OUP user menu

Transport of carboxylic acids in yeasts

Margarida Casal , Sandra Paiva , Odília Queirós , Isabel Soares-Silva
DOI: http://dx.doi.org/10.1111/j.1574-6976.2008.00128.x 974-994 First published online: 1 November 2008

Abstract

Carboxylic acid transporters form a heterogeneous group of proteins, presenting diverse mechanisms of action and regulation, and belonging to several different families. Multiple physiological and genetic studies in several organisms, from yeast to mammals, have allowed the identification of various genes coding for carboxylate transporters. Detailed understanding of the metabolism and transport of these nutrients has become more important than ever, both from a fundamental and from an applied point of view. Under a biotechnological perspective, the increasing economic value of these compounds has boosted this field of research considerably. Here we review the current knowledge on yeast carboxylate transporters, at the biochemical and molecular level, focusing also on recent biotechnological developments.

Keywords
  • membrane transport
  • carboxylic acid
  • transporter
  • yeast
  • metabolic engineering

Introduction

Carboxylic acids comprise a diverse group of organic compounds participating in several cellular processes and being key actors in overall cell functionality. Given their biological significance, intensive work has been carried out in yeasts in an attempt to characterize in detail the mechanisms of transport of these molecules across cell membranes.

As weak organic acids, carboxylic acids partially dissociate in aqueous systems, establishing an equilibrium between undissociated, uncharged molecules and their anionic form(s), according to their pKa and to the pH of the solution. This property influences cell behaviour significantly, especially the mechanisms by which the molecules can cross biological membranes.

Essentially, the transport mechanisms of carboxylic acids can be divided in two main groups: energy-independent (passive) or energy-dependent (active) (Fig. 1). In the first case, simple diffusion and/or facilitated diffusion (a channel or a permease) involves the transport of the uncharged/undissociated form of the acid. At low pH the undissociated form is favoured and, being lipid-soluble, it may cross the cell membrane by simple diffusion. Once inside the cell, the neutral pH leads to the dissociation of the acid. The acid anion cannot readily diffuse out of the cell and accumulates. The protons released acidify the cytoplasm and can alter several metabolic pathways (Krebs et al., 1983; Pampulha & Loureiro-Dias, 1990, 2000; Holyoak et al., 1996). Additionally, the intracellular protons can also influence free radical production, leading to severe oxidative stress, which is a major component of weak organic acid stress found in the yeast Saccharomyces cerevisiae under aerobic conditions (Piper et al., 2001).

Figure 1

Schematic representation of reported mechanisms of transport of carboxylic acids across the yeast plasma membrane. The equilibrium between the undissociated and dissociated form of the acid depends on the medium pH and on the acid pKa(s). The undissociated form (XCOOH) can cross the membrane through passive transport mechanisms: simple diffusion or facilitated diffusion (a permease or a channel). The anionic form (XCOO) enters the yeast cell by permeases acting as secondary active transporters, such as proton symporters. Once inside the cell, and due to the neutral intracellular pH, the acid tends to dissociate and accumulate. The anionic form accumulated in the cytoplasm is exported by a pump, a primary active transport mechanism with ATP cost.

Regarding the energy-dependent processes (Fig. 1), two very distinct systems have been described: pumps that are associated with stress response processes, extruding the anion form of the acid accumulated in the cytoplasm, and permeases, typically associated with metabolic processes that use distinct regulatory networks involved in the utilization of nonfermentable carbon sources.

With the advent of gene-sequencing technologies, an increasing number of transporters have been identified. All these transmembrane solute transporters are currently classified using a nonambiguous classification system, the Transport Classification system (TC; http://www-biology.ucsd.edu/~msaier/transport), based on both function and phylogenic criteria. The current TC distinguishes around 600 phylogenetic families of transporters. Several families were grouped in large superfamilies such as the Major Facilitator Superfamily (MFS, TC 2.A.1), which is ubiquitous and accounts for almost half of the solute transporters. In S. cerevisiae, the first eukaryote organism to have its genome fully sequenced (Goffeau et al., 1996), it is estimated that up to 30% of the proteome consists of membrane proteins (Nelissen et al., 1997; Paulsen et al., 1998; De Hertogh et al., 2002) and nearly 10% are responsible for the transport of small molecules through the plasma membrane (Paulsen et al., 1998; Van Belle & André, 2001). A recent study has determined the existence of 2480 transporters belonging to 204 phylogenetic subfamilies within five complete genomes of the hemiascomycetes yeasts S. cerevisiae, Candida glabrata, Kluyveromyces lactis, Debaryomyces hansenii and Yarrowia lipolytica (De Hertogh et al., 2003).

This review will focus on the most relevant physiological and molecular studies regarding the mechanisms and regulation of carboxylic acid transport across the yeast plasma and mitochondria membranes. We also discuss the role of carboxylate transporters in recent biotechnological advances in winemaking and lactate production.

Yeast carboxylic acid metabolism

Yeasts can use a wide range of nutrients as carbon and energy sources, although monosaccharides, namely glucose and fructose, are preferentially utilized (Barnett, 1976; Gancedo, 1998; Barnett & Entian, 2005). In natural environments and in many industrial applications, the availability of preferable substrates is often scarce. By taking advantage of alternative metabolic pathways that use carbon sources such as ethanol, glycerol, amino acids or carboxylic acids, cells are able to overcome such limitations. According to their ability to use carboxylic acids from the Krebs cycle as sole carbon and energy sources, yeasts can be divided into Krebs-positive or Krebs-negative species (Barnett & Kornberg, 1960).

Saccharomyces cerevisiae is an example of a Krebs-negative species, as it is able to grow on media containing monocarboxylic acids as the only source of biomass and metabolic energy but not on di- or tricarboxylate intermediates of the Krebs cycle. Other examples are Schizosaccharomyces pombe and Zygosaccharomyces bailii, which are able to metabolize malic acid but only in the presence of a fermentable carbon source (Baranowski & Radler, 1984; Osothsilp & Subden, 1986; Sousa et al., 1996). In contrast, species such as Kluyveromyces marxianus, K. lactis, Hansenula anomala, Candida sphaerica and Candida utilis are Krebs-positive species, as they are able to use intermediates of the Krebs cycle as sole carbon an energy source (Zmijewski & Macquillan, 1975; Côrte-Real et al., 1989; Côrte-Real & Leão, 1990; Cássio & Leão, 1993; Queirós et al., 1998).

Glucose regulation in Krebs-negative yeasts

During the course of evolution, S. cerevisiae has, like other microorganisms, developed efficient regulatory systems to allow rapid changes in metabolic pathways. When glucose is available it is preferentially used over other nutrients. This phenomenon, designated by yeast catabolite repression, is achieved in part by inhibition of the synthesis of enzymes required for the utilization of alternative carbon sources, such as those involved in gluconeogenesis, the glyoxylate cycle, the Krebs cycle and the respiratory chain (Barnett, 1976; Gancedo, 1998). Derepression of these metabolic pathways is required for the efficient utilization of acetate, lactate and pyruvate, as well as an active transport system to cross the plasma membrane (Cássio et al., 1987; Casal et al., 1996). Glyoxylate bypass is prominent when cells grow in acetate as this pathway provides carbon compounds necessary for biosynthesis. Lactate is first oxidized to pyruvate by l-lactate cytocrome c oxidoreductase and then decarboxylated by the pyruvate dehydrogenase complex to acetyl-coenzyme A (CoA) (Fig. 2).

Figure 2

The main metabolic pathways of oxidative and gluconeogenic metabolism in Saccharomyces cerevisiae. Some genes involved in the utilization of nonfermentable carbon sources are indicated: STL1, glycerol proton symporter (Ferreira et al., 2005); JEN1, lactate transporter; ADY2, acetate transporter; ALD2, aldehyde dehydrogenase; GUT1, glycerol kinase; GUT2, mitochondrial glycerol-3-phosphate dehydrogenase; FBP1, fructose-1,6-bisphosphatase; PFK1,2, phosphofructokinase; PCK1, phosphoenolpyruvate carboxykinase; DLD1,3, d-lactate dehydrogenase; CYB2, l-lactate cytochrome-c oxidoreductase; ACS1,2, acetyl-CoA synthetase; ICL1, isocitrate lyase; MLS1, malate synthase; PDC1,5,6, pyruvate decarboxylase; PYC1,2, pyruvate carboxylase.

