A long noncoding RNA promotes cellulase expression in Trichoderma reesei

An undescribed gene locus influences cellulase expression in T. reesei QM9414

Commonly, phenotypic investigation of strains generated by targeted mutagenesis requires a reference strain carrying solely the marker gene. For this reason, the amdS marker gene enabling utilization of acetamide [50] was randomly integrated into the genome of T. reesei QM9414 (ATCC 26921). To our surprise, subsequent characterization studies of the 18 resulting reference strains revealed certain phenotypic characteristics of three candidates. Enzymatic assays of liquid fungal cultures grown on xylan showed that β-glucosidase activity is significantly lower in these strains compared to the parent strain QM9414, i.e, about 59% (Fig. 1A). Moreover, also the cellulase activity was reduced about 50% in cultures that were induced on lactose (Fig. 1B).

Open image in new window

Since enzymes acting on the degradation of cellulose were affected in these strains, a crucial role of the disrupted genomic region in regulation of cellulose degradation was hypothesized. Hence, the site targeted by the amdS marker gene was identified via inverse PCR. Interestingly, sequence analysis revealed that—although a random integration strategy was followed—the amdS gene was integrated at the same locus in all three strains. This finding was verified by Southern blot analysis applying a locus-specific probe (Fig. 1C). The reason for the preference of this target site remained unclear; however, the experiment resulted in the identification of a hitherto undescribed region of the genome probably involved in regulation of PBDE expression.

Discovery of HAX1, a long noncoding RNA

Based on the genome sequence of T. reesei QM6a accessible in the Joint Genome Institute (JGI) database [51], the site targeted by the amdS marker gene was assigned to an intergenic region on scaffold 14. It is flanked upstream by a 2-isopropylmalate synthase (Protein ID 79495), and downstream by a poorly described coding region referred to as “hypothetical protein” (Protein ID 108999); both of them are minus-strand encoded (Fig. 2a). Remarkably, the targeted region seems to lack any homologues. Genome conservation analysis displayed in Fig. 2a indicates only slight sequence similarities between T. reesei to a noncoding region on the Trichoderma virens genome and no homolog regions in Nectria haematococca or Trichoderma atroviridae.

Open image in new window

To find out if the observed cellulase-depleted phenotype of the disruption strains is caused by the interruption of the undescribed genomic region or rather by any impact of the integrated amdS cassette on expression of the adjacent genes, transcript analysis of the genes encoding the hypothetical protein and the 2-isopropylmalate synthase was performed. For this purpose, the three mutants and their parent strain QM9414 were pre-grown and replaced to a cellulase inducing and a repressing condition and compared to a replacement to a medium without carbon source. As given in an additional file, transcript levels of both adjacent genes are the same in QM9414 and the mutant strains (Additional file 1). Thus, an effect of amdS integration on the cellulase activity by altering the expression of the upstream- or downstream genes can be excluded. In addition, a particular effect of the used marker system was ruled out by generation of other disruption strains bearing the hph gene (encoding hygromycin B resistance) instead of the amdS gene at the same locus.

As a consequence, the interrupted region itself was supposed to encode a gene involved in regulation of cellulose degrading enzymes. Hence, gene prediction was performed applying different gene prediction models for in silico analyses of the locus. One used algorithm provided by the gene prediction web server AUGUSTUS [52] assigned the sequence as coding for a gene. An illustration of the predicted, 782 bp-long structural gene is shown in Fig. 2b. It contains two putative introns at 113–201 and 269–325 bp from the ATG. At the 5′-end (untranslated- and coding region), a sequence element rich in Xyr1-binding sites is present.

