Transcriptome and secretome analysis of Aspergillus fumigatus in the presence of sugarcane bagasse

Strains, media, and growth conditions

A. fumigatus AF293, gently donated by the Prof. Dr. Sérgio Akira Uyemura (University of São Paulo, BR), was grown on YAG medium (2% (w/v) dextrose, 0.5% (w/v) yeast extract, 0.1% (v/v) trace elements and 1.8% (w/v) agar) at 37 °C for two days. Spores were harvested and inoculated to a final concentration of 1 × 10⁸ per 50 mL of YNB culture with 1% (w/v) fructose as the carbon source at 37 °C for 16 h (h) in a rotary shaker with agitation at 200 rpm. Afterward, the mycelia were transferred to 1% (w/v) SEB (47.5% cellulose; 9.0% hemicellulose and 34.3% lignin) or 1% (w/v) fructose as the carbon source for 3, 6, 12, 18, 24, 48 and 72 h. Fructose was used as a control in all experimental conditions [26]. Mycelia were harvested by filtration through Whatman grade 1 filters (GE Healthcare, Grandview Blvd. Waukesha, WI, USA), washed once with sterile cool water and kept at − 80 °C until RNA extraction. Supernatants were collected to measure enzymatic activity and determine the secretome. All the experiments described below were performed in three biological replicates.

Enzymatic activity assays

Specific xylanase (endo-1,4-β-xylanase) and cellulose (endo-1,4-β-glucanase) activities were performed with Azo-Xylan (Birchwood) and Azo-CM-Cellulose (both from Megazyme International, Bray, Ireland) as substrates, respectively, according to the manufacturer’s protocols. The enzymatic activities are represented as U mL^− 1. All the reactions were performed in triplicate. The software Mega-Calc™ (Megazyme International) was used to determine the enzymatic activities.

Enzymatic activities were also measured by the dinitrosalicylic acid (DNS) assay [27]. Cellulase activities were measured with β-glucan and low-viscosity carboxymethylcellulose (CMC) as substrates, and xylanase activities were measured with the xyloglucan. Briefly, 20 μL of the supernatant from the samples grown in presence of 1% SEB for 24, 48 and 72 h were mixed with 30 μL of sodium acetate buffer 100 mM (pH 5.5) and 50 μL of substrate at 0.5% (w/v) final concentration to achieve a final volume of 100 μL. The reactions were incubated at 40 °C for 5 min for β-glucan and for 10 min for xyloglucan and CMC substrates. The reaction was stopped by adding 100 μL of DNS. All the reactions were performed in triplicate. The calculation of enzyme activities was based on a corresponding standard containing glucose. One unit (U) of enzymatic activity was defined as the amount of enzyme needed to liberate 1 μmol of reducing sugars per minute.

RNA isolation and cDNA synthesis

Fungal biomass was harvested at different times from SEB or fructose culture conditions, and mycelia were ground in liquid nitrogen using a mortar and pestle. Total RNA was purified by using the “Direct-zol™ RNA MiniPrep” kit according to the manufacturer’s instructions (Zymo Research, Irvine, CA, EUA) using the on-column DNAse treatment. RNA integrity was confirmed with a bioanalyzer by using the “Agilent RNA 6000 Nano” kit (Agilent Technologies, Santa Clara, CA, EUA) and the “Plant Total RNA Nano” protocol. RNA was quantified on a Qubit® 2.0 fluorimeter (Thermo Fisher Scientific, Waltham, MS, EUA) with the Qubit® RNA BR Assay kit (Thermo Fisher Scientific, Waltham, MS, EUA). cDNA was synthesized from 1 μg of mRNA using SuperScript® II Reverse Transcriptase (Invitrogen, Carlsbad, CA, EUA).

Library preparation and RNA sequencing

RNA sequencing libraries were prepared using the “TruSeq Stranded mRNA HT Sample Prep” kit (Illumina, San Diego, CA, EUA), mRNA enrichment was performed using magnetic beads coupled with oligo (dT). Sequencing was carried out in the HiSeq 2500 system (Illumina, San Diego, CA, EUA) at the NGS facility located at the Brazilian Bioethanol Science and Technology Laboratory (CTBE), Campinas, SP, Brazil.

