Dissecting Cellular Function and Distribution of β-Glucosidases in Trichoderma reesei

ABSTRACT Trichoderma reesei has 11 putative β-glucosidases in its genome, playing key parts in the induction and production of cellulase. Nevertheless, the reason why the T. reesei genome encodes so many β-glucosidases and the distinct role each β-glucosidase plays in cellulase production remain unknown. In the present study, the cellular function and distribution of 10 known β-glucosidases (CEL3B, CEL3E, CEL3F, CEL3H, CEL3J, CEL1A, CEL3C, CEL1B, CEL3G, and CEL3D) were explored in T. reesei, leaving out BGL1 (CEL3A), which has been well investigated. We found that the overexpression of cel3b or cel3g significantly enhanced extracellular β-glucosidase production, whereas the overexpression of cel1b severely inhibited cellulase production by cellulose, resulting in nearly no growth of T. reesei. Four types of cellular distribution patterns were observed for β-glucosidases in T. reesei: (i) CEL3B, CEL3E, CEL3F, and CEL3G forming clearly separated protein secretion vesicles in the cytoplasm; (ii) CEL3H and CEL3J diffusing the whole endomembrane as well as the cell membrane with protein aggregation, like a reticular network; (iii) CEL1A and CEL3D in vacuoles; (iv) and CEL3C in the nucleus. β-glucosidases CEL1A, CEL3B, CEL3E, CEL3F, CEL3G, CEL3H, and CEL3J were identified as extracellular, CEL3C and CEL3D as intracellular, and CEL1B as unknown. The extracellular β-glucosidases CEL3B, CEL3E, CEL3F, CEL3H, and CEL3G were secreted through a tip-directed conventional secretion pathway, and CEL1A, via a vacuole-mediated pathway that was achieved without any signal peptide, while CEL3J was secreted via an unconventional protein pathway bypassing the endoplasmic reticulum (ER) and Golgi.

IMPORTANCE Although b-glucosidases play an important role in fungal cellulase induction and production, our current understanding does not provide a global perspective on b-glucosidase function. This work comprehensively studies all the b-glucosidases regarding their effect on cellulase production and their cellular distribution and secretion. Overexpression of cel3b or cel3g significantly enhanced b-glucosidase production, whereas overexpression of cel1b severely inhibited cellulase production on cellulose. In addition, overexpression of cel3b, cel3e, cel3f, cel3h, cel3j, cel3c, or cel3g delayed endoglucanase (EG) production. We first identified four cellular distribution patterns of b-glucosidases in Trichoderma reesei. Specially, CEL3C was located in the nucleus. CEL3J was secreted through the nonclassical protein secretion pathway bypassing endoplasmic reticulum (ER) and Golgi. CEL1A was secreted via a vacuole-mediated conventional secretion route without a signal peptide. These findings advance our understanding of b-glucosidase properties and secretory pathways in filamentous fungi, holding key clues for future study.
pathway to the plasma membrane (30,31). Most unconventional secretion pathways are stimulated by stress that may compromise the functional integrity of the classical secretory pathway and trigger the demand for effective alternatives. Secreted proteins utilizing unconventional protein secretion pathways have been found in fungi, such as chitin synthases of Neurospora crassa (32), a cellobiase of the agaric fungus Termitomyces clypeatus (33), effector proteins during biotrophic invasion of Magnaporthe oryzae (27), and endochitinase Cts1 of Ustilago maydis (34). It has been suggested that cellulase excretion in T. reesei might be assisted through unconventional protein secretion pathways, though mostly it occurs via the conventional ER-Golgi secretion pathway (19). Nevertheless, our knowledge of the molecular basis of the nonclassical secretion process in filamentous fungi is still limited.
The effect of the expression of these b-glucosidases on cellulase production of T. reesei using cellulose as the carbon source was investigated by comparison with the parent strain T. reesei RUT-C30 (Fig. 1B to F). Both the FPase activity (Fig. 1B) and the pNPCase activity (the CBH activity) (Fig. 1C) were reduced in all the recombinant strains. For strains Rcel3B, Rcel3E, Rcel3F, Rcel3H, Rcel3J, Rcel3C, and Rcel3G, the CMCase activity (the CMC activity) was lower than that of RUT-C30 at 168 h but rose quickly in the following 72 h to the same level as or a higher level than that of RUT-C30 at 240 h, demonstrating delayed endoglucanase production in these T. reesei transformants (Fig. 1D). The pNPGase activity at 168 h was increased by 121%, 93%, and 156% in strains Rcel3B, Rcel3F, and Rcel3G, respectively, was decreased by 52.7%, 99.4% and 14.5% in strains Rcel1A, Rcel1B, and Rcel3D, respectively, and was almost unchanged for Rcel3C, Rcel3E, Rcel3H, and Rcel3J (Fig. 1E). Specifically, the pNPGase activity in strains Rcel3B and Rcel3G was notably increased by 2.33-and 1.08-fold, respectively, at 240 h, demonstrating that the overexpression of cel3b and cel3g can significantly increase the b-glucosidase production, which was also observed for cel3a (21,22). The total protein concentration of the supernatant stayed almost unchanged in strains Rcel3E, Rcel3F, Rcel3H, Rcel3J, Rcel3C, Rcel3G, and Rcel3D but decreased in Rcel3B, Rcel1A, and Rcel1B (Fig. 1F). It is worth noting that all the cellulase activities and the total extracellular protein concentration were decreased sharply in strain Rcel1B, followed by strain Rcel1A. It seems that the overexpression of CEL1B-DsRed in T. reesei inhibits the cellulase production severely.