Transcript formation of this potential gene further referred to as “hax1” was analysed via reverse transcription quantitative PCR (RT-qPCR) (Fig. 2c). To cover the whole structural gene, transcript levels of five overlapping fragments (each of them approximately 200–300 bp in length) were determined in separate PCRs. In general, transcript levels of hax1 were quite low. However, transcript formation could be verified for the 5′-end of the gene (PCR 2) as well as for the 5′-upstream region (PCR 1). Lower levels could also be detected in case of the central region including the second intron (PCR 4), but no amplicons resulted for the central region comprising both introns (PCR 3) and for the 3′-end (PCR 5). This analysis suggested that not the whole predicted gene, but mainly, the 5′-terminal region is transcribed. Therefore, the generation of the potential translation product Hax1 is rather unlikely. This assumption is further supported by the detection of a transcript in the 5′-region that comprises at least a part of the putative first intron (Fig. 2c, PCR 2), suggesting that posttranscriptional splicing events do not take place. In addition, the analysis of the codon usage [53] of the protein encoding sequence displayed in an additional file, neither supports the presence of a typical T. reesei gene, nor does it indicate an origin in lateral gene transfer from another organism (Additional file 2). Finally, a protein BLAST of the deduced amino acid sequence as well as a nucleotide BLAST of the hax1 gene to the pool of coding elements gathered in the NCBI archive [54] did not result in any hits.

Altogether, there are no indications for the generation of a translation product, even if transcript formation could be verified. These findings suggest that the region targeted by amdS insertion is a ncRNA. To concretely define the boundaries of this RNA, 3′ and 5′ rapid amplification of cDNA ends (RACE) were performed. 5′ RACE revealed a frequent occurrence of a 5′-end 127 nt upstream of the ATG. Classical 3′ RACE applying the supplied Oligo dT-Anchor primer for reverse transcription did not yield any specific products. Yet, enrichment of HAX1 from QM9414 RNA extracts using a biotinylated hax1-specific DNA probe and streptavidin-linked magnetic beads followed by poly(A)-tailing and 3′ RACE defined the major 3′-end to be located 173 nt downstream of the ATG. This end could also be verified via HAX1 enrichment and 3′ RACE, skipping the additional poly(A)-tailing step.

The resulting RNA is 299 nt in length, hence classifying it as a lncRNA. Since reverse transcription using the Oligo dT-Anchor primer for 3′ RACE analysis gave a specific product, HAX1 is supposed to be 3′ polyadenylated. According to in silico structure prediction using the RNAfold Web server [55], it is characterized by a distinct secondary structure formation (Fig. 2d), which is a typical property of lncRNAs. To this end, the function of the lncRNA HAX1 as a regulatory factor of PBDE expression was supposed.

No phenotypic effects by hax1-disruption in the T. reesei wild-type strain QM6a

As the disruption of hax1 led to a decreased cellulase expression in T. reesei QM9414, the same locus was disrupted in the wild-type strain QM6a by directed integration of the amdS gene. For this purpose, a disruption cassette comprising the 5′- and 3′-flanking regions of the targeted site for integration of the amdS marker gene was produced. Protoplast transformation led to integration of the cassette into the genome of QM6a_∆tmus53 by homologous recombination. Four of the obtained disruption strains (QM6a_Dhax1_1, _2, _3, and _8) were chosen for phenotypic analysis. Again, strains were grown in liquid media containing lactose for induction of cellulase expression. The detected enzymatic activities in the resulting culture supernatants are given in Fig. 3. Surprisingly, in HAX1 disrupted strains, cellulase activity was not reduced compared to the parent strain. To learn if this apparent lack of the characteristic phenotype is caused by a failure in transcript formation, RT-qPCR was performed. Amplicons using primers for PCR 2 (compare Fig. 2c) appeared after 28.3 cycles for cultures grown without carbon source, and 27.3 cycles for cultivation on 1% d-glucose, yielding average relative logarithmic transcript levels of 0.00 and 0.26, respectively (see Additional file 3). These results demonstrate that, though transcript formation of hax1 is given, the disruption of the locus had no effect on cellulase activity in case of the T. reesei wild-type strain QM6a. This raised the question if the role of the lncRNA HAX1 might differ in the various strains of T. reesei.

