The introduction of the fungal d-galacturonate pathway enables the consumption of d-galacturonic acid by Saccharomyces cerevisiae

Background Pectin-rich wastes, such as citrus pulp and sugar beet pulp, are produced in considerable amounts by the juice and sugar industry and could be used as raw materials for biorefineries. One possible process in such biorefineries is the hydrolysis of these wastes and the subsequent production of ethanol. However, the ethanol-producing organism of choice, Saccharomyces cerevisiae, is not able to catabolize d-galacturonic acid, which represents a considerable amount of the sugars in the hydrolysate, namely, 18 % (w/w) from citrus pulp and 16 % (w/w) sugar beet pulp. Results In the current work, we describe the construction of a strain of S. cerevisiae in which the five genes of the fungal reductive pathway for d-galacturonic acid catabolism were integrated into the yeast chromosomes: gaaA, gaaC and gaaD from Aspergillus niger and lgd1 from Trichoderma reesei, and the recently described d-galacturonic acid transporter protein, gat1, from Neurospora crassa. This strain metabolized d-galacturonic acid in a medium containing d-fructose as co-substrate. Conclusion This work is the first demonstration of the expression of a functional heterologous pathway for d-galacturonic acid catabolism in Saccharomyces cerevisiae. It is a preliminary step for engineering a yeast strain for the fermentation of pectin-rich substrates to ethanol. Electronic supplementary material The online version of this article (doi:10.1186/s12934-016-0544-1) contains supplementary material, which is available to authorized users.


Background
Citrus pulp and sugar beet pulp are pectin-rich wastes that are produced in considerable amounts in various countries. Citrus pulp results from the production of orange juice concentrate. Its production is concentrated mainly in Brazil and the USA, which have a share of 54 and 26 % of the global market, respectively [1]. In the 2014/2015 harvest, 1.4 million tons of orange juice concentrate (65° Brix) were produced in these two countries [1], and this amount would result in the production of about 1.9 million metric tons of citrus pulp (dry matter). Likewise, large amounts of sugar beet pulp are generated during the extraction of sugar from sugar beets in temperate countries, especially Russia, the USA and several European countries [2]. In the 2013 harvest, a total of 246 million metric tons of sugar beet were produced worldwide [2], resulting in 12.3 million metric tons of beet pulp (dry matter).
Citrus pulp and sugar beet pulp are usually sold for incorporation in cattle feed, but the costs of drying these wastes make this application barely profitable [3]. On the other hand, these wastes are potentially important sources of carbohydrates that can be used as raw materials in biorefineries for the production of bio-based chemicals and biofuels. In fact, there is a rising demand