Characterization of the recombinant T. reesei strains. The inserted copy number and the mRNA level of each b-glucosidase gene in the corresponding recombinant strain were measured by quantitative PCR (qPCR) (Fig. 2). The inserted copy number was 1 for genes cel3h, cel3j, cel3c, cel1b, cel3g, and cel3d, 2 or 3 for cel3e and cel1a, 3 for cel3b, and 11 for cel3f ( Fig. 2A). Correspondingly, the mRNA levels of genes cel3b, cel3e, cel3f, cel3h, cel3j, cel1a, cel3c, cel1b, cel3g, and cel3d in the corresponding recombinant strains were 1.3, 11.2, 3.6, 36.1, 7.8, 8.8, 3.3, 3.6, 22.4, and 13.6, respectively (Fig. 2B). The mRNA levels of genes cel3b, cel3e, cel3f, cel3h, cel3j, cel3c, cel3g, and cel3d were increased remarkably compared to RUT-C30. In contrast, the mRNA levels of genes cel1a and cel1b stayed almost unchanged in strains Rcel1A and Rcel1B at 168 h, but were increased by 53.3% and 55.4%, respectively, at 120 h (Fig. S2). Furthermore, the transcriptional levels of cellulase genes encoding cellobiohydrolase CBH1 (CEL7A), endoglucanase CMC (CEL7B), and b-glucosidase BGL1 (CEL3A) were reduced along with the overexpression of cel3b, cel3e, cel3j, cel3c, and cel3d, while they stayed constant with the overexpression of cel3f and cel3g (Fig. 2C). In strains Rcel1A and Rcel1B, even though the mRNA levels of genes cel1a and cel1b were not changed markedly, the transcriptional levels of cellulase genes CBH1, CMC, and BGL1 were decreased notably. In strain Rcel3H, the overexpression of cel3h severely inhibited the expression of b-glucosidase BGL1 (Fig. 2B) but caused no noticeable change in the expression of CBH1 and CMC (Fig. 2C). Though only one copy of cel3h was inserted, strain Rcel3H displayed the highest mRNA level of cel3h (Fig. 2C). The glucose concentrations in the fermentation supernatant of each recombinant strain were also checked (Fig. 2D). The extracellular glucose concentration was reduced significantly in the recombinant strains at 72 h, which was not that significant at 120 h and 168 h. Obviously, the overexpression of each b-glucosidase gene decreased the cellulase production at the transcription level (Fig. 2C). The decreased cellulase production was not due to central carbon catabolite repression (CCR), as glucose in the supernatants of the recombinant strains was not increased compared to that of RUT-C30 (Fig. 2D).
Cellular distribution of b-glucosidases. The cellular distribution of different b-glucosidases was observed by confocal laser scanning microscopy (CLSM) using the recombinant strains cultivated in TMM plus 2% cellulose ( Fig. 3 and Fig. S3). Large sums of dense red fluorescent punctuates, probably secretion vesicles, were found throughout the cytoplasm of strains Rcel3B, Rcel3E, Rcel3F, and Rcel3G, while no fluorescence was observed on the cell membrane/wall of these strains. In strains Rcel3H and Rcel3J, red fluorescence was observed in a reticular network with brighter nodes. In these two strains, the red fluorescence was diffused on the endomembrane and the cell membrane/wall. In particular, the red fluorescence was also found in the septa of  Function and Location of b-Glucosidases in T. reesei ® strain Rcel3H. In strain Rcel3C, a series of red fluorescent dots were distributed in the cytoplasm, whose granular fluorescent pattern was totally different from that found in strains Rcel3B, Rcel3E, Rcel3F, and Rcel3G but similar to that of the reported nuclear proteins in T. reesei, such as CLP1 (35), Rce1 (36), MAT1-2-1 (37), XYR1 (38), and CRE1 (38). This result suggests that the protein CEL3C-DsRed is possibly in the nucleus. In strain Rcel1A, globular fluorescent bodies were found, which were much larger than those of strains Rcel3B, Rcel3E, Rcel3F, Rcel3G, and Rcel3C, indicating that CEL1A-DsRed was probably located in vacuoles. CEL3D-DsRed was also mostly located in vacuoles with protein aggregation, unlike that from CEL1A-DsRed diffusing throughout the whole vacuoles.
The red fluorescence appeared at 48 h in strains Rcel3B, Rcel3E, Rcel3F, Rcel3H, Rcel1A, and Rcel3G, 72 h in Rcel3J and Rcel3C, 96 h in Rcel3D, and 216 h in Rcel1B (Fig. S4). The cellulase production was almost abolished in strain Rcel1B (Fig. 1), leading to the extremely poor growth of strain Rcel1B on cellulose. Therefore, the red fluorescence of protein CEL1B-DsRed was not observed until 216 h, which was very weak.
