Optimization of cellulolytic enzyme components through engineering Trichoderma reesei and on-site fermentation using the soluble inducer for cellulosic ethanol production from corn stover

Background Cellulolytic enzymes produced by Trichoderma reesei are widely studied for biomass bioconversion, and enzymatic components vary depending on different inducers. In our previous studies, a mixture of glucose and disaccharide (MGD) was developed and used to induce cellulase production. However, the enzymatic profile induced by MGD is still not defined, and further optimization of the enzyme cocktail is also required for efficient ethanol production from lignocellulosic biomass. Results In this study, cellulolytic enzymes produced by T. reesei Rut C30 using MGD and alkali-pretreated corn stover (APCS) as inducer were compared. Cellular secretome in response to each inducer was analyzed, which revealed a similar enzyme profile. However, significant difference in the content of cellulases and xylanase was detected. Although MGD induction enhanced β-glucosidase production, its activity was still not sufficient for biomass hydrolysis. To overcome such a disadvantage, aabgl1 encoding β-glucosidase in Aspergillus aculeatus was heterologously expressed in T. reesei Rut C30 under the control of the pdc1 promoter. The recombinant T. reesei PB-3 strain showed an improved β-glucosidase activity of 310 CBU/mL in the fed-batch fermentation, 71-folds higher than that produced by the parent strain. Meanwhile, cellulase activity of 50 FPU/mL was detected. Subsequently, the crude enzyme was applied for hydrolyzing corn stover with a solid loading of 20% through separate hydrolysis and fermentation (SHF) and simultaneous saccharification and fermentation, respectively, for ethanol production. Better performance was observed in the SHF process, through which a total of 119.9 g/L glucose was released within 12 h for concomitant ethanol production of 54.2 g/L. Conclusions The similar profile of cellulolytic enzymes was detected under the induction of MGD and APCS, but higher amount of cellulases was present in the crude enzyme induced by MGD. However, β-glucosidase activity induced by MGD was not sufficient for hydrolyzing lignocellulosic biomass. High titers of cellulases and β-glucosidase were achieved simultaneously by heterologous expression of aabgl1 in T. reesei and fed-batch fermentation through feeding MGD. We demonstrated that on-site cellulase production by T. reesei PB-3 has a potential for efficient biomass saccharification and ethanol production from lignocellulosic biomass. Electronic supplementary material The online version of this article (10.1186/s13068-018-1048-5) contains supplementary material, which is available to authorized users.


Background
Lignocellulosic biomass has attracted great attention for producing biofuels and chemicals due to its abundance and renewability [1]. However, the high cost of cellulase production is the major bottleneck for the economic feasibility [2]. Trichoderma reesei has been widely studied owing to its capacity to secrete large amounts of proteins. Among various T. reesei strains, the well-known hypercellulase producer T. reesei Rut C30 is of particular interest for the industrial production of cellulases [3].
Cellulase production by T. reesei requires induction. Cellulolytic enzyme cocktails from T. reesei with different compositions were obtained when different inducers were used, which lead to different biomass hydrolysis efficiency [4]. Usually, lignocellulose is a natural inducer for cellulase production by T. reesei. However, the low titer and productivity of cellulases remain as challenges [5]. We previously developed a low-cost soluble inducer (mixture of glucose and disaccharide, MGD), which enabled high cellulase titers and productivity [6]. However, the profile of cellulolytic enzymes induced by MGD has not been investigated in detail.
T. reesei cellulases normally have low β-glucosidase activity (BGA), which causes the accumulation of cellobiose that resulted in the inhibition of the production of endo-and exo-cellulases [10,11]. In our previous study, improved β-glucosidase production of T. reesei Rut C30 engineered by an artificial transcription factor leads to improved glucose release [11]. Increased BGA was also achieved by overexpression of either heterologous or endogenous β-glucosidase genes using the cellulase or xylanase promoters, but unfortunately, reduced expression of cellulase or xylanase genes was observed [12][13][14]. Employment of strong constitutive promoter for bgl1 gene overexpression in combination with the feeding of soluble inducer [6] may be an alternative strategy for improving BGA without comprising the production of other cellulolytic enzymes. However, such attempt has not been reported so far.
In this study, the profile of cellulolytic enzymes induced by MGD was investigated in detail. In addition, to further improve BGA in T. reesei, gene (aabgl1) encoding β-glucosidase 1 in Aspergillus aculeatus was overexpressed in T. reesei Rut C30 under the constitutive pyruvate decarboxylase promoter pdc1. The recombinant strain was used for cellulase production by fed-batch fermentation through feeding the soluble inducer. Furthermore, cellulosic ethanol production was performed by Saccharomyces cerevisiae through separate hydrolysis and fermentation (SHF) or simultaneous saccharification and fermentation (SSF) using the crude cellulases. To our best knowledge, this is the first time that high titers of both cellulases and β-glucosidase have been achieved by the fed-batch fermentation. The lignocellulolytic enzyme cocktail reported in this study showed a great advantage in the saving cost for the supplementation of commercial β-glucosidase and facilitate cellulosic biomass bioconversion.

