| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
J. Biol. Chem., Vol. 278, Issue 34, 31988-31997, August 22, 2003
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
From the
Genencor International, Inc., Palo Alto, California 94304 and the
Fungal Genomics Laboratory, North Carolina State
University, Raleigh, North Carolina 27695-7251
Received for publication, May 7, 2003 , and in revised form, June 3, 2003.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
|
|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
The principal component of biomass is cellulose. It consists of polymers of
-1,4-linked glucose residues that are organized into higher order
fibrillar structures (8). In
addition to cellulose, biomass contains large quantities of hemicelluloses,
heteropolysaccharides composed of two or more monosaccharides such as
D-xylose, L-arabinose, D-mannose,
D-glucose, D-galactose, and
4-O-methyl-D-glucuronic acid
(1).
Among the most prolific producers of biomass-degrading enzymes is the
filamentous fungus Trichoderma reesei. The cellulase activity
produced by T. reesei is composed of a complement of endoglucanases
(EGI/Cel7B, EGII/Cel5A, EGIII/Cel12A, EGIV/Cel61A, and EGV/Cel45A) and
exoglucanases (the cellobiohydrolases, CBHI/Cel7A, and CBHII/Cel6A) that act
synergistically to break down cellulose to cellobiose (glycosyl
-1,4-glucose)
(9–14).
Two -glucosidases (BGLI/Cel3A and BGLII/Cel1A) have been identified that
are implicated in hydrolyzing cellobiose to glucose
(15,
16). An additional protein,
swollenin (encoded by the gene swo1), has been described that
disrupts crystalline cellulose structures, presumably making polysaccharides
more accessible to hydrolysis
(17). The four most abundant
components of T. reesei cellulase CBHI/Cel7A, CBHII/Cel6A, EGI/Cel7B,
and EGII/Cel5A together constitute greater than 50% of the protein produced by
the cell under inducing conditions and can be secreted in excess of 40 g/liter
(18).
The regulation of cellulolytic enzyme expression in T. reesei is
complex and only partially understood. Transcription of the major components
of cellulase (CBHI/Cel7A, CBHII/Cel6A, EGI/Cel7B, EGII/Cel5A, EGIII/Cel12A,
EGIV/Cel61A, and EGV/Cel45A) is induced not only by cellulose but also by a
variety of disaccharides including lactose, cellobiose, and sophorose
(glycosyl -1,2-glucose)
(13,
19,
20). Induction by these
molecules is antagonized by the presence of the preferred carbon sources,
glucose and fructose. Sophorose is by far the most potent inducer of cellulase
expression (21). However, it
is unclear whether this high potency is an innate characteristic of the
molecule or whether other disaccharides are less effective because they are
more readily cleaved by cellular glucosidases. The balance between intact
disaccharides and inhibitory monosaccharide cleavage products might then
influence the transcriptional state of these genes. In accordance with this
notion both lactose and cellobiose fail to fully induce cellulase gene
expression when present at high concentrations.
In addition to enzymes with cellulolytic activity, a number of enzymes have
been identified in T. reesei that degrade hemicellulose
(17,
22–31).
These enzymes include four xylanases (Xyn1, Xyn2, Xyn3, and Xyn4) and
mannanase (Man1), which cleave the xylan and mannan main chains of
hemicellulose. Acetyl xylan esterase (Axe1), 
-glucuronidase (Glr1), and
arabinofuranosidase (Abf1) digest side chains containing acetyl,
methylglucuronic acid, and arabinose moieties, respectively. Additionally,
enzymes that digest oligosaccharides derived from hemicellulose have been
identified. These are -xylosidase (Bxl1) and three
-galactosidases (Agl1, Agl2, and Agl3).
Among the hemicellulases that have been studied, all except arabinofuranosidase are expressed at a substantially higher level when T. reesei is grown in medium containing cellulose than when it is grown in medium containing non-inducing carbon sources such as sorbitol (28, 32). Additionally, expression of many of the genes encoding these enzymes, with the notable exception of man1, is induced by xylans. Sugars such as sophorose, arabitol, xylobiose, cellobiose, and galactose also induce expression to varying extents, particularly of enzymes that degrade substrates related to these sugars (32). The molecular mechanisms by which T. reesei senses the composition of the extracellular milieu and modulates the expression of these enzymes are unknown. Moreover, although expression of some of these genes is modified by certain substrates, it is unclear to what extent each gene has a unique regulatory apparatus and to what extent expression of these genes are coupled among themselves and with the cellulases via sharing of regulatory pathways. The mechanisms by which the cellulase and hemicellulase genes are regulated are likely to influence the ecological niches that T. reesei occupies and are of interest in the commercial production of these enzymes.
The very large quantity of biomass-degrading enzymes synthesized by T. reesei requires a significant investment of cellular resources. Evidence suggests that a primary means by which the cell manages these demands is to regulate transcription of the genes encoding these enzymes according to the availability of different carbon sources. However, the degree to which gene products involved in other cellular processes, such as secretion, must be regulated to accommodate the substantial burden of cellulase biosynthesis has not been systematically investigated.
