Uncovering CAZyme Induction in the Thermophilic Fungus Thermoascus Aurantiacus

Background: Filamentous fungi are excellent lignocellulose degraders, which they achieve through producing carbohydrate active enzymes (CAZymes). CAZyme production is highly orchestrated and the application of –omics methods such as RNA-Seq has greatly expanded understanding of this important biotechnological process. The thermophilic fungus Thermoascus aurantiacus secretes high amounts of highly active thermostable enzymes that enable saccharications at higher temperatures; however, the genome-wide response to CAZyme induction is not understood. Results: A fed-batch system with plant biomass-derived sugars D-xylose, L-arabinose and cellobiose established that these sugars induce CAZyme expression in T. aurantiacus. The C5 sugars induced both cellulases and hemicellulases, while cellobiose specically induced cellulases. A minimal medium formulation was developed to enable RNA-seq studies of T. aurantiacus with these inducers. It was found that D-xylose and L-arabinose strongly induced a wide variety of CAZymes, auxiliary activity (AA) enzymes and carbohydrate esterases (CEs), while cellobiose facilitated lower expression of mostly cellulase genes. Furthermore, putative orthologues of different unfolded protein response genes were up-regulated during the C5 sugar feeding together with genes in the C5 sugar assimilation pathways. Conclusion: This work has identied two additional CAZyme inducers for T. aurantiacus, L-arabinose and cellobiose, along with D-xylose. A combination of biochemical assays and RNA-seq measurements established that C5 sugars induce a suite of cellulases and hemicellulases, providing a path to produce a broad spectrum thermotolerant enzymatic mixture for deconstruction of plant biomass.


Introduction
Carbohydrate active enzymes (CAZymes), are vital for the conversion of plant polysaccharides to biofuels and bio-based chemicals (1). Cellulose and hemicellulose are the most abundant polysaccharides in plant biomass and thus are an immense untapped carbon pool for biotechnological applications. Through using CAZymes such as cellulases and xylanases, lignocellulosic plant biomass can be deconstructed into simple sugars that can be further converted into biofuels and bioproducts (2).
The thermophilic fungus Thermoascus aurantiacus is a notable host for thermostable CAZyme production (3). The enzymes of this fungus were found to be more heat stable and effective at deconstructing lignocellulose than enzymes from other thermophilic fungi and demonstrated the release of sugars from pre-treated biomass at comparable levels to the commercial enzymatic mixture CTec2 at 50°C. Notably, the enzymatic mixture from T. aurantiacus lost only half of its activity during sacchari cation of pre-treated switchgrass at 70°C, while CTec2 was inactive at this temperature (4). Moreover, T. aurantiacus secretes large amounts of a lytic polysaccharide monooxygenase (LPMO), which depolymerizes cellulose and has been extensively studied and shown to be the most abundant protein secreted by T. aurantiacus (5). Recently, procedures for genetic transformation, sexual crossing and gene editing using CRISPR/Cas9 have been established for this self-crossing organism, which opens many possibilities to harness the ability of T. aurantiacus to secrete highly active and thermostable CAZymes (6-8). However, little is known about CAZyme induction in T. aurantiacus, which is vital for designing improved strain engineering and bioprocess strategies.
Cellulase and xylanase induction has been intensively studied in a few lamentous fungi (9)(10)(11)(12). Trichoderma reesei has historically been the most prominent fungus for commercial cellulase production (13,14). The cellulases of T. reesei are induced in the presence cellulose as well as simple sugars like cellobiose, lactose and sophorose. While cellobiose induces cellulases in a variety of lamentous fungi, including the model fungus Neurospora crassa, cellulase induction by lactose and sophorose is limited to a few fungi besides T. ressei (15). For N. crassa and T. ressei, xylose induces xylanase production; however, xylose was found to induce both cellulases and xylanases in the industrially used host Aspergillus niger (16). While less is known about the inducing effect of L-arabinose, another major plant cell wall sugar, it was reported that this sugar induced xylanases in N. crassa and T. reesei (17)(18)(19). These ndings indicate that sugars related to plant cell wall degradation have very different effects on CAZyme secretion, even in closely related fungal species.
