Genomic analysis of the secretion stress response in the enzyme-producing cell factory Aspergillus niger

Many species of filamentous fungi such as Aspergillus niger are effective secretors of, mainly, hydrolytic enzymes to facilitate their saprophytic lifestyles by providing substrates from polymeric organic materials. This high capacity secretory system has driven the exploitation of filamentous fungi as cell factories for provision of enzymes used in a wide variety of applications [1]. Since the availability of gene-transfer systems, several fungal species have become potentially excellent hosts for achieving commercial yields of heterologous proteins. However, the yields of recombinants enzymes are often lower than desired, and this is especially so when the donor organism is not a fungus [2]. Many approaches have been used to overcome the bottlenecks to achieving high-secreted yields of heterologous proteins from fungi [2, 3] but the levels do not reach the same level as the best native protein.

Several data suggest that bottlenecks mainly exist at the post-transcriptional level, and most probably within the secretory pathway [4, 5]. The ER orchestrates the folding and some post-translational modifications of proteins that reside in, or pass through, the endomembrane system of a eukaryotic cell. In expression systems, the large flux of proteins being translocated into the ER generates a need to enhance the efficiency of protein folding and transport as well as the quality control of the synthesized proteins. Increased flux of proteins through the ER, especially those which do not fold correctly, or at least with the required kinetics, lead to the induction of stress responses that are collectively called protein secretion stress or ER stress. Mechanisms that allow the cell to sense the state of the lumen and to respond to ER stress conditions have been characterized in detail in the yeast Saccharomyces cerevisiae and in mammalian cells [6]. The first of these mechanisms is the unfolded protein response (UPR), a complex intracellular signaling pathway that increases the transcriptional activity of a number of genes involved in protein folding, glycosylation and transport [6, 7]. The UPR also appears to be intimately linked to the ER-associated (protein) degradation (ERAD) pathway [7, 8], in which misfolded proteins in the ER lumen are retrotranslocated (dislocated) through the translocon to the cytoplasm, and are ubiquitinated and degraded by the proteasome [9]. Recently, a novel type of feed-back mechanism, termed repression under secretion stress (RESS), has been suggested in filamentous fungi [10, 11]. It functions during secretion stress and down-regulates the transcript levels of some genes. A gene array study under ER stress conditions has recently been published with A. nidulans where ca. 30% of the predicted genes were represented on the array [12]. They described the first attempt to analyze in part the influence of the production and secretion of a heterologous protein on the cellular transcript profile of a filamentous fungus. A complete analysis of ER stress responses in the yeast S. cerevisiae, a fungus that does not secrete proteins as effectively as A. niger has also been described [7].

The ER surveillance system continuously coordinates the activity and participation of the processing and degradation pathways for unfolded proteins. Upon accumulation of unfolded proteins in the ER lumen, the UPR is activated, reducing the amount of new protein translocated into the ER lumen, increasing dislocation of proteins from the lumen of the ER and, then, their degradation, and bolstering the protein-folding capacity of the ER. Thus, understanding the process of folding and the stress responses in filamentous fungi may hold the major key to improving their use as cell factories for production of recombinant proteins. A genome-wide expression analysis of these secretion-related stress responses has never been reported in industrially-exploited species, since genome sequence data has not previously been available. Recently, the sequencing and annotation of the genome of A. niger have been completed [13]. DNA GeneChips have been made available and provide an unprecedented resource for exploring expression profiles in response to particular environmental cues. Here, we report the first gene expression analysis studies showing the response to various secretion stresses and assessing the breadth of the UPR in A. niger.

This is the first complete analysis of ER stress in A. niger. It has been made possible by the availability of the complete annotated genome sequence [13] and genome-wide GeneChips, which include built-in control sequences. The global ER stress response has already been reported in S. cerevisiae [7] but equivalent studies in the filamentous fungi have been hampered by the lack of complete genome sequences and the non-availability of gene arrays. Such bottlenecks are being removed and we have already seen the description of ER stress responses in Trichoderma reesei [25] and A. nidulans using arrays that cover approximately one third of the predicted open reading frames [12]. A. nidulans is scarcely exploited for its capacity to secrete enzymes because other filamentous fungi, including A. niger, have proved to hold advantages in terms of the range and yield of secreted enzymes. Previous analyses of secretion stress in A. niger have relied either on analysis of specified target genes [2, 26] or have used cDNA subtraction libraries [18, 27] which, in contrast to GeneChips, reveal only a small fraction of differentially expressed genes.