Microorganisms and culture media
The T. reesei Rut C30 strain was purchased from the ARS (NRRL) Culture Collection, and the seed and fermentation medium for T. reesei Rut C30 in shake flask were described elsewhere [6]. Escherichia coli DH5α was used as host strain to construct various plasmids. Agrobacterium tumefaciens AGL-1 was used for A. tumefaciensmediated transformation (AMT) of T. reesei. E. coli and A. tumefaciens were cultured on LB (Luria-Bertani) medium. The commercial Angel Yeast Saccharomyces cerevisiae (Angel Yeast Co., Hubei, China) was used for cellulosic ethanol fermentation. MGD and alkali pretreatment corn stover (APCS) were prepared as described in our previous study [6].
feeding MGD. We demonstrated that on-site cellulase production by T. reesei PB-3 has a potential for efficient biomass saccharification and ethanol production from lignocellulosic biomass.

Preparation of two kinds of inducers
The soluble inducer MGD was prepared as described previously, which was produced from 600 g/L glucose transglycosylated by β-glucosidase [6]. Another inducer is pretreated corn stover, which was produced from corn stover pretreated by 2% NaOH for 90 min [15]. The chemical compositions of APCS were determined to be 62.6% cellulose, 21.4% hemicellulose, and 8.2% lignin of the dry mass (DM).

Cellulase and β-glucosidase production by fed-batch fermentation
The cultivation medium and conditions for seed and hyphal growth were the same as described in our previous study [6]. A 7 L fermenter was used for cellulase and/ or β-glucosidase production with the initial volume of 3 L. When MGD was used as inducer for cellulase and/or β-glucosidase production, the media and culture conditions were consistent with our previous reports [6]. When APCS was used for cellulase production, the fermentation conditions were the same with that of MGD except that the APCS was fed at 8 h intervals at a rate of 0.8 g/L/h [16].

Component analysis of the lignocellulose cocktails by different inducers
Samples taken from fermenters were centrifuged at 6000 rpm for 5 min at 4 °C, the supernatant was collected and filtrated through a 0.2 μm filter, then it was concentrated using a lyophilizer, and the protein concentration was measured by the Bradford method (Sangong, China). Totally 100 μg protein for each sample was mixed with fivefold volume of cold acetone at − 20 °C for 1 h, then the protein solution was centrifuged at 12,000 rpm for 15 min. The precipitate was collected, and incubated at 60 °C for 1 h, and then 50 μL dissolution buffer and 4 μL reducing reagent were added in the precipitate. The 2 μL cysteine-blocking reagent was added into the protein solution at room temperature for 10 min, followed by the centrifugation with 10 Da ultrafiltration tube at 12,000 rpm for 20 min. Then the ultrafiltration tubes were purged by 100 μL dissolution buffer and centrifuged at 12,000 rpm for 1 min for three times. Protein digestion was performed by adding 50 μL sequencing-grade trypsin (50 ng/μL) to an ultrafiltration tube and incubated at 37 °C for 12 h. After ultrafiltration tube was centrifuged at 12,000 rpm for 20 min, the filtrate with the peptides was gathered for protein labeling and MS analysis.
The iTRAQ (isobaric tags for relative and absolute quantification) labeling of peptides obtained from the test samples induced by MGD or APCS was performed using an iTRAQ reagent Multiplex kit (Applied Biosystems, Foster City, CA) according to the manufacturer's protocol. The LC-MS/MS analysis of fractionated labeled sample was performed with a QStar Elite mass spectrometer (Applied Biosystems/MDS Sciex) coupled with online microflow HPLC system (Shimadzu). The peak areas of the iTRAQ reporter ions reflect the relative abundance of the proteins in the samples. Sequences were mapped based on the reference genome of T. reesei Rut C30, obtained from the Joint Genome Institute (JGI) Genome Portal (http://genom e.jgi.doe.gov/pages /searc h-for-genes .jsf?organ ism=Trire RUTC3 0_1).