Induction of CAZymes by T. aurantiacus is less well understood than in the previously mentioned species. It was initially reported that T. aurantiacus secreted the highest level of cellulases when grown on alkali pre-treated bagasse and xylan (10) while weak cellulase activity was found on Avicel cellulose. To identify further inducers for T. aurantiacus, batch cultivations with D-xylose, cellobiose and L-arabinose were performed (10). High xylanase secretion of T. aurantiacus was reported for batch-cultures of xylan and non-metabolizable methyl β-D-xylopyranoside (MXP), but not xylobiose and D-xylose (11,20).
Recently, experiments using controlled feeding of xylose to T. aurantiacus cultured in shake asks and bioreactors demonstrated substantial cellulase and xylanase induction that was higher than batch cultures with xylan (12). Studies performing similar fed-batch experiments using T. reesei found strong CAZyme secretion induction with sugars related to plant cell wall degradation (18, 21). Therefore, continuous feeding helps to circumvent two challenges associated with batch cultures: 1) the high initial amount of sugar rst causes carbon catabolite repression (CCR) (22); the sugar is then consumed rapidly, leading to no further induction and 2) the presence of impurities in highly puri ed plant polysaccharides (e.g. cellulose contaminated with residual xylan) can lead to fundamentally different outcomes compared to using puri ed sugar substrates at low concentration (12). Lastly, fed-batch studies can be of great utility for -omics experiments to re ne bioprocess and genetic engineering strategies for generating high enzyme secreting fungal strains (23)(24)(25).
The goal of this study was to test CAZyme inducers of T. aurantiacus by fed-batch cultivation to screen for enzymatic activities and to reveal how gene expression is affected by these inducers.

Strains and Culture Conditions
Thermoascus aurantiacus ATCC 26904, which was obtained from the American Type Cell Culture Collection and maintained on potato dextrose agar (Fluka ® ) by inoculating fresh plates with a piece of a grown culture. Cultures were incubated for 2 days at 50 °C to enhance growth and were then shifted to 45°C to prevent desiccation of the plates.

RNA-Seq experiments
For RNA-Seq experiments, T. aurantiacus pre-cultures were grown as described above with the minimal medium containing 25 mM (NH 4 ) 2 SO 4 and 30 mM trisodium citrate and the mycelia from these precultures isolated and inoculated into fed batch shift experiments as described above, The peristaltic pump was calibrated to release 50 mg l -1 h -1 of either D-xylose, L-arabinose or cellobiose during 16 h to four shake ask replicates for each sugar. In addition to these cultures, 4 cultures were set up as a no carbohydrate control and 4 cultures containing 2% (w/v) D-glucose served as a control treatment for carbon catabolite repressing conditions. After 16 h, the feed was stopped and 10 ml of each culture were rapidly ltered over a lter paper, peeled off and snap frozen with liquid nitrogen. These frozen mycelia were ground with a mortar and a pestle and stored at -80 °C. The RNA extraction was performed with Maxwell RSC system (Promega, Madison, WI, USA) and the Plant RNA kit (Promega, Madison, WI, USA) according to the fungal RNA-extraction protocol. Brie y, aliquots of extracted mycelium were transferred into designated tubes and 400 ul of Homogenization solution and 200 ul of Lysis buffer were added and vortexed and the extraction was initiated. The puri ed RNA was eluted in nuclease free water. RNA quality was assessed with a NanoDrop spectrometer and the quality with the RNA 6000 Nano kit (Agilent, Waldbronn, Baden-Württemberg, Germany). to generate a table containing the counts per gene. Furthermore, 500 bases were added to the coordinates of the 3' end of each transcript (or until the start of the next gene) to improve annotation (from only 21 % of the reads aligned, to 77 %). Genes with mean expression levels lower than 1, were ltered which resulted in a table of 8591 genes (instead of the original 8798 genes). This minimal ltering step ensured that genes with low expression such as transcriptional regulators remained in the data set. The differential expression analysis was performed with the DESeq2 package in R (64, 65). Heat maps were created using (pheatmap(sampleDistMatrix)) with the pheatmap package (66) and the Morpheus online tool (https://software.broadinstitute.org/morpheus/).

Protein analysis
Culture ltrates for protein analysis were recovered by transfer of the culture broth to a spin column (Brand), which was centrifuged at 8000 rpm for 3 min. The ltrate from the column was then used for protein assays.