Recent developments in genomic and proteomic techniques have uncovered the mechanisms of glucose signal transduction in S. cerevisiae, revealing the role of this sugar as a regulator in addition to its importance as a nutrient (for reviews see Gancedo, 1998; Carlson, 1999). Snf1p, a protein kinase, is the key element for glucose repression by regulating the activity of transcription factors required for the derepression of most genes downregulated by glucose. In the presence of glucose Snf1p is inactive, becoming active when glucose is limiting. The best characterized transcription factor involved in glucose repression is Mig1p, which binds to the promoters of a variety of genes and stops their transcription (Ostling et al., 1996; Wilson et al., 1996). When Snf1p is active, Mig1p is inactivated by phosphorylation and is delocalized from the nucleus to the cytoplasm (De Vit et al., 1997).

The Snf1p protein kinase also regulates Cat8p, a zinc cluster transcriptional activator that is essential for growth of yeast on nonfermentable carbon sources (Hedges et al., 1995). When glucose is absent, Cat8p is phosphorylated (Rández-Gil et al., 1997), and functions as a positive regulator for transcription of nonfermentable carbon-source-associated genes, such as the gluconeogenic enzymes, fructose-1,6-biphosphatase, phosphoenolpyruvate carboxykinase and isocitrate lyase (Hedges et al., 1995). Expression of the CAT8 gene is subject to glucose repression, mainly caused by Mig1p, and is upregulated by transcription factors Hap2/3/4/5, a global activator complex of respiratory gene expression (Rández-Gil et al., 1997).

When glucose is consumed, S. cerevisiae starts using respiratory substrates such as ethanol as a carbon source for aerobic growth (Fig. 2), leading to a diauxic growth curve. In the diauxic shift the mRNA level of c. 700 genes increases, while the mRNA level of c. 1000 genes declines (De Risi et al., 1997). Cat8p is highly involved in the reprogramming of carbon metabolism during the diauxic shift (Haurie et al., 2001). The highly induced genes comprise those coding for aldehyde dehydrogenase (ALD2) and acetyl-CoA synthase (ACS1,2), to produce acetyl-CoA. By contrast, pyruvate decarboxylase (PDC1,5,6) is decreased and pyruvate carboxylase (PYC1,2) is induced, redirecting pyruvate to oxaloacetate, which supplies the Krebs cycle and gluconeogenesis.

Krebs-positive vs. Krebs-negative yeasts

The Krebs-positive species C. utilis (Cássio & Leão, 1993), C. sphaerica (Côrte-Real et al., 1989), H. anomala (Côrte-Real & Leão, 1990), K. lactis (Zmijewski & Macquillan, 1975) and K. marxianus (Queirós et al., 1998) can use malic acid and other Krebs cycle intermediates as sole carbon and energy source. The Krebs-negative species S. cerevisiae (Rodriguez & Thornton, 1990), Z. bailii (Baranowski & Radler, 1984) and S. pombe (Osothsilp & Subden, 1986) can consume malic acid only in the presence of glucose or another assimilable carbon source. In contrast to dicarboxylate transporters found in Krebs-negative species, most found in Krebs positive are glucose repressed, presenting high affinity for the substrates, and can transport intermediates of the tricarboxylic acid cycle (Table 1) (Camarasa et al., 2001).

View this table:
Table 1

Transport systems for carboxylic acids in yeast

SpeciesSpecificityMechanism of uptake/regulationReference(s)
Monocarboxylate permeases
C. albicansd-Lactic, l-lactic, pyruvic and propionic acidProton symport/induced and glucose repressedSoares-Silva et al. (2004)
C. utilisd-Lactic, l-lactic, pyruvic, acetic and propionic acidProton symport/induced and glucose repressedLeão & van Uden (1986), Gerós et al. (1996)
D. anomalaAcetic, propionic, formic and sorbic acidProton symport/induced and glucose repressedGerós et al. (2000)
K. lactisLactic and pyruvic acidActive transport/strain dependent glucose regulationLodi et al. (2004), Queirós et al. (2006)
K. marxianusd-Lactic, l-lactic, pyruvic, acetic, propionic, glycocolic and bromoacetic acidUniport/partially induced and glucose repressedFonseca et al. (1991)
S. cerevisiaed-Lactic, l-lactic, pyruvic, acetic and propionic acidProton symport/induced and glucose repressedCássio et al. (1987)
Acetic, propionic and formic acidProton symport/induced and glucose repressedCasal et al. (1996)
T. delbrueckiid-Lactic, l-lactic, pyruvic, acetic, formic and propionic acidProton symport/induced and glucose repressedCasal & Leão (1995)
Acetic, formic and propionic acidProton symport/induced and glucose repressedCasal & Leão (1995)
Z. bailiiAcetic, propionic and formic acidProton symport/induced and glucose repressedSousa et al. (1996)
Acetic acidProton symport/constitutiveSousa et al. (1996)
Dicarboxylate permeases
C. sphaerical-Malic, d-malic, succinic, fumaric, oxaloacetic and α-ketoglutaric acidProton symport/induced and glucose repressedCôrte-Real et al. (1989)
C. utilisl-Malic, d-malic, succinic, fumaric and α-ketoglutaric acidProton symport/induced and glucose repressedCássio & Leão (1993), Saayman et al. (2000)
H. anomalal-Malic, d-malic, succinic, fumaric, oxaloacetic, α-ketoglutaric, malonic and maleic acidProton symport/induced and glucose repressedCôrte-Real & Leão (1990)
K. lactisl-Malic, succinic, fumaric and α-ketoglutaric acidActive transport/strain dependent glucose regulationZmijewski & Macquillan (1975)
K. marxianusl-Malic, d-malic, succinic and fumaric acidProton symport/induced and glucose repressedQueirós et al. (1998)
S. pombel-Malic, d-malic, succinic, oxaloacetic, malonic and maleic acidProton symport/constitutiveOsothsilp & Subden (1986), Camarasa et al., (2001); Grobler et al., (1995); Sousa et al., (1992); Saayman et al. (2000)
Z. bailiil-Malic and d-malic acidFacilitated diffusion for nondissociated form/induced by glucose and repressed by fructoseBaranowski & Radler (1984)
Tricarboxylate permease
C. utilisCitric and isocitric acidProton symport/induced and glucose repressedCássio & Leão (1991)
General permease
C. utilisd-Lactic, l-lactic, pyruvic, l-malic, d-malic, fumaric, α-ketoglutaric, succinic, glycine and glutamic acidFacilitated diffusion/induced and glucose repressedCássio & Leão (1991, 1993)

Concerning dicarboxylic acid metabolism in Krebs-negative yeasts, Z. bailii and S. pombe can efficiently degrade malic acid in the presence of glucose, whereas S. cerevisiae consumes only a small amount of malic acid during wine must fermentation. This feature has been attributed to the absence of a specific transporter, and to the low effectiveness of metabolism in S. cerevisiae (Kuczynski & Radler, 1982; Salmon, 1987; Radler, 1993). The acid enters the cell, being decarboxylated to pyruvate (Rodriguez & Thornton, 1990; Radler et al., 1993) by the NAD+-dependent malic enzyme. Very distinct Km values for malic enzyme have been estimated, 10 mM for S. pombe, 3.2 mM for Z. bailii and 50 mM for S. cerevisiae (Fuck et al., 1973; Kuczynski & Radler, 1982; Viljoen et al., 1994, 1999; Boles et al., 1998). Pyruvate is then oxidatively decarboxylated to acetaldehyde, which is reduced to ethanol. Malic acid can also be transformed into succinate, via fumarate (Fig. 3) (Kuczynski & Radler, 1982).

Figure 3

Yeast metabolic pathways for malic acid utilization.