Construction of the aabgl1 expression cassette
The primers used in this study are listed in Additional file 1: Table S1. The pdc1 promoter (1500 bp) and pdc1 terminator (1000 bp) were amplified from T. reesei Rut C30 genome by PCR with the corresponding primers. Similarly, aabgl1 gene (GenBank: D64088.1) and hygromycin B cassette were amplified from S. cerevisiae MNII/cocδBEC3 genome [17] and pAN7-1 (GenBank: Z32698.1), respectively. The resistance gene of hygromycin B cassette was inserted into pCB301 (GenBank: AF139061.1) digested by SpeI, yielding the backbone vector pCBHYGB. The amplified pdc1 promoter, aabgl1 gene, and pdc1 terminator were fused to pCBHYGB using the seamless cloning master mix according to the manufacturer's instructions (Sangon Biotech, Shanghai, China), generating the vector pCBHYGB-PB (Additional file 3: Figure S1A) which was verified by sequencing.

Transformation of T. reesei Rut C30 and qPCR analysis
The vector pCBHYGB-PB harboring the overexpression cassette of aabgl1 was transformed into T. reesei Rut C30 using A. tumefaciens-mediated transformation (ATMT) method previously developed [18]. Twenty transformants were verified by PCR with the primer aabgl1-F/R and evaluated in shake flasks. qPCR analysis of the transformant T. reesei PB-3 was carried out according to the previous work [6], and the primers are listed in Additional file 1: Table S1. Transcription level of sar1 was used as an internal control.

Determination of copy numbers by qPCR analysis
The genomic DNA of T. reesei transformants, isolated using fungal DNA extraction kit (Sangon Biotech, Shanghai, China), was used as template for qPCR analysis. The aabgl1 copy number in transformants was determined as described by Mason et al. [19]. The tef1α (translation elongation factor 1-alpha, NCBI Reference Sequence: XM_006963994.1) gene was used to represent a single copy gene [14].

Enzymatic saccharification in shake flasks
The crude cellulases induced by different inducers from T. reesei Rut C30 were collected by centrifugation (6000 rpm, 5 min). The enzymes were mixed with APCS and then re-suspended in HAc-NaAc buffer (0.2 M and pH 4.8) with an enzyme loading of 3 mg protein or 20 FPU/g APCS and a solid loading of 5% (w/v). Reaction solutions were incubated in a shaking bath at 100 rpm, 50 °C. The samples were collected at different time points, followed by centrifugation at 8000 rpm for 2 min, and the supernatants were subjected to analysis of glucose concentration.

Ethanol fermentation in bioreactor using in-house cellulases
SHF and SSF processes were carried out in 2.5 L fermenter (Kobio Tech Co., Korea) at a total biomass loading of 20% DM (200 g solid/L H 2 O). The pH value was maintained at 4.8 by adding 0.2 M H 2 SO 4 or aqueous ammonia automatically. Angel yeast (1 g/L) was used for ethanol fermentation containing 1 g/L yeast extract and 1 g/L peptone. In SHF process, APCS was degraded by the recombinant T. reesei PB-3 crude enzyme with enzyme loading of 30 FPU/g biomass at 50 °C for 108 h, followed by ethanol fermentation at 30 °C. The hydrolysis was started with 7% APCS uploading, and four feedings were carried out at 12, 24, 36, and 48 h, respectively, each with 4, 4, 4, and 1% solid uploading. Conditions for SSF were almost like SHF except that the fermentation temperature was maintained at 37 °C throughout the process and commercial yeast was inoculated at 12 h in SSF.