Bioinformatic analysis of RNA-Seq data

FastQC [28] was used to check the quality of the sequencing reads visually. Removal of the remaining adapter sequences and quality trimming with a sliding window of size 4, minimum quality of 20, and length filtering (to keep reads with a length of at least 60 bp) were performed with Trimmomatic v0.32, [29]. Clean reads were screened against a database of ribosomal RNA with the aid of SortMeRNA [30]. High-quality reads were mapped in a strand-specific manner by using TopHat2 [31] against the genome sequence of A. fumigatus Af293 obtained from ASPGD [32, 33]. The number of exon-exon junctions at different levels of read subsampling was employed to confirm sequencing saturation with RSeQC [34]. Mapping of the reads to the features of the exons were summarized at the gene level by using the function featureCounts from the Rsubread v1.12.6 package [35] in R v3.0.2 [36] and the annotation file in GFF3 format from ASPGD. Differential gene expression was analyzed with edgeR [37] in R [36]. Briefly, genes with at least one CPM (counts per million) in at least three samples were kept for analysis, which was equivalent to removing genes with low and noisy expression. The expression values were normalized by the trimmed mean of M-values (TMM) method to account for differences in the composition of RNA [38]. After the dispersion was estimated and the biological coefficient of variation was computed, the differentially expressed genes were called by fitting a negative binomial model with generalized linear models (GLS) that included factors for the TMM and the dispersion estimates [37]. A likelihood ratio test was performed to provide a p-value for differential expression. The p-values were adjusted for multiple testing by the method of Benjamini-Hochberg, to control the false discovery rate (FDR) [39]. The full R script used for the analysis and the raw count matrix are available in Additional file 1: Figure S1 and Additional file 2: Table S1, respectively. Genes with FDR values lower than 0.05 and log2-fold changes greater than 1.0 or lower than − 1.0, i.e., a difference of twice the expression level in either direction, were considered differentially expressed.

Supernatant analysis by SDS-PAGE

Supernatants (50 mL) from SEB or fructose culture conditions were lyophilized until completely dry and re-suspended in 2 mL of buffer (Tris-HCl 50 mM, pH 6.8; 1 mM DTT; and 1 mM protease inhibitor), and 15 μL was separated by 10% SDS-PAGE (110 V, 90 min). The proteins were visualized by staining with 0.1% Coomassie Brilliant Blue R250 (w/v), which was followed by destaining with 45% methanol and 10% acetic acid solution (v/v). The protein concentration was determined by Bradford’s Assay (Bio-Rad Protein Assay Hercules, CA, EUA) [40]. Prior to mass spectrometry, all the bands from the SDS-PAGE gels were manually excised, reduced, alkylated, digested with trypsin, and purified (Promega, Madison, WI, EUA - V5111) according to a previously described method [41].

Identification of proteins by coupled system of the LC-MS/MS type

Peptides were sequenced on a Synapt G2 HDMS (Waters, Milfords, MS, EUA) mass spectrometer coupled to a UPLC NanoAcquity system with 1D technology (Waters, Milfords, MS, EUA) and captured by a C18 Symmetry column (5 μm, 180 μm × 20 mm) (Waters, Milfords, MS, EUA). The peptides were separated by using a 2–90% acetonitrile gradient in 0.1% formic acid and an HSS T3 analytical column (1.8 μm, 75 μm × 100 mm) (Waters, Milfords, MS, EUA) with a flow of 300 μL min^− 1 for 120 min. The data were acquired on a Waters Synapt G2S Q-TOF mass spectrometer equipped with a NanoLockSpray (Waters, Milfords, MS, EUA). The experiments were performed in the HDMSE mode (data-independent analysis). The mass spectra were processed with the ProteinLynxGlobalServer (PLGS) software version 3.1. The proteins were identified by comparison to the Aspergillus UNIPROT database (207,966 proteins) [42]. The defined parameters were automatic tolerance for precursors and ion products, minimum of three corresponding ion fragments per peptide, minimum of seven corresponding ion fragments per protein, trypsin missed cleavage, carbamidomethylation as a fixed modification, oxidation of methionine as a variable modification, and 4% FDR peptide.

Protein analysis

Protein sequences were analyzed with the BLAST (basic local alignment search tool) software (http:ncbi.nlm.nih.gov/Blast.cgi). The subcellular localization of proteins was predicted by YLoc (interpretable subcellular localization prediction) (abi.inf.uni-tuebingen.de/Services/YLoc/webloc.cgi) [43], and the presence of signals due to peptides of the secreted proteins was predicted by SignalP v.4.0 (http://www.cbs.dtu.dk/services?SignalP/) [44]. Additionally, Secretome Pv2.0 (http://www.cbs.dtu.dk/services/SecretomeP/) was used to define the proteins that were secreted by the non-classic pathway [45].