To correlate the timing at which fluorescence appears to the timing of the expression level of each b-glucosidase, the mRNA level of each b-glucosidase gene at the time when fluorescence appears and 24 h before the fluorescence appearance was measured (Fig. S4). Overexpression of cel3b, cel3e, cel3f, cel3j, cel3c, or cel3d was observed in the respective recombinant strains at both time points. Overexpression of cel3h and cel1a was found only at the time of fluorescence appearance, while overexpression of cel3g was observed only at 24 h before the fluorescence appearance. It seems that there is no direct connection between intracellular fluorescence intensity and the transcriptional level of each b-glucosidase (Fig. S4). For instance, although the mRNA level of cel3c in strain Rcel3C was extremely higher than that of cel3b in strain Rcel3B at 48 h, its fluorescence intensity was weaker than that of cel3b.
Identification of extracellular b-glucosidases. To identify whether the tested b-glucosidases are intracellular or extracellular, real-time monitoring of fluorescence intensity in fermentation supernatant was performed for all the recombinant strains and the parent strain RUT-C30 using a fluorescence microplate reader (Fig. 4). No fluorescence was detected in the supernatant of RUT-C30 and its recombinant strain that expressed DsRed alone (19). The fluorescence intensity in the supernatant of strains Rcel1B, Rcel3C, and Rcel3D was similar to that of RUT-C30 and almost stayed unchanged during the whole fermentation process, which suggested that proteins CEL1B-DsRed, CEL3C-DsRed, and CEL3D-DsRed were not secreted into the culture medium. Therefore, CEL3C and CEL3D are intracellular proteins. However, the expression of CEL1B-DsRed was too weak to determine whether CEL1B was intracellular or extracellular, due to almost no cell growth of strain Rcel1B on cellulose. In contrast, the fluorescence intensity in the supernatant of the other seven recombinant strains, Rcel3B, Rcel3E, Rcel3F, Rcel3H, Rcel3J, Rcel1A, and Rcel3G, was increased gradually along with the fermentation time, reaching 109-, 79-, 76-, 21-, 10-, 34-, and 55-fold that of RUT-C30 at 168 h, respectively. This suggested that all these seven DsRed-labeled b-glucosidases were secreted into the extracellular space, indicating that they are extracellular. Except for CEL1A-DsRed, the dynamic fluorescence intensity of these extracellular b-glucosidases had little change at the first 48 h, began to rise in the next 24 h, and went up rapidly from 72 h to 168 h during the cultivation process, showing a similar trend of protein secretion as that of CEL7A (CBH) and CEL7B (CMC) (19). The fluorescence intensity of CEL1A-DsRed was not remarkably increased until 96 h, similar to that of the major extracellular b-glucosidase CEL3A (19). This suggested differences in effective secretion time of these extracellular b-glucosidases. In this way, we showed that b-glucosidases CEL3B, CEL3E, CEL3F, CEL3H, CEL3J, CEL1A, and CEL3G are extracellular, while CEL3C and CEL3D are intracellular. According to their fluorescence intensity in the supernatant, the extracellular b-glucosidases were classified into two groups, the high-secretion (CEL3B, CEL3E, CEL3F, and CEL3G) and the intermediatesecretion (CEL3H, CEL3J, and CEL1A) groups.
The extracellular b-glucosidases CEL3B, CEL3E, CEL3F, CEL3H, and CEL3G were secreted through tip-directed protein trafficking pathways, while CEL3J and CEL1A were secreted through non-tip-directed pathways. A detailed cellular distribution of subapical and apical compartments of the identified extracellular b-glucosidases in T. reesei mutants was imaged to investigate their individual protein trafficking pathways. The proteins CEL3B-DsRed, CEL3E-DsRed, CEL3F-DsRed, CEL3H-DsRed, and CEL3G-DsRed distributed along the hyphal tube in gradient-like fashion, with higher concentrations at apical regions ( Fig. 5 and Fig. S5). This indicated that these five b-glucosidases were secreted to the extracellular space via tip-directed vesicle trafficking, which was the same as CEL3A (19). Among these five b-glucosidases, only CEL3H-DsRed was located abundantly on the cell membrane/wall, septa, and organelle membranes (Fig. 5), implying that the secretion of CEL3H-DsRed might be assisted by both tip-and non-tip-directed protein trafficking pathways. The other two extracellular b-glucosidases, CEL3J-DsRed and CEL1A-DsRed, did not show gradient fluorescent distributions from the apex to the subapex, indicating that their transport might occur via non-tip-directed secretion pathways. Formation of vacuoles occurred in both the apical and subapical segments of the hyphae in strain Rcel1A.
