Introduction

Pretreatment of lignocellulosic biomass generates inhibitory compounds that interfere with microbial growth and fermentation and poses a significant challenge for economical cellulosic biofuels production. Remediation of inhibitory compounds by physical and chemical means has been determined to be too expensive for use in practice (Liu and Blaschek 2010). A bioabatement method was able to remove aldehyde inhibitors such as 2-furaldehyde (2-furancarbaldehyde; furfural); however, additional sugar and carbon source were consumed, and most abatement agents lack fermentation capability (Nichols et al. 2010). Tolerant ethanologenic yeast strains were found to be able to convert furfural and 5-(hydroxymethyl)-2-furaldehyde [5-(hydroxythyl)-2-furancarbaldehyde; 5-(hydroxymethyl)-2-furfural; HMF], representative inhibitors for biomass pretreatment, into less toxic compounds furanmethanol (FM) and furan-2,5-dimethanol (FDM; 2,5-bis-hydroxymethylfuran) while producing normal yields of ethanol (Liu et al. 2004; 2005; 2008b; Liu 2006; Talebnia and Taherzadeh 2006; Martin et al 2007). The identification and clarification of FM and FDM as metabolic conversion products of furfural and HMF suggested that the attached aldehyde functional group on the furan ring is the toxic cause but not the furan since numerous furan compounds are not toxic to yeast (Liu et al. 2004; 2008b; Liu 2006). Commonly termed as furan inhibitors inherited by historical reasons, furfural and HMF are in fact aldehyde inhibitors. The conversion of the aldehyde functional group into an alcohol form reduces the chemical toxicity. This clarification has led to an attempt to classify the inhibitors by the chemical functional groups to facilitate mechanism studies of the in situ detoxification (Larsson et al 1999; Klinke et al 2004; Liu and Blaschek 2010). The current classification of inhibitors contains aldehydes, ketones, phenols, and organic acids commonly associated with lignocellulose hydrolysates and biomass pretreatment procedures (Fig. 1). The new classification of the inhibitors has facilitated discoveries of new genes and new functions of known genes. For example, a newly described aldehyde reductase enzyme encoded by ARI1, a previously uncharacterized ORF YGL157W of Saccharomyces cerevisiae, possessed reduction capabilities toward at least 14 aldehydes including common lignocellulose-derived inhibitors such as furfural, HMF, vanillin, and cinnamaldehyde (Liu and Moon 2009). Since the discovery of the in situ detoxification of the fermentation inhibitors by tolerant yeast, significant progress has been made in understanding mechanism of the detoxification in the last decade (Palmqvist and Hahn-Hägerdal 2000; Liu 2006; Liu et al 2008a). This review presents our current knowledge of molecular mechanisms involved in the yeast tolerance and in situ detoxification of inhibitors, mainly aldehydes, derived from lignocellulosic hydrolysates. Improved understanding of yeast tolerance and the detoxification provides insight into dissection of tolerance mechanisms and the strategies for more tolerant strain development for economical biofuels production.

Fig. 1
figure 1figure 1

A classification of inhibitory compounds to fermentative microorganisms derived from biomass pretreatment and lignocellulose hydrolysates based on chemical functional groups of aldehydes, ketones, organic acids, and phenols

Full size image

Reduction enzymes

Collective functions of multiple reduction enzymes

Understanding the importance of detoxification of the aldehyde functional group has allowed recognition of numerous enzymes possessing aldehyde reductase activities that contribute to the detoxification of the aldehyde inhibitors associated with lignocellulose pretreatment such as furfural, HMF, cinnamaldehyde, and vanillins. It is well established that the detoxification of the aldehyde inhibitors is attributed to multiple enzymes involved in reduction activities coupled with cofactors nicotinamide adenine dinucleotide phosphate (NADPH) and/or nicotinamide adenine dinucleotide (NADH) (Morimoto and Murakami 1967; Nemirovskii et al 1989; Villa et al. 1992; Wahlbom and Hahn-Hägerdal 2002; Liu et al. 2004; 2008b; Nilsson et al 2005; Liu 2006; Petersson et al. 2006; Liu and Moon 2009; Almeida et al. 2008; Liu and Blaschek 2010; Moon and Liu 2011; Fig. 2). Most in vitro enzyme assays for reduction of furfural and HMF were evaluated using whole-cell protein extracts; however, a few examples using partially purified proteins are available (Liu and Moon 2009; Moon and Liu 2011). Some notable enzymes such as alcohol dehydrogenase ADH7, ADH6, and ADH1, aldehyde dehydrogenase ALD4, and methylglyoxal reductase GRE2 and GRE3 have been demonstrated to possess efficient aldehyde reduction activities (Table 1). Comparative proteomic analysis of an industrial yeast strain suggested Adh5p and Adh1p as the catalytic agents for furfural reduction (Lin et al. 2009a). Protein extracts from individual gene clones often show distinct cofactor preference. However, whole-cell protein extracts from a tolerant ethanologenic yeast display strong aldehyde reduction activities with either NADH or NADPH and do not appear to have a strong cofactor preference (Liu et al. 2008b). A single gene deletion of the related reductase does not appear to significantly affect the detoxification capacity in yeast. It is likely that yeasts are able to respond globally and at multiple complex levels in biotransformation of the aldehyde inhibitors. Transcriptome analysis indicated many reductase genes were immediately induced by the toxic treatment such as ADH7, ARI1, GRE2, and ALD4 (Ma and Liu 2010). Among these, ADH7 can have a greater than 30- to 80-fold increase in transcript abundance after the HMF addition at 10 min and 1 h, respectively. As demonstrated by 13C-labeled metabolic flux and transcription study, ADH7 and ORF YKL071W, and possibly four other reductases, are associated with the yeast resistance to the furfural challenge (Heer et al. 2009).