The protein pro le was investigated with SDS-PAGE gels (8-16% Tris-glycine mini gel; Invitrogen, Carlsbad, CA, USA). Accordingly, 15 ul of the ltered culture broth was mixed in a 5:1 ratio with loading dye (Laemmli buffer/2-mercaptoethanol (4:1)) and boiled for 5 min at 95 °C. The mixture was loaded together with 5 ml of the Novex sharp prestained protein standard molecular weight markers (Thermo Fisher Scienti c, South San Francisco, CA USA) and separated for 65 min at 140 V. The gel was then stained with SimplyBlue safe stain (Thermo Fisher Scienti c, South San Francisco, CA USA).
Protein concentrations were measured with the Bradford Protein Assay (Bio-Rad, Hercules, CA, USA). The assay reagent was diluted 1:5 and 200 mL were added to a 96 well plate together with 5 ml of culture ltrate. Bovine gamma globulin (Thermo Fisher Scienti c, South San Francisco, CA USA) was added as a standard (0 -2 g/L) (12). The absorbance measurement was performed at 595 nm.
CMCase and xylanase activity was measured with the DNS method (67) with 1% (w/v) sodium carboxymethylcellulose (CMC) or beechwood xylan as substrate in 50 mM sodium acetate buffer pH 5. For the assay, culture ltrates were diluted either 1:20 or to 50 mg protein / mL and 10 ml were added in triplicates to a 96 well PCR plate together with 70 ml of the polysaccharide substrates. Monomeric sugar standards (D-glucose for CMCase and D-xylose for xylanase) ranging from 0 -12.5 uM was added in duplicates to each plate. All assays were performed in PCR cyclers for 30 min at 65 °C. The reactions were stopped via adding 80 ul of the DNS reagent and boiling for 5 min at 95 °C. The amount of released sugars was measured at λ = 540 nm. One unit of enzyme activity (U/mL) was de ned as amount of released sugar (nmol) per time (min) per volume of culture supernatant (mL). Assays for p-nitrophenyl β-D-cellobioside (pNPC) and p-nitrophenyl β-D-glucopyranoside (pNPG) were performed as previously described in (68).

Arabinose and cellobiose induce CAZymes in T. aurantiacus
Previously, fed-batch induction experiments led to the identi cation of xylose as a inducer of CAZymes in T. aurantiacus (12). During this previous fed-batch study, beechwood xylan and a variety of celluloses (Avicel, Sigmacell, bacterial cellulose) also induced CAZyme production in T. aurantiacus (12). Since both beechwood xylan and the plant-derived celluloses (Avicel, Sigmacell) contained xylose, the inductive effects of other sugars besides xylose in these substrates could not be deconvoluted (15). Therefore, we chose two additional lignocellulose-derived sugars to test as inducers: L-arabinose, which constitutes ~ 10% of beechwood xylan, and cellobiose, which is a common cellulase inducer and the product of cellobiohydrolase, (12). T. aurantiacus was grown rst in a seed culture in D-glucose medium supplemented with soy meal peptone for 48 h and shifted to two sets of shake asks; one set containing each individual sugar (0.5% w/v D-xylose, cellobiose and L-arabinose, referred to as batch culture set) and another set of asks where the same amount of those sugars were continuously fed (fed-batch set) over a 3 day period (Fig. 1). The batch culture set also included triplicate beechwood xylan (0.5% w/v) cultures to compare the puri ed sugars with a complex substrate that previously found to be a good CAZyme inducer.
SDS-PAGE of culture ltrates demonstrated that the characteristic cellulase bandspreviously observed for T. aurantiacus (cellobiohydrolase, endoglucanase, and lytic polysaccharide monooxygenase) (3) were clearly visible for the fed-batch cultures, which had higher total protein titers than batch cultures (Fig. 1A-B). The three sugar feeds induced different enzymatic activities. High xylanase activities were found during C5 sugar feeding, while cellobiose fed-batch led only to a small increase in xylanase activity in the culture broth (Fig. 1C). Interestingly, cellobiosidase activity followed the same trend of high activity during C5 sugar feeding (Fig. 1D) while endoglucanase activities, measured through the CMCase assay, were in a comparable range for both the cellobiose and C5 fed-batch cultures (Fig. 1E). In contrast, the cellobiose fed-batch and batch cultured displayed higher beta-glucosidase activity than the C5 sugar cultures (Fig. 1F). The ability of the supernatants from the fed-batch cultures to release glucose from Avicel was also tested. In these experiments, the supernatant from the cellobiose-grown cultures had 72% higher glucose release than the xylose-grown cultures and 114% higher glucose release than the arabinosegrown cultures (Fig. 1G).