The extracellular b-glucosidases CEL3B, CEL3E, CEL3F, CEL3H, CEL1A, and CEL3G entered the ER, while CEL3J did not. ER-tracker with green fluorescence was utilized to stain the seven strains Rcel3B, Rcel3E, Rcel3F, Rcel3H, Rcel3J, Rcel1A, and Rcel3G to determine if extracellular b-glucosidases enter the ER for protein modification and processing. Except for strain Rcel3J, there was a partial overlap of green and red fluorescence in the other six strains, leading to yellow fluorescence as pointed out by the white arrows in Fig. 6 and Fig. S6, indicative of the colocalization of the six DsRed-labeled b-glucosidases with the ER. This result showed clearly that CEL3B, CEL3E, CEL3F, CEL3H, CEL1A, and CEL3G entered the ER, while CEL3J did not. Interestingly, the green fluorescence of Rcel3J was much weaker than that of the other recombinant strains, demonstrating that the functional integrity of the ER was compromised severely in strain Rcel3J.
Cellular distribution of b-glucosidases under native conditions. To prove that the observed b-glucosidase localizations in BGL-DsRed overexpressed strains reflect those of the native b-glucosidases, we further studied the cellular distribution of a couple of b-glucosidases under the control of their own promoter by integrating fluorescence gene DsRed at their endogenous locus. On the basis of their transcriptional levels in T. reesei RUT-C30 (Fig. 2B), the 10 studied b-glucosidases can be categorized into three groups-cel1a and cel1b with a relative mRNA level of .1 to the housekeeping gene sar1 (39); cel3e, cel3d, and cel3c with a 0.1 , relative mRNA level of ,1; and cel3g, cel3h, cel3b, cel3j, and cel3f with a relative mRNA level of ,0.1. Therefore, cel1a, cel3c, and cel3f, representing each group of b-glucosidases, respectively, were selected for Function and Location of b-Glucosidases in T. reesei ® cellular distribution investigation under native conditions. To achieve this, plasmids for fusing fluorescence gene DsRed to each b-glucosidase at its own endogenous locus through homologous recombination were constructed and transformed individually into T. reesei KU70 (Fig. 7). KU70 was obtained from RUT-C30 by deleting gene ku70 for efficient homologous recombination (40) (Fig. 7A). We obtained 3, 1, and 7 transformants for cel1a, cel3c, and cel3f, respectively, which were named Ncel1A1-3, Ncel3C1, and Ncel3F1-7. The cellulase activities and the secreted protein in these recombinant strains Ncel1A1-3, Ncel3C1, and Ncel3F1-7 changed little compared with the parental strain T. reesei KU70, demonstrating that the integration of fluorescence gene DsRed did not affect the cellulase activities. Cellular observation by CLSM showed that the red fluorescence in strains Ncel1A1-3 appeared at 48 h and was located mainly in vacuole (Fig. 7C), which was the same as that observed in strain Rcel1A (Fig. 3). The fluorescence intensity of culture supernatant in Ncel1A was increased along with time ( Fig. 7D), suggesting that CEL1A could be secreted to the extracellular space. Therefore, CEL1A was proved to be extracellular and distributed in vacuole independent of random overexpression of CEL1A-DsRed or integration of DsRed at the endogenous locus of CEL1A, suggesting that the observed localizations under CEL1A overexpression conditions reflect those of the native CEL1A. Unfortunately, for strains Ncel3C1 and Ncel3F1-7, red fluorescence could not be detected either intracellularly or extracellularly during a 7-day fermentation process, possibly due to the low transcriptional level of cel3c or cel3f (Fig. 2B). It seems that cel3c and cel3f must be overexpressed to study their cellular distribution. BFA inhibited cellulase secretion in Rcel3B, Rcel3E, Rcel3F, Rcel3H, Rcel1A, and Rcel3G, but not Rcel3J. To further study the secretion pathways of the seven extracellular b-glucosidases, the corresponding mutant strains were treated with 10 mg/ml brefeldin A (BFA), which interferes with Golgi body-dependent trafficking (32,41). As indicated by the unchanged red fluorescence in the mycelia of all BFA-treated strains compared to the untreated ones (Fig. S7), the expression of each DsRed-tagged b-glucosidase in the corresponding recombinant strain was not affected by BFA at the tested concentration, similar to that of DsRed-tagged CMC, CBH, and CEL3A (19). However, the fluorescence intensity of the supernatant showed different characteristics among the BFA-treated strains (Fig. 8). For strains Rcel3B, Rcel3E, Rcel3F, Rcel3H, Rcel1A, and Rcel3G, the fluorescence intensity in the supernatant was decreased notably in the presence of BFA throughout the whole fermentation process (Fig. 8A to D, F, and G), suggesting significantly reduced secretion of these extracellular b-glucosidases. Meanwhile, the pNPGase activity, the FPase activity, and secreted protein concentration were also decreased to various degrees in these recombinant strains treated with BFA ( Fig. 8A to D, F, and G). Interestingly, the fluorescence intensity in the supernatant of strain Rcel3J was not affected by the addition of BFA (Fig. 8E), which implied that CEL3J secretion was insensitive to BFA. It seems that the secretion of CEL3J was via an alternative secretion pathway bypassing the Golgi. The more interesting thing is that the pNPGase activity in the BFA-treated strain Rcel3J was similar to that of the untreated strains. This result suggests that the major extracellular b-glucosidase production was also not sensitive to BFA at all in Rcel3J, showing different secretion behavior from that in strains Rcel3B, Rcel3E, Rcel3F, Rcel3H, Rcel1A, RBGL (19), RCMC (19), and RCBH (19). It seems that the expression of CEL3J-DsRed might impart BFA resistance to the cellulase secretion in T. reesei. Function and Location of b-Glucosidases in T. reesei ® CEL3H and CEL3J were located on the cell membrane but not the cell wall. Our result indicates that both CEL3H and CEL3J were probably localized on the cell membrane/wall ( Fig. 2 and Fig. 5). To determine whether they are on cell membrane or cell wall, a green fluorescent dye, GC-PEG-cholesterol-FITC, for cell wall imaging was employed (42). As shown in Fig. 9, green fluorescence was observed on the cell wall of Rcel3H and Rcel3J, which was not overlapped with the red fluorescence of either CEL3H-DsRed or CEL3J-DsRed, as demonstrated by a lack of yellow fluorescence in the merged figures. This result suggests that CEL3H and CEL3J were distributed on cell membrane rather than cell wall.