Analytical methods
Total reducing sugars, filter paper activity (FPA), BGA, and xylanase activities were assayed according to the methods described elsewhere [20][21][22]. The cellulosic composition of both corn stover and APCS were determined via the analytical procedures recommended by NREL. The concentration of glucose and ethanol was detected by HPLC (Waters 410, Waters, MA) equipped with an Aminex HPX-87H column (300 mm × 7.8 mm, Bio-Rad, Hercules, CA). The column oven temperature was set at 65 °C. Samples were eluted with 0.004 M H 2 SO 4 at a flow rate of 0.6 mL/min. The glucose and ethanol yields were calculated as follows: Glucose yield (%) = Glucose g × 0.9 Cellulose in substrate g × 100% Ethanol yield (%) = Ethanol g Glucose in substrate g × 0.51 × 100% All results are presented as the average values of three independent experiments with standard deviations at significance p < 0.05.

Cellulase overproduction by fed-batch culture
Fed-batch fermentation strategy was employed for T. reesei cellulase production using MGD or APCS as inducer.
The fermentation process consists of two steps, namely, rapid growth process in the excess of the substrate, followed by cellulase production phase performed in fed-batch mode under carbon limitation [23]. Figure 1 illustrates the time course of the fed-batch culture of T. reesei Rut C30. The results of MGD ( Fig. 1a) were in accordance with our previous study [6]. It was of interest to see that cellulase yield and productivity of MGD (144 h) were 13.87 and 10.40 times, respectively, higher than that of APCS at 108 h ( Fig. 1b). Obviously, MGD presents great advantage for cellulase production in T. reesei, and the characteristics of the enzymatic cocktail were further studied. The two types of crude enzymes Cel-MGD and Cel-CS achieved from MGD to APCS feeding strategies showed great difference in cellulase, β-glucosidase and xylanase activities ( Table 1). The potential of the two enzymes for biomass degradation was further evaluated. Under the same protein concentration, the specific activities of cellulase and β-glucosidase of Cel-MGD were 0.74 and 1.44 times higher than that of Cel-CS, respectively, which exhibit the robust ability in cellulose degradation. However, the hemicellulose conversion would be more favorable in Cel-CS due to the higher specific xylanase activity (9.95 times) than that of Cel-MGD.

Hydrolysis efficiency of cellulases cocktail in T. reesei Rut C30
The saccharification ability of Cel-MGD and Cel-CS was further examined during time course hydrolysis of APCS. Interestingly, opposite results were obtained when the crude enzymes were evaluated using the same protein (3 mg/g APCS) or the same FPA (20 FPU/g APCS) loading under the same reaction conditions (Fig. 2). When 3 mg protein/g APCS was employed (Fig. 2a), 24.49 g/L glucose (70.46% glucose yield) was released using Cel-MGD, which was higher than that of 21.65 g/L glucose   (Table 1), indicating the effectiveness of Cel-MGD for APCS hydrolysis. Surprisingly, the glucose concentration (27.82 g/L) produced by 20 FPU Cel-CS/g APCS was higher than that released by Cel-MGD (26.01 g/L) (Fig. 2b). The same FPA meant the same amount of cellulase, ignoring the xylanase and other glycoside hydrolases in cellulolytic cocktails. Cellulase produced by T. reesei comprises three major enzyme components: cellobiohydrolases, endoglucanases, and β-glucosidases [8], along with some other elements involved in cellulose degradation, for instance, the expansin-like protein swollenin (SWO1) and GH61 polysaccharide monooxygenases (PMOs), which have been shown to enhance lignocellulose degradation [9]. Therefore, different components in cellulases may be the reason for the above results. Aiming to further optimize the crude enzymes induced by MGD, the lignocellulolytic enzyme profiles of Cel-MGD and Cel-CS were further compared by secretome analysis, and described in the below section. The β-glucosidase (BG) activity is low in cellulase cocktails by T. reesei, which leads to disaccharide accumulation, resulting in a reduced cellulose degradation efficiency [12][13][14]24]. It was interesting to find that higher BGA was obtained in Cel-MGD than Cel-CS. Extra BG (FPA:BGA = 1:2) was added into Cel-MGD and Cel-CS to evaluate whether enough BG is present in the crude enzymes (Fig. 2). It was found that the released glucose was improved 25.84% using Cel-MGD + BG compared with the crude enzyme, indicating that the T. reesei cellulase using MGD as inducer was still weak in BGA.

