Saccharification of wheat straw by A. niger
The wheat straw used was composed of 37±1.69% cellulose and 32±1.2% hemicelluloses, 22±0.1% lignin and, after ball milling, the substrate retained 25±0.76% crystallinity (data are the mean of three replicates and are shown ± standard deviation). In order to determine the time at which degradation of the wheat straw by A. niger
had begun to take place, the monomeric sugar content of the culture supernatant was analysed by HPLC. shows that, prior to inoculation, in minimal media containing 1% straw, the concentration of free monomeric sugar present in the liquid fraction was 76±0.9 µM. After 12 h of incubation, control samples, which had not been inoculated with A. niger
, contained similar levels of each sugar. In the A. niger
cultures, the total concentration of free monomeric sugar present in the liquid fraction increased to 166±26.9 µM, showing that degradation of wheat straw polysaccharides had begun to take place. There were also changes detected in the proportions of individual sugars (). Xylose, arabinose and galactose levels were increased by a statistically significant level, whilst glucose levels were not. Two non-exclusive hypotheses could explain this observation; i) the hemicellulose fraction of the lignocelluloses substrate is degraded first, and/or ii) glucose is preferentially imported by the fungus. Indeed a requirement for glucose depletion prior to xylose utilisation, when both sugars are present, has been observed in Aspergillus nidulans
. After 24 h of incubation, levels of free sugar had not increased any further suggesting that the balance of degradation and sugar uptake had reached a steady state. RT-PCR showed that transcription of several genes encoding glycoside hydrolases, (endoglucanase; eglA
, cellobiohydrolase; cbhA
and the endoxylanases; xynA
) that are transcriptionally activated in response to xylose by the xylanolytic regulator XlnR, were highly induced at this point, when compared to expression in the Glucose 48 h cultures (data not shown). These two time-points were therefore chosen for RNA-seq analysis (Glucose 48 h and Straw 24 h). After 24 h incubation in the straw media, the particles of straw were in intimate association with A.niger
mycelia (). It is possible therefore that the responses seen are due not only to the presence of inducing molecules, but also due to the physical interaction between the fungal mycelia and the straw.
To investigate the repressive effect of glucose on expression of degradative enzymes, glucose was added exogenously to the wheat straw-incubated 24 h cultures to a final concentration of 1% (w/v). Samples were taken for RT-PCR analysis after 30 min, 1, 2 and 5 h. For all genes tested the levels of hydrolase expression decreased over this time course, reaching for most of the genes a similar level to that seen in the Glucose 48 h cultures after 5 hours of exposure to the exogenously added glucose (data not shown). Therefore, 5 hours after the addition of glucose to the straw cultures was selected as the final time point for RNA-seq analysis (Straw+Glucose 5 h). These represent three physiologically distinct conditions; long term growth under glucose repressing conditions, growth in the presence of an inducing lignocellulosic substrate and growth in the presence of the inducing substrate and glucose simultaneously. RNA was extracted from triplicate independent cultures in each condition for RNA-seq analysis.
The wheat straw-induced transcriptome of A. niger
Reads were mapped to the A. niger
ATCC 1015 genome sequence 
as it is phylogenetically very close (based on the β-tubulin sequence) to the N402 strain used in this study, and RPKMs (Reads Per Kilobase of exon model per Million mapped reads) were calculated for each annotated gene. The ATCC 1015 gene model is thought to under-predict the true number of genes present in the A. niger
so, in order to extract the maximum amount of data from our transcriptome sequencing, reads were also mapped to the CBS 513.88 sequence 
, which has a greater number of predicted genes. Approximately 2.5% more reads were successfully mapped to the ATCC 1015 genome than the CBS 513.88, reflecting the closer relationship between this strain and N402 used in this study. The CBS 513.88 genome model contains 4213 genes not included within the ATCC 1015 genome model and 939 of these genes were found to have an RPKM of 1 or more in at least one of the conditions tested in our transcriptome and are therefore very likely to be present within the ATCC 1015 genome also. A full list of the gene expression for the 4213 genes is included in supplementary material (Table S1
). Eight of these genes encoded potential polysaccharide degrading enzymes or demonstrated a transcriptional pattern in the CBS 513.8 of interest to this study and so were added to the ATCC 1015 gene model and their RPKM values from the ATCC 1015 mapping were calculated.