A low variation (a technical standard deviation of 0.16 and a biological standard deviation of 0.25 on average for the signal log ratio of 'present' genes) was shown for GeneChip replicates (data not shown). Generally, dependent on growth conditions, 5000–6000 genes were found to be expressed, with 10–20 genes detected as false positives in a single experiment. In any global transcriptomic study it is essential to have biological replicates that provide assurance of the validity of the conclusions reached. The biological replicates used in this study showed a small variation compared to the experimentally induced variation except for the tunicamycin study where the experimentally induced variation was small. Both for the chemostat-cultivated tPA strain and the shake-flask-cultivated tunicamycin-exposed cells, less than 4% and 1% of the expressed genes, respectively, were determined as changed by at least 1.5-fold.

We confirmed that each of the stress conditions led to induction of the UPR as judged by the transcriptional induction of A. niger genes known to be affected by ER stress, bipA [28] and pdiA [29]. In addition, we confirmed splicing of the 20bp intron in the A. niger hacA mRNA [10, 16]. It has recently been established that the UPR does not just involve the simple switch based around the synthesis of the mediating transcription factor but that there is further complexity under some conditions [30–32]. In A. niger [10], Trichoderma reesei [11] and S. cerevisiae [33] it is known that ER stress can lead to the transcriptional down-regulation of some genes encoding secreted proteins and this effect, termed repression under ER stress (RESS) [11], may be independent of the UPR [10]. ER stress due to over-expression of membrane proteins can elicit the transcriptional up-regulation of bipA without apparent splicing of the hacA intron in A. niger [34]. This effect, together with RESS, indicates that there is complexity in ER stress responses in filamentous fungi. Our results confirm the existence of the RESS mechanism in A. niger since the transcription of several major secreted proteins encoding genes was clearly repressed by at least 2 treatments. Moreover, in common with the data generated in Arabidopsis thaliana [35], many repressed genes in A. niger encode membrane proteins and transporter proteins that may be essential for the maintenance of cellular ion homeostasis. Thus, the list presented in the additional file 6 contains several putative zinc, iron, calcium and manganese transporters and the zinc-regulated transcription factor Zap1, which were mainly repressed by t-PA production. Previous studies showed that both calcium [36] and zinc [37] are required for ER function in yeast and that their deficiency induces the UPR.

Analysis of the impact of secretion stress on the genes encoding components of the secretory system in A. niger is summarized in Fig. 5. Few of the translocon genes (encoding components of both the co- and post-translational translocation system), which were largely not represented in the A. nidulans arrays [12], were induced under two or more stress conditions. The signal sequence recognition system was transcriptionally unaffected except in one component but there was more response from components of the signal peptidase complex. Several of the translocon genes and one component of signal peptidase were induced under UPR in S. cerevisiae [7]. A. niger homologues of the S. cerevisiae SEC11, SPC2 and SPC3 were all up-regulated under ER stress. Of the predicted ER-resident chaperones, bipA was induced under all conditions as expected [28]. The calnexin-encoding clxA gene [38], where the production of prochymosin was previously shown to induce its transcription, was shown here to be additionally induced by both DTT and tunicamycin. A homolog of the S. cerevisiae LHS1 gene, lhsA, has not been previously described in A. niger but was annotated in the genome of A. niger and represented on the GeneChips. As with bipA, lhsA was transcriptionally up-regulated under all 3 conditions compared to controls. In S. cerevisiae, the chaperone cycle involves products encoded by KAR2 (encodes a Bip-like chaperone) and LHS1 (where the ATPase activities of these two Hsp70p chaperones are coordinately regulated) as well as nucleotide exchange activity provided by Sil1p and DNAJ proteins such as Scj1p and Jem1p [39]. The ER-resident chaperone and foldase system was generally induced under UPR conditions in S. cerevisiae even if KAR2 and PDI1 failed to meet the stringent criteria applied [7]. Although lhsA has been identified here as a stress-responsive gene, as was at least one DNAJ protein, no candidate genes have been found for the nucleotide exchange factor homolog of yeast SIL1. On the presumption that this functionality exists, it may reside in a protein with low sequence identity to Sil1p or in another component, e.g. LhsA. A putative ortholog of a mammalian p58-encoding gene (similarity to human p58 with e value of 4e-59, showing 32% amino acid identity over 450 residues) was induced by both t-PA and tunicamycin. P58 is involved in translational regulation in mammalian cells, its induction is mediated by ATF6 and it plays a role in regulating the PERK/eIF2alpha/ATF4 pathway [40]. Homologs of PERK and ATF4/6 appear to be absent from A. niger so the role of the putative p58 in A. niger is intriguing.

