OUP user menu

Overexpression of the aldose reductase GRE3 suppresses lithium-induced galactose toxicity in Saccharomyces cerevisiae

Claudio A. Masuda, Jose O. Previato, Michel N. Miranda, Leandro J. Assis, Luciana L. Penha, Lucia Mendonça-Previato, Mónica Montero-Lomelí
DOI: http://dx.doi.org/10.1111/j.1567-1364.2008.00440.x 1245-1253 First published online: 1 December 2008

Abstract

In Saccharomyces cerevisiae, lithium induces a ‘galactosemia-like’ phenotype as a consequence of inhibition of phosphoglucomutase, a key enzyme in galactose metabolism. Induced galactose toxicity is prevented by deletion of GAL4, which inhibits the transcriptional activation of genes involved in galactose metabolism and by deletion of the galactokinase (GAL1), indicating that galactose-1-phosphate, a phosphorylated intermediate of the Leloir pathway, is the toxic compound. As an alternative to inhibiting entry and metabolism of galactose, we investigated whether deviation of galactose metabolism from the Leloir pathway would also overcome the galactosemic effect of lithium. We show that cells overexpressing the aldose reductase GRE3, which converts galactose to galactitol, are more tolerant to lithium than wild-type cells when grown in galactose medium and they accumulate more galactitol and less galactose-1-phosphate. Overexpression of GRE3 also suppressed the galactose growth defect of the ‘galactosemic’gal7- and gal10-deleted strains, which lack galactose-1-P-uridyltransferase or UDP-galactose-4-epimerase activities, respectively. Furthermore, the effect of GRE3 was independent of the inositol monophosphatases INM1 and INM2. We propose that lithium induces a galactosemic state in yeast and that inhibition of the Leloir pathway before the phosphorylation step or stimulation of galactitol production suppresses lithium-induced galactose toxicity.

Keywords
  • lithium
  • galactosemia
  • aldose reductase
  • galactokinase
  • Saccharomyces cerevisiae
  • inositol monophosphatase

Introduction

Lithium is a drug that has been used for the last five decades to treat bipolar disorder. Several direct biochemical targets that are inhibited at therapeutic concentrations of lithium have been described. Mg2+-dependent enzymes have been shown to be inhibited by lithium as its hydrated radius is similar to that of Mg2+. Some of these enzymes share a structural motif [Asp–Pro–(Ile/Leu)–Asp–Gly/Ser)–Thr/Ser] responsible for lithium binding, such as inositol monophosphatases (IMPases), inositol polyphosphate-1-phosphatase, fructose-1,6-bisphosphate phosphatase and bisphosphate nucleotidase (York et al., 1995). Another important enzyme inhibited by lithium is phosphoglucomutase, which utilizes Mg2+ as a cofactor but does not possess the structural motif shared by the enzymes mentioned above (Ray et al., 1978; Rhyu et al., 1984; Masuda et al., 2001).

Phosphoglucomutase is a central enzyme in carbohydrate metabolism and, particularly for this work, is essential for galactose metabolism (Douglas et al., 1961; Oh & Hopper et al., 1990; Boles et al., 1994). As galactose enters the cell, it is metabolized by the Leloir pathway (Kosterlitz et al., 1943; Leloir et al., 1951; Kalckar et al., 1957), which consists of a series of reactions catalyzed by galactokinase, galactose-1-phosphate uridyltransferase (GALT) and UDP-galactose-4-epimerase (GALE) that transform galactose to glucose-1-phosphate (Glc-1-P). In yeast, the GAL1, GAL7 and GAL10 genes encode these enzymes, respectively (Robichon-Szulmajster et al., 1958; Douglas & Hawthorne et al., 1964). After Glc-1-P is formed, phosphoglucomutase then converts it to glucose-6-phosphate (Glc-6-P), which is further metabolized through glycolysis (Douglas et al., 1961; Boles et al., 1994).