In this study, aspects of these questions are addressed by determining the transcriptional effects of cellulase inducers on a genomic scale. Over 5100 cDNAs from T. reesei were partially or fully sequenced. Twelve cDNAs encoding new enzymes with putative roles in biomass degradation were discovered. Microarrays were used to examine the regulation of these and previously identified genes encoding biomass-degrading enzymes in the context of the extensive repertoire of newly identified genes. The results presented shed light on the coregulation of cellulases and hemicellulases and the mechanisms by which the cell copes with synthesizing very large quantities of these secreted enzymes.
|
EXPERIMENTAL PROCEDURES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
For Northern blot analysis, 
1 x 108 spores were
inoculated into 50 ml of minimal medium supplemented with 5% glucose and grown
at 30 °C for 24 h. Mycelia were collected by centrifugation, washed in
carbon-free medium, and resuspended to an optical density of 0.3 in 50 ml
of minimal medium supplemented with 5% glucose, 2% avicel (FMC), 2% glycerol,
or 2% glycerol containing 1 mM sophorose (Sigma). Cultures were
grown in flasks with vigorous agitation for 20 h. For microarray analysis,
inoculation was performed as above. Cultures were grown in triplicate
overnight. Mycelia were collected by centrifugation and washed with minimal
medium containing 2% glycerol. They were then resuspended to an OD of
0.15–0.2 and grown for 9 h (OD 0.5). The cultures were then divided in
half, and 0.01 volume of 100 mM sophorose was added to 1 flask of
each pair immediately and 10 h later. The cultures were grown an additional 2
h to an OD 2–4.
cDNA Library and Sequencing—T. reesei strain QM6a mycelia were grown in baffled flasks at 30 °C for 24 h in YEG medium with vigorous aeration. 5 ml of this culture was added to 50 ml of the following media and grown under the following conditions: Vogel's liquid medium, 2% avicel, 3 and 6 days; Vogel's liquid medium, 2% Solka floc (International Fiber Corp., North Tonawanda, NY), 3 and 6 days; Vogel's liquid medium, 2% wheat bran (Skidmore Sales and Distributing Co., Inc., West Chester, OH), 3 and 6 days; Vogel's liquid medium, 2% beet pulp (D&D Ingredients Distributors, Inc., Delphos, OH), 6 days; Vogel's liquid medium, 2% glucose, 24 h; Vogel's liquid medium, 2% lactose, 24 h; Vogel's liquid medium, 2% xylose, 24 h; Vogel's liquid medium, 2% fructose, 24 h; Vogel's liquid medium, 2% maltose, 24 h; Vogel's liquid medium, no carbon source added, 24 h; Vogel's liquid medium, no nitrogen source added, 24 h; Vogel's liquid medium, 2% phosphoric acid swollen cellulose, 3 days; YEG medium, 42 °C, 1.5 h; YEG medium, 20 mM dithiothreitol, 1.5 h; YEG, room temperature, closed container with no agitation (anoxia), 1.5 h; solid state, 15 g of wheat bran, 1 g of proflo, 1 g of solkafloc, 30 ml of water, 6 and 7 days; solid state, 15 g of beet pulp, 1 g of proflo, 1 g of solkafloc, 30 ml of water, 9 days.
RNA was prepared from the mycelia by grinding under liquid nitrogen with a mortar and pestle and extracting using Trizol reagent (Invitrogen) according to manufacturer's instructions. cDNA libraries were constructed by Invitrogen in the vector pREP3Y, which is a derivative of pREP3X (33) containing additional restriction sites at the multiple cloning site. ESTs1 were generated by sequencing cDNA clones from the 5' end. Template DNA was extracted in a 96-well format using a modified alkaline lysis protocol. Sequencing reactions were performed following standard Big Dye (Applied Biosystems) protocols for a 0.25x reaction. Cycle sequencing was performed over 35 cycles (96 °C for 10 s; 50 °C for 5 s; and 60 °C for 4 min) in an Applied Biosystems GenAmp 9700 thermocycler. DyeEx 96-well plates (Qiagen) were used for dyeterminator removal. Samples were sequenced using an ABI 3700 capillary sequencer (Applied Biosystems).
Fermentation—Duplicate fermentations were run for each of
the strains displayed in Fig.
3. 0.8 liters of minimal media containing 5% glucose was
inoculated with 1.5 ml of frozen spore suspension. After 48 h, each culture
was transferred to 6.2 liters of the same media in a 14-liter Biolafitte
fermenter. The fermenter was run at 25 °C, 750 rpm, and 8 standard liters
per min of air flow. One hour after the initial glucose was exhausted, a 25%
(w/w) lactose feed was started and was fed in a carbon-limiting fashion to
prevent lactose accumulation. The concentrations of glucose and of lactose
were monitored using a glucose oxidase assay kit or a glucose hexokinase assay
kit with -galactosidase added to cleave lactose, respectively
(Instrumentation Laboratory Co., Lexington, MA). Samples were obtained at
regular intervals to monitor the progress of the fermentation. Samples
obtained before (20–35 g/liter glucose) and just after glucose
exhaustion, and 24 and 48 h after the lactose feeding commenced were used for
microarray analysis.
|
Isolation of RNAs, Labeling, and Hybridization—Mycelia were harvested by filtration through miracloth (Northern blots and fermentation samples) or by vacuum filtration through Whatman No. 1 paper and were quick frozen in liquid nitrogen. For Northern blotting, RNA was prepared as described above for construction of cDNA libraries. Polyadenylated RNA was selected 2 times using Oligotex (Qiagen). Blotting was performed using a NorthernMax-Gly Kit (Ambion). 32P-Labeled probes were generated using a DECAprime Kit (Ambion). Hybridization was performed using ULTRAhyb Ultrasensitive Hybridization Buffer (Ambion).