RPKM values for all genes were calculated for each of the separate biological replicates, as well as for the combined mapping of all three, (Table S2
). Table S3
shows that inter-replicate reproducibility was extremely high (R-squared values of >0.9). The results shown within the main text are from the combined mapping scores. Three statistical significance tests were applied to changes in gene expression, the Likelihood Ratio Test 
, Fisher's Exact Test 
, and an MA-plot-based method with Random Sampling model 
. The results of these tests for each gene are listed in Table S2
. All gene inductions discussed within the text had a p-value of <0.001 for all three statistical tests.
Expression of CAZy genes
Deconstruction of plant cell wall polysaccharides is mediated by enzymes of three major classes: Glycoside Hydrolases (GH), Carbohydrate Esterases (CE) and Polysaccharide Lyases (PL). The Carbohydrate Active Enzyme database (CAZy - http://www.cazy.org
) subdivides enzymes within these three classes into families based on their related activity and sequence. CAZy defines 281 of these enzymes within the CBS 513.88 genome, all but two of which are also predicted in the ATCC 1015 genome model (CBS 513.88 annotation: An14g07390 and An14g07420). The ATCC 1015 genome encodes 246 predicted GHs, representing 51 families; 25 CEs representing 9 families and 8 PLs representing 2 families.
After 48 hours of growth in minimal media with 1% glucose, CAZy genes represent approximately 3 percent of total mRNA, with the glucoamylase glaA (GH15 family) accounting for the majority (over 65 percent) of this (). SDS-PAGE of the culture supernatant revealed the presence of a highly predominant protein band, which was identified by tandem MS as GlaA (data not shown). Twenty-four hours after the mycelia were transferred to straw, expression of CAZy genes made up more than 19 percent of total mRNA. This is a strong over-representation of the CAZy group of genes, as they represent only ~2.5 percent of the coding genome. Thirty of the induced CAZy genes reached an expression level above 50 RPKM. They represent 14 families of GH, 2 of CE and 1 of PL ().
The diverse categories of CAZy genes expressed during exposure to straw reflect the complexity of the carbohydrates present within the substrate. However, it is interesting to note that around 65 percent of the mRNA from the CAZy group at this time-point is from genes encoding just 5 families of enzyme, GH7, 11, 61 and 62 (cellobiohydrolases, xylanases, polysaccharide monooxygenases and arabinofuranosidases, respectively) and CE1 (acetyl xylan esterases) ( and ). Proteins from each of these categories, except GH61, were also identified within culture supernatants by SDS-PAGE and tandem MS (data not shown). The fact we did not identify any GH61 proteins, that play a role in the oxidative cleavage of recalcitrant plant biomass 
, amongst the major bands could be due to a discrepancy between transcript and protein levels, or simply a technical issue in detection, such as the protein not staining well, or the protein being attached to the substrate or the fungal membrane, whilst only the supernatant was analysed. Based on transcript abundance, the 5 categories of encoded enzyme might provide the bulk of activities required for the degradation of straw. Catabolite repression, by addition of glucose to the straw cultures, exerts strong repression upon CAZy gene expression; after 5 hours, CAZy gene mRNA is reduced to only 1% of total mRNA. The glucoamylase glaA
(GH15) becoming again the most expressed CAZy gene under this condition ().
The degradative response is sequential and triggered by carbon starvation
To establish the order of induction of the genes that were highly expressed by 24 h, a time-course experiment was performed. RNA was extracted 0.5, 1, 2, 3, 6, 9 and 12 h after the switch from glucose to straw, and the expression of genes of interest was measured using RT-PCR (Figure S1
). The results show the cellobiohydrolase cbhB
to be induced after 6 hours of exposure to straw, whilst cbhA
do not appear to be induced until the 9 hour time point. These timings were verified, and shown statistically significant, using quantitative RT-PCR (Figure S2
). To investigate the regulatory basis for the differential response of these genes, expression under the glucose and straw conditions was examined in strains deleted for either the gene encoding the xylanolytic activator XlnR, that mediates xylose-induction of some GHs and esterases, or CreA, which mediates wide-domain carbon catabolite repression 
. All three of the glycosyl hydrolases showed a statistically significant (p-value of >0.01 in an equal variance, one-tailed T-test) dependence upon XlnR for maximal induction at 24 h in straw (Figure S3
), which is typical of hydrolases in A. niger