Inhibition of phosphoglucomutase by lithium in galactose-grown yeast leads to growth arrest and to accumulation of galactose-1-phosphate (Gal-1-P), Glc-1-P, galactitol and glycerol (Bro et al., 2003) and, probably, some of these metabolites become toxic to cells at high concentrations. The accumulation of Gal-1-P and galactitol resembles the cellular phenotypes observed in the human disease galactosemia, which is in fact a group of autosomal recessive disorders of galactose metabolism (Bosch et al., 2006; Fridovich-Keil et al., 2006). Three types of galactosemia are encountered in humans. Classical galactosemia is caused by a deficiency in GALT enzyme that causes elevated levels of galactose, Gal-1-P and galactitol and diminished amounts of glycosylated proteins. It leads to infant death if not treated properly and the first symptoms are jaundice, hepatosplenomegaly, hepatocellular insufficiency, food intolerance, hypoglycemia, renal tubular dysfunction, muscle hypotonia, sepsis and cataract (Bosch et al., 2006; Fridovich-Keil et al., 2006). Galactokinase deficiency causes galactosemia type II and mutations in this gene lead to accumulation of galactitol. This metabolic disease is more benign and causes cataract and galactose intolerance. Galactosemia type III is caused by mutations in the GALE gene and its deficiency impairs galactose metabolism (Walter et al., 1999). Only the peripheral form of GALE deficiency appears benign (Timson et al., 2006). The existence of these three types of galactosemia and its different outcomes provide evidence that accumulation of Gal-1-P or a derivative is the main cause of cellular toxicity that leads to clinical complications. Using the yeast Saccharomyces cerevisiae as a model we characterized Gal-1-P as the metabolite that causes the galactose-specific lithium-induced growth arrest and identified genes that suppress its toxicity.

Materials and methods

Strains, plasmids and culture conditions

Saccharomyces cerevisiae wild-type strain BY4741 (MATa, his3Δ1, leu2Δ0, met15Δ0, ura3Δ0, and GAL2+) was used as the control strain. Strains deleted in gal1, gal2, gal4, gal7, gal10, gal80, inm1 and gre3 were obtained from MATa deletion library (Open Biosystems). Cells were grown at 30 °C in YP medium (1% yeast extract, 2% Bacto peptone and +0.003% methionine), containing 2% glucose, 2% galactose or 2% glycerol, and in synthetic dextrose (SD) medium (0.67% yeast nitrogen base, 2% glucose and appropriate auxotrophic supplements, such as 0.003% uracil, leucine, histidine, methionine, lysine, adenine and tryptophan). When lithium was added to YPGal, the medium was supplemented with 0.003% methionine to avoid the toxic effects of 3′-phosphoadenosine-5′-phosphate (PAP) accumulation, induced by lithium (Dichtl et al., 1997). Escherichia coli strain XL1-Blue was grown at 37 °C in LB (0.5% yeast extract, 0.5% NaCl, 1% Bacto tryptone) and used as a host for plasmid propagation. Plasmid pGRE3 obtained from Open Biosystems (clone identity: YSC3867-9520270), which contains the GRE3/YHR104W ORF cloned under control of the GAL1 promoter, was used to transform BY4741, Δgal7, Δgal10, Δinm1/inm2 and Δgre3 strains.

Double disruption of INM1 and INM2

Disruption of the entire coding region of the INM2 gene on the Δinm1 strain (Open Biosystems) was carried out by one-step gene replacement using the HIS3MX6 module (Brachmann et al., 1998) and the primers inm2-kan5′: 5′-GAAGCTCCACAAGCCCTTGTAAAGTAAACTTAGATTGAAGCGATGCGTACGCTGCAGGTCGAC-3′ and inm2-kan3′: 5′-TTTCTTAGATGTAGCTTATTTAACGACACCTGAGGGGGGCATTTAATCGATGAATTCGAGCTCG-3′. Genomic DNA from a selected Δinm1/inm2 clone was examined for disruption by PCR using primers INM2 KO (sense): 5′-GACATTCCAAAGCCAGTAACTAAAA-3′; INM1 KO (sense): 5′-AGAGAACTTTCTTCTGCTTCCTTTC-3′ and kanB (antisense): 5′-CTGCAGCGAGGAGCCGTAAT-3′ as reported (Brachmann et al., 1998).

Extraction and analysis of yeast metabolites

Extraction of metabolites was carried out based on previously published protocols (Gonzalez et al., 1997). Briefly, yeast cells were grown as described and aliquots of the culture were collected using rapid vacuum filtration in 0.45-μm millipore filters. Cells were immediately transferred to 80% ethanol at 80 °C and incubated for 3 min. Samples were lyophilized, resuspended in water and centrifuged at 20 000 g for 20 min at 4 °C. The supernatant was used to quantify metabolites.

