D-Galactose induces cellulase gene expression in Hypocrea jecorina at low growth rates

The pantropical ascomycete Hypocrea jecorina (anamorph Trichoderma reesei) is used industrially to produce various extracellular enzymes including cellulases and hemicellulases. The major extracellular protein produced is cellobiohydrolase I (Cel7A) and therefore its promoter is also used to drive heterologous protein production (Penttilä, 1998). Cellulases are usually formed on insoluble cellulose-containing materials, but their formation can also be induced by several mono- and disaccharides such as L-sorbose, sophorose and lactose (Aro et al., 2005). Lactose (1,4-O-β-D-galactopyranosyl-D-glucose) is a rather unusual inducer for cellulases as it is not expected to occur in the natural environment of H. jecorina. However, as a by-product of the dairy industry it represents an attractive renewable carbon source for industrial enzyme production with H. jecorina, but its slow metabolism and the fact that cellulase yields are lower than on cellulose still limit its use (Andreotti et al., 1980). Elucidating the mechanism by which lactose induces cellulase formation would therefore be helpful to improve its industrial use.

Lactose metabolism in H. jecorina is initiated by extracellular hydrolysis by β-galactosidases (Seiboth et al., 2005). The resulting D-glucose enters the glycolytic pathway directly whereas the D-galactose moiety can either be phosphorylated by galactokinase to galactose 1-phosphate and enter the Leloir pathway (Frey, 1996) or be reduced by an aldose reductase to galactitol (Fig. 1; Seiboth et al., 2004). Deletion of the galactokinase-encoding gene gal1 drastically reduced cellulase expression during growth on lactose, while deletion of the galactose-1-phosphate uridylyltransferase gene (gal7), which encodes the next step in the Leloir pathway, had no effect on cellulase induction (Seiboth et al., 2002a, 2004). These results were interpreted such that either D-galactose, bound to galactokinase, or D-galactose 1-phosphate is necessary for induction. In batch cultures, however, growth on D-galactose alone did not lead to cellulase induction, not even in the carbon-catabolite-derepressed mutant strain Rut-C30 (Seiboth et al., 2004), and strains which overexpressed the main extracellular β-galactosidase showed faster growth on lactose but were unable to form cellulases (Seiboth et al., 2005). Thus, if either D-galactose or D-galactose 1-phosphate is the actual inducer of cellulase formation, their action is limited to the slow growth on lactose which is due to the low extracellular β-galactosidase activity.

(31K): Fig. 1. Pathways of D-galactose catabolism in H. jecorina. The classical Leloir pathway (right) and a second pathway (left) which was recently described for H. jecorina (Seiboth et al., 2004; Pail et al., 2004) and A. nidulans (Fekete et al., 2004) are shown. Results from these studies were summarized to draft this second hypothetical pathway. ? indicates steps which have still to be verified.

To test the hypothesis that D-galactose or D-galactose 1-phosphate would act as inducers of cellulase gene expression when D-galactose catabolism occurs at a low rate, we performed chemostat cultures of H. jecorina on D-galactose at different dilution rates. Fungal strains and cultivation conditions.
H. jecorina QM9414 (ATCC 26921) and the uridine-auxotrophic, pyr4-negative mutant TU-6 (Gruber et al., 1990b) were maintained on malt extract agar supplemented with uridine (10 mM) when required. For chemostat cultivations, cultures were pre-grown in 250 ml Erlenmeyer flasks on a rotary shaker (250 r.p.m.) at 30 °C in the following medium (pH 5): 8 g NH₄H₂PO₄ l¹, 7 g Na₂HPO₄ l¹, 4 g KH₂PO₄ l¹, 1 g CaCl₂ l¹, 1 g MgSO₄ l¹, 20 ml l¹ trace elements solution (containing, per litre: 250 mg FeSO₄.7H₂O, 80 mg MnSO₄.H₂O, 70 mg ZnSO₄.7H₂O, 85 mg CoCl₂) with 10 g glycerol l¹ and 0·1 g peptone l¹ as a carbon source. Mycelia were harvested by filtration after 36 h, washed with cold tap water and then transferred into a fresh medium with the respective carbon source without peptone. Chemostat cultivations were carried out in a 2·5 l glass bioreactor (Inel) with a working volume of 2 l, equipped with one six-blade Rushton disc turbine impeller. Operating conditions were pH 5·0, 30 °C, 300 r.p.m. and 0·3 vvm (volumes of air per volume of liquid per minute). To minimize water and substrate losses, the waste gas was cooled in a reflux condenser connected to an external cooling bath (4 °C). The feeding medium contained carbon source at 3 g l¹, a concentration low enough to make the culture carbon-limited. Prior to cultivations, glass parts of the reactor were treated with the anti-adhesive agent Sigmacot (Sigma) to avoid fungal wall growth. Polypropylene glycol 2000 (Union Carbide Chemicals & Plastics) was used as antifoam agent; a few drops were injected into the reactor once a day through a membrane filter (Millipore). Onset of steady-state in the culture was established when no changes in biomass dry weight were observed in three successive samples taken over a period of three residence times (defined as the reciprocal of the dilution rate, e.g. 20 h for D=0·050 h¹). To exclude any regulatory effects on cellulase gene expression by light as observed recently (Schmoll et al., 2005), cultivations were performed in constant darkness.

