For storage, T. reesei strains CL847 21 and Rut-C30 (ATCC 56765) cultures were grown on plates of Potato Dextrose Agar (Difco Laboratories, USA) at 30°C. After sporulation, the spores were resuspended in a sterile NaCl (9 g L^-1)-glycerol 20% solution and stored at -80°C. Frozen spores were used to inoculate a Fernbach flask containing 250 mL of culture medium (glucose 30 g L^-1; corn steep 2 g L^-1; (NH₄)₂SO₄ 1.4 g L^-1; KOH 0.8 g L^-1; H₃PO₄ 85% 4 mL L^-1; phthalic acid, dipotassium salt 5 g L^-1; MgSO₄.7H₂O 0.3 g L^-1; CaCl₂ 0.3 g L^-1; FeSO₄.7H₂O 5.0 mg L^-1; MnSO₄.H₂O 1.6 mg L^-1; ZnSO₄.7H₂O 1.4 mg L^-1; CoCl₂.6H₂O 2.0 mg L^-1). Cultivation was carried out at 30°C with stirring at 110 rpm. After 72 h, 100 mL of medium broth was used as an inoculum for bioreactor culture. Fermentation of T. reesei was carried out in a 4 L bioreactor under culture conditions previously described by Pourquié and Warzywoda, 1993 17. The cellulase production was performed in two steps. In the first step, a growth phase, with 2 L starting medium containing 35 g L^-1 of lactose as carbon source, 27°C and pH regulated at 4.8 (with 6 M ammonia) was conducted. The air flow was adjusted at 0.5 VVM and initial stirring was set at 500 rpm. This parameter was gradually increased to maintain pO₂ above 40% oxygen saturation. In the second step, when initial lactose was depleted, a fed-batch phase was initiated. During this phase, a 250 g L^-1carbon source solution was injected at a 4 mL h^-1 rate. The feeding solution composed of either 60% lactose and 40% xylose (W/V) or only lactose. Samples were collected periodically to determine the biomass, carbon and protein concentrations. For both strains, the initial lactose was depleted after 30 h of cultivation. At this stage of the culture, the biomass dry weight concentrations were between 15 to 18 g L^-1 and remained steady during the whole fed-batch phase for all cultures whatever conditions were tested. No carbon source accumulation was observed during the whole fed-batch phase.

CL847 strain cultivations were performed in triplicate for the lactose as the only carbon source condition, and in duplicate for the mixed lactose-xylose condition. Only one production was carried out for the Rut-C30 strain.

Lactose was assayed by high-performance liquid chromatography on a 7.8 × 300 mm² HPX-87P column (Biorad) maintained at 85°C, using a Varian Prostar Model 350 HPLC equipped with a refractive index detector. Eluant was helium-degassed distilled water at a flow rate of 0.4 mL min^-1. Quantification was performed using a solution of 1 g L^-1 of lactose as external standard.

Biomass concentration was assayed using a gravimetric method. A culture volume is filtered with a vacuum pump on a dried and preweighed GF/C glass fiber membrane (Wathman). After washing with distilled water, membranes are dried for 48 h at 105°C and weighed.

Samples were collected around 160 h after start of cultivation. At this stage, protein concentration for all cultures was around 30 g L^-1. The culture supernatants were harvested by centrifugation for 15 min at 10,800 g and 4°C. The supernatants were further clarified on a glass fiber filter GF/F (Whatman, Maidstone, UK) and concentrated and diafiltered against 10 times their volumes of Milli-Q water using a 5 kDa membrane (Amicon system, Millipore Bradford, USA) to eliminate salts. Total protein concentrations were determined in duplicates using the Bio-Rad Dc protein assay kit (Bio-Rad). Aliquots of extracellular protein samples were stored at -80°C for 2DE gel experiments. The same amount of proteins (200 μg) was used for each 2D gel, regardless for the initial supernatant concentration.

