Filamentous fungi are important cell factories for large‐scale enzyme production. However, production levels are often low, and this limitation has stimulated research focusing on the manipulation of genes with predicted function in the protein secretory pathway. Single deletion of four genes with predicted function in different metabolic processes improved the production of an Aspergillus nidulans homologous xylanase.
Microbial enzymes find a myriad of applications in many fields such as chemical, fermentation, agricultural, textile, pharmaceuticals and food production (Singh et al. , 2019). Choosing a suitable expression system is critical for high‐yield enzyme production, and many bacteria, filamentous fungi and yeasts have been commonly used to express recombinant enzymes (Demain and Vaishnav, 2009). In the bioenergy field, techno‐economic analysis demonstrated that enzyme costs are much higher than generally considered. Therefore, the optimization of cell factories for large‐scale processes can contribute to reducing the costs of enzyme production (Klein‐Marcuschamer et al. , 2012).
Aspergillus is one of the most studied genera of filamentous fungi due to the medical, food spoilage and industrial relevance of some species (Gabrielli et al. , 2014; Frisvad et al. , 2018; Taniwaki et al. , 2018). Several Aspergilli have a long history as producers of plant polysaccharide modifying and degrading enzymes, which allows broad utilization of carbon sources from different substrates. Aspergillus nidulans is a genetic model that has been extensively studied regarding metabolism, cellular development and regulation, contributing to the understanding of eukaryotic cell biology and molecular processes (Brandl and Andersen, 2017). The main advantage of filamentous fungi, in relation to other microbial cell factories, is the potential to produce high quantities of extracellular enzymes (Brandl and Andersen, 2017; de Vries et al. , 2017). Differently from prokaryotes, which have simpler pathways of protein secretion and is frequently essential for the organism fitness and pathogenicity (Maffei et al. , 2017), the secretion of proteins in eukaryotes, such as filamentous fungi, involves a more complex machinery in which N‐glycosylation can influence considerably (Dai et al. , 2013).
Random mutagenesis generally used for strain breeding may introduce many mutations into genomic DNA, resulting in higher productivity, though mutations are mostly unknown. On the other hand, rational genetic manipulation of target genes is time‐consuming (Nevalainen and Peterson, 2014). Strain engineering by deletion of extracellular protease genes (Zhang et al. , 2014), deletion of autophagic genes (Yoon et al. , 2013), overexpression or deletion of genes associated with the secretory pathway (Schalén et al. , 2016) and manipulation of unfolded protein response (UPR) pathway (Yu et al. , 2017) have been commonly explored as a rational approach to increase protein production in fungal systems. Furthermore, improvement of protein secretion has also been obtained by introducing multiple copies of heterologous genes (Liu et al. , 2014), by using native or artificial strong regulators (Zhang et al. , 2017), gene fusion to a well‐secreted homologous protein, using native signal sequences and others (Schalén et al. , 2016).
An important drawback when using fungi as cell factory is that the expressed heterologous protein can be lost or stuck in the secretory pathway due to issues in processing, post‐translational modifications (PTMs) or misfolding, thus inducing the endoplasmic reticulum (ER) stress. ER stress activates UPR to alleviate stress and restore homeostasis, promoting cell survival and adaptation (Heimel, 2015; Sun and Su, 2019). Under low levels of unfolded proteins, Ire1 is bound to BiP that is an ER‐resident Hsp70 chaperone. However, in cases of high levels of unfolded proteins, BiP is dissociated from Ire1, resulting in the Ire1 oligomerization and trans‐autophosphorylation. Consequently, Ire1 conformation is altered, which enables the ability to bind in cis‐acting elements present in the promoters of UPR‐target genes, known as UPREs. Ire1 homodimer activates the RNase domain that has hac1u (uninduced) as its substrate, removing an unconventional intron resulting in the hac1i (induced) mRNA. This results in the translation of Hac1p, the bZIP transcription factor responsible for the restoration of ER homeostasis (Krishnan and Askew, 2014). Sims et al. (2005) showed that approximately 23% of genes upregulated in A. nidulans recombinant strain producing chymosin were also upregulated when UPR was chemically induced by dithiothreitol (DTT). In addition, transcriptional analysis of mutant yeast strains with improved secretion of a heterologous α‐amylase showed a series of altered cellular processes, and the balancing of amino acid biosynthesis seemed to be particularly important (Huang et al. , 2017). Approximately 600 strains of Neurospora crassa carrying single deletion in genes with predicted functions in the secretory pathway were investigated for alterations in secretion, and seven strains hyperproducing cellulolytic enzymes were identified. Mutants implicated the loss of the Sterol Regulatory Element Binding Protein (SREBP) pathway (Reilly et al. , 2015; Qin et al. , 2017).