Open Access
Microbial Cell Factories *Correspondence: ext-alessandra.biz@vtt.fi; peter.richard@vtt.fi 3 VTT Technical Research Centre of Finland Ltd, P.O. Box 1000, 02044 VTT Espoo, Finland Full list of author information is available at the end of the article for ethanol, especially in Brazil and the USA, and biorefineries using citrus wastes or sugar beet pulp could meet part of this demand [4,5]. However, hydrolysates obtained from pectin-rich wastes have high contents of d-galacturonic acid, which represents 18 % (w/w) of citrus waste hydrolysates [6] and 16 % (w/w) of sugar beet pulp hydrolysates [7]. Unfortunately, this uronic acid is not catabolized by Saccharomyces cerevisiae, which is the microorganism of choice for ethanol production.
Conversely, many microorganisms that are able to use d-galacturonic acid do not produce ethanol in appreciable amounts. Escherichia coli, for example, has the isomerase pathway for d-galacturonic acid catabolism but, due to low ethanol and inhibitor tolerance, it is not the preferred organism for ethanol production. Alternatively, ethanol production with S. cerevisiae is already a highly productive and robust process. This yeast is preferred for commercial scale ethanol production for several reasons, including the resistance to contaminants, bacteriophages, inhibitors and low pH [8]. It also tolerates higher osmotic pressures, enabling the use of a concentrated culture medium, and greater concentrations of ethanol. Considering this, the expression of a heterologous pathway for catabolism of d-galacturonic acid into S. cerevisiae, instead of engineering bacteria for ethanol production, is the preferred path for industrial ethanol production from hydrolysates containing d-galacturonic acid.
Different approaches have been used to engineer S. cerevisiae for this purpose. Huijes et al. [9], for instance, introduced by integration into the chromosomes the five genes of the bacterial isomerase pathway (uxaC, uxaB, uxaA, kdgK and kdgA), which converts d-galacturonic acid to pyruvate and glyceraldehyde-3-phosphate. Although Huijes et al. [9] demonstrated by qPCR that all genes were transcribed, only two enzymes of the pathway showed detectable activity after integration in S. cerevisiae. In another approach, Souffriau [10] expressed the fungal pathway for d-galacturonic acid catabolism in S. cerevisiae by cloning the four genes of the pathway (gar1, lgd1, lga1 and gld1, from Trichoderma reesei) in two plasmids with the gap-DH-ADH1 bidirectional promoter. In this work the activity of all the heterologous enzymes were detected in the cell lysate. Despite this, the recombinant yeast strain did not consume d-galacturonic acid.
In the present work, we extended the work of Soufriau [10] by integrating to the yeast chromosomes the four genes of the reductive fungal d-galacturonic catabolic pathway; in addition, we also integrated of the recently discovered d-galacturonic acid transporter from Neurospora crassa, in order to engineer a yeast that is able to use d-galacturonic acid as a carbon source. This work represents an important step in the construction of a S. cerevisiae strain able to produce ethanol from d-galacturonic acid.

Pathway assembly and gene expression
The fungal pathway for catabolism of d-galacturonic acid was expressed in Saccharomyces cerevisiae strain CEN. PK111-61A, with the genes being selected from three different filamentous fungi, Aspergillus niger, Trichoderma reesei and Neurospora crassa.
As a first step, four genes encoding the catabolic d-galacturonate pathway in A. niger (gaaA, gaaB, gaaC and gaaD) were integrated into the yeast chromosomes, with each one being expressed under strong and constitutive yeast promoters (PGK1 or TPI1). The corresponding enzyme activities of gaaA, gaaC and gaaD were detected in the cell lysate. The corresponding activity of gaaB, namely l-galactonate dehydratase, was not detectable, even when a codon-optimized gaaB ORF was used. This result was unexpected since gaaB activity was demonstrated when expressed in a multi copy expression vector [11]. Considering this, lgd1, an l-galactonate dehydratase encoding gene from T. reesei, was used instead. The resulting strain (H4531), having the complete pathway (gaaA, gaaC, gaaD and lgd1) integrated, was then retested for the activity of all four enzymes. The enzymatic activities obtained from H4531 are listed in Table 1.
During the course of the study, a d-galacturonic acid transporter (coded by gat1) was described in N. crassa [12]. Although a previous study reported that native S. cerevisiae is able to import d-galacturonic acid when grown at acidic pH values [13], the introduction of a transporter might improve the intake, especially at higher pH values. For this reason, gat1-gfp fusion protein gene was integrated into H4531, resulting in the strain H4535. The location of GAT1 was confirmed by fluorescence microscopy, with the green fluorescence of GFP being observed in the plasma membrane.