Open image in new window

HAX1-length differs in different T. reesei strains

As the resulting phenotype from the disruption of hax1 indicated differences in various T. reesei strains, 5′ and 3′ RACE for HAX1 was performed also in QM6a and Rut-C30. In both strains, 3′ RACE of HAX1 enriched from total RNA extracts verified the major 3′-end located 173 nt downstream of the ATG, as defined before in QM9414. In contrast, the 5′-ends of the lncRNA turned out to be different in the strains. Whereas in QM9414, the major transcription start point was located 127 nt upstream of the ATG, the most frequent 5′-end in QM6a was found 89/90 nt upstream of the ATG yielding a shorter transcript of 262 nt in length. Only rarely, longer versions of HAX1 (5′-end up to − 151 nt) were detected in QM6a, just as in QM9414. However, the highest abundance of long HAX1 versions was found in the hypercellulolytic strain Rut-C30. There, the major 5′-end is located − 255 nt of the ATG, albeit also in this case sporadically both longer (e.g., − 266 nt) and shorter (e.g., − 160 nt) versions were detected. This allows the possibility that not only the presence or the absence of HAX1, but also a particular length can have a distinct regulatory impact. To learn if the different levels of cellulase activities could result from variations in the genomic organisation of the hax1 locus rather than from HAX1 length, the locus was sequenced in QM9414 and Rut-C30 and aligned to the sequence of QM6a. Both the sequence of the locus and the position relative to the adjacent genes were the same in all three strains (see Additional file 4), hence pointing to the transcript length as the important difference. Provided that the RNA length is a relevant property, we investigated if it might be determined by a difference in RNA stability of the HAX1 versions. We could not observe degradation of any of the three HAX1 versions up to 2 h after the transcription was stopped by the transcription inhibitor DRB (see Additional file 5). Thus, we exclude that the different lengths are a result of different stability, but rather of a different transcription start site.

The respective ends of the major HAX1 versions identified in QM6a (hax1_QM6a), QM9414 (hax1_QM9414), and Rut-C30 (hax1_Rut-C30) are marked on the schematic illustration in Fig. 4a. Abundance of the transcript versions in the different strains was tested by a qualitative PCR using cDNAs as templates and a limited number of cycles to enable an estimation of the quantities. As expected, hax1_QM6a can be detected in all three strains, whereas the longer transcript versions hax1_QM9414 and hax1_Rut-C30 are mainly amplified from QM9414 and Rut-C30, respectively (Fig. 4b). Levels of hax1_Rut-C30 were very low, giving only faint bands. Albeit this version is detectable in all three strains, it is predominantly detectable in samples derived from Rut-C30 compared to QM6a and QM9414, especially on sophorose as a cellulase inducing condition (Fig. 4b).

Open image in new window

To estimate if HAX1_QM6a, HAX1_QM9414, and HAX1_Rut-C30 could fulfil different functions, in silico structure prediction was performed using the RNAfold Web server [55]. The secondary structures of minimized free energy are displayed in Fig. 4c. All of them are characterized by a lncRNA-typical formation of duplexed RNA. They are composed of a stem forming a hairpin-rich region at one end and splitting into two stem loops on the other end. In HAX1_QM6a and HAX1_QM9414, most of the hairpins are quite well conserved; however, in HAX1_QM9414, an additional hairpin is breaking the structure of the main stem. For HAX1_Rut-C30, in contrast, the overall structure is rather different. In fact, the three versions of HAX1 only have one stem loop in common. This supports the assumption that those RNAs can act differently.

