernatant and 4.5 mL acetonitrile aqueous solution containing 1% formic acid were placed in a 10 mL centrifuge tube, vortexed for 1 min and then dried with nitrogen. To each centrifuge tube, 0.1 mg PSA and 1.5 mL acetonitrile aqueous solution (3:7, v:v) containing 0.1% formic acid were added to achieve a constant volume; the mixture was vortexed for 1 min and then subjected to ultrasound in an ultrasonic generator for 5 min prior to centrifugation at 10000 rpm for 5 min. Then 1 mL of the supernatant was ltered through a 0.22 µm nylon lter before analysis using HPLC-MS/MS.
RNA extraction and quality testing.
Total RNA was extracted from cultivated PDB media containing 10 µg/mL STC using TRIzol® reagent according to the manufacturer's instructions (Invitrogen) and genomic DNA was removed using DNase I (TaKara). Then RNA quality was determined using a 2100 Bioanalyser (Agilent) and quanti ed using the ND-2000 (NanoDrop Technologies).

RNA-seq library construction and sequencing.
An RNA-seq transcriptome library was prepared using the TruSeq TM RNA sample preparation kit from Illumina (San Diego, CA) with 5 μg of total RNA. Messenger RNA was isolated according to the polyA selection method using oligo (dT) beads and then fragmented by fragmentation buffer rstly. Secondly double-stranded cDNA was synthesized using a SuperScript double-stranded cDNA synthesis kit (Invitrogen, CA) with random hexamer primers (Illumina). The synthesized cDNA was subjected to end-repair, phosphorylation and 'A' base addition according to Illumina's library construction protocol. Libraries were size selected for cDNA target fragments of 200-300 bp on 2% Low Range Ultra Agarose followed by PCR ampli cation using Phusion DNA polymerase (NEB) and 15 PCR cycles. After quanti cation using TBS380, the paired-end RNA-seq sequencing library was sequenced using the Illumina HiSeq xten (2 × 150bp read length).
Mapping reads and sequence assembly.
The raw paired-end reads were trimmed and quality controlled by SeqPrep (https://github.com/jstjohn/SeqPrep) and Sickle (https://github.com/najoshi/sickle) with default parameters. Clean reads were separately aligned to reference genomes using the orientation mode of TopHat (http://tophat.cbcb.umd.edu/version 2.0.0) software 62 .The mapping criteria of bowtie were as follows: sequencing reads should be uniquely matched to the genome allowing up to two mismatches, without insertions or deletions. The regions of genes were expanded following depths of sites and the operon obtained. In addition, the whole genome was split into multiple 15 kbp windows that shared 5 kbp. Newly transcribed regions were de ned as having more than two consecutive windows without overlapped regions of the gene, where at least two reads mapped per window in the same orientation.
Differential expression analysis and functional enrichment.
To identify DEGs (differentially-expressed genes) amongst paired samples, the expression level of each transcript was calculated according to the fragments per kilobase of exon per million mapped reads (FRKM) method. RSEM (http://deweylab.biostat.wisc.edu/rsem/) was used to quantify gene abundances 63 . The R statistical package software EdgeR (Empirical analysis of Digital Gene Expression in R [http://www.bioconductor.org/packages/2.12/bioc/html/edgeR.html]) was used for differential expression analysis 64 . In addition, functional enrichment analysis including GO and KEGG were done to identify which DEGs and metabolic pathways were signi cantly enriched in GO terms compared with the whole-transcriptome background (using the Bonferroni-corrected P-value ≤0.05 to determine signi cance). GO functional enrichment and KEGG pathway analysis were achieved using Goatools (https://github.com/tanghaibao/Goatools) and KOBAS (http://kobas.cbi.pku.edu.cn/home.do) 65 .

Data avalilability
All data generated or analysed during this study are included in this article.