Immobiline DryStrips (18 cm, pH 4–7, Amersham Biosciences) were rehydrated overnight at room temperature with 200 μg of proteins diluted in rehydration solution (DeStreak solution, Amersham Biosciences) supplemented with 2% (v/v) 4–7 IPG buffer and 2.8 mg mL^-1 dithiothreitol to a final volume of 350 μL. Isoelectric focusing was performed on a Multiphor II system at 20°C with a 3-phase gradient program: 500 V for 1 Vh, 3500 V for 3 kVh and 3500 V for 27 kVh. Following isoelectric focusing, each strip was equilibrated for 10 min in 10 mL of SDS equilibration buffer (50 mM Tris-HCl pH 6.8, 6 M urea, 30% (v/v) glycerol, 1% (w/v) SDS, a trace of bromophenol blue) containing 25 mM dithiothreitol. A second equilibration step was then performed in the same SDS equilibration buffer containing 250 mM iodoacetamide instead of DTT. The strips were then loaded onto 12% homogeneous acrylamide gels and sealed with 0.5% (w/v) agarose in SDS running buffer (25 mM Tris base, 192 mM glycine, 0.1% (w/v) SDS). The second dimensional separation was performed using an Ettan™ DALT system (Amersham) at 0.5 W/gel and 16°C overnight, followed by 17 W/gel for 3 h. After electrophoresis, the acrylamide gels were either silver-stained for spot picking experiments or stained with Biosafe Coomassie Stain (Biorad) for comparative analysis experiments.

For comparative studies, each culture sample was independently prepared and used in 2DE in triplicates. To allow an unbiased comparative analysis, Coomassie Blue staining was used instead of silver staining (Biosafe Coomassie Stain, Biorad). The amount of proteins used for 2DE (200 μg) and Coomassie Blue staining was the best compromise between spot-detection sensibility and coloration saturation (data not shown).

Each sample was analyzed in triplicate. Gels were scanned on a calibrated GS800 scanner (Biorad). Images were analyzed using ImageMaster II software (GE Healthcare) using the following workflow. After automatic spot detection, artifacts such as dust or cracks on gels were manually eliminated, and then the weaker spots (individually < 0.05% of the whole gel volume) were eliminated. Remaining spots were then automatically linked to reference spots on a synthetic reference gel to allow comparison between samples.

All samples were analyzed in duplicate and mean values were calculated. Overall cellulase activity of the samples was measured as Filter Paper (FP) activities using the IUPAC-recommended procedure 24. Endoglucanase activity was assayed as CMCase activity with CMC (Aqualon) as substrate in 50 mM acetate buffer (pH 4.8) for 30 min at 50°C. Xylanase activity was measured with Oat Spelt Xylan (Sigma) as substrate in the same conditions. For all three activities, sugar release was assayed via the dinitrosalicylic acid method using glucose or xylose as the standard. β-glucosidase activities were determined using 4-nitrophenyl-β-D-glucopyranoside with paranitrophenol as the standard 25.

Cellulase and hemicellulase production is dependent on fungus cultivation conditions 26. It has been demonstrated that the production of the main cellulases of Trichoderma is transcriptionally regulated and carbon source-dependent 813. In order to obtain the fullest complement of the hemicellulolytic enzymatic system, T. reesei was grown under conditions promoting the production of both cellulases and hemicellulases. Thus, T. reesei was cultivated on a lactose-xylose medium in fed-batch fermentation, since this medium is known to induce the production of both cellulases and hemicellulases in T. reesei 2728.

Total extracellular proteins from the culture supernatant were separated by 2DE. Preliminary investigations using pH 3 to 10 IPG strips revealed that most proteins had pIs < 7. Thus, IPG strips ranging from pH 4 to 7 were chosen for detailed expression analyses to improve the resolution of the proteins spots and facilitate further quantification of individual protein species. The resulting protein maps are shown in Figure 1. Ninety-five distinct protein spots were detected on the 2D gel after staining. The distribution of the protein spots showed that most strongly secreted proteins had an isoelectric point below 6 and a molecular weight above 43 kDa.

Among the 95 protein spots, 36 were identified by MALDI-TOF mass spectrometry (Table 1). To increase the amount of identified proteins, 18 additional spots were analyzed by nanoLC-MS-MS, resulting in the identification of nine further proteins (Table 2). Absence of reliable identification of the remaining protein spots is due to small amounts of biological material and/or post-translational modifications known to affect identification 18. In most cases, molecular masses observed on 2D gels were higher than the expected masses calculated from the protein sequences, probably because of glycosylation. Several protein spots were assigned to the same protein, suggesting the presence of numerous isoforms and/or degraded forms (Tables 1 and 2).