Overexpression of hax1 rescues the cellulase-depleted phenotype in QM9414_Dhax1 strains

To understand the distinct impact of HAX1_QM6a, HAX1_QM9414, and HAX1_Rut-C30, all three versions were overexpressed in two of the QM9414 HAX1 mutant strains (i.e., QM9414_Dhax1_8 and _14). As an initial control experiment, the absence of the three hax1 versions was tested in the two recipient strains by RT-qPCR (see Additional file 6). The expression constructs comprise the native bgl1 (β-glucosidase 1) promoter [56] and the respective version of hax1, which are flanked by the pyr4 (encoding the orotidine 5′-phosphate decarboxylase) upstream- and downstream regions. Consequently, homologous integration at the pyr4 locus resulted in a 5-fluoroorotic acid (5-FOA) resistant and uridine auxotroph phenotype. The usage of the bgl1 promoter, of which the transcription start point is known, secured that only hax1 is expressed. For characterization, the overexpression strains and the two parent strains were cultivated in liquid culture on 1% lactose for 72 h. Supernatant samples were taken after 37, 48, 60, and 72 h. Cellulase activity was increased in all overexpression strains compared to the parent strains after 48 h of cultivation (Fig. 5A). Moreover, differences resulting from overexpression of the distinct hax1 version became evident: overexpression of hax1_QM6a has less impact on cellulase activity than hax1_QM9414, and overexpression of hax1_Rut-C30 yielded the highest cellulase activity. Even though the results slightly differed between the two strain lines, exactly the same trend could be found. As shown for QM9414_Dhax1_14, this trend remains constant during the whole time course of the experiment (Fig. 5B) and can also be detected at the final time point if referred to the biomass (Fig. 5C).

Open image in new window

Overexpression of hax1 enhances cellulase activity in T. reesei QM6a

To learn if the expression of hax1_QM6a, hax1_QM9414, and hax1_Rut-C30 can change cellulase expression in the T. reesei wild-type strain QM6a, the before used constructs were introduced into QM6a_∆tmus53 and integrated at the pyr4 locus by homologous recombination. Subsequently, the resulting strains overexpressing either of the hax1 versions as well as the parent strain (QM6a_∆tmus53) and a reference strain (QM6a_∆tmus53_∆pyr4) were characterized via enzymatic assays. Again, a time course experiment was performed growing the strains on 1% lactose for cellulase induction. The enzymatic activities of samples prepared after 37, 48, 60, and 72 h were determined. As observed before for hax1 overexpression in QM9414_Dhax1, the cellulase activities are importantly higher in all QM6a_OEhax1 strains compared to the parent and reference strain (Fig. 6A). Again, depending on the expressed version of HAX1, different levels of cellulase activities were achieved: the overexpression of the longest HAX1 version identified in Rut-C30 had the largest impact on cellulase activity. This is true for different times of cultivation as well as for values referred to the final biomass (Fig. 6A, B). Interestingly, in the case of the wild-type strain QM6a, the effect of overexpressing hax1 was even more pronounced than in the case of QM9414. For example, the cellulase activities of strains overexpressing hax1_Rut-C30 in QM9414_Dhax1 were threefold higher compared to the parent strain, whereas in QM6a_OEhax1_Rut-C30, they were even fivefold increased. Conclusively, the introduction of additional hax1 copies (especially the long version hax1_Rut-C30) did not only restore cellulase expression in hax1-disrupted strains, but it suggests itself as a tool for enhancing cellulase production.

Open image in new window

Fungal strains and growth conditions

For carbon source replacement experiments, mycelia were pre-cultured in 250 ml of Mandels–Andreotti (MA) medium [65] supplemented with 1% (w/v) glycerol as the sole carbon source on a rotary shaker (180 rpm) at 30 °C for 24 h. A total of 10⁹ conidia per litre (final concentration) were used as the inoculum. Pre-grown mycelia were washed, and equal amounts were resuspended in 20 ml MA medium without any carbon source (reference condition) or MA medium containing 1% (w/v) d-glucose or 1.5 mM sophorose. Samples were taken after 3 h of incubation from three biological replicates.