d-galacturonic acid consumption
Both engineered strains, H4531 and H4535, were cultivated for 5 days under aerobic conditions in YP medium, supplemented with 12 g L −1 of d-galacturonic acid. The S. cerevisiae strain CEN.PK113-1A, which does not have auxotrophies, was used as a control. Even after adaptive laboratory evolution, the recombinant strains, as well as the control strain grew poorly and did not consume d-galacturonic acid in this condition. A second fermentation was therefore done, with the difference that the medium was supplemented with 80 g L −1 of d-fructose (Fig. 1). This time, 20 % of the d-galacturonic acid was consumed by the yeast strain H4535, which expressed the four genes of the reductive pathway and the transporter for d-galacturonic acid. Notably, most of this consumption was observed in the first 24 h, and corresponds to the complete utilization of the d-fructose by the yeast. Consumption of d-galacturonic acid was negligible by the control strain and by the recombinant strain expressing the reductive pathway, but lacking the transporter. We could also exclude that d-galacturonic acid was simply converted to l-galactonate or other galacturonate pathway intermediates since these metabolites were not detected in neither in the culture medium nor in the cell extract. Besides the clear difference in the d-galacturonic acid consumption, every other parameter measured, such as ethanol, biomass, pH and d-fructose, did not differ significantly among the strains.
To further analyse the role of d-fructose as a co-substrate that enables d-galacturonic acid consumption, an additional fermentation was performed. Here d-fructose was added in two batches of 40 g L −1 , being the second at 72 h (Fig. 2). With this experiment was possible to observe that an additional galacturonic acid consumption followed the addition of fructose in 72 h.
Finally, to prove that the galacturonic acid was definitely being catabolized by the engineered yeast, and not only converted to e.g. l-galactonate, the fate of this sugar was monitored by NMR, using the uniformly labeled d-[UL-13 C 6 ] galacturonate as substrate. For that, a fermentation was carried out with 40 g L −1 d-fructose and 4 g L −1 of d-galacturonic acid, with the labeled substrate corresponding to half of the total amount of d-galacturonic acid. Again, a control strain without the pathway was fermented in the same conditions. Figure 3a, b shows the NMR spectrum for the supernatant of this fermentation at 24 h, with a corresponding to the spectrum of the supernatant from the control strain and B from the engineered strain with the fungal pathway and the transporter. In Fig. 3b, there is a clear signal of uniformly labeled glycerol showing the 13 C-13 C scalar coupling fine  and H4535 contains the genes gaaA, lgd1, gaaC, gaaD and gat1. The fermentation was carried out with d-fructose as co-substrate, with a second load of d-fructose being added at 72 h of fermentation. d-galacturonic acid consumption was monitored over time. Error bars represent the standard error of the mean structure, meaning that the carbon backbone was originally from the labeled d-galacturonic acid. In Fig. 3a this fully labeled glycerol is not detected in the spectrum. Glycerol is a product of the d-galacturonic acid pathway and of sugar catabolism in yeast, and was detected by HPLC for both control and engineered strains in the previous experiments. The detection of uniformly 13 C-labeled glycerol already confirms that d-galacturonic acid is metabolized by the engineered strain, and not by the control. However, to further assure that the d-galacturonic acid is being metabolized, the yeast cells were collected after 96 h of fermentation and hydrolyzed. The spectrum of the resulting amino acids was analyzed for both the control strain (Fig. 3c) and the engineered strain (Fig. 3d). Also here we observed the 13 C-13 C scalar coupling fine structure in the sample of the engineered strain, proving that the labeled carbon is coming from the d-galacturonic acid. The alfa carbon of several amino are 13 C-labeled and have 13 C-labeled neighbors in the engineered strain, but not in the control strain, which is a proof that d-galacturonic acid is catabolized and ends up in the biomass of the engineered strain.