. Interestingly, a smaller scale induction of cbhB
in the straw media was still observed in the ΔxlnR
strain. This induction is mediated by alleviation of CreA repression, since in the ΔcreA
was expressed at a significantly higher level than the wild-type during growth on glucose, as assessed by qRT-PCR, whilst cbhA
were not (Figure S4
). This observation led to the hypothesis that the earlier induction seen for cbhB
is triggered, not by the presence of an inducing sugar such as xylose, but instead by the absence of an available carbon source. Transferring mycelia from glucose (48 h) to media completely devoid of carbon source for 24 h had a significant inductive effect upon cbhB
expression, whereas it did not induce cbhA
). As only a very low concentration of free xylose is present in the straw media until several hours after incubation with A. niger
(), this might explain why the induction of cbhB
occurs earlier than the induction of cbhA
CbhA and CbhB are the only two GH7-family cellobiohydrolases in A. niger
. CbhB contains a family 1 Carbohydrate-Binding-Module (CBM), whilst CbhA has no CBM 
. CBMs aid the attachment of the enzymes containing them to the complex polysaccharide surfaces of intact cell walls 
. It could be speculated that the CBM-containing enzyme is induced early because it plays an important role in targeting the relatively intact plant cell wall. Whilst the enzyme lacking the CBM is induced later, once soluble oligosaccharides have been released. Six other genes that were induced after 24 h in straw also encoded proteins containing CBM domains. Two of these genes, TID_205580 (encoding a member of the GH family 5, containing a CBM 1) and abfB
(an arabinofuranosidase containing a CBM 42), showed the same pattern of expression and regulation as cbhB
; i.e. de-repression in a ΔcreA
strain, induction 6 h after transfer to straw, and induction by carbon source starvation (Figures S3
). This CreA de-repression-dependent/XlnR activation-independent mechanism of induction may allow a subset of hydrolases that are targeted more specifically to the intact plant cell wall to play a “scouting” role under carbon starvation conditions, testing for available complex polysaccharides and liberating small quantities of sugar and oligosaccharides which trigger the subsequent larger scale induction of the genes themselves and the remaining majority of hydrolases by XlnR. Such a system would allow the organism to probe the surrounding environment for complex substrates when undergoing carbon starvation, without over-committing resources, until the presence of a degradable substrate is sensed from the release of inducing sugars. The expression of the “scouting” enzymes themselves, as well as the majority of hydrolases, which are completely dependent upon an active form of XlnR for expression, are then induced to extremely high levels (around one-fifth of the total mRNA inside the cell 24 hour after the transfer to straw, ). None of the genes identified here as being induced early are predicted to encode enzymatic activities capable of releasing xylose from wheat straw, which would lead to XlnR activation. However, our preliminary data (not shown) indicates that the genes shown here are only part of a larger subset, which may encode such activity. Defining the full subset of hydrolase genes that are induced concurrently with cbhB
is the subject of ongoing study.
Of the non-CAZy genes that are responsive to straw, both the putative lipase gene (TID_173684) and esterase, estA
(TID_51662), are induced by the 6 h time point (Figure S1
); they therefore form part of the early response, along with cbhB
and the GH5 family member TID_205580. The lipase-encoding gene shares the same CreA-dependent expression pattern as cbhB
). The genes encoding the Pth11 homologue and HsbA are also induced by a lack of carbon source but their induction shows dependence upon neither XlnR nor CreA (data not shown). The extremely early expression of the homologue of pth11
, after 3 h, may indicate a role in the early stages of the signalling pathway and is the subject of current investigation. The concurrent timing of expression of hsbA
with the majority of hydrolases, at 9 h, raises the possibility of a role in recruiting hydrolases to the surface of the straw, i.e. analogous to the recruitment of degradative enzymes to the surface of certain plastics 
. The gene encoding the hydrophobin HfbD was induced later, at 12 h. The possible functionality of surface binding proteins in the recognition of surfaces, and the possibility that those proteins recruit hydrolytic enzymes to the surface, is intriguing and may provide new avenues to enhance the action of hydrolases in the saccharification of biomass in the generation of biofuels and the synthesis of other chemicals within a bio-based economy.