Construction of a cbh2 : : goxA reporter strain.
A 621 bp region of the H. jecorina cbh2 upstream region was amplified from H. jecorina QM9414 genomic DNA by PCR using the primers cbh2XhoFw1 (5'-CAGTCTCGAGGAATTCTAGGCTAGGTATGCGAG-3') and cbh2XbaRv (5'-GACTTCTAGAGGTGCAATACACAGAGGGTG-3') to introduce XhoI and XbaI restriction sites. The amplification protocol consisted of an initial denaturation step of 1 min at 94 °C, followed by 30 cycles with a denaturation (1 min, 94 °C), an annealing (45 s, 57 °C) and an elongation step (1 min, 72 °C), followed by a final 7 min elongation step at 72 °C. The amplicon was cloned upstream of the Aspergillus niger glucose oxidase gene (goxA) into pSJ3 (Mach et al., 1999) using the XbaI and XhoI restriction sites and sequenced. Then the H. jecorina pyr4 gene was inserted as a 2·7 kb SalI fragment (Gruber et al., 1990a) into the XhoI site, resulting in pCHG1. Transformation of H. jecorina TU-6 with pCHG1 was performed via the protoplast method described by Gruber et al. (1990b). Single integration of the reporter cassette was verified by Southern blotting and hybridization with the goxA gene (data not shown). The reporter strain was designated CHG1. We have tested two other reporter strains with single reporter gene integration under selected conditions which essentially confirmed the cellulase gene expression results obtained for CHG1.

Enzyme assay.
For the determination of glucose oxidase (Gox) activity in the culture supernatants, samples were first dialysed using a Spectra/Por 3 dialysis membrane (Serva) with a MWCO of 3500 against the assay buffer for 24 h. Gox activity was quantified spectrophotometrically as described previously (Mach et al., 1999). One unit (1 U) of activity is defined as the amount of enzyme required to oxidize 1 µmol D-glucose min¹ at pH 5·8 and 25 °C. Protein concentration was determined by a modified Lowry method (Peterson, 1983), using BSA as protein standard.

Electrophoretic and immunological techniques.
The protocols described by Ausubel et al. (2005) were used for SDS-PAGE with 10 % polyacrylamide gels and subsequent Western blotting. CBHI/Cel7A and CBHII/Cel6A were detected with the aid of monoclonal antibodies (Mischak et al., 1989).

Analysis of monosaccharides and polyols.
The concentration of D-galactose, D-glucose, galactitol and lactose in the medium was determined by HPLC analysis, using the H⁺ exchange column Aminex HPX-H⁺ (Bio-Rad). As mobile phase 10 mM H₂SO₄ at 55 °C was employed with isocratic elution and refractive index detection.

Determination of fungal dry weight.
Mycelial dry weight was determined by withdrawing two 20 ml aliquots from the medium, followed by suction filtration through a glass wool filter and drying to a constant weight at 80 °C. Data were averaged and deviated by not more than 14 %.

Reproducibility.
Results are means of three to five measurements. The data were analysed by Sigmaplot (SPSS Inc.) and standard deviations for each procedure were determined. The SD values were always less than 14 % of the means.