Discussion
In this work, we demonstrate for the first time a functional catabolic pathway for d-galacturonate consumption by S. cerevisiae. For this purpose, the fungal d-galacturonic acid pathway and the d-galacturonic acid transporter were integrated into the yeast chromosomes and the corresponding enzyme activities were shown in the cell lysate, and with the d-galacturonic acid transporter being located in the cell membrane. The fermentation trials with d-fructose as co-substrate showed that the engineered strain was, furthermore, able to catabolize d-galacturonic acid. The tracing of the 13 C from uniformly labeled d-galacturonic acid showed that these carbons end up in products of the fermentation, particularly glycerol, and in the amino acids of the biomass. The fungal pathway for d-galacturonic acid catabolism was recently discovered by Hilditch et al. [14] and confirmed by Martens-Uznova et al. [15]. It consists of four enzymes, with two reduction steps (Fig. 4). The first enzyme of the pathway, d-galacturonate reductase (EC 1.1.1.365), coded by gaaA in A. niger, converts d-galacturonate into l-galactonate, using NADPH or NADH as the electron donor. l-galactonate dehydratase (EC 4.2.1.146), coded by gaaB in A. niger and by lgd1 in T. reesei, converts l-galactonate into 2-keto-3-deoxygalactonate. The third enzyme, 2-keto-3-deoxygalactonate aldolase (EC 4.1.2.B7), coded by gaaC in A. niger, cuts 2-keto-3-deoxygalactonate between carbons 3 and 4, producing pyruvate and l-glyceraldehyde. The pyruvate can enter the citric acid cycle, while the l-glyceraldehyde is converted to glycerol by glyceraldehyde reductase (EC 1.1.1.372, encoded by gaaD in A. niger) using NADPH as the electron donor.
Recently, Benz [12] described the d-galacturonic acid transporter, which was suggested to act as an H + /dgalacturonic acid symporter. The corresponding gene was discovered in Neurospora crassa, and to further confirm its functionality, the transporter was also co-expressed in yeast with the first enzyme of the fungal pathway, galacturonate reductase (gaaA).
Souffriau [10], similarly to what we have done, also expressed the genes of the fungal pathway from Trichoderma reesei (gar1, lgd1, lga1 and gld1) in S. cerevisiae. Differently from our approach, the four genes were expressed from two plasmids with the gap-DH-ADH1 bidirectional promoter. Despite the fact that the activity of all the enzymes was shown in the cell lysates, the recombinant yeast strain did not grow on d-galacturonic acid. Further adaptive laboratory evolution, in a medium containing both glycerol and d-galacturonic acid, did not lead to mutants that were able to use d-galacturonic acid.
Similarly to Souffriau [10], we chose the reductive pathway for introduction into S. cerevisiae, since previous work showed that the individual enzymes could be expressed in this yeast [8,[15][16][17]. However, our approach differs from that of Souffriau [10] in four ways: First, we integrated the genes into the yeast genome, instead of using plasmid vectors; second, we used genes from both T. reesei and A. niger; third, we also integrated a d-galacturonic acid transporter [12]; and, fourth, we used d-fructose, instead of glycerol, as a co-substrate for growth of the recombinant yeast. Interestingly, both Souffriau [10] and we detected the activities of the d-galacturonic acid pathway enzymes in the cell lysate, but our use of hexose sugar as a co-substrate, rather than the glycerol used by Souffriau [10] and also the use of a d-galacturonic acid transporter may be the reasons why we were able to demonstrate d-galacturonic consumption. Additionally, the use of the A. niger d-galacturonate reductase GAAA, which accepts both NADH and NADPH as a cofactor, instead of the T. reesei reductase, which strictly uses only NADPH, could also have enabled the observed d-galacturonic acid consumption.
Although the engineered strain was able to catabolize d-galacturonic acid, the strain is still not optimized for industrial processes using pectin-rich waste hydrolysates. Due to the fact that the fungal pathway requires reducing power that is also needed for ethanol production, further engineering of the strain and optimization of the cultivation conditions are still needed to address the redox balance of this process.

Conclusion
In this work, we describe for the first time a strategy for the introduction of a functional pathway that resulted in d-galacturonic acid catabolism in Saccharomyces cerevisiae. This strategy was based on the integration into the yeast chromosomes of not only the four enzymes of the fungal pathway of d-galacturonic acid catabolism, but also the recently described d-galacturonic acid transporter. This represents the first step in the construction of a strain of S. cerevisiae that would be able to produce ethanol from d-galacturonic acid. Such a strain would find application in a citrus/sugar beet waste biorefinery, in which pectin-rich wastes would be hydrolyzed and then fermented to produce ethanol.