For direct cultivations for the β-glucosidase assays, 30 ml MA medium containing 1% (w/v) oat spelt xylan (Sigma-Aldrich, St. Louis, MO, USA) were inoculated with 10⁸ conidia per litre (final concentration) and incubated at 30 °C and 250 rpm for 66 h. For direct cultivations for cellulase assays, the Dhax1 strains were incubated for 48 h and 70 h at 30 °C and 180 rpm in 100 ml MA medium containing 1% (w/v) α-d-lactose inoculated with 10⁹ conidia per litre (final concentration). For time course experiments, the OEhax1 strains were incubated for 72 h and 10 ml samples were taken after 37, 48, and 61 h.

Plasmid construction

Construction of p5-A-3 used for disruption of the hax1 locus in the T. reesei wild-type strain QM6a was done as follows. The 452 bp 5′- and the 1066 bp 3′-flanks of the targeted site for insertion of the amdS marker gene were amplified using the primers locus 5′for SalI/locus 5′rev HindIII and locus 3′for Acc65I/locus 3′rev XbaI, respectively, and cloned into pGEM-T (Promega, Madison, WI, USA). Subsequently, the 5′-fragment was released with HindIII and SalI and ligated into the vector pAMDS (a derivate of pUC19 containing the 3864 bp amdS gene [23]) digested with the same enzymes. Finally, the 3′-fragment released with Acc65I and XbaI was introduced into the plasmid p5-A resulting from the first cloning step, downstream of the amdS marker gene. The final construct p5-A-3 contains the hax1 5′- and 3′-flanks interrupted by the amdS marker gene in forward orientation.

For generation of the hax1 overexpression constructs pCD-∆pyr4-Pbgl1-hax1_QM6a, pCD-∆pyr4-Pbgl1-hax1_QM9414 and pCD-∆pyr4-Pbgl1-hax1_Rut-C30, the respective hax1 versions were amplified using chromosomal DNA of T. reesei QM6a as template. The following primers were used: hax1 for_QM6a_BcuI and hax1 rev_3′QM6a for the 262 bp hax1_QM6a; hax1 for_QM9414_BcuI and hax1 rev_3′QM6a for the 299 bp hax1_QM9414 and hax1 for_Rut-C30_BcuI and hax1 rev_3′QM6a for the 428 bp hax1_Rut-C30. The purified PCR products were blunt-end ligated into pJET1.2 (Thermo Scientific, Waltham, MA, USA) and the appropriate orientation was verified by digest with BcuI and XbaI. In a next step, a 997 bp fragment of the bgl1 promoter was PCR-amplified using the primers Pbgl1 for_Kpn2I and Pbgl1 rev-NheI, then digested with Kpn2I and NheI, and subsequently cloned into the respective pJET-hax1 vectors digested with Kpn2I and BcuI. To prevent positioning of the hax1 gene and the foreign 5′-untranslated region of bgl1 next to each other, the terminal point of the promoter fragment was chosen to be equal to the transcriptional start point previously defined for this gene [56]. For construction of the final hax1 overexpression cassettes, the Pbgl1-hax1 fusion products were isolated from the plasmids by digestion with Kpn2I and XbaI, extracted from a gel, and introduced into BcuI/Kpn2I-digested pCD-∆pyr4 (carrying the cbh2 terminator [66]) in forward orientation.

Fungal transformation

Accidental disruption of hax1 (Dhax1) in QM9414 occurred by usage of 9 µg of the vector pAMDS (a derivate of pUC19 containing the 3864 bp amdS gene) in an optimized protocol for particle bombardment [67]. Selection for positive transformants carrying the amdS gene, was performed on minimal medium (72.9 mM KH₂PO₄, 10.2 mM sodium citrate, 1% (w/v) d-glucose, 20 ml/L trace element solution (see MA medium [65]), and 1.5% agar noble dissolved in ultrapure H₂O) containing acetamide. After autoclaving, acetamide, CsCl and MgSO₄ were added from sterile-filtered stock solutions to final concentrations of 10, 10, and 24.3 mM, respectively.