Different protein profiling using two inducers by analysis of T. reesei secretome
More glucose was released in APCS hydrolysis by Cel-CS than by Cel-MGD when the same FPA was used for evaluation (Fig. 2b), which means different hydrolysis efficiency of the crude enzymes using APCS and filter paper (FP) as substrates. The biomass hydrolysis rate is one of the major limitations towards the commercialization of second-generation biofuel, which contained various rate-limiting factors (enzyme adsorption, crystallinity, accessibility, reactivity, enzyme clogging, and degree of polymerization) [25]. In this study we focused on enzyme reactivity by optimizing enzyme compositions.
The iTRAQ system was employed for quantitative analysis of the secretome of T. reesei Rut C30 induced by MGD or APCS. In total, 242 proteins were identified using Maxquant after applying the cutoff of unused protein score > 2, which corresponds to a 99% confidence level and a false discovery rate < 1.0%. The proteins identified were sorted with N-terminal Sec-dependent secretion signal using SignalP 3.0 and classified according to the biological functions. A total of 143 secreted proteins with signal peptides were identified, which were sorted into cellulases, hemicellulases, peptides, chitinase, phosphatases, transport, and hypothetical proteins. Other proteins without any predicted signal peptides were probably detected due to non-conventional mechanisms of secretion, or by possible cell lysis and/or death, which was also proposed in the previous studies [26,27]. Interestingly, there are similar components for exocellular proteins induced by the MGD and APCS, in which 57.3% proteins were putative glycoside hydrolases (78 proteins) and esterases (4 proteins) that are related to lignocellulosic biomass degradation. In the previous secretome analysis of T. reesei Rut C30 during solid fermentation, similar amounts to secreted proteins and lignocellulose degradation-related proteins were identified (135 and 68, respectively) [27]. In another secretome study of T. reesei Rut C30 grown on lignocellulosic biomass (solid-state culture), 84 glycosyl hydrolases were identified by iTRAQ [26]. However, the protein diversity detected in this study is not comparable with the previous studies [26,27], which may be due to different culture mode. All the identified proteins are listed in Additional file 2: Table S2. Fold change (FC) of main cellulosic enzymes' abundance differences detected common to the Cel-MGD and Cel-CS are listed in Table 2.
The same components of the extracellular proteins from T. reesei Rut-C30 induced by two different inducers (MGD and APCS) may be due to the presence of wheat bran in the culture medium, which contains cellulose and was supplemented as nutrients. However, our recent study [28] suggests that APCS induced higher xylanase activities, which is consistent with the current secretome data (Table 2), and also with much higher xylanase activity we detected in Cel-CS (Table 1). Significant differences in the ratio of cellulases and hemicellulases were revealed. Higher quantities of two cellobiohydrolases CBH1, CBH2, as well as EG2, EG3, and EG5 were detected in Cel-MGD. In addition, there are four β-glucanases (BGL1, BGL2, and CEL3j and CEL3e) detected by iTRAQ analysis, and the ratios (MGS/CS) for iTRAQ results are 1.38, 1.71, 2.59, and 2.59, respectively. The higher amount of these proteins is consistent with the higher FPase and BG activities in Cel-MGD as shown in Table 1. It is well-known that multiple enzymes must work together to degrade cellulose into glucose. Therefore, Cel-MGD showed stronger cellulose degradation ability than Cel-CS under the same protein dosage (Fig. 2a). Moreover, degradation of lignocellulose demands more accessory proteins such as SWO1, glucuronoyl esterase (CIP1 and CIP2), and AA9 lytic polysaccharide monooxygenases (LPMOs) [9,29]. In consistence with more glucose release by Cel-CS when the same FPA was used, there were more auxiliary proteins in Cel-CS, including two LPMOs (CEL61a and CEL61b), which were 2.81 and 1.43 times higher in Cel-CS than that in Cel-MGD, respectively. In addition, SWO1, CIP1, and CIP2 contents reached about 3.36, 1.55, and 1.56 folds in Cel-CS to that in Cel-MGD. LPMOs are capable of oxidizing recalcitrant polysaccharides, which showed a synergistic effect with lignocellulose on degradation of a range of cellulosic substrates [9,30]. Therefore, Cel-CS was more effective on APCS hydrolysis than Cel-MGD under the same conditions of FP degradation. Other glycoside hydrolases in Cel-MGD are significantly lower than that in Cel-CS, such as GH2 (candidate β-mannosidase), GH16 (candidate glucan endo-1,3(4)-β-D-glucosidase), GH17 s (candidate glucan endo-1,3-β-glucosidases), GH30 (candidate endo-β-1 6-galactanase), GH36 (α-galactosidase), GH71 (candidate α-1,3-glucanase), GH76 (candidate α-1,6-mannanase), GH79 (candidate β-glucuronidase), GH92 (candidate α-1,2-mannosidase), and chitinase (Additional file 2: Table S2). Induction of cellulase production in T. reesei by the mixture of MGD and APCS was tested in our previous work, and better degradation effect was observed when optimized ratio of MGD and APCS was employed [28]. The data of secretory proteins in this study further revealed new players in cellulose degradation, especially auxiliary proteins secreted in response to APCS which are in low abundance when MGD was employed as inducer. We have tested the effects of overexpression of swo1, cip1, and cip2 in T. reesei Rut C30 using the eno1 promoter, but no improvement of cellulase production was observed (data not shown). Addition of accessory proteins may be an alternative way. It was reported that adding small amount of lytic polysaccharide monooxygenase AA9 (formerly named GH61) as an accessory enzyme enhanced hydrolysis of pretreated polar and corn stover [31]. In another study, supplementation of pectinase and α-l-arabinofuranosidase increased hydrolysis efficacy of pretreated sugarcane bagasse [32]. These students emphasized the importance of optimization of enzyme compositions for efficient lignocellulose degradation.