Strains
The S. cerevisiae strain CEN.PK111-61A was used as a parental strain for the engineered strains. The prototrophic CEN.PK113-1A was used as a control in the fermentations. All the plasmids were produced in E. coli TOP10 cells. Table 2 shows the original and engineered S. cerevisiae strains used in this work.

Plasmid construction and gene integrations
Plasmids are listed in Table 3. Primers used for plasmid construction are listed in Table 4. All the PCR reactions were performed using Phusion High Fidelity (Finnzymes, Finland) and Phusion High Fidelity buffer or GC Buffer (Finnzymes, Finland). For bacterial and yeast colony PCR, DyNAzyme (Finnzymes, Finland) was used instead. For yeast colony PCR, cell disruption was carried out using Zymolyase 100 T (Seikagaku Biobusiness, Japan). Ligations were done using T4 DNA ligase (Promega,   USA). Recombination in yeast for plasmid assembly or genome integration was done using Lab Transformation kit (Molecular Research Reagents Inc., USA), following the protocol of Gietz et al. [21]. Recombination in vitro was done using Gibson Assembly ® Master Mix (New England Biolabs, USA).

gaaC integration
The gaaC integration cassette was designed to have the gaaC gene under the PGK1 promoter (fragment 3), the flanking regions of S. cerevisiae HIS3 gene (fragments 1 and 4) used as homologous regions for genome integration, and the S. pombe HIS5 gene (yeast marker, fragment 2) assembled circularly in the pRS426 vector (fragment 5), which contains a URA3 marker. gaaC was directly amplified from A. niger ATCC 1015 genome. It was inserted into the PGK1 promoter region of B1181 plasmid using Gibson Assembly. The resulting plasmid, B5430, contains the PGK1/gaaC component (fragment 3) of the gaaC integration cassette and was digested with BglII to obtain fragment 3. Fragment 1, which corresponds to the-1000 bp downstream region of S. cerevisiae his3, and fragment 4, which corresponds to the upstream +1000 bp region of his3, were amplified from the S. cerevisiae (H3488) genome. HIS5 (fragment 2) was amplified from the S. pombae genome. The pRS426 vector was linearized by digestion with NotI (integration vector). The five fragments from the gaaC integration plasmid were assembled by recombination in yeast. The URA3 marker was used for the selection of the assembled plasmid and the transformants were selected in SCD-URA plates. The assembled plasmid was introduced to E. coli by electroporation for multiplication and digested with NotI for the isolation of the cassette (fragments 1 to 4). The integration of the gaaC cassette into the parental strain CEN.PK111-61A was done by recombination. The HIS5 marker was used for selection of the gaaC-integrated yeast and transformants were selected in SCD-HIS and confirmed by colony PCR. Activity was confirmed using the cell lysate. A colony expressing gaaC was stored as H4362.

gaaD integration
The gaaD was amplified from cDNA and restricted with BglII and ligated to the B2159 plasmid in the TPI promoter region. gaaD under TPI1 promoter was amplified and ligated to the vector pRS405 (containing LEU2 yeast marker) to form B5555 integration vector. The gaaD integration cassette with the LEU2 marker was obtained by the linearization of B5555 using AflII. The gaaD cassette was integrated into the gaaC-expressing yeast strain H4362, in the leu2-3112 locus. Transformants were selected in SCD-LEU and confirmed by colony PCR. The activity of the gaaD gene was confirmed using the cell lysate. A colony expressing gaaC and gaaD was stored as H4410.