As expected, most of the identified proteins were related to biomass degradation and were assigned to cellulases and hemicellulases. Cellobiohydrolases Cel7A and Cel6A were the two most abundantly secreted proteins. These proteins are known to account for 70 to 80% of the total T. reesei cellulases 429, consistent with the high intensity of the corresponding protein spots observed on the gel. The only β-glucosidase identified on the gel was BGLI, in accordance with reports of the other β-glucosidases being either intra-cellular, membrane-anchored, or playing only a minor role in cellulose hydrolysis 6. Four out of the five known endoglucanases were also identified, but one of them, endoglucanase Cel61A, was only identified with a single peptide and thus should be considered provisional (Table 2). The minor endoglucanase Cel45A of T. reesei 30 was not identified, probably because of its highly acidic pI. This secretome analysis also revealed the expression of the ORF_27554 product annotated as candidate endoglucanase in the T. reesei genome database. The similarity between the observed and predicted molecular weight for this new endoglucanase (Table 1) suggests that this protein is only sparsely glycosylated. In addition, the product of the gene cel74a 6 was detected on the protein map. This enzyme, formerly endoglucanase VI, has been characterized as a xyloglucanase. The observation that several spots matched to this protein supports previous data that there are multiple isoforms of this enzyme 31.

We also identified some major components of the hemicellulolytic system of T. reesei: β-xylosidases, xylanases and arabinofuranosidase (Table 1). Three out of the four known xylanases were identified. The last xylanase, XYNIII, focalizes at a pH around 8 and is outside the range of our pH 4 to 7 gels (data not shown). Furthermore, we did not identify any galactosidases, which is surprising given that these proteins are purported to be induced by lactose 32. It is not unlikely that this protein corresponded to one of the minor unidentified spots. The present study highlights the production of a putative arabinofuranosidase (ORF_55319, ABFIII Table 1). As stated previously for the putative endoglucanase, the close correlation between the observed and predicted molecular weight of this putative arabinofuranosidase suggests that this enzyme is also sparsely glycosylated. T. reesei is thus able to produce at least three different arabinofuranosidases. Two of them have already been described in a purification study (ABFI) 33 and in cDNA analysis (ABFII) 13. This work provides evidence for the production of both ABFII and a novel third α-L-arabinofuranosidase (ORF_55319, ABFIII) not reported previously. Apart from cellulases and hemicellulases, non-hydrolytic proteins including CIPI, CIPII and swollenin were identified (Table 2), providing new evidence for the production and secretion of these proteins. CIPI and CIPII were discovered during the T. reesei genome sequencing program as proteins with a cellulose-binding domain, but no other functional domain, such as a glycosyl hydrolase domain, could be found 13. However, a recent phylogenetic analysis suggests they share close relationships with cellulases, which adds support to the potential roles of these genes in biomass degradation 14. Proteolytic enzymes such as trypsin were also found at low levels (0.2%). This may explain the presence of some altered proteins, especially Cel6A, whose molecular weight in the gels was lower than expected (Figure 1). Heterogeneity of cellobiohydrolases on PAGE-SDS has already been reported, and explained by glycosylation and proteolysis 19.

In total, 22 biomass-degrading enzymes were identified on our gels, to be compared with the previous study of Vinzant et al (2001) 19 where only 10 enzymes were identified. An analysis of the lactose-xylose 2D gel image by ImageMaster II software indicated that the identified proteins account for 83% of all visible proteins in the gel in terms of spot volume. This percentage rises to 93% for the secretome of T. reesei CL847 grown on lactose alone.

Rut-C30 has for decades been the reference cellulase-overproducing strain in academic publications. This strain, like CL847, has been obtained through random mutagenesis and subsequent screening. The last common ancestor of these two strains is the reference T. reesei strain QM6a. CL847 was further evolved from strain QM9414. Enzymatic activities vary significantly between these two strains (Table 3). In the same culture conditions, Rut-C30 has a slightly higher FPase and CMCase-specific activity, while xylanase and β-glucosidase activities are significantly higher for CL847 strain (respectively ×2.7 and ×1.5). Two mutations were identified in Rut-C30. Firstly, a mutation in cre1, a gene encoding a transcription factor mediating glucose repression for cellulase production was discovered. This frameshift mutation leads to a truncated protein that might account for some increase in cellulase production in this strain 34. Secondly, a frameshift mutation was observed for glucosidase II alpha subunit, leading to defective extracellular protein glycosylation 35. However, it is almost certain that these are not the only mutations affecting this strain. As Rut-C30 grown on lactose and xylose failed to produce cellulases, secretomes of Rut-C30 and CL847 were produced with lactose as the only carbon source.