Carbon-limited chemostat cultivation of H. jecorina
A series of constant-mass, carbon-limited, chemostat-type continuous cultures of the glucose oxidase reporter strain H. jecorina CHG1 were performed at five different dilution rates ranging from 0·075 h¹ to 0·015 h¹ with D-galactose, lactose, D-glucose or galactitol, as well as an equimolar mixture of D-galactose+galactitol and D-galactose+D-glucose, respectively, as carbon source. Similarly to the majority of reports on fungal chemostat cultures such as the lactose-limited cultures of the catabolite-derepressed H. jecorina Rut-C30 mutant (Pakula et al., 2005), four to five residence times were needed to achieve steady state under these conditions. Cultures exhibited some pellet formation at D=0·015 h¹ with every carbon source tested. In case of the lactose-, D-glucose- and D-galactose-limited cultures, the steady-state biomass concentration was 1·48±0·20 g l¹ in all cultures irrespective of the dilution rate. Cultures at the two lowest dilution rates (0·030 and 0·015 h¹) also displayed the lowest biomass concentration range. However, due to the relatively high standard error (14 %) of the protocol used for determining steady-state biomass concentration, these differences were not considered significant. The residual steady-state concentrations of D-glucose, D-galactose and lactose in the medium were 0·080·10 mM, 0·100·12 mM and 0·090·12 mM, respectively. Consistent with the results of Pakula et al. (2005), we could not detect D-glucose or D-galactose in the lactose-grown cultures. In galactitol-limited cultures, the steady-state biomass concentration was 1·36±0·17 g l¹, with a residual galactitol concentration of 0·080·09 mM. In accordance with the general theory of chemostat culture, carbon utilization rates (mmol carbon source consumed per gram of biomass formed per hour) for all carbon sources tested increased linearly with the dilution rate. The calculated growth yield (grams of biomass formed per gram of carbon source consumed) was between 45 and 49 % for all cultures, which correlates well with previously published data from various fungal carbon-limited cultures (e.g. Pirt & Callow, 1960). There were thus apparently no qualitative differences in the overall metabolism of H. jecorina at different dilution rates compared to other fungi, and we therefore considered the system appropriate for the purpose of this study.

Effect of dilution rate on H. jecorina cbh2 gene expression on D-galactose and lactose
Chemostat cultures of the cbh2 : : goxA reporter strain H. jecorina CHG1 were performed at five different dilution rates ranging from 0·075 h¹ to 0·015 h¹ with D-galactose or lactose as limiting carbon source. Table 1 shows that on D-galactose, cbh2 promoter activation, measured by glucose oxidase activity as reporter gene, did not occur at D between 0·075 and 0·030 h¹, but was clearly detectable at D=0·015 h¹. These findings were further supported by the detection of the cbh2-encoded Cel6A protein in the culture fluid only at 0·015 h¹ but not at higher dilution rates. Also the cbh1-encoded major cellulase Cel7A of H. jecorina was only detectable at this dilution rate (Fig. 2a).

Table 1. Glucose oxidase activity in the supernatant of carbon-limited chemostat cultures of H. jecorina CHG1 at different dilution rates One unit (1 U) of glucose oxidase activity is defined as the amount of enzyme required to oxidize 1 µmol D-glucose min1 at pH 5·8 and 25 °C.

(35K): Fig. 2. Cellulase formation in carbon-limited chemostat cultures of H. jecorina CHG1 at different dilution rates: Western blots of Cel7A and Cel6A accumulation in the supernatant of chemostat cultures with D-galactose (a) and lactose (b) as carbon source. Equal volumes (corresponding to 300 µl supernatant) were loaded per lane.

Using lactose as a limiting carbon source, however, cbh2 promoter activation as well as the appearance of the Cel6A and Cel7A proteins (Fig. 2b) in the culture filtrates were detected already at higher dilution rates: both reporter gene expression and the two proteins were already detected at D=0·042 h¹. At higher dilution rates a washing out of the mycelia was observed, indicating that the maximal specific growth rate was exceeded. Levels of the two cellobiohydrolases increased when the dilution rate was further lowered. At D=0·015 h¹ cbh2 gene promoter activation was fourfold as high as on D-galactose.

Effect of dilution rate on H. jecorina cbh2 gene expression on D-glucose
In theory, cbh2 gene expression at D=0·015 h¹ could be the result of a specific D-galactose induction or due to a general derepression at the low growth rate. To discriminate between these two possibilities chemostat cultures were also performed with D-glucose as the limiting nutrient at low dilution rates. In contrast to the results obtained with D-galactose, no cbh2 promoter activation was detected under these conditions, which was also supported by the absence of the Cel6A protein. Cel7A whose basal level is known to be under carbon catabolite repression by Cre1 (Zeilinger et al., 2003) was detectable only at the lowest dilution rate, 0·015 h¹ (Table 1, Fig. 3a). These findings prove that the triggering of cbh2 gene expression by D-galactose at d=0·015 h¹ is not due to a general carbon catabolite derepression but to a specific induction by D-galactose.