Schizosaccharomyces pombe cells cultivated in a medium containing malic acid are unable to perform gluconeogenesis as under these conditions the malic enzyme is not present to convert malate to pyruvate, and cell growth is thus impaired (Viljoen et al., 1999). In S. pombe the mae2 gene encodes the malic enzyme which, as in S. cerevisiae (MAE1/YKL029c) (Boles et al., 1998), is expressed in anaerobic glucose-grown cells (Viljoen et al., 1994). However, whereas in S. pombe the malic enzyme is located in the cytosol, in S. cerevisiae Mae1p is located in the mitochondria. The role of the malic enzyme in S. cerevisiae has not yet been clarified, although it has been postulated that it supplies pyruvate during growth on ethanol and acetate (Boles et al., 1998).

More recently, in the Krebs-positive yeast C. utilis, a malic enzyme gene CME1 was cloned and shown to be subjected to glucose repression and substrate induction (Saayman et al., 2006). It is predicted to be located in the cytoplasm and its amino acid sequence shows homology to the malic enzymes of S. pombe and S. cerevisiae.

Carboxylic acid uptake in yeasts

The mechanisms involved in the uptake of weak organic acids in several yeast species are summarized in Table 1.

Uptake of monocarboxylic acids

No work had been published on membrane transport of lactic acid in yeasts, until Leão & van Uden (1986) reported a study on C. utilis in 1986. Lactic acid-grown cells of this yeast display activity for an accumulative electroneutral proton-monocarboxylate symport, shared by l-lactate, d-lactate, pyruvate, propionate and acetate, inducible and subjected to glucose repression. Similar transporters were later described in the plasma membrane of other yeasts such as S. cerevisiae, Torulaspora delbrueckii, Candida albicans and K. lactis (Cássio et al., 1987; Casal & Leão, 1995; Lodi et al., 2004; Soares-Silva et al., 2004; Queirós et al., 2006). The only exception was found in lactic acid-grown cells of K. marxianus where a monocarboxylate uniporter is present (Fonseca et al., 1991). In these reports it was demonstrated that undissociated lactic acid enters the cells by simple diffusion. For all these cases the presence of a proton symporter mechanism for the dissociated form of the acid rather than a facilitated diffusion was confirmed by a range of evidence. The evaluation of kinetic parameters at different extracellular pH demonstrated that the Km values when expressed as the concentration of undissociated acid varied more than 100-fold between pH 3.0 and 6.0 whereas a much lower variation was found for the undissociated acid. Additionally, above pH 6.0 the amount of the undissociated form of the acids is negligible and thus the observed kinetics could not be assigned to these molecules. The presence of a proton symporter mechanism is further supported by the effects of the protonophore carbonyl cyanide m-clorophenylhydrazone (CCCP) in the accumulation and initial uptake rates of monocarboxylic acids, and also the observed proton movements during the initial uptake of the acids that follow Michaelis–Menten kinetics.

Saccharomyces cerevisiae cells growing in acetate or ethanol display activity for another monocarboxylate permease with a different pattern of regulation and specificity. The system is also a proton symporter, although it is shared by acetate, propionate and formate but not by lactate or pyruvate (Casal et al., 1996). Similar permeases were identified in T. delbrueckii, Z. bailii and Dekkera anomala (Casal & Leão, 1995; Sousa et al., 1996; Gerós et al., 2000), although the D. anomala permease also accepts sorbate. Additionally, in Z. bailii a proton symporter specific for acetic acid is operating even in glucose-grown cells (Sousa et al., 1996).

Uptake of di- and tricarboxylic acids

A different group of transporters is involved in the uptake of dicarboxylates. The first carboxylate transporter reported was found in K. lactis (Zmijewski & Macquillan, 1975), which is shared by malate, succinate, fumarate and α-ketoglutarate. Since then, additional dicarboxylate transporters have been described for other yeasts, namely C. sphaerica (Côrte-Real et al., 1989), H. anomala (Côrte-Real & Leão, 1990), C. utilis (Cássio & Leão, 1993) and K. marxianus (Queirós et al., 1998). In all these species the dissociated form is transported across the plasma membrane through a proton symport mechanism, which is induced by the substrate and subjected to glucose repression, except for K. lactis where glucose repression is strain dependent (Queirós et al., 2006). Schizosaccharomyces pombe and Z. bailii also have dicarboxylate permeases (Baranowski & Radler, 1984; Osothsilp & Subden, 1986) but they present a different regulatory pattern, requiring the presence of an assimilable carbon source such as glucose, as both are Krebs-negative species (Baranowski & Radler, 1984; Viljoen et al., 1999). In Z. bailii dicarboxylates are transported by a facilitated diffusion mechanism for the undissociated acid whereas S. pombe presents a dicarboxylate/proton symporter (Baranowski & Radler, 1984; Osothsilp & Subden, 1986).

Few tricarboxylate transporters have been characterized thus far in yeast. A low- and a high-affinity transport system subjected to glucose repression was identified in C. utilis (Cássio & Leão, 1991). The high-affinity transporter is specific for citrate and isocitrate and behaves as a proton symporter. A facilitated diffusion mechanism was described for the low-affinity system, transporting a broad range of substrates including mono-, di- and tricarboxylates, as well as glycine and glutamate (Cássio & Leão, 1991, 1993).

Genes coding for carboxylate permeases in yeasts

Despite the vast amount of physiological data accumulated on yeast carboxylate transporters over several decades, the genes coding for transporters remained unknown for a long time. Because of the importance of lactate and pyruvate uptake/extrusion in mammalian metabolism, the first carboxylate transporter genes that have been studied belong to the mammalian monocarboxylate porter (MCP) family (Kim et al., 1992; Garcia et al., 1995). We next focus on carboxylate transporter encoding genes identified thus far in distinct yeast species, listed in Table 2. A short review on the MCP family is also given later in this paper.

View this table:
Table 2

Genes encoding carboxylate transporters in yeasts

Gene nameORF or accession no.Carrier substratesReference(s)
Monocarboxylate permeases
ScJEN1YKL217wLactate/pyruvateCasal et al. (1999)
KlJEN1XM_454682.1Lactate/pyruvateLodi et al. (2004), Queirós et al. (2006)
CaJEN1XM_711015.1Lactate/pyruvateSoares-Silva et al. (2004)
ScADY2YCR010cAcetatePaiva et al. (2004)
Dicarboxylate permeases
KlJEN2XM_455537.1Succinate/malateLodi et al. (2004), Queirós et al. (2006)
SpMae1NM_001020205Succinate/malate/malonateGrobler et al. (1995)
Mitochondrial carriers
CTP1YBR291cCitrateKaplan et al. (1995)
SFC1YJR095wSuccinate/fumaratePalmieri et al. (1997)
OAC1YKL120wOxaloacetatePalmieri et al. (1999)
DIC1YLR348cDicarboxylateKakhniashvili et al. (1997)
ODC2YOR222wOxidicarboxylatePalmieri et al. (2001)
ODC1YPL134cOxidicarboxylatePalmieri et al. (2001)
Efflux pumps
PDR12YPL058CSorbate, benzoate, propionate, acids resultant from amino acid catabolismPiper et al. (1998), Bauer et al. (2003)
Channel
FPS1YLL043WAcetateMollapour & Piper (2007)

Monocarboxylate permeases

The lactate/pyruvate permease Jen1p

The first gene encoding a fungal monocarboxylate permease was identified in S. cerevisiae. The complementation of a mutant unable to transport lactate revealed that the JEN1/YKL217w gene codes for the lactate/pyruvate permease (Casal et al., 1999); it belongs to the MFS (TC 2.A.1.12.2) and contains 12 putative transmembrane segments (TMSs). A Jen1–green fluorescent protein (GFP) fusion showed that, upon induction, the protein was localized at the plasma membrane (Paiva et al., 2002). The function of Jen1p was questioned for some time, as Jen1p has no homology to any monocarboxylate permeases described in mammals or to the Escherichia coli LctP (TC 2.A.14.1.1) lactate permease. In contrast, it has homology to the E. coli sialic acid transporter NanT (TC 2.A.13.1.1), suggesting that it could be a regulator rather than a permease (Halestrap & Price, 1999; Makuc et al., 2001). Sialic acids are monocarboxylated monosaccharides, normally present in mammal sugar side chains of glycoproteins, glycolipids and glycosaminoglycans whose carboxylic group is deprotonated at physiological pH (Vimr et al., 2004). These doubts were dissipated by the heterologous overexpression of Jen1p in Pichia pastoris, allowing the measurement of lactate transport in reconstituted plasma membrane vesicles (Soares-Silva et al., 2003). These results demonstrated that Jen1p is a permease and confirmed the involvement of the proton motive force as well as the specificity of permease substrates determined in S. cerevisiae cells.