Here, a data bank of A. nidulans differentially expressed (DE) genes under chemically induced ER stress was generated by RNA‐seq using DTT and tunicamycin (Tm). Twelve target DE genes with predicted function in the secretory pathway were selected for single‐gene deletions in A. nidulans recombinant strains producing homologous and heterologous xylanases. After the single deletion of four genes with predicted function in different metabolic processes, the production of the homologous xylanase was increased, whereas the production of the heterologous xylanase remained unaltered. In addition, we investigated the presence of UPREs in the promoters of the deleted genes since those elements were described in Aspergillus niger (Mulder et al. , 2006), but not in A. nidulans.
Initially, two A. nidulans strains producing recombinant endo‐xylanases (EC 126.96.36.199) were constructed to evaluate the expression and production of recombinant enzymes classified in the same glycoside hydrolase (GH) family but from different microbial sources. The gene encoding for the heterologous xylanase (tpet_0854) was isolated from the hyperthermophilic bacterium Thermotoga petrophila, while AN7401/xlnE was selected as a homologous model. These xylanases share 53% of amino acid sequence similarity and belong to the family GH10 (Fig. 1A), and both enzymes have two predicted N‐glycosylation sites (Fig. S1).
Both genes were cloned into the pEXPYR expression vector and transformed into A. nidulans (Segato et al. , 2012). xlnE was more expressed (Fig. 1B), and XlnE was more secreted than Tpet_0854 (Fig. 1C and Fig. S2), despite the higher copy numbers of tpet_0854 (5 copies) compared with the single copy of xlnE in the genomes of the recombinant strains (Fig. 1D) (Appendix S1). Although the native gene was present in the genome and the reinsertion of the xlnE in random locus was carried out, the transformed xlnE was not inserted into the native site (Fig. S3). Moreover, no native xylanase production was detected in all the enzyme production experiments carried out in minimal media supplemented with maltose. Only the recombinant XlnE was produced since it is under the control of the glucoamylase promoter.
To generate a database of genes induced by chemical stress, the expression of canonical UPR markers, such as bipA and hacA (hacAi and hacAu ), was analysed by quantitative PCR (qPCR) (Appendix S1). The expression levels of these genes were higher at early stages (2 h) than at later periods (8 h) after chemical stress induction (Fig. S4), and thus short exposition periods were selected to perform RNA‐seq analysis (Appendix S1). Cellular stress was induced with Tm and DTT, which prevent N‐glycosylation of newly synthesized proteins and the formation of disulphide bonds, respectively (Fan et al. , 2018; Yoo et al. , 2018). A total of 294 million reads were generated, more than 94% of the filtered reads were mapped to the A. nidulans FGSC A4 reference genome, and a total of 10,774 genes were analysed (Table S1). For DTT, 1905 and 1172 genes were DE at 2 and 8 h, respectively, while 312 and 1862 genes were DE at 2 and 8 h of exposition to Tm respectively (Fig. S5A and B, Table S2).
Gene ontology (GO) enrichment of upregulated genes was performed for functional analysis (Appendix S1). The GO term oxidoreductase activity was enriched in the 2 h DTT treatment. Catalytic activity was found highly enriched, and lipid metabolic process, carboxylic acid metabolic process, organic acid metabolic process, cellular amino acid metabolic process and oxoacid metabolic process were moderately enriched in the 8 h DTT treatment. Regarding downregulated genes, the GO terms organonitrogen compound metabolic process, lipid metabolic process, catalytic activity and cytoplasm were enriched in 2 h DTT, whereas oxidoreductase activity and extracellular region were enriched in 8 h DTT. Additionally, the GO term secondary metabolic process was enriched in the upregulated genes in the 2 h Tm treatment, whereas intracellular was enriched in 8 h Tm (Fig. S5C).