CEL3C was a nuclear protein. It is worth noting that CEL3C-DsRed was distributed as a series of red dots in strain Rcel3C (Fig. 5) as found for other nuclear proteins in T. reesei, such as CLP1 (35), Rce1 (36), MAT1-2-1 (37), XYR1 (38), and CRE1 (38), suggesting its possible localization in the cell nucleus. The import of nuclear protein into the cell nucleus is mediated by the nuclear localization signal (NLS) that has a basic amino acid cluster rich in lysine or arginine. CEL3C contained NLS ( 324 VLPLSKKKKL 334 ) as predicted by NLS Mapper (http://nls-mapper.iab.keio.ac.jp/cgi-bin/NLS_Mapper_form.cgi). Protein homology modeling of CEL3C on SWISS-MODEL (http://swissmodel.expasy.org/) revealed that the basic amino acid residues KKKK in the NLS were exposed to the surface of the threedimensional (3D) structure of CEL3C (Fig. 10A), benefiting from nuclear import. The cell nucleus of Rcel3C was stained blue by the nuclear dye DAPI (49,6-diamidino-2-phenylindole), which was colocalized with the red fluorescence of CEL3C-DsRed, generating pink Function and Location of b-Glucosidases in T. reesei ® fluorescence in the merged picture (Fig. 10B). This further shows that CEL3C was targeted to the nucleus of hyphal cells.

DISCUSSION
Early studies have demonstrated that b-glucosidases are involved in the regulation of cellulase biosynthesis in T. reesei, affecting the expression and secretion of cellulase both in amounts and in speed (20,21,25), which is further supported by our findings here. The endoglucanase production was delayed in all the recombinant strains expressing different DsRed-tagged b-glucosidases, except for strains Rcel1A, Rcel1B, and Rcel3D, while the pNPCase activity was decreased in all the recombinant strains. Based on the different impacts on overexpression of different cellulase components, the 10 investigated b-glucosidases were classified into three types (Fig. 11)-CEL1A, CEL1B, and CEL3D, whose overexpression repressed all the cellulase activities, including pNPGase, pNPCase, and CMCase; CEL3B, CEL3F, and CEL3G, whose overexpression decreased the pNPCase activity, increased the pNPGase activity, and did not reduce the CMCase activity; and CEL3C, CEL3E, CEL3H, and CEL3J highlighted in black in Fig. 11, whose overexpression decreased the pNPCase activity and left the pNPGase activity unchanged, without compromising the CMCase activity. The overexpression of BGL genes cel1a, cel1b, and cel3d decreased the pNPGase activity (Fig. 1E), which might lead to cellobiose accumulation. This, in turn, could cause product inhibition of upstream cellobiohydrolase CBH and endoglucanase CMC, repressing both the transcription levels and the protein levels of CBH1 and CMC (Fig. 2C, Fig. 1C and D), thus slowing down the cellulose hydrolysis (43,44). In contrast to the reduced pNPCase activity, the CMCase activity was increased at 240 h in strains Rcel3B, Rcel3E, and Rcel3C in spite of a delay, indicating that the overexpression of b-glucosidases affected different cellulase components in different ways. It seems that the different cellulase components can be regulated separately through unknown intricate mechanisms, which requires more study. Therefore, further study on when and how cellulase components  can be tuned separately might be an interesting research direction to reveal the complicated fungal regulatory networks for fine-tuning the expression and secretion of cellulase (45,46).
The overexpression of cel3b and cel3g remarkably increased extracellular b-glucosidase production, similar to cel3a (19). A neighbor-joining phylogram of the putative b-glucosidases revealed that CEL3B and CEL3G shared high protein sequence homology to the major extracellular b-glucosidase CEL3A (47). Proteins CEL3A and CEL3B showed similar catalytic properties, both of which exhibited high specific activity against cello-oligosaccharides and were capable of catalyzing transglycosylation reactions from cellobiose, indicating that CEL3B is probably an alternative to CEL3A. In contrast, CEL3G only showed high substrate specificity toward pNPG. Unlike CEL3A (19), CEL3B and CEL3G were not on the cell membrane.