Development of minimal medium for T. aurantiacus protein production
The fed-batch experiments identi ed D-xylose, L-arabinose and cellobiose as inducers of CAZyme production. RNA-Seq experiments were designed to study gene expression patterns under these growth conditions to describe the cellular responses to these sugars. However, the T. aurantiacus induction experiments described above were performed with soy meal peptone, a complex nitrogen source containing sugars. Using peptones can pose challenges when using systems biology tools, since considerable differences among manufacturers and also batch effects for the same product can be found (26). While peptones were found to enable high protein production of T. aurantiacus, using a de ned medium appeared more favorable for RNA-Seq experiments. Therefore, a minimal medium with a de ned nitrogen source was designed in this study. Previously, T. aurantiacus cultivations have been performed on media containing complex nitrogen sources such as yeast extract or peptones (4,9,10,20,(27)(28)(29)(30)(31)(32). In only one study, T. aurantiacus was cultivated in Vogel's minimal glucose medium (11). This medium supported poor growth in our hands, so the McClendon's medium, which has been used for the above described CAZyme production was adapted to employ (NH 4 ) 2 SO 4 as the sole nitrogen source for a de ned minimal medium to study T. aurantiacus induction.
T. aurantiacus growth on D-glucose with McClendon's medium with soy meal peptone as the nitrogen source was compared with growth with (NH 4 ) 2 SO 4 ( Fig. 2A-B). Growth of T. aurantiacus with (NH 4 ) 2 SO 4 as the nitrogen source in liquid medium was < 30% of the growth with soy meal peptone as measured by mycelial biomass. Growth with (NH 4 ) 2 SO 4 resulted in a drop in pH after 3 days from the initial of pH 5.5 to 2.5, while growth on soy meal peptone after 3 days resulted in pH of 6. The poor growth on (NH 4 ) 2 SO 4 was attributed to the drop in pH, which is consistent with previous studies demonstrating optimal growth and enzyme production by T. aurantiacus at pH ≥5 (12). To prevent the drop in pH during cultivation, the (NH 4 ) 2 SO 4 -containing medium was buffered with 25 mM sodium citrate. The citrate buffer maintained the pH around 5 for the submerged T. aurantiacus culture and doubled the fungal growth compared to adding (NH 4 ) 2 SO 4 alone. This minimal medium was tested in a fed-batch experiment with D-xylose, cellobiose and L-arabinose as described before (Fig. 2). Although the overall protein production was lower in the minimal medium than in medium amended with soy meal peptone, the enzyme activity patterns were consistent between both tests. Therefore this improved minimal medium was used to assess gene expression patterns during induction.
Differential gene expression analysis of T. aurantiacus The RNA-Seq study was performed using the fed-batch system described above. T. aurantiacus was grown in D-glucose minimal medium and then shifted to shake asks containing minimal medium, where D-xylose, cellobiose and L-arabinose were added through continuous feeding. Additionally, a starvation condition (only minimal medium without a carbon source) and a CAZyme-repressing condition (high Dglucose minimal medium) were added as controls. RNA-Seq was performed using 3 biological replicates for each of those conditions.
A Venn diagram was generated to uncover genes differently expressed in D-glucose medium and the sugar fed-batch conditions compared to no carbohydrate medium to investigate the effects of those sugars on gene expression (Fig. 3). All up-regulated genes of the D-xylose, cellobiose, L-arabinose and Dglucose treatments compared to no carbohydrate medium (logFC > 1, pval < 0.05) were used for this analysis. All numbers in parenthesis are T. aurantiacus protein IDs from the JGI Mycocosm database unless otherwise stated.