Galactitol was identified by its retention time and quantified by peak area using myoinositol as the internal standard on the samples analyzed by high pH anion-exchange chromatography with a pulsed amperometric detection (HPAEC-PAD) on a Dionex LC 20 system, equipped with a CarboPac MA1 column (4 mm × 250 mm). The material was eluted with 700 mM NaOH at a flow rate of 0.5 mL min−1. Mannose, galactose, glucose, mannitol, galactitol and glucitol were used as carbohydrate standards. Gal-1-P content was measured using a coupled system. 10 μL samples were incubated at 30 °C in 250 μL buffer containing 50 mM Tris-HCl, pH 7.5, 3.5 mM MgCl2, 0.66 mM EDTA, 1.0 mM NAD+ and 2 U mL−1 galactose dehydrogenase (Sigma). After depletion of the galactose content, 2 U mL−1 of alkaline phosphatase (Sigma) was added and NADH formation was monitored spectrophotometrically at A340 nm.

Aldose reductase activity

To measure aldose reductase activity, a yeast protein extract was prepared essentially as described (Masuda et al., 2001), except that after the lytic step, the cell lysate was centrifuged at 3000 g for 10 min at 4 °C and the supernatant was further centrifuged at 20 000 g to avoid contamination with mitochondrial NADPH reductases. The supernatant was collected and used in the enzymatic assay. Aldose reductase activity was measured at 30 °C by determining NADPH oxidation at A340 nm using a Spectramax M5 spectrophotometer (Molecular Devices). The reaction mixture contained 50 mM sodium phosphate buffer, pH 5.0, 1.0 mM dithiotreitol, 0.120 mM NADPH and 300 mM galactose. The reaction was started by the addition of 40 μg of protein extract. A control experiment was run in parallel without the addition of galactose. Initial velocity was determined by subtracting the rate of formation of NADPH in the presence of galactose from that measured in the absence of galactose. Protein was determined by the Lowry method using bovine serum albumin as a standard (Lowry et al., 1951).

Results

Lithium toxicity in galactose

The growth inhibition caused by lithium in galactose medium could be explained by three different mechanisms. The first mechanism is a starvation-like effect because inhibition of phosphoglucomutase activity by lithium decreases the flux of galactose to Glc-6-P, reducing the fermentation rate (Masuda et al., 2001). The second mechanism could be the accumulation of phosphorylated intermediates that might deplete the cell of free phosphate, thus inhibiting important metabolic reactions that are dependent on phosphate (Thevelein & Hohmann et al., 1995). A third mechanism could be the toxic effect of the metabolites Gal-1-P and Glc-1-P that accumulate inside the cell in the presence of lithium. We examined the toxicity of lithium in the presence of increasing concentrations of galactose and demonstrated that lithium is more toxic to wild-type yeast cells as the concentration of galactose in the medium is increased (Fig. 1a). Even in the presence of glycerol, another carbon source whose metabolism is not directly inhibited by lithium, galactose becomes toxic in the presence of lithium (Fig. 1b). These results show that the effect of lithium on the growth rate of cells grown in medium containing both glycerol and galactose is not a starvation-like effect, but it is probably due to the accumulation of a toxic compound (Masuda et al., 2001; Bro et al., 2003).

1

Galactose is toxic in the presence of lithium. (a) Saccharomyces cerevisiae strain BY4741 was grown to the stationary phase and seeded at a concentration of 107, 106 or 105 cells mL−1 (from left to right) onto plates containing YP medium in the absence or presence of 30 mM lithium and increasing concentrations of galactose. (b) BY4741 was grown as above but medium was supplemented with 2% glycerol. Growth was recorded after 4 days at 30°C.

In order to test the hypothesis that a toxic metabolite is generated by galactose metabolism in the presence of lithium, we performed a growth test with a strain deleted of the GAL4 gene, which encodes a transcriptional activator of galactose-metabolizing genes (Hashimoto et al., 1983; Keegan et al., 1986; Lohr et al., 1995). As expected, the reduced galactose metabolism in the gal4 strain makes this strain more tolerant to lithium in galactose-containing medium (Fig. 2). In contrast, deletion of the Gal4p repressor, the GAL80 gene, renders expression of galactose-utilizing genes constitutive (Lohr et al., 1995) and does not overcome toxicity. Another result supporting this hypothesis is the lithium tolerance in galactose-containing medium observed in the gal1-deleted strain that lacks galactokinase, the enzyme responsible for the phosphorylation of galactose to Gal-1-P (Schell & Wilson et al., 1977) (Fig. 2). Deletion of the high-affinity galactose permease gene, GAL2, which reduces, but does not inhibit galactose uptake completely, did not reduce toxicity at the concentration of lithium tested. On the other hand, deletion of GAL7 (GALT) (Douglas & Hawthorne et al., 1964; Segawa & Fukasawa et al., 1979) or GAL10 genes (GALE) (Thoden et al., 2001), which induce accumulation of Gal-1-P even in the absence of lithium, is highly toxic to cells growing in galactose medium (Fig. 2). These results suggest that a phosphorylated intermediate of the Leloir pathway, probably Gal-1-P or a derivative, is toxic in yeast as it is proposed for human cells (Fridovich-Keil et al., 2006).