On the other hand, the overexpression of cel1b completely hampered the induction of cellulase production by cellulose, resulting in almost no growth of T. reesei. This was consistent with early findings that the simultaneous absence of cel1a and cel1b resulted in a marked increase of the extracellular b-glucosidase activity at 12 h (25). More interestingly, gene cel1b is predicted to contain 6 putative binding sites for the carbon catabolic repressor CRE1 that represses cellulase production (48). Nevertheless, the expression of cel1b was increased in T. reesei Dcre1 with high cellulase production and was downregulated in both T. reesei Dxyr1 (48) and ZC121 (49) with fairly low cellulase production on cellulose. In addition, it was found that knockout of gene cel1b in strain TU-6 (the uridine-auxotrophic derivative of T. reesei QM9414) caused only a slight decrease of intracellular pNPGase activity on cellulose (25). The contradictory findings from these previous studies and our study might be explained by the fact that the appropriate level of cel1b expression is required for efficient cellulase production. Also, the overexpression of CEL1A-DsRed reduced the cellulase production moderately in T. reesei on cellulose. Previous research showed that the CMCase, pNPCase, and FPase activity on day 7 have little change growing on cellulose when cel1a was deleted in FIG 11 Cellular distribution and protein secretion pathways of b-glucosidases CEL3B, CEL3E, CEL3F, CEL3H, CEL3J, CEL1A, CEL3C, CEL1B, CEL3G, and CEL3D in T. reesei. CEL3B, CEL3E, CEL3F, CEL3H, CEL3J, CEL1A, and CEL3G were extracellular b-glucosidases. CEL3B, CEL3E, CEL3F, and CEL3G were secreted through the tipdirected conventional ER-Golgi secretion pathway. CEL3H was located on the cell membrane and endomembrane and was secreted through tip-and non-tip-directed conventional ER-Golgi secretion pathways. CEL1A was in vacuole and secreted via the vacuole-mediated conventional ER-Golgi secretion pathway. CEL3J was distributed on the cell membrane and endomembrane and secreted via an unconventional protein pathway bypassing the ER and Golgi. The intracellular b-glucosidases CEL3C and CEL3D were in the nucleus and vacuole, respectively. CEL1B was observed in the cytoplasm with extremely low expression. The names of b-glucosidases are marked with different colors according to the impact of their overexpression on cellulase expression, which are categorized into three types-CEL1A, CEL1B, and CEL3D, marked in green, whose overexpression repressed all the cellulase activities, including pNPGase, pNPCase, and CMCase; CEL3B, CEL3F, and CEL3G, marked in red, whose overexpression decreased the pNPCase activity, increased the pNPGase activity, and did not reduce the CMCase activity; and CEL3C, CEL3E, CEL3H, and CEL3J, highlighted in black, whose overexpression decreased the pNPCase activity, and left the pNPGase activity unchanged, without compromising the CMCase activity.
Function and Location of b-Glucosidases in T. reesei ® QM9414 (50). It was found that knockout of gene cel1a or cel1b did not affect the expression of cellulase but delayed the expression of cellulase (25). cel1a and cel1b are the highly induced genes in T. reesei growing on cellulose or sophorose (51), whose encoded proteins exhibited significant transglycosylation activity when incubated with relatively high concentrations of glucose and cellobiose (25).
Based on the confocal imaging result by CLSM, the cellular distribution patterns of b-glucosidases can be divided into four classes: (i) CEL3B, CEL3E, CEL3F, and CEL3G, forming clearly separated protein secretion vesicles in the cytoplasm; (ii) CEL3H and CEL3J, diffusing the whole endomembrane as well as cell membrane with protein aggregation, like a reticular network; (iii) CEL1A and CEL3D in vacuoles; and (iv) CEL3C in the nucleus. The type II cellular distribution is reminiscent of the major cellulase components CEL3A, CBH, and CMC (19).
Among the 10 tested b-glucosidases, CEL3B, CEL3E, CEL3F, CEL3H, CEL3J, CEL1A, and CEL3G were identified as extracellular, CEL3C and CEL3D as intracellular, and CEL1B as unknown. According to the signal peptide in the N-terminal sequences of the protein (http://www.cbs.dtu.dk/services/SignalP/) (52), b-glucosidases CEL3B, CEL3E, CEL3F, CEL3H, and CEL3G are predicted to be extracellular, which was confirmed by our result. Nevertheless, both CEL3G and CEL1A were reported to be intracellular in some early studies (14,53). When overexpressed under the constitutive Ppdc promoter or the inducing Pcbh1 and Pxyn2 promoters, the level of CEL3G was low and was not detected in the fermentation supernatant by Western blotting (14). In contrast, when the constitutive promoter Ptef1 was utilized for its overexpression, the expression of CEL3G was very high and was found to be extracellular (14). The protein level of CEL3G was good both in cytoplasm and in the supernatant as observed by CLSM (Fig. 3) and a microplate reader, respectively (Fig. 4). It is probably the case that the lack of secretion in the previous study was due to its low protein concentration in T. reesei. CEL1A has also been reported to be intracellular by Western blotting (53). This inconsistency might be due to the relatively poor secretion level of CEL1A at the early fermentation stage as demonstrated by the low fluorescence intensity of CEL1A-DsRed on day 3 (Fig. 4), leading to the unsuccessful detection of the extracellular CEL1A on day 3 by Western blotting as performed in the literature (53). In addition, the cel1a homologue, bgl4 from Humicola grisea, which shared 73.1% identity, was also proved to be extracellular (54). Therefore, we believe that CEL3G and CEL1A are extracellular b-glucosidases. As shown here, it is more reliable to investigate whether protein is extracellular or intracellular by monitoring the dynamic fluorescence intensity during the whole cultivation process. CEL1B does not have the signal peptide for protein secretion and was previously reported to be intracellular (47), which could not be confirmed in our study owing to the fact that only very weak red fluorescence was observed in strain Rcel1B at the end of fermentation (216 h).