In all of the microarray experiments performed, the relative expression levels were determined between two RNA samples derived from growth under two different conditions. RNA was prepared using a FastRNA (Red) Kit (Qbiogene). Two aliquots of total RNA were taken from each sample. One aliquot was labeled with cyanine 3-CTP, and the other aliquot was labeled with cyanine 5-CTP (PerkinElmer Life Sciences) using a fluorescent linear amplification kit (Agilent Technologies). 250 ng of cyanine 5-labeled RNA derived from mycelia grown under one condition of interest were combined with 250 ng of cyanine 3-labeled RNA derived from mycelia grown under a second condition of interest, and the pooled RNAs were hybridized to the microarrays using an in situ hybridization kit (Agilent Technologies). The relative fluorescent intensities of cyanine 5- and cyanine 3-labeled species bound to each probe of the microarray were determined using an Agilent Technologies microarray scanner and software. The log of the ratio (log ratio) of the two fluorescent species bound to each of the probes reflects the relative expression levels of the cognate genes in the two samples (34, 35). To avoid possible bias resulting from differences in incorporation of the dyes, duplicate microarrays were hybridized in which reciprocal dye combinations were used. Thus, for example, in sophorose induction experiments, cyanine 5-labeled RNA derived from sophorose-induced cultures was combined with cyanine 3-labeled RNA from uninduced cultures; and in replicate microarrays, cyanine 3-labeled RNA derived from sophorose-induced cultures was combined with cyanine 5-labeled RNA from uninduced cultures. A total of 6 microarrays was performed for each condition in the sophorose induction experiments (duplicate reciprocally labeled arrays for triplicate cultures). In the lactose induction experiments 4 microarrays were performed for each time point (duplicate reciprocally labeled arrays for duplicate fermenters). For all figures the mean values across replicates are presented.
Microarray Design—60-mer oligonucleotides corresponding to the assembled ESTs were designed and synthesized in situ by Agilent Technologies.
Bioinformatics and Data Analysis—Sequence chromatograms were
assigned base quality values by Phred (version 0.990722,
www.phrap.org).
Sequences containing >100 bases with Phred quality values 
20 or an
average base quality 12 were retained. High quality sequences were
assembled using the PhredPhrap (version 0.990329) script provided by Consed
(version 11.0) (36). Contigs
were virtually translated in 6 reading frames and were annotated using the
BioSCOUT system from LION Biosciences as in Andrade et al.
(37). Briefly, biasdb
(37) and Seq
(38) were used to identify
compositionally biased regions and regions of low compositional complexity,
respectively. For each sequence, homology search results using FASTA
(39) and BLAST
(40) were pooled, and a
lexical analysis procedure was applied to the descriptive field of the set of
homologs (37). Annotations
were classified as direct, clear, tentative, marginal, or unknown according to
the following criteria: 10<–70,
10–10 to 10–70,
10–4 to 10–10, 0.1 to
10–4 and >0.1 for BLAST; and >500, >130,
>90, and >0 for FASTA, respectively. Selected sequences were further
analyzed by aligning them with similar genes using ClustalW
(clustalw.genome.ad.jp/).
All tools were used with default parameters. Secretion signal sequence
prediction was performed according to Ref.
41. Glycoside hydrolase
families and carbohydrate-binding module families are were assigned as in Ref.
42. To identify genes involved
in protein processing and secretion, the sequences of gene products of
interest from other organisms were compared by BLAST to the translated ESTs.
Microarray data were quantified using Feature Extraction software (Agilent
Technologies). The data were visualized and analyzed using Genespring version
4.2 (Silicon Genetics). Clustering was performed using a standard correlation
and Genespring default settings.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
The predicted coding regions of the EST set were compared with all publicly available sequence data bases. Twelve new sequences were identified encoding proteins with significant similarity to known enzymes whose substrates are commonly found in biomass. Full-length sequences corresponding to these gene products were determined. The genes encoding these sequences along with relevant previously identified enzymes from T. reesei are listed in Table I.
|
Among the newly identified sequences were 3 new predicted endoglucanases Cel74a, Cel61b, and Cel5b. The expected amino acid sequences of these gene products correspond to glycoside hydrolase (GH) families 74, 61, and 5, respectively, based on alignments with other members of these families (43). All of these sequences contain a signal sequence, suggesting that they are secreted proteins. Cel74a is the first member of GH family 74 to be found in T. reesei, and it includes a C-terminal carbohydrate-binding domain of the CBM1 family (42). Unlike the previously known T. reesei member of the GH61 family (EGIV/Cel61a), Cel61b has no carbohydrate-binding motif. Cel5B has a C-terminal hydrophobic region and a consensus sequence suggesting that it may be modified by a glycosylphosphatidylinositol anchor attached to serine 411 (44). This gene product may therefore represent a membrane-bound endoglucanase.
Five new putative -glucosidase sequences were also discovered. As
indicated in Table I, they fall
into GH families 1 and 3. Cel3b and Cel3e contain presumed signal sequences
suggesting that they encode secreted enzymes.
Two predicted hemicellulolytic enzymes including an acetyl xylan esterase (Axe2) and an arabinofuranosidase (Abf2) were also found. Like Cel5b, Axe2 is predicted to contain a glycosylphosphatidylinositol membrane anchor. The glycoside hydrolase and carbohydrate esterase families corresponding to each of these sequences are listed in Table I. Sequence alignments for the predicted gene products with representative members of glycoside hydrolase families are available in the Supplemental Material. Two additional sequences designated Cip1 and Cip2 were found and each contain a fungal-type cellulose binding domain and a signal sequence. The predicted amino acid sequences of these genes do not show compelling sequence similarity to any of the currently defined GH families. However, Cip1 is very similar to an unassigned putative secreted hydrolase from Streptomyces coelicolor (45) (TrEMBL accession number O69962), and Cip2 is highly similar to a multidomain esterase from Ruminococcus flavefaciens that contains an acetyl xylan esterase domain outside the region of sequence similarity with Cip2 (46) (GenBankTM accession number AJ238716 [GenBank] ).