Glucose repression acts at different levels in JEN1 regulation. In the presence of glucose there is no detection of JEN1 gene mRNA (Casal et al., 1999), as the transcription factors Mig1p and Mig2p, involved in glucose repression, downregulate the JEN1 gene (Bojunga & Entian, 1999). Glucose regulation also acts at the posttranscriptional level. Addition of a pulse of glucose to cells expressing the JEN1 gene promotes an increase in mRNA degradation, resulting in a decrease of JEN1 mRNA half-life from 15 min to 3–4 min (Andrade & Casal, 2001; Andrade et al., 2005). A mechanism was proposed for the glucose-accelerated JEN1 mRNA decay, in which a second transcript of JEN1 starting at position +391 was correlated with the glucose-triggered mRNA degradation. In a strain overexpressing the +391 JEN1 transcript, JEN1 mRNA half-life is always diminished, even in the absence of glucose. Therefore, Andrade et al. (2005) proposed that the alternative JEN1 transcript is a glucose sensor that, in the presence of even small amounts of glucose, accelerates JEN1 mRNA decay. In addition to Mig1p and Mig2p, other proteins regulate the transcriptional level of the JEN1 gene. Cat8p is necessary for full derepression during a shift from fermentative to respiratory growth (Bojunga & Entian, 1999). The Hap2/3/4/5 complex is necessary for derepression of the JEN1 gene in medium containing lactate (Lodi et al., 2002). The protein kinase Snf1p, which plays an essential role in the release of glucose-repressible genes from glucose repression (Carlson et al., 1999), is also involved in the control of JEN1 transcription (Lodi et al., 2002).

Regulation also occurs at the protein level. Addition of glucose to cells expressing a Jen1–GFP fusion at the plasma membrane promotes the rapid internalization of the protein (Paiva et al., 2002), in an End3p-dependent manner, and subsequent targeting to the vacuole for degradation. Like many plasma membrane proteins, downregulation of Jen1–GFP fusion protein requires its conjugation to ubiquitin (Paiva et al., 2002).

Jen1p structure/function studies have been performed. An alignment among hemiascomycetes JEN1 homologues showed the existence of the conserved domain 379NXX[S/T]HX[S/T]QDXXXT391, located in the seventh TMS (Fig. 4) (Soares-Silva et al., 2007). Mutation of the conserved residues revealed the involvement of this region both in transport capacity and in substrate affinity. Substitution in amino acid residues N379, H383 or D387 resulted in a major reduction of lactate and pyruvate uptake, but conserved measurable acetate transport. Mutants Q386 and T391 resulted in no or moderate changes in Jen1p transport capacities for lactate, pyruvate and acetate, although modification in the affinity for acids was observed. The existence of a charged interaction between amino acids H383 and D387 contributing to Jen1p protein structure was evidenced, as the double mutation H383D/D387H while behaving as a total loss-of-function allele for lactate and pyruvate uptake recovers the kinetic parameters of Jen1p for acetate transport. The conserved sequence 379NXX[S/T]HX[S/T]QDXXXT391 is also found in Jen1p homologues of archea, bacteria and filamentous fungi species (Soares-Silva et al., 2007). Although none of these proteins has ever been functionally characterized, the presence of this conserved domain points to their role as carboxylate transporters.

Figure 4

Conserved domain 379NXX[S/T]HX[S/T]QDXXXT391 of ScJen1p. Multiple sequence alignment of yeast Jen1p homologues available in the databases and the Escherichia coli sialate transporter (NanT) was built via the clustalw program (Corpet et al., 1988). The underlined amino acids correspond to the seven TMS according to modulation studies.

A structural model for Jen1p was constructed by homology threading based on the LacY structure, demonstrating that the amino acids identified in this domain are facing the Jen1p internal pore and have the potential to interact with substrate and cosubstrate. The overall approach combining bioinformatic tools, multiple alignment and three-dimensional modelling, with site-directed mutagenesis was adequate for the identification of domains crucial for structure/function relationships in this permease (Soares-Silva et al., 2007).

JEN1 homologous genes have been described for several species, such as C. albicans, Metarhizium anisopliae and K. lactis (Fang et al., 2003; Lodi et al., 2004; Soares-Silva et al., 2004). CaJen1p is also a monocarboxylate/proton symporter, glucose repressed and requires the presence of an inducer to be expressed. It has substrate specificity similar to ScJEN1, with the exception of acetate, which behaves as a noncompetitive inhibitor, thus suggesting an independent uptake system, possibly using common proton cosubstrates. Confirmation of CaJen1p transport function was done by heterologous expression in an S. cerevisiae jen1Δ strain. Due to the altered C. albicans genetic code (Santos et al., 1993) CaJen1p was only functional in S. cerevisiae after site-directed mutagenesis of S217 to restore the original protein sequence. Additionally, CaJEN1 transcription is activated by CaCAT8. Recently, it was shown that the CaJEN1 gene and its homologue CaJEN2 are upregulated, 5.5- and 159.5-fold, respectively, during phagocytosis (Piekarska et al., 2006), as well as the l-lactate dehydrogenase enzyme (Cyb2). So far the function of CaJEN2 remains to be determined but it is likely to be involved in the uptake of carboxylic acids as it exhibits 43% identity to CaJEN1.

The functional analysis of two JEN1 homologues found in K. lactis showed the existence of mono- and dicarboxylate permeases (Lodi et al., 2004). The heterologous expression of both KlJEN1 and KlJEN2 in a S. cerevisiae jen1Δ strain confirmed that the KlJEN1 codes for a monocarboxylate transporter shared by lactate and pyruvate (Lodi et al., 2004; Queirós et al., 2006). Information regarding KlJEN2 will be provided further below.

A study on the evolution of Jen1 homologue transporters reported the existence of 35 putative JEN1 orthologues in 17 different fungi, 13 hemiascomycetes and seven euascomycetes (Lodi et al., 2007). A third phylogenic Jenp species called PreJen1 has been proposed among older hemiascomycetes such as D. hansenii and Y. lipolytica, the latter containing up to six JEN genes of undetermined function.

Using the information that resulted from the complete sequencing of various yeast genomes a phylogenetic tree is presented in Fig. 5 containing Jen1p orthologues and paralogues, found in selected yeast species, relevant from a biotechnological/biomedical point of view. Two clusters are clearly distinguished, Jen1 and Jen2. The proteins functionally characterized as monocarboxylate transporters are grouped in Jen1 cluster, whereas the only dicarboxylate transporter yet described, KlJen2, is in a distinct branch. This approach allows the prediction of the functional role of uncharacterized transporters. Interestingly, in Y. lipolytica six homologues are found forming a distinct group with totally unknown function.

Figure 5

Phylogenetic tree of ScJen1p homologues. The sequences used were obtained from Génolevures, except for the Pichia stipitis and Candida albicans homologues which were obtained by homology search with blastp (http://www.ncbi.nlm.nih.gov/blast/Blast.cgi). Accession numbers are indicated in Fig. 4. The permeases with characterized function are highlighted in bold, and are detailed further in Table 2.