Although CEL1A and CEL3J do not contain the signal peptide for protein secretion, we do find them to be extracellular (Fig. 4), indicating that these two proteins are secreted by the unconventional secretion pathways (31). However, CEL1A-DsRed was partially overlapped with ER (Fig. 6), and its secretion was notably inhibited in the presence of BFA (Fig. 8). Moreover, CEL1A was distributed mainly in the vacuoles without enrichment in the apical region (Fig. 5). It seems that CEL1A containing no classical signal peptide was transported to the extracellular medium via the conventional protein pathway involving the ER and Golgi apparatus, which was vacuole-mediated rather than tip-directed. There is a hypothesis that the Golgi-dependent vacuole-mediated secretory route exits in filamentous fungi for regulatory protein secretion (55)(56)(57). In this nontip conventional pathway, proteins are secreted via the fusion of vacuoles with lateral plasma membrane in mature hyphal segments. Similarly, BGL-DsRed, CBH-DsRed, and CMC-DsRed were also observed to reside in the lumen of globular vacuoles (19) in addition to xylanase II (XYN II) (58) and endoglucanase I (59). It is supposed that the cellulase secretion by T. reesei might involve vacuoles (58,60). These hydrolytic enzymes are stored in the vacuoles in a poorly active form to avoid the deleterious effect of their hyperproduction in T. reesei (60). However, unlike BGL, CBH, and CMC, CEL1A does not contain any signal peptide for secretion. How CEL1A can enter the ER without signal peptide is unknown. The leaderless proteins such as fibroblast growth factor 9 (Fgf-9) (61) and an oxidoreductase (POR) from Cryphonectria parasitica (62) have been reported to enter the ER and traverse the Golgi complex in a similar manner to other constitutively secreted proteins, even though they do not contain signal peptide, a transmembrane region, or Golgi localization information (62).
CEL3J was localized to the cell membrane ( Fig. 9) and was also extracellular (Fig. 4). It could reach the plasma membrane via an unconventional protein pathway that did not involve the ER and Golgi apparatus (Fig. 11), as demonstrated by its lack of colocalization with ER tracker (Fig. 6) and resistance to BFA (Fig. 8). What is more, the total b-glucosidase secretion in strain Rcel3J was also BFA-insensitive. Although the expression level of cel3j in strain Rcel3J was much higher than that in RUT-C30, it was still lower than that of cel3e in Rcel3E, cel3h in Rcel3H, cel1a in Rcel1A, cel3g in Rcel3G, and cel3d in Rcel3D (Fig. 2B). However, only the overexpression of cel3j affected ER integrity in strain Rcel3J, whose protein secretion was insensitive to BFA via bypassing the ER. Based on this, we do not think these results regarding an alternate secretion pathway of CEL3J were due to the overexpression of cel3j. A similar phenomenon has been reported in Termitomyces clypeatus, where the cellobiase titer was increased by 17% in the presence of 50 mg/ml BFA, although the total protein secretion was inhibited by 30%. It is suggested that an alternative secretory pathway was operative to deliver cellobiase to the vacuoles for secretion without passing the Golgi (33). In another study, the secretion of the cytoplasmic effectors in the rice blast fungus Magnaporthe oryzae was BFA-resistant, suggesting their secretion is via the unconventional Golgi-independent pathway (63). The secretion of these cytoplasmic effectors is facilitated via the biotrophic interfacial complex, which has not been identified in T. reesei.
It is unexpected to find that one of the tested b-glucosidases was located in the nucleus, namely, CEL3C. Nevertheless, the role of CEL3C is still unknown. Both CEL3C and CEL3D are intracellular, having XYR1-binding sites in the promoter regions (64). They were low in abundance and may have been somewhat more highly induced by cellulose than by sophorose in QM6a (51). They behaved like aryl b-glucosidases, showing high preference for pNPG (47). Nevertheless, unlike CEL3C, CEL3D was mainly located in vacuoles with protein aggregation. It seems that some of the 11 b-glucosidases in T. reesei are not mainly for cellulose degradation/cellulase production. For a better understanding of the function of these b-glucosidases, it is worthwhile to explore how the recombinant strains overexpressing each b-glucosidase behave under other conditions, such as noncellulose carbon sources such as lactose, glucose, galactose, and glycerol, which is undergoing extensive study in our lab.