(19K): Fig. 3. Cellulase formation in carbon-limited chemostat cultures of H. jecorina CHG1 at different dilution rates: Western blots of Cel7A and Cel6A accumulation in the supernatant of chemostat cultures with D-glucose (a) and an equimolar mixture of D-galactose and D-glucose (b). For comparison the supernatants of the chemostat cultures with lactose (Lac) and D-galactose (Gal) as carbon source (D=0·015) were loaded. Equal volumes (corresponding to 300 µl supernatant) were loaded per lane.

Influence of dilution rate on cellulase gene expression in chemostat cultures containing equimolar mixtures of D-glucose and D-galactose
the increased levels of formation of cel7a and cel6a during growth on lactose in comparison to D-galactose, and their occurrence at higher dilution rates on lactose, pointed to a clear difference between the induction by lactose and D-galactose. We tested whether this is due to lactose itself or to the presence of the two monomers D-glucose and D-galactose. Chemostat cultures were therefore performed with an equimolar mixture of the two monosaccharides. The results are shown in Table 1 and Fig. 3(b), and indicate that, identical to the culture on D-galactose alone, cbh2 promoter activation did not occur at D=0·0750·030 h¹, but was clearly detectable at D=0·015 h¹. The accumulation of Cel7A and Cel6A was also consistent with this finding and paralleled the accumulation pattern seen on D-galactose alone. These data therefore indicate that it is the lactose disaccharide which is responsible for the observed difference in cellulase formation, and, moreover, that the simultaneous presence of D-glucose and D-galactose does not influence the induction of cellulases by D-galactose at low growth rates.

Effect of dilution rate on H. jecorina cbh2 expression on equimolar mixtures of D-galactose and galactitol
In H. jecorina, catabolism of D-galactose can proceed via at least two different pathways, i.e. the classical Leloir pathway and a pathway involving formation of galactitol as an intermediate (Seiboth et al., 2004). In order to find out whether this second pathway contributes to cellulase formation on lactose, we performed chemostat cultivations with galactitol and equimolar mixtures of D-galactose and galactitol as carbon source, thus forcing catabolism of at least half of the substrate through this pathway. Maximal growth rate was exceeded for both conditions at D=0·075 h¹, which resulted in a washing out of the mycelia. However, cbh2 promoter activitation as well as expression of Cel7A and Cel6A was observed at 0·030 h¹ on the mixture of D-galactose and galactitol, whereas no cellulases were formed at this dilution rate on D-galactose or galactitol alone (Fig. 4, Table 1). We could not achieve a steady state with the mixture of galactitol and D-galactose or galactitol alone at D=0·015 h¹, which was the lowest growth rate for lactose or D-galactose as limiting carbon source.

(33K): Fig. 4. Cellulase formation in carbon-limited chemostat cultures of H. jecorina CHG1 at different dilution rates: Western blots of Cel7A and Cel6A accumulation in the supernatant of chemostatcultures with an equimolar mixture of D-galactose and galactitol. For comparison the supernatant of the chemostat culture with lactose (Lac) as carbon source (D=0·030) was loaded. Equal volumes (corresponding to 300 µl supernatant) were loaded per lane.