The acetate permease Ady2p from S. cerevisiae

Despite the vast amount of physiological and biochemical data accumulated on acetate metabolism, the genes coding for acetate permeases were only recently discovered. In E. coli the acetate permease, ActP, belongs to the solute: sodium symporter subfamily (TC 2.A.21.7.2) (Gimenez et al., 2003) and has 14 predicted TMSs. It is highly specific for the transport of short-chain aliphatic monocarboxylates, and to a lesser degree for l-lactate, glycolate and oxalate (Gimenez et al., 2003). Several attempts were made towards the identification of the gene encoding the acetate permease in S. cerevisiae (Paiva et al., 1999; Makuc et al., 2001) first unveiled kinetically by Casal et al. (1996). A comparison between the gene expression profile of glucose-grown cells and cells shifted to a medium containing acetate as the sole carbon and energy source was performed (Paiva et al., 2004). In addition to general responses such as the derepression of gluconeogenic genes and the induction of the glyoxylate and Krebs cycles, previously unreported responses were described that include the activation of the translation machinery, rRNA maturation and mitochondria biogenesis. Many carrier genes were shown to be upregulated, including three genes encoding putative membrane proteins that belonged to the same YaaH family (TC 9.B.33), namely YCR010c/ADY2, YNR002c/FUN34 and YDR384c/ΔT03. These genes are homologous to the GPR1 gene of Y. lipolytica, predicted to have five to seven TMSs. The deletion of YlGPR1 was reported to cause a defect in acetic acid utilization (Tzschoppe et al., 1999). Acetate uptake was measured in deletion mutants for ADY2, FUN34 or ATO3 to assess their role in acetate transport. Only a mutant deleted in ADY2 was strongly affected in the ability to transport acetate, whereas the deletion of the two other GPR1-like genes had no significant effect compared with the wild-type strain (Paiva et al., 2004). In acetic acid-grown cells, the deletion of ADY2 abolished the uptake of acetate and only simple diffusion of the acid could be detected. In a second set of experiments, microarray analysis of the S. cerevisiae ady2Δ strain demonstrated that this gene is not involved in a global transcriptional response, neither to acid stress nor to the adaptation to a poor carbon source such as acetate. These results fit with the activity profile of the unknown acetate permease (Casal et al., 1996), thus indicating that it codes for an acetate transporter (Paiva et al., 2004). ADY2 was first described as being involved in other cellular processes, namely proper ascus formation on sporulation medium, containing acetate as the main carbon source (Rabitsch et al., 2001).

In several global expression studies a similar expression profile for ADY2 and JEN1 was observed. These include the activation of transcriptional response of S. cerevisiae grown under limiting conditions for carbon, nitrogen, phosphorous or sulphur, the activation of gene expression of yeast grown in oleate vs. glucose and in different development stages of colonies (Kal et al., 1999; Boer et al., 2003; Vachova et al., 2004). This similarity is also observed in terms of regulation. During the diauxic shift the expression profiles of ADY2 and JEN1 are very similar and both require Cat8p for expression under these conditions (Haurie et al., 2001). Adr1p, another regulator of a large number of genes involved in the metabolism of nonfermentable carbon sources, also binds to the promoter of both genes (Young et al., 2003). Cat8p binding to the JEN1 or ADY2 promoters is Adr1p independent (Young et al., 2003) and full expression of both genes requires the presence of the Snf1 protein kinase (Tachibana et al., 2005).

Palkova's laboratory found that the expression of YCR010c/ADY2, YNR002c/FUN34 and YDR384c/ΔT03 was concomitant with ammonia production and alkalinization of external medium during S. cerevisiae colony development. The corresponding deleted strains show a lower ammonium export. Based on this fact the authors proposed that these proteins are ammonium/H+ antiporters, extruding ammonium from the cell and importing protons (Palkova et al., 2002).

The true function of ADY2 is still controversial, although data provided by functional analysis and the similarity of regulation patterns and expression profiles with the other S. cerevisiae monocarboxylate transporter, JEN1 gene, supports its role as an acetate permease. Very recently, reinforcing this hypothesis the protein AcpA, orthologous to Ady2p, was described as an acetate transporter in the hyphal fungus Aspergillus nidulans (Robellet et al., 2008).

Dicarboxylate permeases

The first cloning and functional analysis of a gene encoding a carboxylate permease in yeasts was that of Mae1, which codes for a dicarboxylate carrier in S. pombe (Grobler et al., 1995). It corresponds to a 49-kDa protein with 10 transmembrane predicted segments, that has been classified in the TDT family of telurite and dicarboxylate transporters (TC 2.A.16). The Mae1 gene encodes a permease for malate and other C4 dicarboxylic acids and behaves as a proton symporter not subjected to glucose repression (Sousa et al., 1992).

Physiological characterization of a S. cerevisiae strain transformed with the S. pombe mae1 gene showed that the monoanionic form of malic acid is actively transported (Camarasa et al., 2001). The transport mechanism is reversible, accumulative and dependent both on the transmembrane gradient of the substrate and on the ΔpH component of the proton motive force (Osothsilp & Subden, 1986; Sousa et al., 1992; Grobler et al., 1995). Maleic, oxaloacetic, malonic, succinic and fumaric acids inhibit malate transport, suggesting that these compounds share the same carrier. Previously, Saayman et al. (2000) demonstrated that although fumarate and α-ketoglutarate bind to the carrier, they are not actively transported into the cell, neither in S. pombe nor in the S. cerevisiae recombinant strain, in contrast to what is observed with malate (Grobler et al., 1995; Camarasa et al., 2001).

In K. lactis, KlJen2p is required for succinate and d,l-malate transport (Lodi et al., 2004; Queirós et al., 2006). The uptake of succinate is inhibited by fumarate, thus indicating that KlJen2p also binds this substrate. The heterologous expression of KlJEN2 enables S. cerevisiae to uptake succinic and malic acids. The function of this gene was further confirmed by phenotypic analysis of the disrupted mutants and KlJen2–GFP fusion analysis revealed that the chimeric protein is located in the cellular plasma membrane (Queirós et al., 2006). Glucose repression of KlJEN2, as well as KlJEN1, is strain dependent. In K. lactis GG1888, expression of these genes does not require an inducer and both genes were expressed in glucose-grown cells; the opposite occurs in K. lactis PM4-4B where there is no expression of these genes when cells are grown in glucose-containing medium (Lodi et al., 2004; Queirós et al., 2006).

The monocarboxylate porter family

Higher eukaryotes, such as mammals, display diverse carboxylate transporter systems, which exhibit different properties with respect to substrate specificity, affinity, capacity and regulation. For example, red blood and tumour cells depend on the production of lactic acid from glucose to obtain most of the ATP needed under normal physiological conditions. Other tissues are also dependent on this pathway under hypoxia or ischaemia (Halestrap & Price, 1999). Under high rates of glycolysis, cellular lactate accumulates, and there is a need to remove it to prevent intracellular acidification and keep up with the velocity of lactic acid production. On the other hand, lactic acid is rapidly imported in tissues where it may become a major respiratory fuel or gluconeogenic substrate, such as in the liver during the Cori cycle (Garcia et al., 1994). Other monocarboxylates that have important roles in a similar metabolic context include pyruvate, acetate and ketone bodies (Poole & Halestrap, 1993).

In mammals, the transport of lactate in erythrocytes was shown to be specifically inhibited by α-cyano-4-hydroxycinnamate and organomercurials. Jennings & Adamslackey (1982) showed that the transport of lactate obeys Michaelis–Menten kinetics and Poole & Halestrap (1992) reconstituted transport activity of the solubilized protein. The protein was named monocarboxylate transporter 1 (MCT1), and its N-terminal was shown to be identical to that of a putative transporter previously cloned from Chinese hamster ovary cells (Kim et al., 1992). The next carboxylate transporter-encoding gene to be cloned was isolated from hamster liver, and named MCT2 (Garcia et al., 1995). It showed 60% identity with MCT1 and was heterologously expressed in insect cells. Study of the X-inactivating sequence of the X chromosome identified another gene, from the X chromosome, with 27% identity to the human MCT1, designated XPTC (Lafreniere et al., 1994). MCT3 was studied in chicken retinal pigment epithelium and its function was confirmed by heterologous expression in a thyroid epithelial cell line (Yoon et al., 1997). Four human MCT-related sequences were identified and cloned from a human cDNA library (Price et al., 1998), which were named MCT3MCT6, whereas MCT9 was revealed in a human expressed sequence tag database (Halestrap & Price, 1999).