2

Inhibition of galactose metabolism suppresses toxicity. The indicated Saccharomyces cerevisiae strains were grown in YP supplemented with 2% glycerol and the indicated concentrations of LiCl and galactose. Growth was recorded after 4 days at 30°C.

Induction of an alternative pathway of galactose metabolism suppresses lithium-induced galactose toxicity

The treatment of choice for human galactosemia is based on a galactose-free diet. However, the symptoms are not fully reversed by this treatment alone, and so other treatments should be developed. As has been noted previously, induction of alternative pathways of galactose metabolism that do not convert galactose to Gal-1-P might be good candidates for improvement of the outcome of galactosemic individuals (Fridovich-Keil et al., 2006). Of special interest is the reduction of galactose to galactitol by aldose reductases. In S. cerevisiae, six aldose reductases are encoded in its genome, where GRE3 and YPR1 are the most expressed genes. GRE3 is the major xylose-reducing enzyme (Petrash et al., 2001) and is induced by oxidative stress, heat shock, starvation, heavy metals, osmotic and ionic stress including lithium (Garay-Arroyo & Covarrubias et al., 1999). YPR1 is the major arabinose-metabolizing enzyme (Nakamura et al., 1997; Ford & Ellis et al., 2002). None of the six genes have been shown to reduce galactose to galactitol and no obvious phenotype regarding galactose metabolism has been described for their deletion mutants.

Because it has been shown that GRE3 is induced by lithium, we tested whether its overexpression could result in lithium tolerance in galactose-containing medium (Fig. 3). We transformed a wild-type strain or a gre3-deleted strain with a multicopy plasmid carrying GRE3 under the control of the GAL1 promoter and observed that GRE3 overexpression suppresses the growth arrest induced by lithium in the presence of galactose (Fig. 3).

3

GRE3 overexpression confers lithium tolerance. BY4741, Δgre3 and the isogenic strains harboring the pGRE3 plasmid were plated onto YP containing 2% galactose in the absence or presence of 30 mM lithium. Growth was recorded after 4 days at 30°C.

Effect of GRE3 overexpression in galactosemic strains

Because overexpression of GRE3 could suppress the galactose-specific growth arrest caused by lithium, we tested whether it would also suppress the galactose-specific growth defect of the known ‘galactosemic strains’Δgal7 and Δgal10. We show that both mutants can grow in YP medium lacking a carbohydrate source or in glucose, but addition of galactose to YP becomes toxic. Surprisingly, growth is rescued by overexpression of GRE3 (Fig. 4).

4

GRE3 overexpression suppresses the growth defect in galactose of galactosemic strains. BY4741, Δgal7 and the isogenic strains harboring the pGRE3 plasmid were plated onto YP medium or YP medium containing either 2% glucose or 2% galactose. Growth in YPD medium was recorded after 1 day and in YP and YP +2% galactose after 4 days at 30°C.

This result suggests that overexpression of GRE3 could suppress the growth inhibition in medium-containing galactose by decreasing the accumulation of Gal-1-P, probably by deviation of galactose metabolism to the formation of galactitol. In order to test this hypothesis, we first compared the total galactose reductase activity in cell-free extracts from wild type, wild-type overexpressing GRE3 (WT-GRE3) and in Δgre3 strains. Results indicate that GRE3 is responsible for about 55% of the total galactose reductase activity in the wild-type strain on comparing wild-type and Δgre3 strains. We also confirmed that overexpression of GRE3 leads to a further increase (1.7-fold) of galactitol reductase activity compared with wild type (Fig. 5).

