Production of ethanol from l-arabinose by Saccharomyces cerevisiae containing a fungal l-arabinose pathway


*Corresponding author. Tel.: +358 (9) 456 7190; Fax: +358 (9) 455 2103, E-mail address:


The fungal pathway for l-arabinose catabolism converts l-arabinose to d-xylulose 5-phosphate in five steps. The intermediates are, in this order: l-arabinitol, l-xylulose, xylitol and d-xylulose. Only some of the genes for the corresponding enzymes were known. We have recently identified the two missing genes for l-arabinitol 4-dehydrogenase and l-xylulose reductase and shown that overexpression of all the genes of the pathway in Saccharomyces cerevisiae enables growth on l-arabinose. Under anaerobic conditions ethanol is produced from l-arabinose, but at a very low rate. The reasons for the low rate of l-arabinose fermentation are discussed.


The two most widespread pentose sugars found in our biosphere are d-xylose and l-arabinose [1]. Pathways for pentose catabolism are relevant for micro-organisms living on decaying plant material, but also in biotechnology when cheap raw materials such as plant hydrolysates are to be fermented to ethanol. Common in the catabolism of these two pentose sugars in all micro-organisms is that the sugar is converted to d-xylulose 5-phosphate. However, the pathways to convert l-arabinose and d-xylose to d-xylulose 5-phosphate are distinctly different in bacteria and fungi. In bacteria, d-xylose is converted to d-xylulose by an isomerase and then phosphorylated by xylulokinase. In bacteria, l-arabinose is first converted to l-ribulose by an isomerase, then phosphorylated by ribulokinase and the l-ribulose 5-phosphate is then converted to d-xylulose 5-phosphate by an epimerase. In fungi, these pentose sugars go through oxidation and reduction reactions before they are phosphorylated by xylulokinase. d-Xylose is first reduced to xylitol by an reduced nicotinamide adenine dinucleotide phosphate (NADPH)-consuming reaction. Xylitol is then oxidised by an NAD+-consuming reaction to form d-xylulose. In fungi, l-arabinose goes through four redox reactions. Two oxidations are coupled to NAD+ consumption and two reductions to NADPH consumption (Figs. 1 and 3).

Figure 1.

The fungal and bacterial pathways for the utilisation of the pentose sugars d-xylose and l-arabinose.

Figure 3.

Redox cofactor requirement in l-arabinose catabolism. l-Arabinose conversion to equimolar amounts of CO2 and ethanol is redox neutral, i.e. anaerobic fermentation to ethanol should be possible. However, the conversion of l-arabinose to d-xylulose requires NADPH and NAD+ and produces NADH and NADP+. NADPH is mainly regenerated in the oxidative part of the pentose phosphate pathway, where the reduction of NADP+ is coupled to CO2 production. The abbreviations are: G6P, glucose 6-phosphate; F6P, fructose 6-phosphate; X5P, d-xylulose 5-phosphate; GAP, d-glyceraldehyde 3-phosphate.

Several attempts have been made, by means of genetic engineering, to generate strains of Saccharomyces cerevisiae able to ferment these pentose sugars. The expression of the bacterial pentose pathways in yeast has had only limited success in the past. The xylose isomerases expressed in S. cerevisiae (or in Schizosaccharomyces pombe) have shown no or only very low activity [2–8]. In an attempt to overexpress the bacterial l-arabinose pathway, the individual enzymes showed activity when produced in S. cerevisiae; however, the yeast strain carrying all three bacterial genes neither grew on l-arabinose nor fermented it to ethanol [9].

The expression of the fungal d-xylose pathway has been more successful: in S. cerevisiae it led to growth on d-xylose as well as to ethanol production from d-xylose [10–13]. The expression of the fungal l-arabinose pathway has not been possible until recently because not all genes of this pathway were known.

All enzymes of the fungal d-xylose pathway can also be used in the l-arabinose pathway. The first enzyme is an aldose reductase (EC This enzyme has been characterised after purification from S. cerevisiae[14] and from Pichia stipitis[15]. Both enzymes have similar activities with d-xylose and l-arabinose. Also, the corresponding genes for the S. cerevisiae enzyme (GRE3[14]) and the P. stipitis enzyme (XYL1, [16]) are known. The d-xylulose reductase (EC and the xylulokinase (EC are also involved in both pathways. Genes coding for d-xylulose reductases are known in various fungi [17,18]. Genes coding for xylulokinase are known for S. cerevisiae[19] and P. stipitis[20].