For the D-xylose feed condition, no genes speci c for xylan and C5 sugar metabolism were found except for a putative alpha-glucuronidase (36875) and an F-box protein orthologue (Prot. ID: 43769) of the A. nidulans gene fbxA. Deletion of fbxA caused reduced secretion of xylanases in its native host and impaired CCR, which was indicated by resistance of the fbxA mutant to the D-glucose analog 2-deoxy-Dglucose (34). Additionally, an unknown sugar transporter (Prot. ID: 55223) was up-regulated. A putative transcriptional regulator with highest similarity to the A. nidulans sexual development regulator nsdD was found here, which was recently found to be vital for cellulase and xylanase in P. oxalicum (35).
When the genes speci c for both C5 sugar feeds were investigated, we found two of the most highly secreted enzymes of T. aurantiacus: the xylanase (Prot. ID: 1236) and endoglucanase (Prot. ID: 65156) which have been identi ed earlier by proteomics (3). Also, a putative beta-xylosidase (Prot. ID: 64461), xylithol dehydrogenase (Prot. ID: 66263) and D-xylose dehydrogenase (Prot. ID: 63693) were identi ed, indicating expression of a variety of genes related to xylan hydrolysis and the D-xylose and L-arabinose catabolism of T. aurantiacus. The metabolism of those two C5 sugars has been extensively investigated for the related fungus A. niger (36, 37). Orthologous of all necessary enzymes for D-xylose and Larabinose catabolism were identi ed in T. aurantiacus, except for the L-arabinose reductase, the rst enzyme needed for L-arabinose assimilation (Supplement Fig. 1). All the genes in the T. aurantiacus Dxylose/L-arabinose assimilation pathway were highly up-regulated during L-arabinose feed, while Dxylose caused up-regulation of the same genes except the putative L-xylulose reductase orthologue. Another pathway that appeared to be up-regulated during the C5 sugar feeds was the unfolded protein response (UPR) (Supplement Fig. 2). Several components of this pathway, namely a bipA (38) orthologue (Prot. ID: 46710) was up-regulated in the L-arabinose condition and clxA (39) (Prot. ID: 65748) was highly expressed in the L-arabinose and D-xylose condition. We found that other UPR related genes, such as the orthologue of the regulator that activates UPR, hacA (40) (Prot. ID: 7916), and the protein disul deisomerase pdiA (41) (Prot. ID: 64656), were highly expressed during D-xylose and L-arabinose feeding relative to all other conditions (Supplement Fig. 2). UPR genes are often expressed during fungal enzyme secretion to counteract protein folding stress (42).
Intriguingly, the cellobiose feed condition did not display any plant cell wall degradation related CAZymes that were speci c for this condition. However, genes up-regulated during D-xylose, cellobiose and Larabinose feed were the main secreted cellobiohydrolase (Prot. ID: 41785) and a beta-glucosidase (Prot. ID: 38776). Therefore, the expression of the main secreted xylanase, endoglucanase and cellobiohydrolase together with several xylan degrading enzymes happened in the C5 sugar conditions, while the fungus in the cellobiose condition appeared to only express the cellobiohydrolase at high levels this time point. Lastly, a serine carboxypeptidase (Prot. ID: 57314) was identi ed up-regulated in all sugar feeds, which had highest sequence similarity to protH of A. niger. This enzyme is predicted to be secreted, which could thus be involved in degradation of secreted CAZymes.

Expression analysis of glycoside hydrolases, auxiliary family enzymes and carbohydrate esterases
The analysis of expression patterns was extended to compare expression of related CAZyme proteins. The carbohydrate active enzymes (CAZy) database classi es enzymes relevant for the degradation, modi cation or creation of glycosidic bonds (43). The JGI mycocosm protein portal contains those CAZyme annotations for T. aurantiacus, where 323 CAZy genes are annotated in the T. aurantiacus genome (44). Enzymes belonging to the glycoside hydrolases (GH), auxiliary activity (AA) and carbohydrate esterases (CE) have been found to be most important for plant cell wall deconstruction (15). Five CAZymes (Prot. ID: 3070, AA 9; 41785, GH7; 65156, GH5; 1236, GH10; 46699, GH3) have been previously identi ed in T. aurantiacus ATCC 26904 from liquid cultures grown on plant biomass by proteomics analysis (3). Virtually nothing is known about what other genes of this fungus respond to plant polysaccharides and their breakdown products. To uncover the expression trends of those unknown genes, we generated a heat map of the expression trends of the T. aurantiacus GH, AA and CE families.