Regulation of the Endoglucanases and the
-Glucosidases— The previously identified endoglucanases
are induced during growth on media containing cellulose, sophorose, or lactose
(13,
19,
20). To determine whether the
newly discovered endoglucanases are similarly regulated, we examined mRNA
levels for each of these gene products by Northern blot. Two different strains
were used: QM6a, a wild type isolate of T. reesei, and RL-P37, a
strain that has been selected for improved production of cellulolytic enzymes
(47). Mycelia from each of
these strains were grown in flasks in minimal media containing glucose,
crystalline cellulose (avicel), or glycerol as the sole carbon source, or
glycerol supplemented with 1 mM sophorose. As shown in
Fig. 1A, with the
exceptions of egl4/cel61a, cel61b, and cel5b, the
endoglucanases were regulated very similarly to one another. Induction by
sophorose resulted in much higher levels of expression than did growth on
cellulose over the time period examined. In addition, expression of these
genes was substantially higher in the strain RL-P37 than it was in QM6a. The
regulation of egl4/cel61a differed from the other
endoglucanases because while it was also more abundant in RL-P37 than it was
in QM6a, it appeared to be comparably induced by growth in cellulose and by
growth in sophorose. Equivalent expression in cellulose and sophorose was also
observed for cel61b in RL-P37 but not QM6a. In contrast,
cel5b, the putative membrane-bound endoglucanase, displayed much less
dramatic regulation. It was clearly expressed on both glucose and glycerol and
induced only a small amount by either cellulose or sophorose. Moreover,
cel5b expression may have been somewhat higher during growth on
cellulose than during growth on media containing sophorose, particularly in
QM6a.
|
The regulation of -glucosidases was examined in an analogous fashion.
The GH family 1 -glucosidases were induced similarly to the
endoglucanases. In contrast to the endoglucanases, however, cel1b was
only slightly more abundant in RL-P37 than it was in QM6a. Among the family 3
-glucosidases, bgl1/cel3a was also regulated similarly
to the endoglucanases in RL-P37 but displayed little induction in QM6a.
cel3c and cel3d were low in abundance and may have been
somewhat more highly induced by cellulose than by sophorose in QM6a. However,
in RL-P37, sophorose induction of cel3c and cel3d was more
apparent than cellulose-mediated induction. cel3b and cel3e
displayed particularly interesting regulation. In QM6a, the expression levels
of these genes were higher when the cultures were grown on glycerol than when
they were grown on glucose. As a consequence, although both cellulose and
sophorose induced expression, the magnitude of the induction appears greater
when comparing cellulose-grown cultures to glucose-grown cultures than when
comparing glycerol-grown cultures in the presence and absence of sophorose.
Surprisingly, in RL-P37 cellulose-mediated induction of cel3b was
apparent although sophorose appeared to repress expression below the basal
level seen when mycelia are grown on glycerol alone. Thus, although the
selective pressures employed to improve cellulase production in RL-P37 appear
to have generally resulted in increased inducibility and abundance of the
endoglucanases, the regulation of the -glucosidases seems to have been
altered in complex and unexpected ways.
Genome Scale Measurement of Transcriptional Regulation Using Microarrays—To measure the expression levels of all of the ESTs sequenced, microarrays were constructed. Sixty-bp oligonucleotide probes containing unique sequences from within each of the ESTs were designed to query the abundance of their corresponding mRNAs. The oligonucleotide probes were synthesized and arrayed as described in Ref. 34. In all of the experiments performed, the microarrays were used to determine the relative expression levels between two different samples. mRNAs comprising the samples of interest were labeled with Cy5 and with Cy3 fluorescent dyes. Reciprocally labeled pairs of samples were combined and co-hybridized to the arrays. The log of the ratio (log ratio) of the two fluorescent species bound to each of the probes reflects the relative expression levels of the cognate genes in the two samples (34, 35).
To validate the performance of these arrays in measuring T. reesei gene expression, the genes shown in Fig. 1A were analyzed using the microarrays. The magnitude of sophorose induction of the genes was determined by labeling mRNAs from cultures of RL-P37 and QM6a grown in glycerol or glycerol supplemented with sophorose. Samples from sophorose-grown cultures were directly compared with samples of the same strain grown in glycerol. The log ratios of expression in sophorose-grown cultures relative to expression in glycerol-grown cultures are indicated in Fig. 1B. In a second experiment, levels of expression were directly compared across the two strains in the presence and absence of sophorose. mRNA samples from RL-P37 cultures were directly compared with samples from QM6a cultures grown under identical conditions. Fig. 1C shows the log ratios of the expression levels of each of the genes in RL-P37 relative to the expression levels in QM6a. Overall, the log ratios observed using the microarrays are in very good agreement with the differences in the magnitudes of expression observed using Northern blots. However, particularly for some of the lower abundance mRNAs (e.g. cel3e and egl5/cel45a), some discrepancies are observed. These discrepancies arise primarily from variability among cultures (note that the cultures used for the microarray experiments were grown under slightly different conditions than for the Northern blots). Additionally, it was unclear whether the log ratios observed precisely and quantitatively mirror the abundance differences seen by Northern blot over the full dynamic range of expression levels. Therefore, for clarity throughout this paper, we refer to log ratios rather than fold changes in expression.