We have identified the missing two genes for l-arabinitol 4-dehydrogenase and l-xylulose reductase. The first has been purified from the mould Trichoderma reesei (Hypocrea jecorina) and the corresponding gene (lad1) identified [21]. To identify the gene for the l-xylulose reductase, all the other genes of the l-arabinose pathway, i.e. genes coding for aldose reductase, l-arabinitol 4-dehydrogenase, d-xylulose reductase and xylulokinase had been expressed in S. cerevisiae. This strain was then transformed with a cDNA library from T. reesei and screened for growth on l-arabinose. In this way a gene coding for an l-xylulose reductase (lxr1) has been identified [22]. The strain harbouring all the genes of the l-arabinose pathway could grow on l-arabinose [22]. In this paper we were interested if such a strain could also ferment l-arabinose to ethanol.

2Materials and methods


The cultivations were done with S. cerevisiae strain H2561 carrying all the genes of the l-arabinose pathway (XYL1, lad1, lxr1, XYL2 and XKS1) and a control strain H2562, which was identical except that it did not contain the lxr1, but an empty vector instead. XYL1 and XYL2 are from P. stipitis coding for aldose reductase and d-xylulose reductase, respectively. XKS1 is from S. cerevisiae and codes for xylulokinase; lad1 and lxr1 are from T. reesei and code for l-arabinitol 4-dehydrogenase and l-xylulose reductase, respectively. XYL1 (under PGK promoter), XYL2 (under ADH promoter) and XKS1 (under ADH promoter) were integrated into the chromosomes by targeted integration: lad1 and lxr1 (both under TPI promoter) were on two different plasmids. The lad1 was on a plasmid with URA3, the lxr1 on a plasmid with LEU2 as selection marker. The control strain was a similar construct with two plasmids, except that it did not contain the lxr1 gene. The construction of the strains has been described earlier [21,22].

2.2Fermentation conditions

Batch cultivations were performed in two 1.8-l Chemap CMF fermenters (Chemap, Switzerland). The culture volume was 0.9 l during the first 2 days and the medium was synthetic complete medium [23] lacking uracil and leucine, with 2% glucose as the carbon source. After 2 days 0.3 l of synthetic medium with 20%l-arabinose was added to give a final l-arabinose concentration of 5%. The temperature was 30 °C and agitation speed was 500 rpm. During cultivation on d-glucose the medium was sparged with air at a flow rate of 1.0 l min−1 and pH was adjusted to 5.00 with 2-M potassium hydroxide. When the l-arabinose was added the air flow was changed to nitrogen at 0.1 l min−1 to have anaerobic conditions. Samples were taken at different time intervals. They were analysed for dry weight and ethanol content. The dry weight was measured for 10 ml of cell suspension. Cells were harvested by centrifugation, washed twice with water and dried overnight at 95 °C. The ethanol concentration was measured enzymatically using a commercial kit (Roche Diagnostics, Mannheim, Germany). The analysis was done in a Cobas Mira automated analyser (Roche).


3.1l-Arabinose fermentation under anaerobic conditions

Yeast cells were first grown on the medium with d-glucose under aerobic conditions for 2 days to generate biomass. After this time most of the ethanol had been used. Synthetic medium containing l-arabinose was then added to give a concentration of 5%l-arabinose. The fermenter was then switched from air to nitrogen and the fermentation monitored for 70 h. In earlier trials we harvested the cells, washed them in phosphate buffer and resuspended them in the l-arabinose medium before introducing them to the fermenter. We observed that under these conditions the cells lysed so that after 70 h only 50% of the initial biomass remained. Without this harvesting and washing, we found that the biomass remained approximately constant during the 70 h. The l-arabinose was added when most of the ethanol had been used because it was then more accurate to measure small changes in the ethanol concentration. In Fig. 2 the ethanol production is shown during the anaerobic l-arabinose fermentation for the strain where the complete pathway is expressed and for a control strain. As a control we used a construct where only the lxr1 was omitted. The initial ethanol concentration in both cases was below 0.05 g l−1.

Figure 2.

Ethanol production from l-arabinose. At time zero the fermenter is switched to anaerobiosis and l-arabinose at a final concentration of 5% is added. The open triangles show the strain where the complete l-arabinose pathway is expressed. The open squares show the control strain, which is a similar construct lacking the lxr1 gene.