The GH heat map (Fig. 2a) contained two main clusters: cluster 1 showed GHs that were highly upregulated on D-glucose or D-glucose and other conditions, while cluster 2 contained GHs that exhibited low expression on D-glucose. Cluster 1 contained diverse types of GHs with only few predicted types of GH related to cellulose, xylan and pectin deconstruction. Conversely, the main four secreted GH described above were all found in cluster 2 (Fig. 4a). Cluster 2 also contained most of the cellulase, xylanase and pectinase genes and was divided into 2 sub-clusters (Cluster 2.1 and 2.2). Cluster 2.1 contained GHs that were highly up-regulated in no carbohydrate medium and to different degrees up-regulated during the sugar feed conditions. Thus, cluster 2.1 was comprised of GH that most likely resemble starvation responsive genes. The beta-glucosidase that was previously identi ed in the T. aurantiacus supernatant (3) was also found to be strongly expressed during starvation and to a lesser extend during the sugar feed conditions. However, Cluster 2.2 contained GHs that were highly expressed under the sugar feed conditions and showed lower expression during starvation. This cluster appeared to resemble the genes induced by the sugar feed conditions and contains the previously identi ed endoglucanase, cellobiohydrolase and xylanase (marked). Also, most of cluster 2.2 genes are putative orthologues of A. niger GHs linked to plant cell wall deconstruction (Fig. 2a). A. niger orthologues of the GH51 arabinofuranosidase afbA, was found to released L-arabinose from arabinan and an unknown putative GH5 (T.ID: 44708) with highest similarity to exgC in A. nidulans were found highly expressed on D-xylose. Genes highly expressed on L-arabinose showed high similarity to the alpha-glucuronidase aguA, alphaxylosidase axlB and the beta-xylosidase xlnD. In addition to the high expression of the main secreted GH5, 7 and 10, we also found another putative GH5 that had highest similarity to exgA and was highly expressed on both C5 sugars. A putative beta-glucosidase with highest similarity to bgl4 was the only GH highly expressed under all three inducing conditions. Few GHs showed elevated expression on cellobiose with only one gene exclusively expressed under this condition predicted beta-glucuronidase (T-ID: 61003) that had the highest similarity to two uncharacterized A. nidulans genes: AN7089 and AN3992. Genes highly expressed during cellobiose feed in cluster 2.2 were mostly cellulases (GH1, GH3, GH5 and GH7). This is in accordance with our nding that cellobiose feed led to only weak xylanase activity in the culture supernatant while cellulases were con rmed to be produced by SDS-PAGE and activity was high for glucose release from Avicel (Fig. 1). Similar trends were found for the AA heat map (Fig. 4b), where cluster 1 contained AAs that were highly expressed on D-glucose with almost no expression on all other conditions. Cluster 2 was divided in 2 subclusters. Cluster 2.1 contained AA that were highly expressed during starvation including 4 predicted laccases, whose expression is often associated with starvation also in other fungi, and one putative unknown LPMO (T.ID: 2027). Cluster 2.2 contained genes that displayed low expression on D-glucose and no carbohydrate medium, and high expression during the sugar feed conditions. Here, the well-studied T. aurantiacus LPMO (marked) was found highly expressed during D-xylose and L-arabinose and to a less extent on cellobiose. Two predicted laccases showed a similar trend. Interestingly, one further laccase (T.ID: 39797) was highly expressed on D-glucose. This gene has a high similarity to A. niger mcoH and the Trichophyton benhamiae conidial pigment biosynthesis oxidase. Therefore, this laccase might be required for pigmentation and could serve as a color locus for strain engineering efforts to identify targeted integrations of DNA constructs, using the newly developed transformation procedure for T. aurantiacus (6, 7).
The CE expression heat map (Fig. 2c) revealed that all genes displayed virtually no expression on Dglucose, with the exception of a putative chitin-deacetylase (T.ID: 45413). Most CEs showed high expression on both D-xylose and L-arabinose and low expression on cellobiose, consistent with their deconstruction of bonds in hemicellulose. None of these genes had high similarities to genes characterized in Aspergilli. The CEs that might likely be of interest for lignocellulose deconstruction are those four that showed low expression on D-glucose and no carbon and thus seemed to be speci cally expressed on one or both C5 sugars (Fig. 3c, black frame).