As for proteome map construction, samples were taken during the late fed-batch production phase (around 160 h) for each sample. At this stage of the production, protein concentrations were around 30 g L^-1 for both strains.

The 2DE profiles of CL847 and Rut-C30 grown on lactose were very different, in terms of both spot numbers and protein composition (Figure 2). CL847 2DE reveals many more protein spots that Rut-C30, especially in minor spots corresponding to less than 0.5% of total spots volume (Figure 3), and consequently most of these spots are unidentified or correspond to degradation forms absent in Rut-C30. Differences in protein spots representing a higher percentage of the total spot volume are due to the presence of several Cel7A isoforms for CL847, while a single and bigger spot is visible for Rut-C30. In contrast, Cel6A isoform profiles were similar. Protein spot quantitation revealed that Rut-C30 has 10% more total cellobiohydrolases than CL847 (Figure 4A). This is related to a higher Cel7A level in this strain (57.4% in Rut-C30 versus 42.1% in CL847), since Cel6A levels were not significantly different between the two strains. As a consequence, the Cel7A-to-Cel6A ratio is much higher in Rut-C30 than in CL847 (Figure 4B). This is in disagreement with the widespread hypothesis that Cel7A and Cel6A are co-regulated 1636. Nevertheless, we cannot rule out the possibility that this change in ratio could be due to a higher level of degradation of Cel7A in CL847. In contrast with cellobiohydrolases levels, the relative amount of BGLI produced by CL847 is twice as much as compared to Rut-C30 (Figure 4D), which is reflected in β-glucosidases activities for these strains in similar conditions (Table 3). No significant differences were observed for endoglucanases Cel7B and Cel5A. However, this area of the gels is heavily crowded, especially for CL847, and any quantification must be taken with caution (Figure 2). Contrast was more pronounced for minor endoglucanases. CL847 produces around 2% Cel12A, while it is almost undetectable in Rut-C30 samples. Cel61A and Cel74A levels were much higher in Rut-C30 (Figure 4C). These results contrast with those of Foreman et al (2003) 13, where endoglucanase co-expression was observed at the mRNA level, prompting the authors to propose co-regulation of these enzymes. We observed no such events at protein level, but the differences may be due to the different strains and culture conditions used in our work and Foreman's. Another point to highlight is that the EGL and AXEIII are also absent in the both strains cultivated on lactose. The β-xylosidase, BXLI, is only present in Rut-C30, although at a very low percentage (0.2%). This is consistent with reports of a low constitutive expression of the BXLI protein in Rut-C30 10. The only xylanase expressed in Rut-C30 was XYNIV, secreted at a similar level to CL847 (Figure 4F). In contrast, in CL847, XYNIV only figured as one of the minor xylanases. This suggests that expression of xylanases XYNI, XYNII and XYNIV is different in the two studied Trichoderma strains and that these different expression pathways were not equally affected by the mutations that led to the CL847 and Rut-C30 phenotypes. Globally, Rut-C30 has a lower xylan-related enzyme secretion while CL847 secretes a more diversified set of enzymes. Other hemicellulase levels also showed marked differences. While ABFII levels were comparable and ABFIII level was tenfold higher in CL847, ABFI was slightly over-produced in Rut-C30 (Figure 4E). As for xylanases, the results suggest that the role of these proteins is not equivalent and that they are not co-regulated. The only mannanase of T. reesei was expressed twice more in CL847 than in Rut-C30 (Figure 4E). Among non-cellulolytic enzymes, except for the absence of CIPII, there was no significant detectable difference in CIPI and SWO. Finally, trypsin was absent in Rut-C30. The presence of proteases may explain the observation of degraded forms of proteins in CL847 and the much higher number of spots, especially in the low molecular weight region of the gels.

These results fit nicely with data obtained from specific enzymatic activities (Table 3): the lower FPase activity in CL847 can be related to the lower amount of cellobiohydrolases in this strain. While it is difficult to link the CMCase activity differences to any specific protein, since many enzymes exhibit endoglucanase activity, the higher β-glucosidase and xylanase specific activities are consistent with the higher BGLI and xylanases activities in the CL847 cellulase productions (Figure 4).