Protoplast transformation of T. reesei was performed as described previously [68]. QM6a_Dhax1 strains resulted from transformation of 13 µg hax1 5′-amdS-3′ DNA derived from a HindIII/Acc65I-digest of p5-A-3 and extraction from a gel. The template DNA was resuspended in 15 µl sterile dH₂O and used for transformation of 10⁷ protoplasts (in 150 μl) of QM6a_∆tmus53. 100 μl–2 ml of the transformation reaction were added to 20 ml melted, 50 °C warm amdS selection medium supplemented with 1.2 M sorbitol. This mixture was poured into sterile petri dishes.

For overexpression of hax1 (OEhax1), 200 µg of the NotI-digested construct pCD-∆pyr4-Pbgl1-hax1_QM6a, pCD-∆pyr4-Pbgl1-hax1_QM9414, or pCD-∆pyr4-Pbgl1-hax1_Rut-C30 were used for transformation of 10⁷ protoplasts (in 150 μl) of QM9414_Dhax1 or QM6a_∆tmus53. Selection for pyr4 deleted transformants was performed on MEX agar containing 1.2 M sorbitol, 1.5 mg/ml 5-FOA, and 5 mM uridine as described by Derntl and co-workers [66].

Plates were incubated at 30 °C for 3–7 days until colonies were visible.

Genotypic characterization

For an initial identification of uridine auxotroph OEhax1 strains, candidates were grown on MA medium containing 1% glycerol as a carbon source without peptone or uridine. The OEhax1 candidates that were unable to grow under these conditions were tested by PCR using the primers 5pyr4_fwd(BglII) and hax1 rev_3′QM6a.

For Dhax1 strains, homokaryotic strains were generated by three rounds of vegetative spore propagation on selection medium. In case of QM9414_Dhax1 strains, the locus of integration was identified via inverse PCR and verified by Southern blot analysis. In case of QM6a_Dhax1 strains correct integration of the construct was tested by PCR and by sequencing.

Extraction of chromosomal DNA for candidate screening and Southern blot analysis was performed as described previously [65, 66]. For Southern blot analysis, the chromosomal DNA of QM9414_Dhax1 strains was digested with SacII, resulting in a 2960 bp fragment specific for the wild type and a 5178 bp fragment specific for disruption of hax1. The locus-specific biotinylated probe applied for hybridization was derived from a PCR using the primer pair locus for and locus rev and the Long-PCR Enzyme Mix (Thermo Scientific) for amplification.

PCR analysis for screening for Dhax1 candidates was performed using the same polymerase and primers. For verification of OEhax1 transformants via PCR, GoTaq G2 polymerase (Promega) was applied. DNA sequencing was performed at Microsynth. All primers used for candidate screening are listed in Table 2.

Inverse PCR

For the identification of the locus that was targeted by integration of the amdS marker in QM9414_Dhax1 strains, an inverse PCR was performed as follows. 20 µg of chromosomal DNA were digested with either Acc65I or NotI (Thermo Scientific) at a final concentration of 1 U/µl in a total reaction volume of 20 µl according to the manufacturer’s instructions. After heat inactivation, 2 µl of T4 Ligase (Promega, 1–3 U/µl) and the corresponding buffer were added and ligation was performed at 18 °C for 90 min. Subsequently, the ligation was stopped by heat inactivation and 1 µl was applied as template in a 25 µl inverse PCR reaction, initially using the primers amdS inv for and amdS inv rev. For further approaches, the primers locus for and locus rev annealing to the identified regions were applied. All amplifications were performed in an iCycler (Bio-Rad, Hercules, CA, USA) using the Long-PCR Enzyme Mix (Thermo Scientific) and the following program: initial denaturation at 94 °C for 3 min, followed by 30 cycles of 30 s at 94 °C, 30 s at 59 °C and 3 min at 72 °C, and final elongation at 72 °C for 5 min. The DNA fragments were sequenced by MWG Biotech (Ebersberg, Germany).