In this study, we have tested the hypothesis whether the triggering of cellulase formation by lactose is due to induction by D-galactose (or a metabolite thereof) at slow growth rates. This hypothesis was created to explain the previous findings that a galactokinase knockout mutant of H. jecorina is impaired in cellulase induction, yet growth on D-galactose in batch culture does not induce cellulase formation (Seiboth et al., 2004). We therefore grew H. jecorina in chemostat cultures with D-galactose as limiting carbon source at various dilution rates between 0·075 and 0·015 h¹. Experiments reported here indeed showed cellulase gene expression at a very low (0·015 h¹) but not a higher (0·030 h¹) dilution rate. Of course, this cellulase formation at a low growth rate does not necessarily prove induction but could also be due to carbon catabolite derepression (Ilyés et al., 2004). It is therefore essential to compare these results to those of chemostat cultures on D-glucose, and also to compare the formation of the two cellulases Cel7A and Cel6A. Cel7A is encoded by cbh1, whose basal expression level is under the control of the Cre1 carbon catabolite repressor. Therefore, a low level of Cel7A is formed under derepressing conditions even without the addition of an inducer, whereas formation of Cel6A is strictly dependent on the presence of an inducer (Zeilinger et al., 2003). This behaviour is also reflected in the present results, i.e. only Cel7A but not Cel6A was formed at 0·015 h¹ on D-glucose. However, both proteins accumulated to a clearly higher level on D-galactose at 0·015 h¹. Consequently, D-galactose indeed acts as an inducer at a low dilution rate and this effect is therefore not due to carbon catabolite derepression, although the two effects may overlap in the case of cbh1.

While these data are consistent with our hypothesis that D-galactose can act as an inducer at slow growth rates, the higher cbh2 gene expression levels and stronger accumulation of Cel7A and Cel6A in the culture fluid on lactose than on D-galactose indicate that lactose is nevertheless a far superior inducer of cellulase formation to D-galactose. The induction seen by D-galactose at 0·015 h¹ is thus not sufficient to account for the efficient cellulase induction seen on lactose, and lactose must therefore transmit a superior or additional signal to the fungus, which ultimately triggers cellulase gene expression at high levels. An obvious hypothesis to explain this finding would be to postulate a lactose-binding receptor which would then trigger a cellulase induction cascade. At this moment, we can neither prove nor reject this theory. However, we note that this theory would not explain why a mutant with a knockout in the galactokinase gene would be impaired in cellulase induction. We have preliminary evidence that in H. jecorina the enzymic activity of the galactokinase is necessary for cellulase induction (L. Hartl, C. P. Kubicek & B. Seiboth, unpublished results), which is in contrast to the situation in Kluyveromyces lactis and Saccharomyces cerevisiae. In K. lactis an additional signal transduction activity, which is essential for induction of the GAL genes, is encoded by the enzymically active galactokinase (Meyer et al., 1991). In S. cerevisiae these two functions are found on two independent but highly related proteins: Gal1 performs the galactokinase activity and Gal3 is responsible for the signal transduction of D-galactose to the transcriptional activator Gal4 (Bhat & Murthy, 2001).

Alternatively, and by analogy with cellulase induction by cellulose [which is believed to be mediated by transglycosylation products, e.g. sophorose (Sternberg & Mandels, 1979; Vaheri et al., 1979)], it could be postulated that lactose would enter the cells and, either directly or also after transglycosylation, act as inducer of cellulase formation. Such an assumption would however be in conflict with the fact that we have not been able to detect either a lactose permease or an intracellular β-galactosidase activity in H. jecorina (Seiboth et al., 2005), which would both be prerequisites for the operation of such a mechanism. In addition, this theory would not explain the need for galactokinase activity in cellulase induction.

Results from the present work show that, despite hydrolysing it before taking it up into the cell, H. jecorina recognizes the presence of hydrolysed lactose in a different way than the simultaneous presence of D-glucose and D-galactose. What is the difference between hydrolysed lactose and a mixture of D-glucose and D-galactose? D-Galactose is, because of mutarotation, a mixture of the α- and β-anomer whereas D-galactose is released from lactose as the β-anomer. The following enzymic step catalysed by galactokinase is strictly specific for α-D-galactose (Holden et al., 2003) and consequently the β-D-galactose has to be first converted by mutarotation to the α-D-galactose by an aldose 1-epimerase. To this end S. cerevisiae and other yeasts possess an aldose 1-epimerase activity located on the bifunctional Gal10 protein, which also has a UDP-glucose 4-epimerase activity (Thoden & Holden, 2005). This bifunctional composition of Gal10 seems to be yeast specific while in other organisms, including H. jecorina, these enzymic functions are found on two different proteins (Seiboth et al., 2002b). A BLASTX search identifies at least two genes encoding proteins with a high similarity to the yeast aldose 1-epimerase part of Gal10 in the H. jecorina genome sequence database (). We do not know at the moment whether they indeed encode D-galactose mutarotases and are expressed on lactose. But the absence or only a low expression of such an enzyme would considerably delay catabolism of the lactose-derived β-D-galactose via the Leloir pathway because of the time needed for non-enzymic mutarotation, which at 30 °C and pH 6·57·0 takes several hours (Pettersson & Pettersson, 2001). This could eventually lead to an accumulation of β-D-galactose in the mycelia to toxic levels. It is therefore possible that this may be the reason why H. jecorina and other fungi have in addition to the Leloir pathway an alternative pathway for D-galactose catabolism (Fekete et al., 2004; Seiboth et al., 2004), which by analogy with the L-arabinose catabolic pathway uses reductive and oxidative reactions to convert D-galactose into fructose 6-phosphate. The first enzyme in this alternative pathway is an aldose reductase, which is able to use both D-galactose anomers as a substrate to form galactitol.