Manipulations of the ER lumenal environment have been previously examined in Aspergillus with the aim of improving the secreted yield of heterologous proteins. This has been attempted with individual genes such as bipA [41] and pdiA [42] as well as through manipulation of the UPR [43]. Detailed knowledge of the responses of individual genes to different stresses should permit refinements to these approaches that would be more consistently beneficial. The complexity of the chaperone cycle has already been mentioned but the formation of disulfide bonds is another key example. Although the pdiA gene encodes the principal foldase in A. niger [29], other genes encoding members of this foldase/isomerase family have also been described [44, 45]. A fuller description of their responses to different stresses is provided by the GeneChip studies described here and shows that all three foldase genes are up-regulated by all stresses except for tigA by DTT. The contributions to foldase activity provided by the three identified lumenal foldases (PdiA, TigA, PrpA) is not known although, in S. cerevisiae, Pdi1p is known to contribute more activity than the other yeast foldases [46]. In S. cerevisiae, Ero1p is an essential lumenal protein involved in electron transfer during the formation of disulfide bonds and its homolog in A. niger, eroA, was induced under all stress conditions. Interestingly, Ero1p is retained in the lumen by attachment to the ER membrane in a manner that is not wholly understood [47] whereas the A. niger EroA is predicted to contain a conventional C-terminal ER retention sequence. The receptor for retention of C-terminal (H/K)DEL-containing proteins (Erd2p in S. cerevisiae) was induced under ER stress in both A. niger and S. cerevisiae [7]. In other areas, several genes involved in glycosylation of secretory proteins were induced under ER stress in both S. cerevisiae [7] and in A. niger (Fig. 5) and included a homolog of the yeast RFT1 gene that is responsible for translocation of lipid-linked glycan intermediates into the ER [48].

In fungi, the only reported translational control during the UPR concerns the expression of the yeast transcription factor Hac1p [20]. This control is mediated by a base-pairing interaction between an intron at the 3'end and the 5' untranslated region, which represses mRNA translation. Splicing of this unconventional intron is sufficient to relieve this translational block. It has been reported that the HAC1 ^u mRNA in yeast was stable, located in the cytosol and associated with ribosomes, yet did not produce protein, indicating that the ribosomes engaged on the mRNA were stalled and that translation was attenuated at the elongation step [20]. Our results showed that hacA ^u mRNA could also sediment in non-polysomal fractions, suggesting that translation of this mRNA species may be blocked at the translational initiation step in addition to the previously reported elongation step, as also shown in yeast [49]. DTT treatment resulted in translational repression of a large number of genes with functions in ribosomal biogenesis and assembly (Table 3). This result may support the hypothesis of a translational repression as a mechanism of reducing ER throughput. It is known that the abundance of RP mRNAs rapidly decreases when yeast cells encounter stress situations [50, 51]. We also found that several mRNAs encoding secreted proteins were also redistributed from polysomes to monosomes after DTT exposure and the gene glaA encoding glucoamylase, which is a major secreted protein in A. niger, was one of them. Therefore, in addition to the transcriptional down-regulation of glaA due to DTT stress, our results suggest that a post-transcriptional regulatory mechanism negatively affects the translation of glaA mRNA in A. niger. For the first time in filamentous fungi, these results suggest the existence, in addition to the RESS, of another feed-back mechanism that occurs upon ER stress at the translational level and leads to a reduction in the amount of new protein translocated into the ER lumen.

This is the first complete analysis of ER stress in A. niger, a filamentous fungus used commercially for the secreted production of a range of native and heterologous proteins. It has been made possible by the availability of the complete annotated genome sequence and genome-wide GeneChips. We induced ER stress either by chemical treatments of the wild-type cells or by expressing a heterologous protein. Following the induction of ER stress, A. niger cells display a diverse array of adaptative changes in gene expression at both the transcriptional and translational levels. The transcriptional responses to each stress were compared and the overlaps common to these conditions led to the identification of robust sets of induced or repressed genes Thus, a range of transcriptional targets has been identified as putative candidates for secretion stress- related genes, including a large number of genes encoding components of the secretory system. Moreover, by combining polysomal fractionation with DNA GeneChip analyses, we have focused on the translational activity of several individuals mRNAs following DTT exposure. Although the observed polysome profiles were similar under control and ER stress conditions, we confirmed the post-transcriptional control of hacA expression and highlighted that differential translation could also occur during ER stress at least for some genes encoding secreted or ribosomal proteins. Many of the genes identified in this study provide targets for improving A. niger as a cell factory for protein production.