Impact of over-expressing β-glucosidase in T. reesei
Although high cellulase titer was obtained using MGD as inducer, BGA was still not sufficient for efficient lignocellulose hydrolysis. Activity of β-glucosidases in T. reesei is normally low, and therefore commercial β-glucosidase is required for efficient biomass hydrolysis [6]. Synergistic effects of some β-glucosidases and T. reesei cellulase were also reported [33]. Genetic engineering has been successfully applied to increase BGA, but in the all reported cases, cellulose-inducible promoters were used, and the disadvantage of such promoters is the reduced expression of cellulase or xylanase genes [12-14, 34, 35]. It was reported that pdc1 (encoding pyruvate decarboxylase) in T. reesei Rut C30 was expressed at high level and pdc1 promoter could constitutively promote heterologous gene expression using glucose as the sole carbon source [36,37], it is therefore suitable to use pdc1 promoter to drive the expression of target genes by MGD induction owing to abundant glucose exist in MGD [6]. The aabgl1 encoding β-glucosidase 1 (aabgl1) from A. aculeatus with high specific activity was used as the target gene for achieving higher BGA in T. reesei [13]. The plasmid pCBHYGB-PB containing aabgl1 expression cassette was constructed (Additional file 3: Figure S1A), which was then transformed into the parent strain T. reesei Rut C30. Twenty recombinants with genetic stability were selected and verified for aabgl1 cassette by PCR. All of the selected transformants (Additional file 3: Figure  S1B) possessed the BGA of 0.44-3.96 IU/mL by 10 g/L MGD as the inducer in shake flask culture. The highest BGA (3.96 CBU/mL) of the recombinant strains was 26.4 times of that of the control (0.15 CBU/mL). These results demonstrated that the heterologous β-glucosidase gene was efficiently expressed using MGD as a carbon source and the enzyme was continuously secreted into the fermentation broth.
The large difference of BGA among the mutants might be due to the different expression levels of aabgl1. The copy number of aabgl1 gene in the recombinants was determined, which is one of the factors affecting the transcription [14]. Four mutants with large improvement in BGA were selected for copy number analysis using qPCR, and the results showed that the transformants T. reesei PB-3 and PB-4 had four and three copies, respectively, while both PB-14 and PB-17 had two copies of aabgl1 gene in the genome (Additional file 4: Figure  S2). High copy numbers integrated into the genome was consistent with higher expression levels of the heterologous gene. However, T. reesei PB-14 and PB-17 harboring the same copies of aabgl1 have displayed different BGA, which illustrated that the integration locus also affects the expression of the heterologous gene. Previous studies also demonstrated that integration sites play important roles in the elevation of β-glucosidase activity in the recombinant T. reesei strains [38,39]. In our previous studies, integration of the new regulatory gene Trvib-1 in a putative strong integration site resulted in higher enhancement of cellulase production [40]. Another possibility is the position effect which was reported in budding yeast Saccharomyces cerevisiae [41]. However, systematic survey of the effect of integration sites on gene expression levels in T. reesei has still not been performed, which will be an interesting topic for future work.
The recombinant strain T. reesei PB-3 with the highest BGA was selected and investigated further. The expression levels of major cellulase genes (cbh1, eg1, Trbgl1, and aabgl1) in T. reesei Rut C30 and PB-3 were detected using 10 g/L MGD in shake flask culture for 36 h by qPCR analysis (Fig. 3a-d). Moreover, there was a distinct band at about 130 kDa in T. reesei PB-3 through SDS-PAGE analysis which is absent in T. reesei Rut C30, further confirming the expression of AaBG1 (Fig. 3f ). The relative aabgl1 expression level was shown to be about 11% of the cbh1 expression level in T. reesei PB-3, indicating that pdc1 promoter could promote gene transcription efficiently in MGD. However, the expression levels of cbh1, eg1, and Trbgl1 in T. reesei PB-3 were lower than the parent strain T. reesei Rut C30. The possible reason was that higher BGA in the recombinant strain T. reesei PB-3 (26.4 times than that of T. reesei Rut C30) would hydrolyze β-disaccharides into glucose, which resulted in lower induction of cellulose. These results implied that BGA must be controlled at proper level to avoid significant reduction in cellulase gene expression. Until now, studies on promoters of T. reesei are still limited [42]. It will be ideal if the promote activity used to overexpress β-glucosidase was activated at latter fermentation stages after induction of cellulase.