RNA stability assay

The strains QM6a, QM9414, and Rut-C30 were cultivated in 25 ml MA medium containing 1% (w/v) α-d-lactose and 10⁹ conidia per litre (final concentration) at 30 °C and 180 rpm for 24 h. The cultivation was carried out in biological triplicates. Then, the transcription was stopped by addition of 16 µg/ml (final concentration) of the inhibitor DRB. The cultures were further incubated and after 30, 60, 90, and 120 min, 500 µl samples were taken. A reference sample was taken from each culture before addition of DRB. The three biological replicates of each strain were pooled to one 1.5 ml sample. Mycelia were harvested by centrifugation and frozen in liquid nitrogen. Finally, the RNA was extracted, cDNA synthesis was performed and the samples were analysed via quantitative PCR as described in “Transcript analysis” section.

Transcript analysis

Extraction of RNA from fungal mycelia, cDNA synthesis, and quantitative PCRs were performed as described previously [69]. Template cDNAs were diluted 1:100 (QM9414) or 1:20 (QM6a). Analysis was carried out in triplicates. The following PCR protocols were run: 3 min initial denaturation at 95 °C, followed by 50 cycles of 15 s at 95 °C, 15 s at 60 °C and 20 s at 72 °C (for hax1 and act) or 3 min initial denaturation at 95 °C, followed by 40 cycles of 15 s at 95 °C and 120 s at 64 °C (for sar1). Control reactions, data normalization using sar1 and act as reference genes and calculations were performed as published previously [70].

Qualitative PCRs for estimating the abundance of hax1_QM6a, hax1_QM9414, and hax1_Rut-C30 were performed using the Q5 High-Fidelity DNA Polymerase (New England Biolabs, Ipswich, MA, USA) according to the manufacturer’s instructions. The template cDNAs were diluted 1:20. Amplification was performed running the following program: initial denaturation at 98 °C for 30 s, followed by 35 cycles of 10 s at 98 °C, 10 s at 70 °C and 20 s at 72 °C, and final elongation at 72 °C for 2 min. The resulting fragments (262, 299, and 428 bp) were analysed on a 1.5% agarose gel applying a GeneRuler 50 bp DNA Ladder (Thermo Scientific) for size estimation. Primer sequences for all transcript analyses are provided in Table 2.

RACE and enrichment of HAX1

5′ and 3′ RACE was performed using the 5′/3′ RACE Kit, 2nd generation (Roche, Basel, Switzerland). For PCRs, the GoTaq G2 polymerase (Promega) was applied. Primer sequences for RACE are provided in Table 2.

5′ RACE was carried out according to manufacturer’s instructions, applying 0.9–1 µg of DNase I digested RNA extracts for cDNA synthesis in a total reaction volume of 20 µl. For this initial reverse transcription step, the gene specific primer hax1 rev_1.Intron was used. After RNase A digestion and purification with the QIAquick PCR Purification Kit (Qiagen, Hilden, Germany) based on the modified protocol for RACE applications recommended by Roche, a poly(A)-tailing was performed. Subsequently, the hax1 fragments were specifically amplified from the total pool with the Oligo dT-Anchor Primer (included in the kit) and hax1 rev_5′RACE_2 in a first PCR. The resulting product was diluted 1:50 and used as a template for a nested PCR with either hax1 rev_up-Intron or hax1 rev_ 5′RACE_4 and the PCR Anchor primer included in the kit.