The strain with the complete pathway produced about 0.1 g l−1 ethanol during the 70-h period, whereas the ethanol production in the control strain was below our detection limit. The biomass of the strain with the complete pathway was 4 g l−1, i.e. the productivity was 0.35 mg ethanol per g dry weight and hour. The productivity of the control strain was below 0.05 mg ethanol per g dry weight and hour.


4.1The fungal l-arabinose pathway

A fungal pathway for the metabolism of l-arabinose was first described for the mould Penicillium chrysogenum in 1960 by Chiang and Knight [24]. The same pathway has been found in the mould Aspergillus niger[25]. There is also evidence that this pathway is active in the yeast P. stipitis. Shi et al. [26] have found that a deletion in the d-xylulose reductase gene of P. stipitis prevents growth on l-arabinose. This indicates that yeasts and moulds, i.e. fungi, in general use the same pathway for l-arabinose consumption, which is distinctly different from the bacterial pathway. We have combined the genes of this pathway from different fungi, from the yeast P. stipitis XYL1 and XYL2, from the yeast S. cerevisiae XKS1 and from the filamentous fungus T. reesei lad1 and lxr1. We have overexpressed them in S. cerevisiae, which is a yeast species that cannot use l-arabinose. The resulting strain showed activities of all the enzymes of the pathway and could grow on l-arabinose, although at a very low rate. When applying anaerobic conditions we could demonstrate for the first time that in a recombinant strain ethanol was produced from l-arabinose. However, the ethanol production under these conditions was very slow. There are various factors that might limit the l-arabinose fermentation, such as the imbalance of redox cofactors or the l-arabinose transport into the cell.

4.2The imbalance of redox cofactors

The fungal l-arabinose pathway consists of two oxidations and two reductions, i.e. the conversion of l-arabinose to d-xylulose is redox neutral. However, the reductions are NADPH-linked and the oxidations NAD+-linked so that there is an imbalance of redox cofactors. This imbalance of redox cofactors could be solved by a transhydrogenase activity facilitating the equilibration between NADH/NADP+ and NAD+/NADPH. Yeasts are believed not to have such activity [27]. It remains an open question how fungal micro-organisms cope with this cofactor imbalance. It has been suggested [28] that NADPH is mainly regenerated through the oxidative part of the pentose phosphate pathway. The filamentous fungus A. niger exhibited higher activities of glucose 6-phosphate dehydrogenase and 6-phosphogluconate dehydrogenase when growing on pentoses [28]. In the oxidative part of the pentose phosphate pathway, however, the reduction of NADP+ is coupled to CO2 production (Fig. 3). In this case the anaerobic conversion of l-arabinose to CO2 and ethanol is no longer redox neutral, i.e. the fermentation of l-arabinose is inhibited by an accumulation of reduced redox cofactors. A possible strategy to solve this imbalance of cofactors and thereby facilitate l-arabinose fermentation would be to introduce transhydrogenase activity.

4.3l-Arabinose transport into the cell

The ethanol production was about 0.1 g per 70 h at a biomass dry weight of 4 g l−1, which corresponds to 0.13 nmol per mg dry weight and minute. For such an ethanol production rate an l-arabinose uptake rate of 0.08 nmol per mg of dry weight and minute would be required, i.e. only if the maximal uptake rate is less than this order of magnitude, it can be a limiting factor. We are not aware of any information in the literature dealing with the uptake rate of l-arabinose in S. cerevisiae. d-Xylose however, another pentose sugar that is also not a natural substrate for S. cerevisiae, is taken up. Both d-xylose and glucose are transported by facilitated diffusion. The maximal uptake rate for d-xylose has been estimated to be 100–240 nmol (mg min)−1[10]. This is in the same order of magnitude as the maximal glucose uptake rate, which is 200–400 nmol (mg min)−1[29]. The only report about l-arabinose transport in yeast is from Lucas and van Uden [30], who observed that l-arabinose was transported in Candida shehatae by a specific proton symport mechanism.

It remains to be seen whether the l-arabinose fermentation can be stimulated by addressing the problems of redox cofactor imbalance and l-arabinose transport. Addressing the problem of cofactor imbalance can also be beneficial for d-xylose fermentation, since a strain with the l-arabinose pathway would also be able to ferment d-xylose.


This work was supported by the ‘Sustainable Use of Natural Resources’ (SUNARE) programme of the Academy of Finland and the research programme ‘VTT Industrial Biotechnology’ (Academy of Finland; Finnish Centre of Excellence programme, 2000–2005, Project no. 64330).