Analysis of the data obtained in these experiments was extended to include all of the genes encoding putative biomass-degrading enzymes listed in Table I. Fig. 2A shows the results obtained. The data were clustered across three measurements as follows: the magnitude of sophorose induction in RL-P37, the magnitude of sophorose induction in QM6a, and a comparison of the relative levels of expression between RL-P37 and QM6a in sophorose-induced cultures. Clustering across this group of experiments highlights not only the regulation of the genes in the two strains but also the patterns of expression that may have been co-selected with enhanced cellulase production in RL-P37. The clustogram has three primary branches.
One primary branch contains a large cluster of genes that were moderately
to highly sophorose-induced in RL-P37. They were also expressed at a
significantly higher level in RL-P37 than they were in QM6a. This cluster
contains all of the known and putative endoglucanases and exoglucanases, and
all of the known and presumed -glucosidases except cel3b.
Interestingly, a variety of hemicellulases, as well as cip1, cip2,
and swo1 also fall into this cluster. Most of the genes in this large
cluster were induced to some extent in both strains but were a great deal more
induced in RL-P37 than they were in QM6a. However, a subset of genes in this
group, including the xylanases xyn2, xyn3, and xyn4, was not
induced by sophorose in QM6a. Two smaller branches within this cluster showed
somewhat different patterns from the rest of the genes. One subbranch
contained cel5b and bgl2/cel1a, which were induced
approximately equivalently in the two strains and were expressed more highly
in RL-P37 than in QM6a. A second subbranch contains bga1 and
cel3e, which were both induced as much or more in RL-P37 than in
QM6a, but were not more highly expressed in RL-P37 than in QM6a.
The second primary branch of the clustogram consisted of genes that were either not induced or were slightly repressed by sophorose in RL-P37. Interestingly, one of these, cel3b, was highly induced in QM6a. The final primary branch consisted of the single gene abf2, whose expression was unaffected by sophorose.
The pattern of transcriptional regulation in response to sophorose of the complete EST data set is shown in Fig. 2B. The expression profiles observed for the two strains are substantially different. In light of the fact that RL-P37 was derived from QM6a and was selected for its increased capacity for cellulase production, it is likely that many of the differences observed between the two strains influence cellulase synthesis. The full data sets corresponding to Fig. 2, A and B, are available in the Supplemental Material.
To explore further the selective pressures that may have brought about increased cellulase production in RL-P37, we examined ESTs for which the induction or repression in response to sophorose resulted in an absolute level of expression that is either higher or lower in RL-P37 than in QM6a, respectively. In Fig. 2C, ESTs were identified for which the log ratios observed for sophorose induction were higher than 0.3 in QM6a. This set of ESTs was compared with the set that displayed sophorose induction log ratios greater than 0.3 in RL-P37. Eighteen ESTs were induced in common in the two strains. The sets of ESTs that were induced by sophorose in each of the strains were then compared with the set of ESTs that is expressed at a higher level in RL-P37 than in QM6a. Among the 18 ESTs that are induced in common in the two strains, 9 are expressed more highly (log ratio >0.3) in RL-P37 than in QM6a. These ESTs include bgl2/cel1a, cip1, bxl1, a putative aminopeptidase, a putative lactose permease, and four ESTs encoding proteins whose function could not be predicted based on the sequence information available. The majority of the genes tabulated in Table I fall into set e (see Supplemental Material) in Fig. 2B because, although they may be induced in QM6a, the log ratios reflecting the magnitude of their induction were less than 0.3.
A similar analysis was performed for ESTs whose expression was repressed in response to sophorose. Interestingly, there was a set of 4 ESTs that was repressed by sophorose in both strains and whose expression levels were reduced in RL-P37 relative to QM6a. These ESTs may therefore encode proteins whose expression at a relatively high level is detrimental to cellulase production. The identities of the ESTs in each of the sets indicated in Fig. 2, C and D, and their log ratios are available in the Supplemental Material.
Lactose-mediated Induction of Genes Encoding Biomass-degrading Enzymes—Most of the biomass-degrading enzymes listed in Table I are induced by sophorose. These enzymes may be induced via a common sensing and transduction pathway or may be induced via separate pathways. To explore further commonalities in their regulation, the effects of a second inducer, lactose, were examined.
Because lactose inhibits cellulase expression when present at high concentration, mycelia were grown in a fermenter where limited lactose can be supplied at a controlled rate. Initially, cultures were grown in excess glucose to allow rapid cell division and accumulation of cell mass. The cultures were allowed to deplete the medium of glucose. After an hour of carbon deprivation, cultures were fed lactose at a fixed rate that prevented accumulation in the medium. Under these conditions, the cells only undergo significant growth and division during the period when excess glucose is present. Samples of RL-P37 and QM6a mycelia were harvested from the fermenters during growth in glucose, an hour after the glucose had been depleted (carbon deprivation stage) and at 24 and 48 h after lactose feeding commenced. RNA was prepared from each of the samples, and microarray analysis was performed. The expression levels of the biomass-degrading enzymes listed in Table I in lactose and in glucose was determined relative to expression during carbon deprivation (Fig. 3A).