An aromatic amino acid transporter TAT1 that belongs to the MCT family was later identified (Kim et al., 2001). It transports amino acids, but not lactate or pyruvate, through a facilitated diffusion mechanism. Thus, this transporter differs from the other MCTs both for substrate specificity and for transport mechanism, as the uptake is not driven through an H+ gradient. More recently, four new genes that belong to the MCT family, MCT11, MCT12, MCT13 and MCT14, were discovered through homology search (Halestrap & Meredith, 2004). The authors also reported unpublished results concerning Mct8p, which is close to Tat1, and was found to transport thyroid hormones T4 and T3, but not aromatic amino acids, or lactate.

The MCT members belong to the MFS, displaying 10–12 predicted TMS, with their C- and N-terminus located in the cytoplasm, and a loop between TMS 6 and 7 (Poole et al., 1996) and, according to the TC system, belong to the MCP family (TC 2.A.1.13).

Additional MCT homologues were found in S. cerevisiae, Caenorhabditis elegans and Sulfolobus solfataricus and led to the identification of a new subfamily of transporters containing both eukaryote and prokaryote members (Price et al., 1998).

In S. cerevisiae a family named monocarboxylate permease homologues (Mch) comprising five yeast proteins related to the MCT family of animal cells was revealed upon genome sequencing and homology searching: YOR306c, YOL119c, YKL221W, YNL125c and YDL054c (De Hertogh et al., 2002). A compilation of these studies can be consulted in the Yeast Transport Protein database (YTPdb; http://mips.gsf.de/proj/eurofan/eurofan_1/n6/index.html) and in the Yeast Transporter Information (YETI; http://www.yetibio.com/).

A functional analysis of the five Mchp was compared with Jen1p and with the ORF YHL008c product (Makuc et al., 2001), which was proposed to code for an acetate-proton symporter as it resembles the bacterial formate–nitrite transporters (Paulsen et al., 1998). It was found that the Mch members were not involved in the uptake or secretion of monocarboxylates by S. cerevisiae, being neither induced by lactate nor repressed by glucose. In fact, the five members of the MCP family were downregulated in cells grown in medium containing acetate as sole carbon and energy source (Makuc et al., 2001). It was only recently that the true function of one of the Mch members, Mch5p, a facilitator of riboflavin (vitamin B2), was determined experimentally (Reihl & Stolz, 2005).

Although they have a common ancestor, the yeast Mchp do not transport the same substrates as their mammalian MCT equivalents, which transport negatively charged molecules such as lactate, pyruvate and the thyroid hormones. In contrast, Mch5p transports a neutral substrate. Moreover, the majority of mammalian MCT members are proton symporters whereas Mch5p is a facilitator. Figure 6 shows the phylogenetic relationships between the two clades comprising the human MCT members, presently 14, and the S. cerevisiae Mch. The MTC/Mch divergence is here displayed: within the MCT cluster, the four genes (SLC16A1, SLC16A3, SLC16A7, SLC16A8) that are known to be proton cotransporters are in two adjacent subclades, and the two genes presenting a nonmonocarboxylate substrate specificity, MCT8 (SLC16A2) and TAT1 (SLC16A10), cluster in the more distant subclade. The S. cerevisiae Mch form a different clade unrelated to the other members.

Figure 6

Unrooted tree representing the phylogenetic relationship between the MCTs and the Mch. The corresponding amino acid sequences were analysed using clustalw for sequence alignment and phylip for tree plotting (http://evolution.genetics.washington.edu/phylip.html). The Human Genome Organization (HUGO) Nomenclature Committee Database includes more than 40 transporter families of the SLC (Solute Carrier) gene series. The SLC families represent a major portion of the transporter-related genes and additional SLC transporters are constantly being identified (Hediger et al., 2004).

Recently, the first A. nidulans MCT homologue AmcA and two other paralogues, AmcB and AmcC, were identified (Semighini et al., 2004). AmcA has 31% identity to Mch5 and 29% to Mch4 from S. cerevisiae and a topology similar to the members of the MCP family. Although its function remains to be determined the expression of AmcA is increased in medium containing acetate or pyruvate as single carbon source, thus presenting a more similar behaviour to the MCT members.

Carboxylate transporters from the mitochondrial carrier family

The complete sequencing of the S. cerevisiae mitochondrial genome demonstrated that it encodes genes connected with mRNA translation and oxidative phosphorylation, but no mitochondrial transporters (Foury et al., 1998). However, analysis of the nuclear genome (Goffeau et al., 1996) showed the existence of a large family of mitochondrial porters (MP) (TC 2.A.29) which today comprise 34 mitochondrial and one peroxisomal members (el Moualij et al., 1997).

These proteins possess a tripartite structure of three tandem repeated homologous sequences, with about 100 amino acids, each forming two transmembrane helices connected by a loop. The mitochondrial carriers transport a different range of substrates connected largely with mitochondrial metabolism such as the Krebs cycle intermediates (Palmieri et al., 2006). Six carboxylate transporters have so far been identified in the S. cerevisiae genome (Table 2).

One of the first and most studied mitochondrial transporters is the pyruvate carrier, whose existence was first demonstrated in rat liver by specific inhibition of pyruvate uptake by α-cyano-4-hydroxycinnamate (Halestrap et al., 1976). In yeast, the existence of a proton symporter for the uptake of pyruvate and also for lactate was demonstrated in isolated mitochondria by Briquet (1977). Hildyard & Halestrap (2003) analysed strains mutated in all the genes belonging to the mitochondrial carrier family of unknown function and demonstrated that only the strain deleted in the YIL006w gene presented no mitochondrial uptake of pyruvate. Moreover, the YIL006w deletant was not sensitive to α-cyano-4-hydroxycinnamate, also suggesting that the mitochondrial pyruvate transporter was encoded by this gene (Hildyard & Halestrap, 2003). Nevertheless, reconstitution of the purified recombinant protein into phospholipid vesicles demonstrated that the YIL006w gene does not code for a pyruvate transporter (Todisco et al., 2006). Instead it was shown to transport NAD+ as a uniport, as well as other nucleotides in exchange for NAD+. Currently, the gene encoding the mitochondrial pyruvate transporter remains to be identified.

Efflux of carboxylates through the plasma membrane

The concentration of the acid and the pH of the external medium determine whether organic acids can be used as carbon sources or whether they can have deleterious consequences such as alterations in the intracellular pH homeostasis and accumulation of toxic anions, membrane disruption, inhibition of essential metabolic reactions and decrease in ATP yield (Krebs et al., 1983; Eklund, 1985; Pinto et al., 1989; Pampulha & Loureiro-Dias, 1990, 2000; Cardoso & Leão, 1992). However, some yeast species can be agents of food and beverage spoilage as they are capable of adapting and growing in the presence of the highest levels of weak organic acids allowed in commercial food preservation (Piper et al., 1998).

Upon exposure to weak acids, at low pH, S. cerevisiae cells display an adaptative stress response, which involves both the activation of the plasma membrane H+-ATPase, regulating the intracellular pH, pHin and homeostasis (Viegas & Sá-Correia, 1991; Holyoak et al., 1996; Piper et al., 1997; Viegas et al., 1998), and the induction of Pdr12, a plasma membrane carboxylate efflux pump (Piper et al., 1997, 1998). This ATP-binding cassette (ABC) transporter is a member of the Pleiotropic Drug Resistance (PDR) Subfamily (TC 3.A.1.205.3). At low external pH, the Pdr12p deletion mutant was shown to be sensitive to sorbic, benzoic and propionic acids as well as to other carboxylic acids ranging from C1 to C7 (Holyoak et al., 1999). It was demonstrated that Pdr12p is strongly induced by propionic, sorbic and benzoic acids, and actively promotes the efflux of the anionic form of these acids, diminishing their toxic effect (Piper et al., 1998; Holyoak et al., 1999). The specific cellular response to acetic acid stress does not involve the activation of the Pdr12 pump (Bauer et al., 2003). Recently, Pdr12 was described as being involved in the export of fusel acids, resulting from amino acid catabolism, out of the cell (Hazelwood et al., 2006).