For CAZy enzyme identification, the proteins in the secretome were screened with a library of hidden Markov models by using HMMER3 [46] of carbohydrate-active enzymes obtained from dbCAN [47]. Hits were considered positive on the basis of the dbCAN recommendations.

qRT-PCR analysis

After RNA-Seq analysis, 4 DEGs (Differentially Expressed Genes) were selected, including sugar transporters and CAZymes, for qRT-PCR analysis. RNA was extracted and purified as previously described. cDNA was synthesized from 5 μg of RNA using SuperScript® II Reverse Transcriptase (Invitrogen, Carlsbad, CA, USA). Quantitative PCR (qPCR) analyses were performed according to Semighini et al. [48]. The abundance of the respective mRNAs was normalized using β-tubulin probes. The primers for the investigated genes are listed in Additional file 3: Table S2.

Functional enrichment

Genes identified as differentially expressed were analyzed by FunCat functional enrichment [49]. The CAZy proteins from the secretome were classified according to the GO-Slim classifications from the AspGD based on the ontology “Molecular function” (GO Categorization Slim Mapper) [50].

Venn diagrams

The area-proportional Venn diagrams were drawn based on images generated with free online software [51].

Enzymatic analysis

To evaluate the activity of enzymes produced by A. fumigatus in the presence of sugarcane bagasse, we performed enzymatic assays using xyloglucan, β-glucan, and CMC as substrates. Specific endo-1,4-β-xylanase and endo-1,4- β-glucanase activities, were also investigated using Azo-Xylan (Birchwood) and Azo-CM-Cellulose (both from Megazyme International, Bray, Ireland), respectively. The enzymes from supernatants derived from A. fumigatus cultures were capable of hydrolyzing cellulose (CMC) and hemicelluloses (β-glucan and xylan), but no activities were detected in xyloglucans (Fig. 1a-b).

We observed that the activities depended on duration of growth, with maximum activities detected within three days, except for endo-1,4-β-D-xylanase, with peak activity after two days culture and no activities detected prior to 24 h of incubation. Maximum activities of cellulases (0.032 U mL^− 1), endo-1,4-β--xylanase (10.82 U mL^− 1) and endo-1,3-β glucanases (0.77 U mL^− 1) were detected for A. fumigatus while in A. niger (0.002 U mL^− 1; 2.3 U mL^− 1 and 0.4 U mL^− 1, respectively) and T. reesei RUT-C30 (0.0039 U mL^− 1, 0.4 U mL^− 1 and 0.1 U mL^− 1, respectively) are described on the same biomass [4]. These results indicate that A. fumigatus is an excellent producer of an arsenal of hydrolytic enzymes, with activities superior to the hypercellulolytic strain T reesei RUT30-C.

To gain more insight into the hydrolytic enzymes of A. fumigatus specific for sugarcane bagasse breakdown, we selected the cultivation of 24 h (when we detected enzyme activity) because we were interested in the initial process of SEB breakdown. In this time we determinate the transcriptome and the secretome responses of this strain.

Analysis of the transcriptome of A. fumigatus under the influence of sugarcane bagasse as the substrate

To identify potential new genes involved in SEB breakdown, we analyzed the transcriptome by RNA-Seq after 24 h of cultivation. After RNA sequencing, each sample generated approximately 14 to 16 million paired-end reads. RNAseq data were analysed by comparing the mycelium grown on SEB and that grown on fructose. We observed 2227 genes differentially expressed (FDR < 0.05, |log2FC| > 1) in SEB where 1181 were upregulated, while 1045 were upregulated in fructose conditions (downregulated in SEB) (Additional file 4: Table S3). Gene ontology (GO) and the functional catalogue (FunCat) classified the differentially expressed genes functionally in 18 different enriched categories [40]. Two significant categories among upregulated genes were Metabolic Processes (GO:0008152) and Protein Synthesis (GO:0006412) and in downregulated genes Metabolic Processes (GO:0008152) and Energy (GO:0006112) (Additional file 5: Table S4). Genes related to the regulation of C-compound and carbohydrate metabolic processes represent the two GO terms commonly enriched for both up- and downregulated genes in the SEB condition, including genes encoding carbohydrate-active enzymes (CAZymes) and transporters. Given the importance of CAZymes for the degradation of biomass, we directed our efforts toward a better understanding of the transcriptional profile of these enzymes and sugar transporters.