gaaA integration
The gaaA gene was custom-synthetized as a codonoptimized ORF (Genescript). Genescript gaaA plasmid was digested with EcoRI and BamHI and ligated to B2159 plasmid in the TPI1 promoter region. The integration plasmid B5697 was obtained by ligating the TPI1/gaaA fragment to the vector pRS406 (B704), which contains the URA3 yeast marker. The gaaA integration cassette was obtained by the linearization of B5697 using BsmI. The cassette was then integrated into the ura3-52 locus of the yeast strain H4410 (gaaC + gaaD). Transformants were selected in SCD-URA medium and confirmed by colony PCR. Activity was confirmed using the cell lysate. A colony expressing gaaC, gaaD and gaaA was stored as H4425.

lgd1 integration
The integration plasmid B3013 carrying the lgd1 gene ready for integration in yeast was available from previous work [22]. This integration cassette consists of two flanking regions for CAN1, the lgd1 gene under the PGK1 promoter, and the KanMX yeast marker (resistance to G418) between two loxP sites. The plasmid was digested with SacI and KpnI and the integration cassette was transformed to the yeast strain H4425 (gaaA + gaaC + gaaD). Before plating, the yeast cells were incubated in 2 mL of SCD for 1 h. Transformants were selected in YPD + G418 and confirmed by assaying lgd1 activity of the cell lysate. The KanMX marker was then removed by transforming the selected yeast colony with plasmid pSH66, expressing cre recombinase, which removes KanMX yeast marker between the loxP sites [23]. Before plating in YPD + nourseothricin, the yeast cells were incubated in 2 mL of SCD for 1 h. After 4 days, the resulting colonies were re-plated in YPD + nourseothricin. Biomass from this plate was inoculated into 50 mL of YP + 2 % galactose and incubated overnight. The resulting culture was plated in YPD. Some isolated colonies were replated in G418 plates and SCD-URA/-LEU/-HIS medium. After 2 days, none of the colonies grew on G418, confirming that the KanMX marker had been removed. A colony expressing gaaC, gaaD, gaaA and lgd1 was stored as H4531.

gat1-gfp integration
The gat1-gfp gene was custom-synthetized based on the work of Benz [12] as a codon-optimized ORF (Genescript) The gat1-gfp fragment was digested from the Genescript plasmid and ligated to B2159 plasmid, in the TPI1 promoter region. For the integration cassette, the TPI1/gat1-gfp fragment was digested with XmaI. The vector B3531, which contains HO integration region and KanMX yeast marker, was digested with Cfr9I and ligated to the TPI1/gat1-gfp fragment. The gat1-gfp integration cassette was obtained by the digestion of the integration plasmid with XhoI and XbaI and transformed to the H4531 strain. The yeast cells were incubated in 2 mL of YPD for 2 h and then plated in YPD + G418 plates. To confirm the integration of the gat1-gfp gene, cells were incubated in SCD medium for 24 h and analysed using a fluorescence microscope. A colony showing green fluorescence (from the gfp reporter gene) was stored as H4535.

Enzymatic assays
Cell extracts, obtained as follows, were used in all the enzymatic assays. Yeast cells were grown in YPD medium and collected by centrifugation, washed with water and re-suspended in phosphate buffer 50 mM, pH 7 with addition of protease inhibitor Complete EDTA Free (Roche, Switzerland). Cells were disrupted with 0.4 mm diameter glass beads using a bead beater FastPrep (MP Biomedicals, USA) and solid residues were removed by centrifugation. Protein content was assayed by the Bradford method [24]. d-Galacturonate reductase (GAAA) and l-glyceraldehyde reductase (GAAD) activities were assayed in the forward direction by following the decrease in absorbance at 340 nm caused by the oxidation of NADPH. The reaction medium for d-galacturonate reductase contained 10 mM d-galacturonic acid (Sigma-Aldrich, Germany) and 0.2 mM NADPH (Sigma-Aldrich, Germany). The reaction medium for l-glyceraldehyde reductase contained 5 mM l-glyceraldehyde (Sigma-Aldrich, Germany) and 0.2 mM NADPH (Sigma-Aldrich, Germany). The reactions of the d-galacturonate reductase and the l-glyceraldehyde reductase were followed for 5 min.