Cellulase and β-glucosidase overproduction by fed-batch culture
The fed-batch culture was performed for achieving a higher activity of β-glucosidase and cellulase. Figure 4a illustrated the time course of T. reesei PB-3 using MGD as substrate and inducer. BGA was enhanced significantly to 310.1 CBU/mL at 156 h due to the overexpression of the heterologous β-glucosidase, which was more than 71 folds of that in the host strain T. reesei Rut C30. As a result, the maximum β-glucosidase productivity reached 1988.0 CBU/h/L. The cellulase and xylanase activities and extracellular protein secretion of PB-3 were lower than T. reesei Rut C30 because inducible β-disaccharides were degraded with the accumulation of β-glucosidase [6]. However, the cellulase activity was still as high as 50.0 FPU/mL, which was more than 2.94 and 2.2 times of that obtained using cellulose [43] and lactose [44] as inducer, respectively. The high cellulase and β-glucosidase activities obtained simultaneously in T. reesei PB-3 would benefit in situ use of the crude enzymes without concentration for achieving high titer of bioethanol, which would help to reduce the cost involved in product separation.
Extensive studies have been focused on the overexpression of heterologous β-glucosidase in T. reesei, which always employed cellulase or xylanase promoter with cellulose as an inducer, and reduced expression of corresponding genes [12][13][14]. Furthermore, these recombinant strains have not been examined in submerged fermentation through fed-batch culture for β-glucosidase production. Aspergillus sp. has been used in industry by solid-state fermentation. However, Aspergillus sp. contamination associated with the solid-state culture makes the cellulase not suitable for hydrolysis of lignocellulose for biofuels and bio-based chemical production [45][46][47]. On the other hand, the β-glucosidase titer was still low by Aspergillus sp. in submerged fermentation, which showed only 9.3 IU/mL and productivity of 1.78 U/mL/ day [48]. Therefore, submerged culture of T. reesei PB-3 for efficient simultaneous β-glucosidase and cellulase production at large scale has considerable potential for bioconversion of agricultural residues.
To evaluate the hydrolysis ability of the cellulase cocktails of T. reesei Rut C30 or PB-3 on natural cellulosic materials, the crude enzyme from T. reesei Rut C30 (Cel-MGD) or PB-3 (Cel-MGD-BGL) were used to hydrolyze APCS. Cel-MGD was found to be lack of BGA in biomass hydrolysis [6]; although Cel-MGD-BGL had higher BGA, additional β-glucosidase (BG) was also supplemented to examine whether the self-BG was sufficient in this study. Figure 4b showed that 33.82 g/L glucose was released using Cel-MGD-BGL at 72 h with a glucose yield of 97.29%, meanwhile, 33.88 g/L glucose was generated by Cel-MGD-BGL + BG, indicating that the crude Cel-MGD-BGL was efficient for biomass degradation and commercial BG was not required. In addition, the highest amounts of released glucose were 33.47 and 33.40 g/L from Cel-MGD: BG = 1: 3 and Cel-MGD: BG = 1: 2, respectively, but only 23.70 g/L glucose was obtained without additional BG. Moreover, the glucose yield was lower when Cel-MGD: BG < 1:2. More importantly, Cel-MGD-BGL was able to hydrolyze APCS efficiently without extra BG, which was well-matched with Cel-MGD + BG. The effect of the Cel-MGD-BGL loading on hydrolysis performance was investigated at five different dosages (5, 10, 20, 30, and 40 FPU/g APCS). Within 72 h, glucose was completely released under cellulase loading of 20, 30, and 40 FPU/g, but glucose concentration was lower at 20 than that of 30 and 40 FPU/g before 48 h. When the enzymes loading was decreased to 5 and 10 FPU/g, glucose concentration was reduced to 42.8 and 14.63%, respectively, compared to that of 30 FPU/g at 48 or 72 h (Fig. 4c). In summary, the cellulase loading at 30 FPU/g was optimum for cellulosic ethanol production.