We have recently found that deletion of a general aldose reductase led to slower growth on lactose and D-galactose (unpublished data), although some residual D-galactose reductase activity was still present. This indicates that this pathway is operative under the conditions of this work. In order to test whether this pathway also contributes to the inducer formation from lactose, we have grown H. jecorina on an equimolar mixture of D-galactose and galactitol, which resulted in cellulase induction at a growth rate (0·030 h¹) at which otherwise no cellulase formation was detectable on D-galactose alone. Unfortunately, we were unable to cultivate H. jecorina on D-galactose and galactitol at 0·015 h¹, and are thus unable to provide quantitative data for the possible enhancement of cellulase formation over that on D-galactose alone. But we should like to note that a direct comparison of results obtained by growth on lactose and D-galactose/galactitol may be problematic in any case, because galactitol is more reduced than D-galactose, and its presence as a substrate creates a higher demand for maintenance energy. This fact is also reflected in our inability to cultivate H. jecorina on galactitol at D=0·015 h¹ and in the slightly lower steady-state biomass concentration values at higher dilution rates. Nevertheless, these data show that catabolism of galactitol (=operation of the alternative D-galactose catabolic pathway) positively contributes to the induction of cellulase gene expression. Further investigations will be dedicated to finding out how galactokinase and this pathway interact in this process.

The dilution rates reported for lactose are in the same range as those reported for the H. jecorina Rut-C30 (Pakula et al., 2005), with the exception that the lowest dilution rate was 0·022 h¹. Rut-C30 is defective in at least two genes: one encodes the wide-domain regulator Cre1 for carbon catabolite repression (Ilmen et al., 1996), the second the glucosidase II α subunit (Gls2α), a protein which is involved in endoplasmic reticulum quality control (Geysens et al., 2005). In Aspergillus nidulans it was shown that carbon catabolite repression mediated by the orthologous CreA is growth rate dependent and that below a certain growth rate carbon catabolite derepression occurs (Ilyés et al., 2004). Nevertheless, in the carbon-catabolite-derepressed Rut-C30 most Cel7A is produced at a low specific growth rate, with the optimum at 0·031 h¹. Cel7A production decreases with higher dilution rates but also at 0·022 h¹ lower production is found. At low specific growth rates a limitation of the capacity of the secretory pathway was found which prevented an efficient transport of Cel7A through the secretory pathway. It has to be proven if this difference might be due to the truncated gls2α or to other effects. Here we could show that Cel7A and Cel6A gene expression increased with decreasing dilution rates on lactose in CHG1. A comparison of these results indicates that Cre1 is certainly not the only regulator which controls high cellobiohydrolase production, especially during fast growth at high dilution rates.

An interesting side-observation of this study is the finding of much higher cellulase protein concentrations in culture filtrates from lactose- than from D-galactose-grown cultures, significantly exceeding the ratio which would be predictable from the ratio in gene expression measured by glucose oxidase activity. This indicates that induction is not the only mechanism by which lactose promotes cellulase production at high levels and that additional factors are necessary to assure a high production of cellobiohydrolases. The observed difference might be simply the result of the fact that (although of fungal origin) glucose oxidase is a heterologous protein, while the two cellobiohydrolases are homologous proteins. It remains to be tested at which level of gene expression this regulation takes place.

The project was carried out in the framework of an AustrianHungarian Intergovernmental Science and Technology Cooperation Programme (A-5/03). L. K.'s lab is grant-aided by the OTKA (Hungarian Scientific Research Fund; F042602). C. G., C. P. K. and B. S. were supported by the Austrian Science Foundation FWF (P16143). E. F. and L. K. are recipients of an OTKA postdoctoral fellowship (D048617) and a Békésy György fellowship (BÖ98/2002), respectively.