Fermentations trials
In the adaptive laboratory evolution of the transformed yeast, the medium used was YP (10 g L −1 yeast extract, 20 g L −1 peptone) pH 4,5, supplemented with 12 g L −1 d-galacturonic acid. For assaying d-galacturonic acid consumption by transformed yeast in co-fermentation with d-fructose the medium used was YP pH 7 supplemented with 5 g L −1 d-galacturonic acid and with 80 g L −1 d-fructose. For evaluating the role of d-fructose in the d-galacturonic acid catabolism the medium used was initially YP pH 7 supplemented with 5 g L −1 d-galacturonic acid and with 40 g L −1 d-fructose. After 72 h, a concentrated d-fructose solution (40 % w/v) was added to result in the concentration of approximately 40 g L −1 d-fructose. In the 13 C tracing experiments, potassium d-[UL-13 C 6 ] galacturonate (Omicron Biochemicals, Inc., USA) was used for tracing the fate of d-galacturonic acid. The experiment was carried out with 2 g L −1 of the labeled d-galacturonic acid, 2 g L −1 of non-labeled d-galacturonic acid and 40 g L −1 of d-fructose in YP medium (neutral pH). Samples of the supernatant were taken 12 h time intervals and the biomass was collected in 96 h of fermentation. All the fermentations were carried out aerobically using 250 rpm and 30 °C. The CEN.PK 113-1A strain, which does not contain the pathway or transporter, and with similar genetic background, was used as a control in all the experiments.

Adaptive laboratory evolution
Adaptive laboratory evolution is a technique that aims for the selection of microorganisms more adapted to grow in a certain type of medium, for example, media containing toxic compounds or non-metabolizable sugars [26]. In this work, adaptive laboratory evolution was used for the selection of mutants more suited for growing with d-galacturonic acid as a carbon source. This study was carried out with the strains H4531, and H4535, both carrying the reductive fungal pathway, with the H4535 also containing the transporter gene. The cultivation was done in YP medium pH 4.5, supplemented with 12 g L −1 d-galacturonic acid. A dilution of 1:50 was made in fresh medium every 7 days. The cultivations continued for 8 weeks and were analysed by HPLC, as described below.

Chemical analysis
Samples of liquid fermentation broth were taken at 12-48 h or 7-days (adaptive laboratory evolution) intervals. The concentrations of d-galacturonic acid, d-fructose and ethanol were determined by HPLC using a Fast Acid Analysis Column (100 mm × 7.8 mm, BioRad Laboratories, CA, USA) linked to an Aminex HPX-87H organic acid analysis column (300 mm × 7.8 mm, Bio-Rad Laboratories) and 5 mM H 2 SO 4 as eluent with the flow rate of 0.5 ml min −1 . The column was maintained at 55 °C. Peaks were detected using a Waters 410 differential refractometer and a Waters 2487 dual wavelength UV (210 nm) detector.

NMR analysis
The NMR experiments were carried out either at 22 °C (culture supernatants) or at 40 °C (biomass hydrolysate) on a 600 MHz Bruker Avance III NMR spectrometer equipped with a QCI cryoprobe. The culture supernatant samples were prepared by mixing 540 µl of the supernatant with 60 µl of D 2 O (Aldrich, Germany). The HCl hydrolysis of the biomass as well as the 1 H, 13 C HSQC experiments of the aliphatic area with high 13 C resolution are described in detail by Jouhten and Maaheimo [27]. For 13 C decoupling during the acquisition, GARP4 was used.
Authors' contributions AB, JK, HM and PR conceived and designed the experiments. AB, MS-G, JK, HM and PR carried out the experimental work and analyzed the data. AB, JK, HM, DM and PR wrote and revised the paper. NK, DM and PR designed the fundamental concept and participated in the coordination of the study. All authors read and approved the final manuscript.