Fig. 2
figure 2

Conversion pathways of 2-furaldehyde (furfural) and 5-(hydroxymethyl)-2-furaldehyde (HMF) into 2-furanmethanol (FM) and furan-2,5-dimethanol (FDM) coupled with NADH and/or NADPH and catalyzed by multiple reductases

Full size image
Table 1 Genes encoding enzymes possessing aldehyde reductase activities examined by the enzyme specific activity using whole-cell protein extract or partially purified proteins exampled by furfural and HMF for ethanologenic strains of S. cerevisiae
Full size table

Reduction functions of known genes

Important new functions of aldehyde reduction have been discovered for previously reported enzyme-coding genes. For example, yeast clones overexpressing ADH6 and ADH7 displayed high reduction capabilities toward furfural and HMF (Petersson et al. 2006; Liu et al. 2008b). Although they were characterized as alcohol dehydrogenases, the kinetic study of ADH6 and ADH7 showed that their reductive reactions were 50- to 100-fold more efficient than the corresponding oxidations (Larroy et al 2002a, b, 2003). It is possible that ADH6 or ADH7 act as an aldehyde reductase rather than as an alcohol dehydrogenase as their major metabolic function. Cell protein extracts of mutated ADH1 containing yeast strains also showed significant aldehyde reductase activities (Almeida et al. 2008; Laadan et al 2008). ALD4 is a major mitochondrial aldehyde dehydrogenase that is required for growth on ethanol and the conversion of acetaldehyde to acetate using NADP+ or NAD+ as coenzymes. This enzyme is also able to reduce HMF and furfural utilizing NADH as a cofactor. Aldehyde dehydrogenase is known to play an important role in yeast acetaldehyde metabolism. Thus, aldehyde dehydrogenase could be another potential candidate gene for detoxification of aldehyde inhibitors. Similarly, aldo-keto reductase and methylglyoxal-related reductase GRE3 and GRE2 showed aldehyde reduction activities (Liu et al. 2008b; Liu and Moon 2009; Moon and Liu 2011). Xylose reductase from Pichia stipitis and expressed in S. cerevisiae also possessed reduction activities toward furfural and HMF (Almeida et al. 2008). In addition to their significant involvement under stress conditions, these genes appeared to be important candidates facilitating inhibitor reduction. GRE3 has been deleted in an effort to reduce xylitol byproduct production to improve xylose utilization efficiency of yeast (Träff et al. 2001; Kuyper et al. 2005). Considering the significant interaction between inhibitor tolerance and efficient pentose utilization, it is worthwhile to clarify the roles and interplay among the important candidate gene groups for balanced metabolic function in biofuel conversions by yeasts.

New aldehyde reductase genes

A novel gene encoding NADPH-dependent aldehyde reductase, ARI1, was characterized recently (Liu and Moon 2009; Bowman et al. 2010; Saccharomyces Genome Database http://www.yeastgenome.org/). The product of ARI1 is the first purified yeast protein reported as an aldehyde reductase involved in the detoxification of inhibitors of lignocellulose hydrolysates. As mentioned, it has reduction activities toward at least 14 aldehydes including those frequently identified during biomass pretreatment procedures. The optimum performance temperature of the enzyme is 25 °C at pH 7.0. The protein of ARI1 has an approximate molecular mass of 38 kDa and is a member of the subclass “intermediate” of the short-chain dehydrogenase/reductase superfamily with the following typical characteristics: the conserved catalytic site lies at Tyr169-X-X-X-Lys173; an indispensable reduction catalytic tetrad at Asn106, Ser131, Tyr169, and Lys173; and an approved cofactor binding motif at Gly11-X-X-Gly14-X-X-Ala17 near the N terminus. The function of the gene was annotated by conserved functional sequence motifs, gene expression, protein expression, and partially purified protein assays. This newly described gene possibly represents a group of uncharacterized multiple functional genes such as other potential candidates YKL071W, Y62, Y76, Y81, and Y82 (Heer et al. 2009; Liu and Blaschek 2010; Liu and Moon 2009).