The majority of the genes that were induced by sophorose were also induced by lactose. However, four genes that were induced by sophorose were not induced by lactose under these conditions. bga1 was induced equivalently by sophorose in both RL-P37 and in QM6a, but its expression appeared unaffected by either glucose or lactose. axe2 was slightly sophorose-induced in QM6a but was slightly repressed by lactose, particularly in QM6a. cel3c was induced similarly in both strains by sophorose and was repressed, again particularly in QM6a, by glucose and lactose. Finally, cel3b was induced by sophorose in QM6a but not in RL-P37. Its expression was significantly repressed in both strains by both glucose and lactose.
Several of the genes that were induced by sophorose displayed higher levels
of expression, in both glucose and in lactose, than during carbon deprivation.
Among these, most are expressed more highly in lactose than in glucose,
indicating that they are to some degree preferentially expressed in lactose.
Others, such as agl3, abf2, and cel3d, were equivalently
expressed in the two carbon sources indicating that their expression may be
more sensitive to the availability of a carbon source than to its identity.
Nevertheless, the relative induction of these genes is higher in RL-P37 than
it is in QM6a. The -glucosidases cel1b and cel3e are
also sophorose-induced and more highly expressed in RL-P37 than in QM6a, yet
appear to be preferentially induced by excess glucose in this experimental
regimen.
A number of the genes examined were expressed more highly after 1 h of carbon deprivation than they were during the period where excess glucose was present in the medium. Interestingly, more genes displayed this glucose repression (or induction by carbon deprivation) in RL-P37 than in QM6a.
In this experiment, the kinetics of lactose induction appeared to be faster in RL-P37 than in QM6a. Moreover, several genes (e.g. man1, glr1, xyn3, egl1/cel7b, swo1, cbh2/cel6a, egl5/cel45a, cel12a, and xyn4) that showed little or no sophorose induction in QM6a showed clear induction in this strain at the 48-h time point in the lactose-fed fermenters. Because the experiments described above queried only the magnitude of induction by sophorose and lactose, it was unclear whether the absolute expression levels of these genes differed upon induction with the different sugars. To address this question, an experiment was performed in which mRNA from the sophorose-induced cultures was directly compared with mRNA from the fermenter cultures that had been induced for 48 h with lactose. The results of this experiment are shown in Fig. 3B. Remarkably, the final expression level of each of the genes is virtually identical (within log ratio 0.2) in response to the two inducing sugars for both QM6a and RL-P37. Thus despite differences in the experimental regimen and the kinetics of induction, the net expression level of each the genes examined was the same upon induction with either sophorose or with lactose.
Regulation of Genes Involved in Protein Processing and Secretion—Many of the genes encoding biomass-degrading enzymes are induced by lactose and by sophorose more than 10-fold, and some are induced as much as 100-fold. Moreover, the products encoded by these genes accumulate to more than 50% of the total cellular protein (18). The processing and secretion of these gene products are expected to pose a substantial challenge to the secretory pathway of the cell. To ask whether induction of these enzymes produces coordinate changes in the secretory apparatus, a list of 109 T. reesei ESTs and previously identified genes containing open reading frames with significant sequence similarity to known gene products involved in protein glycosylation, folding, secretion, the unfolded protein response, and endoplasmic reticulum-associated protein degradation was compiled. The regulation of these ESTs was examined during cellulase induction.
Fig. 4 shows how this list compares to the sets of genes induced by lactose and by sophorose in QM6a and in RL-P37. Neither of the strains exhibited extensive induction of the identified ESTs. In QM6a, the only ESTs in this set that were induced with log ratio greater than 0.3 by lactose or by sophorose were a gene potentially involved in N-linked glycosylation (annotated as an N-glycosyltransferase), a possible peptidylprolyl cis-trans-isomerase, and an EST (tric039xp19) that contains a region of similarity to Saccharomyces cerevisiae Sec7p. These genes were selectively induced by lactose or by sophorose, respectively, and their log ratios were only slightly greater than 0.3. In RL-P37, a single previously identified gene, encoding protein-disulfide isomerase (48), was induced substantially by both lactose and by sophorose. Twelve genes that participate in a variety of processes including folding, glycosylation, and vesicle transport were induced preferentially by sophorose in RL-P37. An additional gene, encoding Ire1, a transmembrane kinase/nuclease involved in signaling the accumulation of unfolded proteins in the endoplasmic reticulum (49), was induced reproducibly, but with log ratio less than 0.3 by both lactose and by sophorose. The complete data set corresponding to Fig. 4 is available in the Supplemental Material.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
1151 contigs from T. reesei. We did not attempt to
assemble the current sequencing set with the set described by this group
because they derive from different strains. Often genes of related function display commonalities in their regulation. Previous studies (19, 20, 32) have shown that the known cellulase components and some hemicellulases are coordinately induced by certain substrates. The large magnitude in induction of genes encoding cellulase components and the toll on cellular resources used in synthesizing them suggests that the tendency toward tight coordinate regulation is likely to hold true for other genes whose products are involved in biomass degradation or in facilitating their synthesis. We used oligonucleotide-based microarrays to determine the comprehensive expression patterns of the EST set under conditions known to influence cellulase production in T. reesei.