The RNA‐seq data demonstrated to be consistent with previous works (Travers et al. , 2000; Sims et al. , 2005; Guillemette et al. , 2007; Carvalho et al. , 2012). The presence of transcripts related to oxidoreductase activity, mainly at the 8 h chemical treatment, could be partially explained by the role of DTT as a reducing agent. In addition, sustained ER stress was reported to cause oxidative stress in yeast cells by the UPR‐regulated oxidative folding machinery in the ER and mitochondria (Haynes et al. , 2004). This seems to be a ‘side effect’ of cellular response to ER stress, that is the oxidative protein folding intensified by loading of client proteins into the ER somehow results in ROS production. Lipid and cellular amino acid metabolic processes were also enriched, which could be explained by the ER expansion to accommodate higher influx of newly synthesized proteins (Pineau and Ferreira, 2010) and by the supply of cells with amino acids to synthesize proteins involved in the protection against reactive species respectively (Harding et al. , 2003; Herzog et al. , 2013). Moreover, downregulation of extracellular proteins could be associated with the so‐called process repression under secretion stress (RESS), which is a feedback mechanism activated in response to impairment in protein folding or transport (Pakula et al. , 2003).
Genes with predicted function in the secretory pathway were selected using a previously reported secretory model for Aspergillus oryzae (Liu et al. , 2014). A total of 375 homologous genes were found in A. nidulans by using the Aspergillus Genome Database (Asp GD; Table S3). Among them, 114 were upregulated and 53 genes were downregulated under DTT treatment, whereas 27 genes were upregulated and 13 genes downregulated under Tm treatment (Fig. S6A). Target genes for single deletion in A. nidulans were selected based on three criteria: (i) genes up‐ or downregulated in both treatment periods (2 and 8 h); (ii) top 25 genes with the highest or lowest fold‐change; and (iii) genes with no redundant function in the secretory pathway, covering a wide range of biological processes. These rankings resulted in the selection of 12 genes (Fig. S6B and Table S4).
The following step in our strategy was the deletion of the target genes in the A. nidulans strains producing XlnE and Tpet_0854 in order to evaluate their influence on the production of distinct recombinant xylanases. Initially, auxotrophy for uridine and uracil was regenerated by selecting XlnE and Tpet_0854 strains growing in MM with 5‐fluorotic acid (5‐FOA) (Appendix S1). A total of 20‐30 colonies presenting 5‐FOA‐resistance were obtained after three days of cultivation. After monosporic purification, more than 95% of the isolates showed uridine and uracil auxotrophy (pyrG‐) while maintaining equivalent levels of xylanase secretion compared with their parental strains (data not shown). Gene deletions in XlnE and Tpet_0854 pyrG‐ strains were confirmed by diagnostic PCR and Southern blot (Appendix S1) (Fig. S7).
XlnE, ΔpbnA, ΔydjA, ΔtrxA and ΔcypA strains showed 78, 171, 105 and 125% higher xylanase activity than the control strain at 36 h of induction (Fig. 2A), respectively, and even higher after 48 h (Fig. S8) (Appendix S1).
Additionally, we investigated whether the improvement observed in xylanolytic activity resulted because of a specific increase in XlnE production or an increase in total protein secretion. Interestingly, lower protein secretion was observed in the mutant strains, suggesting that single deletions specifically increased the XlnE production (Fig. 2B). On the other hand, no changes in xylanase production, nor in the levels of secreted proteins were observed in the Tpet_0854 mutant strains (Fig. S9).