It is worth noting that the protein localization of b-glucosidases was studied by overexpression of their fusion with DsRed in this study. Sometimes, the protein overexpression might complicate the interpretation of the results when the overexpression of proteins saturates intracellular transport mechanisms, leading to abnormal subcellular localization. However, T. reesei possesses efficient protein synthesis and protein secretion ability (11)(12)(13). Though all the tested overexpressed b-glucosidases in their respective recombinant strains did not replicate normal expression, their transcription levels were no more than 12% of that of CBH1 in RUT-C30 (Fig. 2B). Thus, the overexpression of each b-glucosidase should not be a burden to the cells, avoiding abnormal distribution. Moreover, we have labeled CEL1A with fluorescence protein DsRed at its endogenous locus under its own promoter to prove that the cellular distribution of CEL1A-DsRed under overexpression conditions reflects that of the native CEL1A. Therefore, the cellular distribution of BGL-DsRed fusions that we observed was probably ascribed to each BGL itself fused with DsRed, but not gene overexpression. Early studies regarding protein localization have also used heterologous promoters. For example, heterologous promoters tcu1, CBH1, and CaMV-35S were used for investigating subcellular localization of nuclear proteins (CLP1 [35], Rce1 [36], MAT1-2-1 [37], XYR1 [38], and CRE1 [34]) in T. reesei, SNARE proteins (SNCI, SSOI, and SSOII) (29) in T. reesei, and a unique protein RxLR3 in Phytophthora brassicae (65), respectively. Another study about searching strains with anomalous temporal or spatial protein localization also used heterologous promoter xyl or van (66). Unfortunately, red fluorescence was not observed in strains Ncel3C1 and Ncel3F1-7, which might due to the low transcriptional level of cel3c and cel3f, making it impossible to investigate the cellular distribution of these two proteins under native conditions. For some proteins whose expression is too low to be detected under endogenous promoters, overexpression is required to investigate their protein localization (67,68). Knowledge of cellular localization under overexpressed conditions has the potential under normal conditions to provide insights into the protein localization, giving direction for further investigation. Considering these findings together, we believe that the gene overexpression using promoter CBH1 could not interfere with our result on protein localization of b-glucosidases.
In conclusion, the effect of b-glucosidases on cellulase production was first investigated together with their spatiotemporal localization and secretion by successfully constructing recombinant T. reesei strains expressing red fluorescent protein DsRedtagged versions of each b-glucosidase. The endoglucanase production was delayed in all the recombinant strains expressing different BGL-Dsred except strains Rcel1A, Rcel1B, and Rcel3D. The overexpression of cel1b almost repressed the cellulase production completely, while the overexpression of cel3b or cel3g significantly enhanced extracellular b-glucosidase production. The cellular distribution patterns of b-glucosidases were classified into (i) CEL3B, CEL3E, CEL3F, and CEL3G, displaying protein secretion vesicles in the cytoplasm; (ii) CEL3H and CEL3J, distributing through the whole endomembrane and cell membrane with protein aggregation, like a reticular network; (iii) CEL1A and CEL3D in vacuoles; and (iv) CEL3C in the nucleus. CEL3B, CEL3E, CEL3F, CEL3H, CEL3J, CEL1A, and CEL3G were identified as extracellular, CEL3C and CEL3D as intracellular, and CEL1B as unknown. The extracellular b-glucosidases CEL3B, CEL3E, CEL3F, CEL3H, and CEL3G were secreted through the tip-directed classical secretion pathway, and CEL1A was secreted through via the vacuole-mediated pathway that was achieved without any signal peptide, while CEL3J was secreted via a nonclassical secretion route independent of both ER and Golgi. These findings have implications for our fundamental understanding of b-glucosidases in filamentous fungi.

MATERIALS AND METHODS
Microbial strains and plasmids. Escherichia coli DH5a was used as the cloning host (Vazyme Biotech Co., Ltd., Nanjing, China). T. reesei RUT-C30 (CICC 13052) was applied as a host for expressing DsRed-tagged b-glucosidases. Agrobacterium tumefaciens AGL-1 was utilized as a transfer DNA (T-DNA) donor for fungal transformation. A T-DNA binary plasmid, p-DsRed, containing a red fluorescent proteincoding gene was maintained in our laboratory and used to construct the transformation vectors (19).
To label BGLs at their endogenous locus, we first ligated a short linker (GGTGGCGGTGGCTCGGGT GGCGGTGGCTCG) and DsRed to pXBthg (69) with the ClonExpress MultiS one-step cloning kit (Vazyme Biotech Co., Ltd., Nanjing, China), resulting in plasmid pXBred. Then the upstream (including the promoter and coding sequence) and downstream (including the terminator) fragments of cel1a, cel3c, or cel3f amplified from the genomic DNA of T. reesei RUT-C30 were ligated to the plasmid pXBred at the XhoI and BamHI sites, respectively. The resulting expression vectors were transformed into T. reesei KU70 (70) by the AMT method using hygromycin B as a marker, yielding recombinant strains Ncel1A, Ncel3C, and Ncel3F, respectively. The verification of DNA integration in the recombinant strains was confirmed by PCR (Fig. S8) and sequencing of the PCR product at Sangon Biotech.