Comparisons of the sets of genes that were induced by lactose to those that
were induced by sophorose in these experiments should be interpreted in light
of the experimental regimens used. To measure sophorose induction, expression
levels observed in sophorose-grown cultures were compared with expression
levels when glycerol was used as the sole carbon source. In contrast,
expression levels during the limited lactose feed in a fermenter were
normalized to expression levels during carbon deprivation. It is likely that
these considerations are particularly relevant to the regulation of the
-glucosidases, which as a class showed profound differences in their
regulation by sophorose and by lactose (discussed further below). However,
when a direct comparison was made between the expression levels observed after
sophorose induction to those seen after lactose induction
(Fig. 3B), each of the
genes encoding the putative biomass-degrading enzymes listed in
Table I were expressed
virtually equivalently in both RL-P37 and QM6a. Thus, both induction schemes
appear to ultimately result in the same net levels of expression and
stoichiometry (at the mRNA level) of these enzymes.
Among the cDNAs that were newly identified in this study were three, designated cel74a, cel61b, and cel5b, that appear to encode previously unknown endoglucanases. With the exception of cel5b, all of the endoglucanases were similarly regulated by cellulose, sophorose, and lactose. These results are in agreement with results published for the previously known endoglucanases (20). The endoglucanases were also more dramatically induced and more highly expressed in the strain RL-P37 than they were in QM6a.
Two additional putative glycolytic enzymes, cip1 and cip2, were identified that do not fit into any of the currently defined classes of GHs. The regulation of cip1 among strains with varying cellulase-producing capabilities and across a variety of conditions is indistinguishable from the endoglucanases and particularly the cellobiohydrolase cbh1/cel7a (Figs. 2 and 3).2 Similarly, cip2 has a pattern of expression in common with these genes, particularly in RL-P37. The coregulation of these genes with canonically regulated cellulase components and the fact that they contain distinct cellulose-binding modules furthers the notion that cip1 and cip2 encode previously unrecognized activities with potential roles in biomass degradation.
Five new -glucosidase genes were identified in the EST set bringing
the total number of described T. reesei -glucosidases to 7.
Although it is known that -glucosidase activity is required for complete
digestion of carbohydrates to monosaccharides, the precise functions and
specificities of the enzymes encoded by these genes is not known. Deciphering
the roles of the -glucosidases is further complicated by the fact that
they have been shown to be capable of transglycosylation, thereby converting
sugars that are poor inducers of cellulase gene transcription into more potent
inducers (reviewed in Ref.
52). Barnett et al.
(15) have shown and Takashima
et al. (16) have
suggested roles for bgl1/cel3a and
bgl2/cel1a, respectively, in cellulose saccharification,
although BGLII/Cel1a was later shown to be an intracellular rather than a
secreted protein (53). In
accordance with this function, bgl1/cel3a and
bgl2/cel1a expression were well correlated with the
induction, by both sophorose and lactose, of other genes central to cellulose
degradation. Interestingly, however, these genes differed in their response to
glucose deprivation. The expression of bgl1/cel3a was higher
when glucose was present in excess than when glucose was depleted from the
medium. bgl2/cel1a expression, in contrast, was dramatically
lower in excess glucose.
cel1b, cel3d, and cel3e appeared to be more highly
induced by glucose than by lactose. One interpretation of these results is
that these -glucosidases are less directly relevant to cellulose
degradation than other enzymes of this class. However, it is interesting to
note that these genes are sophorose-inducible and are present at a higher
level in RL-P37 than in QM6a suggesting that their regulation may be tied in
some way to cellulase production.
Several of the hemicellulase-encoding genes (axe2, xyn1, abf1, abf2, agl1, agl2, and agl3) appear to be differentially induced by lactose and by sophorose in either QM6a or in RL-P37 or both. Most notably, abf1, abf2, agl1, and agl2 were substantially more induced in RL-P37 during growth on lactose than they were by growth in sophorose. Previously published results (32) reported that expression of a subset of the currently examined hemicellulases, with the exception of bxl1, were undetectable in cultures where lactose is used as a carbon source. We attribute the discrepancies with the results reported in this study to several factors. The T. reesei strain used in the previous study differed from the ones used in this study, although like RL-P37 it is derived from QM6a. In addition, in contrast to the experiment described in this paper where lactose was fed via a fermenter to minimize accumulation, the previous study used flask cultures in which excess lactose was present. It is also likely that the sensitivity of detection of mRNAs using microarrays by methods used in this study exceed the sensitivity of Northern blots used in previous analyses.2
It has been speculated that in nature low levels of constitutively
expressed cellulases, perhaps in conjunction with a -glucosidase, must
exist to initiate cellulose digestion and generate smaller inducing molecules
(54–56).
These molecules, such as cellobiose and potentially sophorose (generated by
trans-glycosylation), would then mediate the induction of the full complement
of cellulase-encoding genes. Several of the genes encoding biomass-degrading
enzymes displayed regulatory patterns consistent with the possibility that
they play a role in primary induction of cellulase expression. cel5b
mRNA was moderately abundant and was constitutively expressed in glycerol,
glucose, sophorose, and lactose. However, its expression was induced somewhat
by cellulose. In addition, in contrast to the other endoglucanases it was only
slightly more abundant in RL-P37 than in QM6a. It is therefore probably not a
key component of the set of enzymes selected in RL-P37 to digest bulk
cellulose. These characteristics, along with the fact that Cel5b may be
membrane-anchored via a glycosylphosphatidylinositol anchor, make it an
interesting candidate for this role. By analogy, Axe2, which is also predicted
to contain a glycosylphosphatidylinositol anchor, may be involved in primary
induction of some hemicellulases. In addition, several other genes were
moderately to highly increased in abundance when cultures were deprived of a
carbon source (Fig.
3B, genes that appear "repressed by
glucose"). Margolles-Clark et al.