Discussion
The understanding of fungal CAZyme induction is fundamental to enable the e cient conversion of lignocellulose into biofuels and bio-based products. In this study, we compared the CAZyme-inducing effect of D-xylose, cellobiose and L-arabinose through a fed-batch system at shake ask scale using a peristaltic pump. Such systems have been employed before to investigate the effect of highly puri ed CAZyme inducers on enzyme secretion (12,18,21). Here, it was found that continuously feeding D-xylose, cellobiose and L-arabinose caused elevated secretion of cellulases of T. aurantiacus, which was not observed when the same amount of sugar was added in the beginning of the experiment. Notably, we found that while stimulating high enzyme secretion, individual enzymatic activities depended on which sugar was fed to the medium. Xylanase activity was almost absent in the cellobiose feed cultures, while this activity was highly induced in the C5 sugar feed cultures. While overall CMCase activity was similar among all three sugar feeds, we found elevated cellobiosidase activity during C5 sugar feeding and elevated beta-glucosidase activity during cellobiose feeding. Cellobiose feeding also resulted in culture supernatants with an increased ability to release sugars from Avicel. The reason for lower enzyme secretion during batch cultivations might be that simple sugars can cause CCR when present at high amounts, partly mediated through the carbon catabolite repressor Cre-1/Cre1/CreA (22,23,45,46). Typically, secretion of CAZyme for the breakdown of complex sugars, such as polysaccharides, is repressed until all simple sugars are reduced to low levels. However, once the sugars are depleted, there will be no further induction in batch media. Complex lignocellulosic substrates, however, are broken down over longer periods of time and the sugar release thus happens more slowly and continuously, similar to a fed-batch process. Fed-batch has been found to improve many aspects of enzyme production in bioreactors with lamentous fungi compared to batch cultivations, indicating the importance of regulated substrate levels (21,47). We argue that fed-batch tests are vital for understanding the potential of individual inducer molecules, which is not feasible when using complex carbon sources such as cellulose or xylanase.
The high level of CAZyme production when feeding puri ed sugars illuminates previous studies of fungal gene expression and enzymatic activity, which reported that simple sugars such as D-xylose, cellobiose and L-arabinose during batch fermentations weakly induced cellulases and xylanases compared to complex polysaccharides such as xylan and lignocellulosic plant biomass in T. aurantiacus (10), A. niger (16) and Talaromyces versatilis (48). These studies might have found stronger induction effects of the simple sugars on cellulases and xylanases expression when applying continuous feeding as we observed for T. aurantiacus. Nevertheless, it is widely established that complex substrates induce a broader variety of CAZy (15). Feeding mixtures of sugars or hydrolysates may generate more e cacious enzyme formulations without adding solid substrates to the fermentation medium (48-50).
The RNA-seq study established that cellulase and xylanase induction in T. aurantiacus follows a pattern similar to the related fungus A. niger, where D-xylose was found to induce production of these enzymes (15). Putative T. aurantiacus orthologues of most of the characterized cell wall degrading enzymes from A. niger were highly up-regulated during D-xylose and L-arabinose feeding. These highly expressed T. aurantiacus genes have high similarity to A. niger orthologues related to deconstruction of cellulose (exgA, eglC, cbhA/cbhB and bgl4), xylan (xynA, axlB and xlnD) and arabinan (abfA). Notably, for the cellobiose feed, expression of all the respective T. aurantiacus genes was substantially lower compared to the C5 sugar feeds, with only the putative bgl4 orthologue showing equally high expression. Only putative orthologues of exgA, bgl4, eglC and cbhA/cbhB showed increased expression during cellobiose feed. These observations indicated that cellobiose mainly induced cellulase genes. We also found that the prominent T. aurantiacus LPMO (5,51) showed highest expression during C5 sugar feeds and intermediate expression during cellobiose feeding.
The effect of D-xylose feeding was different for the T. reesei CL847 mutant where this sugar strongly induced xylanase activity but very little cellulase activities while cellobiose and lactose strongly induced cellulase activities (21). Furthermore, feeding a mixture of D-xylose and cellobiose or lactose led to reduction in cellulase activities compared to not feeding D-xylose. In contrast, feeding a mixture of Dxylose, D-Glucose and cellobiose did not decrease cellulase production of the T. reesei M3-1 mutant compared to not adding D-xylose to the feed, while xylanase secretion was substantially promoted by adding D-xylose (18). This contrary observation could be due to the different mutant strains used in the respective studies. However, unlike industrial hosts such as Trichoderma sp., D-xylose and L-arabinose alone can be used to produce cellulases and xylanases simultaneously at high amounts in T. aurantiacus.