In order to maintain resistance to weak acids, cells also have to block entrance of the acid's undissociated form, or the molecules would re-enter the cell and dissociate, with the anion being pumped out at the cost of ATP consumption. This cycle would result in a large proton influx and loss of energy. In fact, it was demonstrated that in cells adapted to benzoic acid, the diffusion coefficient of organic weak acids across the plasma membrane is very low, thus reducing the entrance of acids into the cell, and avoiding entry into this costly energy cycle (Loureiro-Dias, 1998).

In addition, yeast cell-wall remodelling in response to weak acid stress was proposed to be mediated by Spi1p, a glycosylphosphatidylinositol-anchored cell wall protein that promotes a decrease in cell wall porosity, resulting in reduced membrane damage and intracellular acidification (Simões et al., 2006).

The role of Fps1 channel on carboxylate transport

In S. cerevisiae the plasma membrane aquaglyceroporin Fps1p was recently identified as a channel that promotes the facilitated diffusion of the undissociated form of acetic acid into the cell (Fig. 1) (Mollapour & Piper, 2007). This is the first example of a channel for carboxylic acids to be reported in yeasts. Fps1 was initially described as being involved in glycerol diffusion, as well as in its efflux/retention upon osmotic shock, by changes in the conformation of the protein, or in response to turgor changes (Luyten et al., 1995; Tamas et al., 1999, 2000). It was also identified as a mediator for the influx of arsenite and antimonite (Wysocki et al., 2001).

Mollapour & Piper (2007) demonstrated that an fps1 S. cerevisiae mutant strain is more resistant to acetic acid and presents a lower level of intracellular acetic acid accumulation compared with the wild type. When wild-type cells are exposed to acetic acid, 80–100 mM pH 4.5, Hog1p is activated, leading to inactivation of the Fps1 protein, a condition essential for the acquisition of acetate resistance (Mollapour & Piper, 2006). Hog1p is a kinase belonging to the HOG mitogen-activated protein (MAP) kinase cascade that upon activation was described as being involved in the osmotic stress response. In this case, however, a different cellular response takes place as glycerol-3-phosphate dehydrogenase gene (GPD1) activation and glycerol production, characteristic of osmotic stress, are not present. In glucose-grown cells, upon acetic acid shock, Fps1p is removed from the plasma membrane. This process requires a Hog1p-dependent phosphorylation and ubquitination, followed by Fps1 internalization via endocytosis and consequent degradation in the vacuole (Mollapour & Piper, 2007). Clarification of the role of Fps1 in the overall picture of the metabolism and utilization of acetic acid, and eventually of other carboxylic acids in yeast, is a promising field of research to be further developed.

Biotechnological implications

Lactate synthesis for the production of biodegradable polymers

Lactic acid is the most frequently carboxylic acid found in nature. The production of this acid is of great industrial importance as it is most broadly utilized in the food, cosmetics and pharmaceutical industries. The use of poly-lactic acid as a biodegradable and biocompatible polymer is increasing, and over 100 000 tonnes year−1 of lactic acid are produced. This represents an industrial investment of several million dollars annually (van Maris et al., 2004).

Lactic acid can be produced by chemical synthesis or by carbohydrate fermentation. Around 90% of the literature on lactic acid production is focused on bacterial fermentation. Bacterial species belonging to the genera Lactobacillus, Streptococcus, Leuconostoc and Enterococcus are the most common producers, with Lactobacillus species being the most commonly employed microorganisms at the industrial level (John et al., 2007). Choice of strain depends on the fermentable carbon source used, as well as on the stereoisomer to be produced (Narayanan et al., 2004). During lactic acid fermentation, the pH decreases thereby inhibiting bacterial metabolism. This is overcome by the addition of neutralizing agents to the medium. Consequently, the final product is not lactic acid but its salt. Purification processes are required to regenerate undissociated lactic acid, as pure lactic acid is preferred for industrial applications.

Microorganisms more tolerant to low pH are being genetically engineered for improvement in lactic acid production. This would eliminate the need of neutralizing agents and consequently lower lactate production costs by decreasing the purification steps. Such is the case of several yeasts, such as S. cerevisiae, K. lactis, T. delbruekii, Z. bailii and some fungi from the genus Rhizopus (Dequin & Barre, 1994; Porro et al., 1995, 1999b; Bianchi et al., 2001; Branduardi et al., 2004; Narayanan et al., 2004; Jin et al., 2005).

The first yeast reported to be genetically engineered in order to produce lactic acid was an S. cerevisiae strain. Although this yeast is capable of growing on lactate as only carbon and energy source, this compound is not a product of its metabolism. In order to produce l-lactate, the heterologous expression of l(+)-lactate dehydrogenase (LDH) from Lactobacillus casei (Dequin & Barre, 1994) created a new metabolic pathway where 20% of glucose was transformed to lactate, but ethanol production also occurred. Strategies involving the deletion of key enzymes of this pathway were adopted. Saccharomyces cerevisiae has three pyruvate decarboxylase genes, PDC1, PDC5 and PDC6, but growth is impaired in the corresponding triple-deleted strain (Flikweert et al., 1996). However, a profound decrease in ethanol production was obtained with the simultaneous deletion of PDC1 and PDC5 (Ishida et al., 2006b).

The first example of complete substitution of ethanol by lactate production was accomplished by Porro et al. (1999a) in an engineered K. lactis strain. This strain expressed a heterologous LDH in order to produce lactate. Additionally, its pyruvate decarboxylase gene was deleted to prevent ethanol production. As a result a yield of 1.19 mol lactate mol−1 glucose was achieved. This value is much lower than the theoretical value of 2 mol lactate mol−1 glucose. It is known that the accumulation of lactate inhibits Ldh activity, and consequently decreases the amount of lactate produced (Branduardi et al., 2006). Therefore, a strain overexpressing the JEN1 gene (Porro et al., 1999a) was engineered and shown to double lactate production. Later studies demonstrated that the overexpression of the JEN1 gene results in an increased lactate production only when the amount being produced is below 8–10 g L−1 (Branduardi et al., 2006). In engineered strains producing high levels of lactate, JEN1 overexpression has little effect. The authors postulate that the Jen1 transport capacity is saturated at these lactate concentrations and that another unidentified lactate exporter is activated above these concentrations (Branduardi et al., 2006). As mentioned above, several pumps are able to transport carboxylic acids but there is no reference to lactate being a substrate for any of these transporters. One may expect that lactate efflux requires energy. This would be in agreement with the fact that yeast cells producing high amounts of lactate have a lower biomass yield (Branduardi et al., 2006). Unexpectedly, under anaerobic conditions the S. cerevisiae strains producing lactate are not able to grow (van Maris et al., 2004). In terms of overall energy production, the yield of ethanol- and lactate-producing strains should be the same, as theoretically in both fermentations two molecules of ATP are formed per molecule of glucose, and the redox balance is the same. Nevertheless, a strain producing ethanol is able to grow under oxygen limitation, thus being more efficient than an equivalent lactate-producing strain. This could be explained by a difference in the final product export mechanism. Ethanol diffuses freely across the plasma membrane, while lactate export requires an energy-dependent transporter, as at intracellular pH it would be predominantly in the dissociated form. As a result no net gain of ATP would occur and the growth of the homofermentative lactate-producing strain is impaired (van Maris et al., 2004).

In summary, the engineering of yeast strains for lactate production aims at obtaining a high yield of pure lactate, in conditions that can be used industrially, requiring few purification steps and using cheap raw materials as carbon and energy source. Several of these goals have been achieved and a yeast strain producing exclusively lactate without ethanol contamination has been engineered (Bianchi et al., 2001; Ishida et al., 2006b). The highest yield achieved so far is 82.3 g lactate L−1, corresponding to 81.5% conversion of glucose to lactate (Ishida et al., 2006b), and a optical purity of 99.9% was obtained for both for l- and d-lactate production (Saitoh et al., 2005; Ishida et al., 2006a, c). Recently, the yeast Pichia stipitis was genetically engineered to produce lactate using xylose as carbon and energy source, thus encouraging research on lactate production from lignocellulosic raw material (Ilmen et al., 2007).