We found 197 differentially expressed CAZyme genes classified based on the CAZy database (http://www.cazy.org) [52]. Concerning the 566 CAZyme genes predicted in the A. fumigatus AF293 genome categorized into the different classes (247 GHs, 105 GTs, 96 CEs, 59 AAs, 15 PLs, and 44 CBMs) (Additional file 6: Table S5), we concluded that 35% of CAZyme genes were differentially expressed in our data, which highlights the potential that a wide spectrum of hydrolytic enzymes were produced. However, glycosyl transferases appeared only in a very small percentage (~ 1%), suggesting a secondary role in polysaccharide degradation (Fig. 2a).

Among the 197 CAZymes differentially expressed, 135 genes were upregulated in SEB and 62 were downregulated in SEB (upregulated in fructose), classified into 67 and 41 families, respectively (Additional file 4: Table S3). The classes of upregulated genes were 40% glycoside hydrolases (GH), 10% carbohydrate esterases (CE), 7% carbohydrate-binding modules (CBM), 5% glycosyltransferases (GT), 5% auxiliary activities (AA) and 2% polysaccharide lyases (PL), while downregulated genes represented 11% GHs, 6% CEs, 1% CBMs, 8% GTs, 4% AAs, and 1% PLs.

For plant biomass degradation, many enzymes working synergistically are required for efficiency hydrolysis. For cellulose degradation, endoglucanases (EGs) catalyze the hydrolysis at random positions in less crystalline regions; cellobiohydrolases (CBHs) act on the reducing and non-reducing ends of the chains, releasing cellobiose, which is cleaved into glucose by β-glucosidases [53]. We observed a synergistic upregulation of endoglucanases (GH5, GH12, GH16 and CBM1), cellobiohydrolases (GH6 and GH7) and β-glycosidases (GH1 and GH3). In addition to the cellulose degradation enzymes, 32 genes involved in xylan hydrolysis were also upregulated, e.g., endoxylanases (GH10, GH11 and CBM1), xylosidases (GH3 and GH43) and acetylxylan esterases (CBM1, CE2, CE16) (Fig. 2b).

To validate RNAseq data and get additional information about the expression over time, we have performed qRT-PCR for 4 DEGs that encode enzymes essential to biomass degradation, Afu4g07850 (LPMO), Afu1g14710 (β-glucosidase), Afu6g11610 (1,4-β-D-glucan-cellobiohydrolase) and Afu3g02090 (β-xylosidase), during 3, 6, 12, 18 and 24 h of cultivation in SEB and fructose. The expression profiles of these genes behaved in different ways: Afu1g14710 and Afu3g02090 genes were strongly induced at the beginning (6 h) of the growth in SEB, and their expression decreased after 6 h, while Afu4g07850 had an increasing gene expression during the time course, and Afu6g11610 increased at 24 h (Fig. 3).

Sugar transporters identified during RNA sequencing

Approximately 106 genes encoding sugar transporters have been reported in the Aspergillus genome, and only 88 genes were described as encoding sugar transporters in A. fumigatus strain Z5, which are distributed among the SP, FHS, SHS, and GPH families (the SP family includes 79 genes) [48]. Additionally, the genomes of filamentous fungi also encode large numbers of major facilitator superfamily (MFS) transporters. Among them, 25 transporters were differentially expressed on SEB, classified as encoding MFS hexose transporter, MFS and sugar transporter, UDP-Glc/Gal endoplasmic reticulum nucleotide sugar transporter, nucleotide sugar transporter, hexose transporter protein, high affinity glucose/hexose transporter, MFS glucose transporter, MFS lactose transporter, MFS maltose transporter, and xylose transporter (Fig. 4a).

Orthologous genes in Aspergillus and non-Aspergillus species were identified by sequence analysis according to Aspergillus Genome Database (AspGD; http://www.aspgd.org) [32]. Three orthologous genes encoding possible putative xylose transporters (Afu1g03530, Afu4g14610, and Afu6g14442) and five related to cellobiose transporters (Afu3g01670, Afu6g14500, Afu6g14560, Afu7g05100, and Afu8g04480) have been identified in N. crassa, A. oryzae, A. niger, and A. nidulans (Additional file 7: Table S6).