5

GRE3 converts galactose to galactitol. Galactose reductase activity was measured at pH 5.0 in total protein extracts (40 μg) obtained from WT (BY4741), and the isogenic strains Δgre3 and WT-GRE3 (overexpressing GRE3) grown to the exponential phase in YP containing 2% galactose. Data are presented as a mean±SD of three independent experiments. The statistical result by Student's t-test was: Δgre3 vs. WT, P=0.003 and WT vs. WT-GRE3, P=0.001.

The effect of GRE3 overexpression on galactitol and Gal-1-P accumulation induced by lithium was tested. Cells were pregrown on galactose medium and then treated with 30 mM lithium for 16 h and the galactitol and Gal-1-P accumulation was quantified. In contrast to wild-type cells, where lithium treatment induced the accumulation of both Gal-1-P and galactitol (Fig. 6a and Bro et al., 2003), cells overexpressing GRE3 (WT-GRE3) accumulated intracellular galactitol but not Gal-1-P when treated with lithium (Fig. 6b). It should be noted that overexpression of GRE3 by itself induces accumulation of galactitol independent of lithium (compare Fig. 6a and b). The effect of GRE3 overexpression on galactitol and Gal-1-P accumulation induced by galactose on the Δgal7 strain was also tested. For this purpose, this strain was pregrown on medium containing 2% glucose and then shifted to medium containing 2% galactose for 16 h and the metabolites were determined (Fig. 6c). As expected, Δgal7 cells accumulate more Gal-1-P than WT (Fig. 6a and c). Overexpression of GRE3 led to a reduction of Gal-1-P and induction of intracellular galactitol (Fig. 6c).

6

Overexpression of GRE3 deviates galactose metabolism to galactitol formation. (a and b) Intracellular steady-state levels of Gal-1-P (open bars) and galactitol (gray bars) were measured in cell-free extracts of WT (BY4741) and WT-GRE3 grown to the exponential phase in YP containing 2% galactose (YPgal). Extracts were also measured from half of the cultures to which 30 mM LiCl was added and further incubated for 16 h. (c) Δgal7 and Δgal7-GRE3 were grown to the exponential phase in YPD medium and then incubated for 16 h in YP medium containing 2% galactose, after which intracellular metabolites were measured. The results of Gal-1-P content were statistically significant using Student's t-test in WT+LiCl vs. WT-GRE3+LiCl, P=0.003; Δgal7 vs. Δgal7-GRE3, P=0.028. The results of galactitol content were statistically significant using Student's t-test in: WT+LiCl vs. WT-GRE3+LiCl, P=0.004;Δgal7 vs. Δgal7-GRE3, P=0.007

The decrease in Gal-1-P content induced by GRE3 overexpression could be explained by a downregulation of galactose uptake into the cells. However, the galactose consumption by WT and WT+GRE3 strains during a 24-h growth period was similar to each other (Fig. 7a). Another possibility could be that GRE3 overexpression changes the balance between the two main pathways of galactose utilization in yeast, Leloir and galactitol formation, favoring the latter. In order to verify the precursor relationship between galactose and galactitol, we have grown the Δgal7-GRE3 strain in the presence of 0.5% galactose plus [U14C]-galactose for 20 h and the metabolites from a cell-free extract were analyzed using HPAEC (Fig. 7b). The peak identified as galactitol was quantified using PAD and the radioactivity associated with the peak was measured. It was observed that it contained 48 μM galactitol and 68.5% of the total radioactivity incorporated from galactose by the cells (Table 1). This result shows that galactitol is formed from the galactose obtained from the medium and that it is the main metabolite accumulated inside the cell. Probably, its further conversion is rate limiting in these cells and leads to its accumulation. These results show that the overexpression of the aldose-reductase GRE3 can redirect galactose metabolism to galactitol formation, reducing the accumulation and, consequently, the toxic effect of Gal-1-P formed by the Leloir pathway.

7

Galactose consumption and conversion to galactitol. (a) Growth (○,●) of WT (open symbols) and WT-GRE3 (closed symbols) strains and consumption of galactose (△, ▲) were assayed in a galactose-grown culture at 30°C. (b) Δgal7-GRE3 was grown in glucose to exponential phase and then transferred to 5 mL YP containing 0.5% galactose plus 25 mCi [U-14C]galactose (302.4 mCi mmol−1) and incubated in an aerated bath at 30°C for 20 h. Cells were collected using filtration and a cell-free extract was analyzed using HPAEC.