(32) have described similar
results for a subset of these genes. Elevated expression of these genes in
response to starvation may indicate that they play a role in
"foraging" for complex carbon sources in nature or in induction of
other genes encoding biomass-degrading enzymes. In this hypothetical role, it
is unclear why the complement of genes that is induced by starvation differs
between RL-P37 and QM6a. One possibility is that because the screens used in
developing RL-P37 included a requirement for cellulase production in the
presence of 2-deoxyglucose, glucose-mediated regulation of many of the genes
has been altered. Alternatively, regulation in response to glucose may be, in
some way, mechanistically tied to induction and cellulase overexpression.
A number of studies have shown that endoplasmic reticulum chaperones and foldases are up-regulated by endoplasmic reticulum stress or secretion of foreign proteins in filamentous fungi (reviewed in Ref. 57). A list of T. reesei ESTs was compiled that contains ORFs with significant sequence similarities to known gene products involved in various aspects of protein secretion. Surprisingly few genes from this list were induced by sophorose or by lactose in either QM6a or in RL-P37 (Fig. 4). Saloheimo et al. (48) showed that in T. reesei the protein-disulfide isomerase gene pdi1 is minimally induced by sophorose and induced by 8–10-fold, compared with growth on glucose, during growth on cellulose-containing medium, or after carbon depletion. In our experiments, pdi1 was not induced in QM6a but was the sole gene among secretion-related genes that was induced (log ratio >0.3) by growth in both sophorose and lactose in RL-P37. It is likely that the relative induction level of pdi1 is a function of the quantity of total secreted protein being produced and that the slight differences observed between the two studies reflect strain differences and differences in the growth regimens employed.
Sophorose is the most potent inducer of T. reesei cellulases known. The large number of genes whose expression is altered by sophorose indicates that its direct and indirect effects on the cell are broad. Moreover, the effect of sophorose on expression of numerous genes appears to differ between the two strains examined. RL-P37 was derived from QM6a by mutation and screening for increased cellulase production in the presence of glycerol and subsequently in 2-deoxyglucose (47). Under the fermentation conditions used in this study, cellulase production in RL-P37 was 10-fold higher than in QM6a. Therefore, one might expect that many of the genes whose expression both differs between these two strains, and is modulated by sophorose, encode gene products involved in cellulase production.
mRNAs corresponding to a large number of ESTs were expressed less abundantly when cells were grown in sophorose than when they were grown in glycerol alone. It is possible that part of the role of sophorose in stimulating production of biomass-degrading enzymes is to directly or indirectly ameliorate the effects of a negative regulator. Such an effect could occur at many levels including transcription, sugar transport, signal transduction, or secretion. In Fig. 2D, we examined the set of ESTs whose expression was decreased in the presence of sophorose. Sets d and e in Fig. 2D (see Supplemental Material for the data sets) contain ESTs that are repressed in RL-P37 cultures grown with sophorose and that are expressed at a lower level in RL-P37 than in QM6a. The strong correlation between reduced expression of these genes in response to sophorose and enhanced production of biomass-degrading enzymes may indicate that their expression is detrimental to cellulase production.
Analogously, a set of genes was identified that was both induced by sophorose and increased in expression in RL-P37 relative to QM6a (Fig. 2C, sets d–f, see Supplemental Material). Among this set are the previously identified glycosyl hydrolase-encoding genes known to be important in cellulase degradation. The co-selection in RL-P37 of some of the newly discovered genes in Table I with the known genes encoding cellulase components strengthens the notion that they may have roles in biomass degradation. It is remarkable that among this group are genes that do not have a known function in cellulose degradation but rather are presumed to act on other substrates (e.g. hemicellulose). These genes would not be expected to have a direct benefit in the selection for increased cellulose production that was applied in deriving RL-P37, yet they appear to have been co-selected for enhanced sophorose induction and overexpression. Thus, it seems that a mutation(s) that improved cellulase production concurrently improved the inducible expression of these ancillary genes. These results suggest that this set of genes may have significant regulatory points of convergence across the spectrum of cellular processes involved in carbon sensing, signal transduction, and transcriptional regulation. By the same reasoning, at least some of the other products encoded by genes in Fig. 2C, sets d–f, are likely to be important either in biomass degradation or in the synthesis of enzymes required for this process. Because of the limited sequence information available, we could not definitively annotate many of these ESTs. However, it is interesting to note that among the ESTs in this set are ones containing ORFs with similarity to several sugar transporters, constituents of the secretory apparatus, and to chitinase. These findings and further characterization of the genes will likely have significant implications for the design of industrial processes for commercial production of biomass-degrading enzymes.
|
FOOTNOTES |
|---|
* This work was supported in part by a subcontract from the Office of Biomass
Program of the Department of Energy Office of Energy Efficiency and Renewable
Energy. The costs of publication of this article were defrayed in part by the
payment of page charges. This article must therefore be hereby marked
"advertisement" in accordance with 18 U.S.C. Section 1734
solely to indicate this fact. 
The on-line version of this article (available at
http://www.jbc.org)
contains sequence alignments for the predicted gene products and data sets for
Figs. 2 and
4.
To whom correspondence should be addressed: Genencor International, 925 Page
Mill Rd., Palo Alto, CA 94304. Tel.: 650-846-7635; Fax: 650-621-7817; E-mail:
Pforeman{at}genencor.com.
1 The abbreviations used are: EST, expressed sequence tag; GH, glycoside
hydrolase; ORF, open reading frame.
2 P. Foreman, unpublished observations.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