Compared with bacteria, a major limiting step of lactate production by yeast seems to be the export of lactate. The identification of a novel putative lactate exporter and improving the activity of overexpressed JEN1 could be important for the industrial application of yeast as a lactate-producing organism.

Wine deacidification

Organoleptic organic acids are structural components of great importance for wine quality. Malic and tartaric acids are the main organic acid constituents of grape must and wine, contributing to the ‘fixed acidity’, a parameter that defines the quality of the final product. In certain regions, during colder years, grape cannot reach sufficient maturity, leading to high acidity. Under these conditions, the winemaker can perform a controlled deacidification of the wine, which can be carried out using yeasts capable of degrading malic acid. Deacidification of the wine is also possible by addition of calcium carbonate, potassium bicarbonate or potassium tartarate. However, these chemicals can lead to undesirable secondary effects such as an increase in calcium concentration, which is an instability factor in the wine (Husnik et al., 2006). Therefore, winemakers generally prefer to use biological deacidification. Malic acid can be metabolized by malolactic fermentation, a process comprising its decarboxylation into lactic acid and carbon dioxide, after alcoholic fermentation. Malolactic fermentation is used only on red wines destined for ageing and selected white wines such as Chardonnay, Pinot Blanc and Pinot Gris, as in other white wines malic acid plays a crucial role as a flavour enhancer (Lonvaud-Funel et al., 1999). This reaction is catalysed by the malolactic enzyme, which is NAD+- and Mn2+- dependent and malate induced (Henick-Kling et al., 1993; Lonvaud-Funel & Joyeux, 1994; Lonvaud-Funel et al., 1995). Malolactic fermentation is performed by bacterial strains of the genera Lactobacillus, Leuconostoc, Oenococcus and Pediococcus (Lonvaud-Funel & Joyeux, 1994; Bony et al., 1997), but preferentially by Oenococcus oeni (Nielsen & Richelieu, 1999).

The use of yeasts has been studied as an alternative to malolactic fermentation carried out by bacteria, S. cerevisiae being the ideal candidate. However, as discussed previously, this yeast uses malic acid very inefficiently, due to the absence of a malate permease and to a high value of Km for malic enzyme (Kuczynski & Radler, 1982; Salmon, 1987). Saccharomyces cerevisiae cells grown under anaerobic conditions show some malate dehydrogenase and fumarase activity (Camarasa et al., 2001), yet these pathways degrade only a small amount of malic acid. Malate dehydrogenase uses oxaloacetate preferentially as substrate and is glucose repressed whereas fumarase is inhibited by intracellular organic and inorganic phosphate (Duntze et al., 1968; Neeff & Mecke, 1977; Kuczynski & Radler, 1982). Engineered yeast expressing bacterial malolactic encoding genes have confirmed the functionality of these enzymes in yeast cells (Williams et al., 1984; Ansanay et al., 1996). The malolactic enzyme from Pediococcus damnosusas and Lactobacillus lactis were coexpressed with the malate permease of S. pombe. In both cases the S. cerevisiae transformants presented a very efficient ability to degrade malic acid in synthetic must (Bony et al., 1997; Volschenk et al., 1997a, b; Bauer et al., 2005). When S. pombe malic enzyme was expressed alone in S. cerevisiae, a significant degradation of malic acid was not observed, probably because the limiting factor in malic acid degradation by S. cerevisiae is the absence of a malate carrier (Volschenk et al., 1997b).

The possibility of using maloalcoholic fermentation (Fig. 3) for malic acid degradation has been also explored (for a review see Volschenk et al., 2003). Schizosaccharomyces pombe mae1 and mae2 genes, which respectively encode the malic acid permease and the malic enzyme, were therefore expressed in S. cerevisiae (Volschenk et al., 2001). The recombinant strain completely degraded 8 g L−1 of malic acid in a synthetic must and 6.75 g L−1 in a Chardonnay must with increased ethanol production. The mae1 and mae2 genes were also integrated in the S. cerevisiae genome, and the recombinant strain was able to degrade 5 g L−1 of malic acid in synthetic and in Chenin Blanc musts (Volschenk et al., 2001).

In order to construct a commercial wine yeast strain able to degrade malic acid efficiently, Husnik et al. (2006) integrated in the genome of an industrial S. cerevisiae strain the malate permease gene of S. pombe and the malolactic gene of O. oeni. This expression cassette was flanked by ura3 sequences and was integrated at the URA3 locus of the S. cerevisiae strain, under the control of the PGK1 promoter and terminator. Analysis of the genome, transcriptome and proteome revealed that the transformant strain was identical to the wild type, except for the cloned genes, indicating that their expression had little effect on global gene expression pattern. This strain, named ML01, was considered as GRAS (generally regarded as safe) by the United States Food and Drug Administration (FDA) and was the first genetically engineered yeast to be commercialized for wine production (Husnik et al., 2006).

Prospects

At present a major concern is adaptation to a new world reality, with decreasing petroleum reserves and the consequent development of new technologies based on renewable resources. This has altered industrial awareness and boosted the development of biorefineries, producing not only biofuels but also other chemicals, such as carboxylic acids. Citric, lactic, acetic and propionic acids have been used for many years now in industrial and pharmaceutical companies. They present distinct properties according to their carbon-chain length, molecular structure and functional groups. Besides the vast, well-established applications, new uses for carboxylic acids in industry are emerging such as the utilization of lactate for production of biodegradable polymers. These acids are also starting to be considered as building-block chemicals and can even be substitutes for petroleum-derived chemicals (Sauer et al., 2008). The biological production of these acids has several goals: to improve product yield/concentration at high purity, to decrease purification steps and to use cheap raw materials as carbon sources. Metabolic engineering of several species has enabled the optimization of these processes, although much needs to be done in order to increase competitiveness.

Understanding in detail the mechanisms underlying the transport of carboxylic acids is crucial towards an efficient biological production of these compounds. The intracellular accumulation of the acids produced leads to an inhibition of metabolic pathways and consequently to decreased productivity (Branduardi et al., 2006). Studies on citrate, malate and lactate production have pointed to the existence of exporters for these substrates (Wolschek & Kubicek, 1999; van Maris et al., 2004; Zelle et al., 2008). The identification of these acid exporters could contribute to an overall improvement in productivity.

The number of known integral membrane protein structures is growing, although it is still small compared with other types of proteins. This number is continuously increasing and reflects the efforts of several multidisciplinary teams in optimizing protein production, protein purification and crystal growth that ultimately enable protein structure determination. However, we must point out that the often unrecognized bottleneck of these approaches is still the optimization of heterologous overexpression of membrane proteins in S. cerevisiae or possibly in the longer term in other oversecreting yeasts such as Y. lipolytica or P. pastoris. Since the publication of the MFS transporter structures (Abramson et al., 2003; Huang et al., 2003), our knowledge of structural data has significantly increased and several structural models are now proposed based on the three-dimensional structure of the E. coli lactose permease, LacY, and glycerol-3-phosphate transporter, GlpT (Lagerstedt et al., 2004; Wood et al., 2005; Lemieux et al., 2007).

One may reasonably expect that within the next few years the molecular structure and mechanism of several yeast carboxylate transporters will be unravelled and that engineering of novel yeast strains combining bacterial and yeast metabolic pathways will be achieved and used commercially.

Acknowledgements

We dedicate this review to the memory of Professor Isabel Spencer-Martins. We are grateful to Professor André Goffeau for his stimulating discussions, encouragement to pursue the studies on yeast membrane transporters and critical reading of the manuscript. This work was supported by Portuguese grant POCI/BIA-BCM/57812/2004 (Eixo 2, Medida 2.3, QCAIII – FEDER). I.S.S. received a fellowship from the Portuguese government: SFRH/BPD/22976/2005.

Footnotes

  • Editor: Karl-Dieter Entian

References

View Abstract