The mechanism by which the presence of a complex substrate, which cannot be imported into the cell, is signalled and triggers expression of the requisite hydrolytic enzyme mix, is widely debated and may vary with fungal species. The predominant induction model in fungal systems 
proposes that general basal expression of small quantities of hydrolase begins the degradation of complex polysaccharides, thereby producing inducing compounds that elicit the full transcriptional response. It has alternatively been suggested that some relevant genes are induced by carbon source depletion, and that the derived enzymes might play a foraging role, under starvation conditions 
. Temporally differential expression of complex polysaccharide degrading enzymes and the presence of “scouting” enzymes that do not require the presence of a substrate-derived inducer for expression, but work to release inducing molecules and trigger a larger degradative response, has also been observed for the A. niger
response to pectin 
and starch 
perhaps indicating a conserved response pattern to insoluble substrates. Our data support this model for lignocellulose and we take it further by proposing a succession of events that not only includes the timing of expression of genes encoding hydrolases but also involves sensing proteins and the possibility that surface-binding proteins serve as a scaffold for recruitment of hydrolases. In summary the overall strategy appears to be an induction of a specific, small scale, sensory response by carbon source starvation, mediated at least partially by alleviation of CreA-dependent catabolite repression. This leads, in the event of successful liberation of free sugars from the wheat straw substrate, to the full scale degradative response, which is activated by XlnR ().
Induction model based on the sequential expression of responsive genes.
Natural antisense transcripts (NATs) are RNAs transcribed from a region of the genome that lies antisense (AS) to a gene. They have been found in a number of organisms, including several fungi 
, and play various regulatory roles 
. The number of reads that fell upon the AS strand was counted for each gene in our study and AS RPKM values were calculated in each condition. AS reads accounted for approximately 2% of total reads under the conditions tested (2.41, 1.94 and 2.04% in Glucose 48 h, Straw 24 h and Straw+Glucose 5 h, respectively). Although the vast majority of genes had just a few associated AS transcripts, 521 genes had an AS RPKM of greater than 1 and up to about 120 (Figure S6
and Table S4
In order to confirm that the NATs observed were biologically significant, and not simply an artefact of RNA-sequencing, a specific example, TID_53176, encoding a predicted membrane protein belonging to the GPR1/FUN34/YaaH family, which is required for acetate uptake in A. nidulans
, was chosen for further analysis. The TID_53176 sense transcript is expressed in the straw media, whilst the AS transcript is expressed in the presence of glucose ().
Sense and antisense transcription from TID_53176.
The AS coverage level is high in both glucose conditions, and extends over the full length of the predicted gene including the two introns and extends both upstream and downstream, but does not overlap any neighbouring genes (). The sense coverage, seen in the straw condition, is shorter in length and there is almost zero coverage of the introns, indicating that the vast majority of sense transcripts are fully spliced. shows that RT-PCR, using primers upstream of the first intron and downstream of the second, can distinguish between the larger AS product and the shorter, fully spliced, sense transcript. Since oligo(dT) was used as the primer for cDNA synthesis, the AS transcript must be polyadenylated. To verify the strandedness of the two products, strand-specific RT-PCR was performed using a tagged primer approach and the results confirmed that all of the larger product is generated from antisense transcripts, whilst the smaller product is the only band seen in a sense-specific reaction (). A trace amount of a smaller antisense product can be seen in the antisense-specific assay under straw 24 h conditions. This may represent a true RNA intermediate, or it could be due to the high level of sense transcript self-priming and slight carryover of primer from the cDNA synthesis priming its amplification.
To identify AS transcripts that responded to the change in carbon source we calculated the ratio of antisense
sense expression under Glucose 48 h and Straw 24 h conditions for the 521 genes with an AS RPKM of >1 (Table S4
). Genes where sense transcription is induced on straw but AS predominates on glucose, include examples of transporters and permeases, CAZy enzymes and the putative lipase TID_173684.
This putative lipase TID_173684, one of the most highly induced genes upon exposure to straw, is also one of the genes showing the most marked antisense
sense ratio switch (Table S4
). Under glucose 48 h conditions there is significant expression of both sense and AS transcripts with a 60% greater level of AS (RPKMs of 1 and 1.6 respectively). After the switch to straw, at 24 h there is a large induction of the sense transcript (~1700-fold) and AS transcription is cut to less than a third of the initial level (Figure S7A
). Standard and strand-specific RT-PCR reactions under the same conditions give a similar pattern of bands to that seen above for TID_53176 (). Interestingly, in the ΔcreA
strain sense transcription is seen in both glucose and straw conditions, suggesting that the AS/S ratio switch is regulated either directly or indirectly by CreA. This was confirmed by strand-specific RT-PCR (Figure S7B
). The timeline of induction experiment detailed earlier (Figure S1
) shows that the AS/S switch in the expression of the putative lipase can be seen to occur between 3 and 6 h after the transfer to straw, which is concurrent with the expression of the carbon starvation induced subset of genes (of which the lipase gene is part). This suggests a possible relationship between carbon source responsive regulation by CreA and antisense transcription, providing an interesting area for further study.