In an attempt to address how the single deletions increased secretion of the homologous xylanase, we investigated the presence of UPREs in the promoters of pbnA, cypA, ydjA and trxA genes using the JASPAR database (Khan et al. , 2018). Our analysis of A. nidulans promoters using the UPREs motifs found in Aspergillus niger was inconclusive. Thus, we used the hac1 upstream consensus sequence from S. cerevisiae (profile MA0310.1) as a model and the highest relative score was found for the following motifs: GACACGTC (0.969), AACACGTC (0.937), GCCACGTA (0.955), AACACATA (0.802), ACCACGTT (0.900), CCCACGTT (0.854) and GACACGTA (0.984) in the promoters of bipA, pdiA, hacA, pbnA, ydjA, trxA and cypA, respectively, suggesting that these genes may be regulated by UPR. The CACGT motif was not found in the pbnA promoter, suggesting that this gene is probably independent of hacA ‐mediated UPR (Fig. 3A).
The XlnE mutant strains were cultivated on MM agar plates supplemented with glucose, CMC, xylose, starch or pectin to investigate growth phenotypes as well as in the presence of DTT to evaluate chemical cellular stress resistance. Deleted genes were not essential to carbohydrate metabolism nor to metabolize DTT since all four mutant strains seem to equally grow in the different conditions (Fig. 3B). In addition, mutant strains were grown on hydrothermal‐pretreated sugarcane bagasse in order to quantify the secretion of different carbohydrate‐active enzymes (CAZymes) in the crude supernatants. Arabinofuranosidase, cellobiohydrolase, β‐glucosidase and β‐xylosidase activities were slightly lower in the XlnE mutant strains (except for β‐xylosidase activity in the ΔcypA strain), whereas amylase, pectinase, endoglucanase, xylanase and xyloglucanase activities were lower only in the ΔpbnA mutant (Fig. 3C) (Appendix S1). Furthermore, no morphological differences were detected in the mutant strains hyphae (Fig. S10).
For further understanding of pbnA, ydjA, trxA and cypA functions, a weighted gene co‐expression network analysis (WGCNA) was performed, resulting in the identification of 21 modules (Fig. S11). The pbnA, ydjA, trxA and cypA genes were, respectively, found in the following modules: lightyellow, brown, darkmagenta and lightcyan (Table S5). GO enrichment analysis of the modules demonstrated that these genes have potential functional links with several GO terms involved in a wide range of processes related to protein secretion (Fig. S12).
To date, pbnA was not characterized in A. nidulans, whereas the orthologous in S. cerevisiae (pbn1 ) encodes a chaperone‐like protein, an essential component of the glycosylphosphatidylinositol‐mannosyltransferase I complex. The lightyellow module was especially enriched for proteasome complex activity (Fig. S12A and Table S6). pbn1 physically interacts with prb1 (vacuolar proteinase B) and ecm27 (Na+/Ca2+ exchanger) in S. cerevisiae. The expression of pdiA and bipA was measured to evaluate whether pbnA deletion induced UPR activation or not. The genes pdiA and bipA were not overexpressed in the knockout strain, suggesting that UPR activation is not involved in the higher production of XlnE at least in 36 h (Fig. S13). This result, in addition to the reduction in total protein secretion (Fig. 2B) and the higher XlnE secretion (Fig. 2A), suggests that the cell converged for the XlnE production by reducing the traffic of other proteins in the ER. Such a compensation mechanism could then avoid UPR activation.
A group of chaperones that includes ydj1 is induced in strains overexpressing hac1 (Graf et al. , 2008). ydjA was detected in the brown module (Fig. S12B and Table S6). This result indicates that this gene might have functions on intracellular protein transmembrane transport, protein refolding and ERAD pathway. In S. cerevisiae, ydj1 physically interacts with prd1, which encodes an intracellular proteinase involved in protein degradation (Buchler et al. , 1994). In addition, ydj1 genetically interacts with pbn1 (pbnA in A. nidulans), tsa1 and cne1 (calnexin). Calnexin is a key ER chaperone that assists in folding and subunit assembly of most Asn‐linked glycoproteins passing through ER (Leach and Williams, 2013). Our results suggest ydjA is a chaperone that contributes to the folding of different A. nidulans proteins since ydjA knockout strain showed lower total protein secretion (Fig. 2B).