View this table:
1

HPAEC peaks of cell-free extract from gal7Δ-GRE3 strain

PeakTotal radioactivity (DPM)Total radioactivity (%)
100
200
346951.268.5
49682.214.1
5782.81.1
61616.32.3

An important issue shown on Fig. 6 is that the Δgal7+GRE3 strain grown in galactose accumulates more Gal-1-P than wild type grown in the presence of lithium; however, the former grows in the presence of galactose (Fig. 4) but wild-type growth is inhibited with lithium (Fig. 3). This result seems to show no correlation between Gal-1-P levels and toxicity; however, it must be remembered that lithium inhibits enzymes other than phosphoglucomutase that might exacerbate Gal-1-P toxicity such as IMPases.

Role of IMPases in relieving galactose toxicity

Mammalian IMPases are able to hydrolyze inositol monophosphates and Gal-1-P (Parthasarathy et al., 1997) and overexpression of the human IMPase is able to suppress galactose toxicity in a Δgal7 yeast strain (Mehta et al., 1999). IMPases in S. cerevisiae are encoded by two genes: INM1 and INM2 (also known as IMP1 and IMP2) (Lopez et al., 1999). Their activities are inhibited by lithium with a 50% inhibitory concentration (IC50) of 0.08 mM. Single inm1 or inm2 mutants and the double mutant have no detectable phenotype under normal growth or salt-stress conditions, including lithium stress, when grown in glucose-containing medium (Lopez et al., 1999). The gene INM1 is upregulated in the gal10 mutant treated with galactose (Slepak et al., 2005), while under these conditions GRE3 does not seem to be changed. On the other hand, GRE3 is upregulated in yeast cells grown in medium containing galactose plus 15 mM lithium, while INM1 and INM2 are not (Bro et al., 2003). Thus, it is possible that GRE3 and INM1/INM2 can suppress galactose toxicity in yeast independent of each other.

To test this hypothesis, we first generated a Δinm1/inm2 strain and then transformed it with the plasmid containing the GRE3 gene. Surprisingly, in contrast to the lack of phenotype observed for the Δinm1/inm2 strain grown in glucose-containing medium (Lopez et al., 1999), we observed that this strain is more sensitive to lithium when grown in galactose-containing medium (Fig. 8). Overexpression of GRE3 could still confer lithium tolerance to the Δinm1/inm2 strain growing in galactose-containing medium, indicating that the GRE3 effect is independent from both yeast IMPase genes. However, it is important to note that the tolerance of the Δinm1/inm2 strain overexpressing GRE3 is not higher than that of the WT control (Fig. 8). Because overexpression of GRE3 on the wild-type strain further improves its lithium tolerance (Fig. 3), this result suggests that GRE3 and IMPases could have additive effects. Inhibition of IMPase activity by lithium exacerbate Gal-1-P toxicity and explains the fact that the Δgal7+GRE3 strain has a higher threshold for Gal-1-P toxicity than wild type grown in the presence of lithium, shown in Fig. 6.

8

GRE3 overexpression confers lithium tolerance independent of yeast IMPases. WT(BY4741), Δinm1/2 and Δinm1/2 harboring the pGRE3 plasmid (Δinm1/2+GRE3) were plated onto YP containing 2% galactose in the presence of the indicated LiCl concentrations. Growth was recorded after 4 days at 30°C.

Discussion

We have shown in a previous work that lithium is more toxic to S. cerevisiae when grown in the presence of galactose (Masuda et al., 2001; Bro et al., 2003). In this work, we have observed that the potency of lithium to cause growth arrest has a direct correlation with the galactose concentration in the medium (Fig. 1). We also observed that inhibition of galactose metabolism caused by either the deletion of the transcriptional activator GAL4 or by the deletion of the galactokinase gene GAL1 suppresses lithium-induced growth arrest in the presence of galactose (Fig. 2). These results suggest that the galactose-specific lithium toxicity is dependent on the metabolism of galactose up to the phosphorylation step. Because the interruption of the GALT/GAL7 enzyme, the step immediately after the phosphorylation of galactose in the Leloir pathway, induces galactose toxicity in yeast per se (Douglas & Hawthorne et al., 1964; Segawa & Fukasawa et al., 1979; Ross et al., 2004; Slepak et al., 2005), these results suggest that accumulation of Gal-1-P is cytotoxic under lithium-stress conditions in galactose-growing yeast cells as it is proposed for the Δgal7-deleted yeast strain and classical galactosemia in humans (Slepak et al., 2005; Fridovich-Keil et al., 2006). These results raise an important issue of whether lithium therapy can induce a galactosemic effect in humans. This topic deserves to be explored in more detail in the future.