To analyze the potential xylose transporters, we selected Afu1g03530 (log₂FC = 3.43) and Afu6g14442 (log₂FC = 4.6), which are orthologous to the xtrD xylose transporter of A. nidulans (An0250) [54] and show high similarity to transporters in other fungi, with conserved regions among different species [54–69]. To characterize the expression profile, A. fumigatus was grown in 1% xylose and 1% fructose as a carbon source for the time course (3, 6, 12, 18, 24, and 48 h). Afu6g14442 gene expression was highly induced after 3 and 6 h of cultivation in 1% xylose with an increase in up to 25-fold. On the other hand, the expression of Afu1g03530 increased to 250-, 180-, 25-, 60-, 200-, and 5-fold after 3, 6, 12, 18, 24, and 48 h, respectively (Fig. 4b). These results lead us to speculate that both genes could encode potential xylose transporters, which can be further better characterized.

Characterization of the secretome of A. fumigatus in the presence of sugarcane bagasse

Once the transcriptome was characterized, we analyzed the secreted protein profiles of A. fumigatus cultivated in the same condition by SDS-PAGE and LC-MS/MS. The total protein secreted by the fungi was approximately 300 μg mL^− 1 in SEB versus 112 μg mL^− 1 in the fungi grown on fructose (Additional file 8: Figure S2). In the SEB supernatant, we detected 128 secreted proteins, and only 44 were detected in the fructose supernatant (Additional file 9: Table S7), 27 of which are the same for both conditions (Fig. 5a).

As previously described, in the A. fumigatus genome, 566 CAZyme genes (461 proteins, excluding GTs) are predicted [25]; 271 of them are predicted to be secreted (Additional file 6: Table S5). The presence of a signal peptide in these proteins was inferred using SignalP 3.0 (http://www.cbs.dtu.dk/services/SignalP/) [44]. Approximately 78 proteins identified in SEB correspond to CAZymes in the classes GH (46), CE (8), AA (8), PL (2), and CBM (15), which is 18.61% of the total CAZymes encoded by the A. fumigatus genome. The remaining proteins are classified as proteins of unknown functions (17%), hydrolases/peptidases/proteases/binding proteins (14%), oxidoreductases (6%), transferases (2%), and lyases (2%) (Fig. 5b). We have also identified intracellular proteins, suggesting cell lysis or an unknown mechanism.

We have also identified lignin-depolymerizing enzymes such as catalase-peroxidase, cellobiose dehydrogenase, catalase B, FAD-dependent oxidase, laccase, and Cu-Zn superoxide dismutase. Although the sugarcane bagasse employed here was treated by steam explosion, traces of lignin might have remained in the substrate, which justifies the secretion of these enzymes by the fungus.

We can conclude that the most important CAZymes (GH3, GH5, GH6, GH7, GH10, GH11, GH43, GH62, GH93, CE1, CE16 and AA9 (LPMO)) were secreted and play important roles in biomass degradation. For the first time, new proteins such as GH16 (endo-1,4-beta-glucanase), GH5 (endoglucanase), LPMO (AA9), swollenin and GH3 (β-glucosidase) were identified in the Aspergillus fumigatus secretome, probably because we used sugarcane bagasse as the source of carbon, and these enzymes can be specific to this complex biomass.

Integration of secretomics and transcriptomics

We observed weak correlations between transcriptome and secretome datasets, mainly because we chose the same time (24 h) to isolate mRNA and proteins. Considering that Aspergillus needs at least a few hours to translate mRNA to protein and to secrete it, these data provide an idea about which proteins are transcribed earlier or produced constitutively. Among the 1181 upregulated genes in transcriptome, 63 encoded proteins were detected in the secretome. As the same way, 16 of secreted proteins were identified as downregulated in RNAseq data (Fig. 6). In addition, the weak correlation observed could be the result of the influence of some factors that alter transcription and translation mechanisms [22, 23, 54].

The variation in expression during fungal growth in the presence of SEB was observed for some genes in the qRT-PCR data (Fig. 3). We observed that the gene expression depends on the duration of incubation, which explains the low correlation between the data of the secretome and transcriptome and supports the hypothesis that the approach used in this study is able to provide information about the modulation of gene expression of A. fumigatus in the presence of sugarcane bagasse. Genes encoding enzymes that are upregulated in the transcriptome (the highest log₂FC) and are present in the proteome, as well, are described in Table 3. The main enzymes produced by the fungus are endo-β1,4-xylanase, cellobiohydrolase B, endoglucanase and arabinofuranosidase, which leads us to speculate that A. fumigatus spent its energy on the transcription and production of cellulolytic and xylanolytic compounds, which were probably being secreted for biomass degradation and the release of small sugars.