For 3′ RACE, classical analysis was performed according to manufacturer’s instructions, applying 0.45 µg of DNase I digested RNA from T. reesei QM9414 in a total reaction volume of 20 µl. After reverse transcription with the Oligo dT-Anchor Primer, cDNAs were RNase A digested and used for amplification of hax1 applying the gene specific primers up hax1 for_2 and hax1 for_3′RACE_3 for the initial and nested PCRs, respectively. Further 3′ RACE approaches based on prior enrichment of HAX1 were performed. For this purpose, HAX1 was enriched from 1.25 to 2.1 mg total RNA extract using a biotinylated and HPLC-purified hax1 specific DNA probe (sonde hax_5-Biotin) as well as streptavidin-linked magnetic beads included in the µMACS™ Streptavidin Kit and the corresponding µMACS separator (Miltenyi Biotec, Bergisch Gladbach, Germany) based on manufacturer’s instructions. According to the unusual high calculated melting temperature of the biotinylated probe, initial denaturation was performed at 85 °C for 5 min and annealing to appropriate amounts of streptavidin-linked magnetic beads was done at 70 °C for 15 min. RNase-free TEN buffer (10 mM Tris/HCl, pH 8.0; 1 mM EDTA; 100 mM NaCl₂) and TE buffer (10 mM Tris/HCl, pH 8.0; 1 mM EDTA) were used for binding and washing, respectively. Enriched RNA was eluted with 150 µl RNase-free dH₂O and digested with DNase I (Thermo Scientific). For poly(A)-tailing followed by RACE analysis, the enriched RNA was precipitated with isopropanol, washed with 70% (w/v) ethanol and dissolved in 20 µl RNase-free dH₂O. Poly(A)-tailing was performed using 19 µl of the resulting RNA in a total reaction volume of 25 µl. For proceeding to cDNA synthesis without prior poly(A)-tailing, the RNA was purified using the GeneJET RNA Cleanup and Concentration Micro Kit (Thermo Scientific) as instructed in the protocol for purification of DNase I digested samples. In this case, the RNA was eluted from four columns with 10 µl each and pooled for cDNA synthesis. Reverse transcription and further 3′ RACE analysis steps were performed as described before using the RACE Kit and applying the gene specific primers up hax1 for_2 and hax1 for_3′RACE_3 for the initial and nested PCRs, respectively.

For both, 3′ and 5′ RACE, the final PCR products were extracted from a gel, blunt-end ligated into pJET1.2 (Thermo Scientific) and analysed by sequencing (Microsynth).

Enzyme assays

The β-glucosidase activity was assayed in 50 mM sodium citrate buffer at pH 5.0 using p-nitrophenyl-β-d-glucopyranoside as a substrate. The enzyme assay was performed at 50 °C for 30 min as previously described [71] and the activity was calculated from the absorbance at 405 nm based on the Lambert–Beer law.

The cellulase activity was assayed in 25 mM sodium acetate buffer pH 4.5 using Azo-Cellazyme C tablets (Megazyme, Wicklow, Ireland) as substrate, essentially according to the manufacturer’s instructions. The reaction time was increased to 30 min–5 h until detectable values were obtained. The same reaction times were used for all strains cultivated for the same time period within one experiment, and samples with higher cellulase activity were adjusted by dilution to enable comparison. Cellulase activities given in μ were calculated from the absorbance at 590 nm for 10 min reaction time based on the equation μ = 232.6 * Abs + 5 [72].

One unit is defined as the amount of enzyme required to release 1 µmol of d-glucose reducing-sugar-equivalents per minute under the respective assay conditions. For the final timepoints of cultivation, the enzymatic activities were referred to the biomass dry weight derived from incubation of the harvested mycelia at 80 °C for 24 h.

Structure and sequence analysis tools

Genome analysis and BLAST were performed using the NCBI database [54] and the sequence of Trichoderma reesei QM6a v2.0 accessible in the JGI database [51]. For structural gene prediction, the web server AUGUSTUS described by Stank and Morgenstern was used [52]. Codon usage was calculated applying the online Codon Usage Calculator of Biologics Corp [53]. In silico prediction of RNA secondary structures of minimized free energy was performed using the RNAfold Web server [55] included in the ViennaRNA Package 2.0 [73], provided by the University of Vienna. Displayed structures are based on HAX1 sequences without polyA-tail; however, the addition of a random polyA-tail did not change structures.

Statistical test

For statistical analyses of the data, the program GraphPad Prism 5.00.288 was used to perform ANOVA (P < 0.05) tests and Tukey’s multiple comparison test as posttest.