It has been shown in a galactokinase-deficient mouse model that overexpression of a human aldose reductase induces cataract formation and accumulation of galactitol when these animals are fed with a high-galactose diet. This result supports the cause/effect relationship between galactitol accumulation and cataract formation (Bosch et al., 2006). Inhibitors of aldo-reductases have been suggested for treatment of cataract induced by diabetes, where sorbitol synthesis via reduction of glucose would be inhibited (Ai et al., 2000). This same approach has been suggested to treat cataract induced by galactosemia (Bosch et al., 2006). On the other hand, we have observed that overexpression of the yeast aldose reductase GRE3 relieves the toxic effect of lithium in the presence of galactose and suppresses the galactose-specific growth defect of the gal7- and gal10-deleted strains. This relief is associated with a decrease in the steady-state levels of Gal-1-P and an increase in the galactitol levels accumulated in the cell due to drainage of galactose from the Leloir pathway. These results show that, at least in yeast cells, the activity of aldose reductases is important for relieving galactose toxicity acting as an escape valve for reducing the accumulation of Gal-1-P. Based on these results, treatment of classical galactosemia with aldose reductase inhibitors could decrease galactitol production inhibiting cataract formation but could, at the same time, increase the accumulation of Gal-1-P, leading to the aggravation of other symptoms in a patient. Further work should be carried out to address this issue.

It has been shown previously that deletion of the galactokinase gene in yeast suppresses the gal7- and gal10-derived galactose toxicity (Douglas & Hawthorne et al., 1964; Ross et al., 2004). In this work, we observed that GAL1 deletion can also suppress galactose-induced lithium toxicity in yeast. These results corroborate the hypothesis that the accumulation of Gal-1-P in cells is toxic. It also suggests that inhibitors of galactokinase could alleviate the toxic effects of galactose in yeast, and possibly in patients with classical galactosemia, and could be used as a potential target for treatment of this disease (Bosch et al., 2006).

Mehta (1999) have shown previously that overexpression of a human IMPase gene is able to relieve gal7-derived toxicity in galactose medium. They propose that because IMPase can also use Gal-1-P as a substrate (Parthasarathy et al., 1997), its overexpression would decrease Gal-1-P levels below the toxic threshold. This decrease would then allow the remaining Gal-1-P to be metabolized by an alternative pathway that depends on the UDP glucose/galactose pyrophosphorylase activities of UGP1 and the GALE activity of GAL10 (Mehta et al., 1999). This alternative pathway may also play an important role under the conditions described in our work. When GRE3 is overexpressed in a Δgal7 yeast strain, we observe a reduction in the steady-state levels of Gal-1-P that possibly brings it below the toxic threshold, suppresses the toxicity and enables the metabolism of the remaining Gal-1-P via the UGP1- and GAL10-dependent pathway. However, the participation of other alternative pathways for galactose metabolism cannot be dismissed, as rescue of galactose toxicity by overexpression of GRE3 is not dependent on the presence of GAL10 (Fig. 4) nor INM1 or INM2 (Fig. 8), which are involved in this pathway. A galactose-utilization pathway has been proposed for bacteria and other fungi that includes galactitol formation as an essential step (Szumiło et al., 1981; Fekete et al., 2004; Shakeri-Garakani et al., 2004). Although this pathway has never been reported in S. cerevisiae, our result that overexpression of GRE3 can revert the galactose-specific growth defect of a Δgal10 and Δinm1/inm2 yeast strains is indicative that another pathway, independent of the GALE and IMPase activities, might exist in S. cerevisiae. Further work is ongoing to address whether other galactose utilization pathways are active in S. cerevisiae.

Acknowledgements

This work was supported by grants from Fundação Carlos Chagas Filho de Amparo à Pesquisa do Estado do Rio de Janeiro (FAPERJ) to M.M.L. and C.A.M.; Fundação Universitária José Bonifácio/UFRJ to C.A.M.; PEW Charitable Trust to C.A.M.; and Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) to M.M.L. M.N.M. and L.L.P. are recipients of post-doctoral fellowships and L.J.A. of an undergraduate fellowship from FAPERJ. We acknowledge the Millennium Institute for Vaccine Development and Technology (CNPq, 420067/2005-1).

Footnotes

  • Editor: David Goldfarb

References

View Abstract