trxA encodes thioredoxin (trx1 orthologous in S. cerevisiae) that contributes to protection against reactive oxygen species (ROS; Thön et al. , 2007). Detoxification mechanisms include the production of superoxide dismutases, catalases, peroxiredoxins, glutathione and the thioredoxin system, a dual system composed of thioredoxin and thioredoxin reductase (TrxR; Thön et al. , 2007). trxA was co‐expressed with genes predicted to have function in cell redox homeostasis, fatty acid transport and ubiquitin‐dependent protein catabolism in the darkmagenta module (Fig. S12C and Table S6). trx1 has predicted physical interactions with trx2 (thioredoxin), ero1 (protein disulphide isomerase), met16 (3′‐phosphoadenylsulphate reductase), ahp1 (thiol‐specific peroxiredoxin), tsa1 (thioredoxin peroxidase) and mxr1 (methionine‐S‐sulfoxide reductase) in S. cerevisiae. Analysis of the genetic interactions showed crosstalk between trx1 and pbn1 networks through tsa1 and prb1 in S. cerevisiae. tsa1 genetically interacts with prb1, which, in turn, is physically linked to pbn1. It is likely to suggest that, although TrxA is very important for the folding of several proteins in A. nidulans, it did not affect the XlnE production.
Finally, cypA is a homologue to cpr1 in S. cerevisiae, which encodes a peptidyl‐prolyl cis‐trans isomerase (PPiases) catalysing the cis‐trans isomerization of peptide bonds in N‐terminal proline residues. cpr1 showed physical and genetic interactions with set3, hos4, hos2, sif2 and snt1 in S. cerevisiae. These genes, including cpr1, form the Set3p complex (SET3C), in which the subunits Hos2, Set3, Sif2 and Snt1 form the core complex, whereas Hst1p, Sum1p and Cpr1p are peripherally associated (Pijnappel et al. , 2001). Many of these proteins are histone deacetylases (HDACs), and deacetylation has been identified as a major regulator of eukaryotic gene transcription due to a repressive chromatin structure (Kurdistani and Grunstein, 2003). Then, the SET3C complex is responsible for transcriptional repression of some genes related to the early/middle class of sporulation‐specific genes, including the key meiotic regulator in yeasts (Pijnappel et al. , 2001). cypA was co‐expressed with genes with predicted functions in the hyphal tip, vacuole organization and response to heat (Fig. S12D and Table S6). How the deletion of pbnA, ydjA, trxA and cypA exactly impact XlnE production, but not Tpet_0854, remains to be determined.
Cellular stress was chemically induced in A. nidulans using DTT and Tm. Based on the RNA‐seq database, twelve genes with predicted function in the secretory pathway were deleted in A. nidulans recombinant strains producing homologous and heterologous GH10 endo‐xylanase. Higher enzyme production occurred only in the mutant strains (ΔpbnA, ΔydjA, ∆trxA and ΔcypA) producing the homologous xylanase. These four deleted genes have predicted functions in processes such as proteasome activity, protein refolding, ERAD pathway, hyphal tip, vacuole organization, autophagy, cell redox homeostasis, fatty acid transport and ubiquitin‐dependent degradation. The continuing characterization of pbnA, ydjA, trxA, cypA, as well as other mutants, will enhance our understanding of protein secretion in filamentous fungi, including industrially relevant species.
MPZ, MVR, CRFT, EPA, FJC and AD were supported by FAPESP (São Paulo Research Foundation), grants no. 2014/15403‐6, 2013/24988‐5, 2016/16306‐0, 2017/26315‐9, 2017/10083‐1, 2012/20549‐4 and 2015/50612‐8. AD and CRFT were also supported by CNPq (Brazilian Council for Scientific and Technological Development), grants no. 404654/2018‐5 and 304816/2017‐5 and 420392/2018‐1 respectively. The authors acknowledge Espaço da Escrita – Pró‐Reitoria de Pesquisa – UNICAMP – for the language services provided. Research supported by LNBR – Brazilian Biorenewables National Laboratory (CNPEM/MCTIC) during the use of the NGS ‐ Next Generation Sequencing open access facility.