The genetic regulation of T. aurantiacus CAZyme production at this point remains unknown. In A. niger and T. reesei it was found that both types of enzymes were induced by the same orthologues (XlnR/Xyr1) (15). Notably, XlnR in A. niger was found to induce expression of xlnD and aguA when grown on D-xylose and xylan (16). Both putative CAZyme orthologues were also highly expressed in T. aurantiacus when grown on D-xylose. This nding, together with the recent demonstration that XlnR overexpression in T. aurantiacus caused a substantial increase in xlyanase production compared to the wild type strain (6, 7), suggests that XlnR has a similar role in T. aurantiacus as in A. niger. Additonally, two further orthologues (Prot IDs 43769 and 600557) of the f-box protein gene fbxA of A. nidulans and the sexual development regulator nsdD were identi ed through Venn diagram binning in the D-xylose condition in T. aurantiacus; fbxA was found to be essential for xylanase expression in A. nidulans (34). Recently, the nsdD orthologue was deleted in P. oxalicum, which caused a substantial reduction of cellulase and xylanase secretion (35). Thus, 600557 might have a similar function in T. auranticus and could be another target for strain improvement. Further transcription factors that have been found to mediate CAZyme production were orthologues of carbon catabolite repressor Cre1/Cre-1/CreA (45,46,52), the cellulase/xylanase regulators XlnR/Xyr1, Clr-1/ClrA, Clr2/ClrB and ClbR (15,16,50,53), the amylase regulator AmyR/Col26 (54-56), the nitrogen-and starvation regualtor PacG/Vib1 (55) and further regulators involved in pectin and rhamnose utilization GaaR/Pdr-1/RhaR (57)(58)(59). Putative orthologues for all of those genes were found in T. aurantiacus and their dynamics could be uncovered using our data set. We found that the putative T. aurantiacus orthologues of all of the genes listed above except ClbR and CreA showed statistically signi cant lower expression on D-glucose (pval < 0.05) compared to most other feed and starvation conditions (Supplement Fig. 3). The comparably high expression of those regulators on no carbohydrate medium compared to the sugar feed conditions is not surprising, since some transcription factors such as CRE1 and XLNR were found to be regulated at the protein level through post-translational modi cations in other fungi (60-62). Nevertheless, CCR seemed to affect those regulators in a similar way, potentially mediated by the CreA orthologue, as it has been described for A. niger, N. crassa and T. reesei (15). Therefore, we hypothesize that overexpression of the respective T. aurantiacus genes showing high similarity to known transcriptional CAZyme activators will likely have substantial effects on expression of cellulases, xylanases and pectinases. Furthermore, we found that several gene orthologues related to UPR in A. niger were found to be highly expressed during D-xylose and L-arabinose feed in T. aurantiacus. While it is not certain at this point whether this pathway is stimulated by C5 sugars or the onset of high enzyme secretion induced by those sugars, the UPR pathway may be a promising target for future strain engineering efforts. Lastly, we identi ed several candidate genes in T. aurantiacus that had highest similarity to genes used by A. niger to assimilate D-xylose and L-arabinose. While experimental studies of their activities are required to draw nal conclusions, their high up-regulation in T. aurantiacus during growth on L-arabinose and D-xylose makes them the suitable targets for studying C5 sugar catabolism in this fungus.

Conclusion
This study has identi ed three lignocellulose derived sugars, D-xylose, L-arabinose and cellobiose, as CAZyme inducers for T. aurantiacus. Gene expression analysis indicates that the C5 sugars induce the production of a broad spectrum of CAZymes, including cellulases and xylanases while cellobiose      Gene expression analysis of (A) Glycoside hydrolases, (B) oxidoreductases and (C) esterases related to plant polysaccharide deconstruction and modi cation. Each colored cell represents the average of 3 biological replicates. The known CAZymes of T. aurantiacus that have been shown to be secreted during growth and cellulosic plant biomass that have been veri ed using shotgun proteomics are indicated