Bioethanol production from corn stover using in-house produced T. reesei enzymes
In our previous study, 44.8 g/L ethanol was produced from APCS by Cel-MGD + BG [6]. However, it is important to achieve higher concentration of glucose from APCS to increase product concentration and reduce purification costs [49][50][51]. Therefore, fed-batch strategy was adopted in this study for corn stover hydrolysis with a total biomass uploading of 20% by SHF or SSF process. Figure 5a shows that the glucose was generated consistently during the process of saccharification, and the glucose concentration reached 119.92 g/L at 108 h with the yield of 86.24% by SHF, indicating that APCS could be hydrolyzed efficiently by the crude cellulase Cel-MGD-BGL. Once the S. cerevisiae was added into the fermenter, glucose was metabolized rapidly within 12 h, and the ethanol concentration reached 54.23 g/L with the ethanol yield of 76.47%. When SSF process was applied for bioethanol production, 52.09 g/L ethanol was produced within 96 h and the ethanol yield reached 73.45% (Fig. 5b). Although the ethanol yield was lower by SSF than SHF, the productivity reached 0.54 g/L/h using SSF, which was higher than 0.45 g/L/h in the case of SHF. Compared with SHF, SSF encompasses the hydrolysis and fermentation process within the same vessel, which was more favorable for bioethanol production because of the shorter processing time, lower cost of equipment, and lower contamination risk [52]. However, higher sugar conversions and higher ethanol yield were achieved by SHF in this study because low temperature (37 °C) was not suitable for saccharification of Cel-MGD-BGL.
Comparison of cellulosic ethanol and productivity from corn stover obtained in this study using in-house cellulase with those reports by other commercial cellulase is shown in Table 3 [53][54][55]. The results indicated that in-house cellulase production by the recombinant T. reesei PB-3 was effective to replace commercial enzyme for biomass saccharification. Besides, effective bioconversion for glucose was also achieved in this work. The results suggested that T. reesei PB-3 has great potential for efficient bioconversion of cellulosic biomass.

Conclusions
The secretome of T. reesei RUT C30 with MGD or APCS as inducer was compared, which indicated that cellulolytic enzymes induced by MGD had higher content of major cellulases, but lower BGA still remains a challenge. To further improve the hydrolysis efficiency of the cellulytic enzyme cocktail, T. reesei PB-3 was developed through overexpression of A. aculeatus β-glucosidase gene under a constitutive pdc1 promoter, which substantially improved β-glucosidase production. Furthermore, overproduction of cellulases and β-glucosidase was achieved simultaneously by T. reesei PB3 through the fed-batch fermentation using MGD as the inducer. The crude enzyme was used for the hydrolysis of APCS, and bioethanol fermentation was achieved through SHF and SSF, respectively. To our knowledge, this is the first report that high activities of cellulases and β-glucosidase were achieved simultaneously for lignocellulosic bioconversion, and the crude enzyme can be used for cellulosic hydrolysis and ethanol production more efficiently.
Authors' contributions FWB and XQZ conceived the project. YHL performed the study. YHL, XYZ, and FZ analyzed the data and drafted the manuscript. XQZ, FWB, LCP, DBZ, and AK critically revised the manuscript. All authors read and approved the final manuscript.