Detoxification pathways

Enhanced genetic background

A tolerant yeast is able to withstand challenges of high levels of furfural–HMF inhibitor complex and produce normal yields of ethanol while the parental strain fails to establish a viable culture under the same conditions (Liu et al. 2009). It is clear that the tolerant yeast possesses different genetic mechanisms for in situ detoxification of the toxic compounds that enable active metabolism for ethanol production. Characterization of gene expression dynamics of the tolerant yeast strain suggested that the tolerant yeast appeared to have an inheritable genetic makeup that is distinct from its parental strain. At least 16 gene transcripts involved in glucose metabolism had significantly greater abundance in the inhibitor-tolerant yeast compared with its parental strain even without the inhibitor treatment (Liu et al. 2009). Many of these are key genes involved in glucose metabolic process, NAD(P)H metabolic and regeneration, and transferase activities such as HXK1, HXK2, GLK1, TDH1, TDH3, LAT1, PDC6, ADH4, ALD2, ALD4, ZWF1, SOL3, RBK1, TAL1, NQL1, and PRS2 (Table 2). Some of these key genes displayed as high as four- to sixfold increased abundance, for example, the hexokinase-encoding genes HXK1 and HXK2, glyceraldehyde-3-phosphate dehydrogenase gene TDH1, dihydrolipoamide acetyltransferase component (E2) of the pyruvate dehydrogenase complex gene LAT1, and major mitochondrial aldehyde dehydrogenase gene ALD4 and TAL1 that encodes transaldolase.

Table 2 Gene ontology (GO) categories and terms for significantly induced genes by HMF during the lag phase in S. cerevisiae
Full size table

Reprogrammed regulatory networks and redox balance

Glycolysis and pentose phosphate pathway are closely related pathways in yeast glucose metabolism. This close relationship is of such importance that the two pathways cannot be viewed separately when discussing yeast tolerance and detoxification of the lignocellulose inhibitors. Under the challenge of furfural–HMF complex, yeast is unable to grow, and most genes involved in these pathways are severely repressed. A tolerant yeast strain, on the other hand, demonstrated different expression dynamics and completed ethanol fermentation. Under inhibitor stress, high levels of expression by HXK1, HXK2, and GLK1 appeared to secure the initiation stage of phosphorylation of glucose by these enzyme encoding genes (Liu et al. 2009). Then, the significantly induced expression of ZWF1, SOL3, GND1, and GND2 as well as the repression of glycolytic enzyme phosphoglucose isomerase apparently drive the glucose metabolism toward pentose phosphate pathway. Gene deletion mutations of ZWF1 and GND1 are highly sensitive to furfural and HMF (Gorsich et al. 2006). The enhanced expression of ZWF1 at an early step is key to shifting the glucose metabolism in favor of pentose phosphate pathway over glycolysis (Liu et al. 2009; Fig. 3). Consequently, all other cofactor NAD(P)H regenerating steps involving ZWF1, GND1, GND2, and TDH1 were up-regulated in the tolerant yeast. Aldehyde reduction enzyme encoding genes ALD4, ALD6, ADH6, ADH7, and SFA1 displayed significantly increased transcription at the early time points in the presence of furfural–HMF complex. These accelerated NAD(P)H-dependent reductions of acetaldehyde, furfural, and HMF would generate sufficient NAD+ and NADP+, in return, to provide necessary cofactors needed for oxidative reactions or NAD(P)H regenerations by Zwf1p, Gnd1p, Gnd2p, Tdh1p, and Ald4p. Redox metabolism, in the form of interconversion of the pyridine-nucleotide cofactors NADH/NAD+ and NADPH/NADP+, plays a key role in the yeast metabolism. NADH is required in respiration and fermentative pathway in conversion of pyruvate to CO2 and ethanol. NADPH is mainly required for the synthesis of amino acids and nucleotides and a major source of NADPH production in yeast is through the oxidative phase of pentose phosphate pathway. The up-regulated ZWF1, SOL3, GND1, and GND2 along with enhanced expressed TDH1 are important for NAD(P)H regenerations to supply necessary cofactors needed for acetaldehyde conversion and reduction of furfural and HMF. Thus, a NAD(P)+/NAD(P)H-dependent redox balance is well-maintained in the altered pathways for the in situ detoxification of furfural and HMF by the tolerant yeast.

Fig. 3
figure 3

A schematic illustration of glucose metabolic pathways and conversion of furfural and HMF by tolerant S. cerevisiae NRRL Y-50049 inferred by metabolic profiling analysis and quantitative mRNA expression analysis compared with its wild-type strain NRRL Y-12632. Black arrowed lines and letters indicate normal or near-normal levels of reactions, expressions, or pathways; green indicates enhanced, and red for repressed expressions, reactions, or pathways. Bold lines and letters indicate the levels of expression and pathways are statistically significant. Key steps of enhanced NAD(P)H regenerations are circled in blue and significant aldehyde reductions circled in orange. Interactions of cofactor regeneration and balanced utilization pathways are linked by dotted lines

Full size image

Under the inhibitor challenge, tolerant yeast also appeared to be able to achieve NAD(P)H regeneration through a short path to the “TCA cycle”: tricarboxylic acid cycle. This process involves many genes in amino acids metabolism pathways closely related to the TCA cycle, including both induced genes such as CHA1, ALT1, PUT1, PUT2, and CAR1, and repressed genes such as ARG1, ARG3, ARG4, ARG5, ARG6, ARG7, ARG8, LYS4, LYS14, and LYS20 (Ma and Liu 2010). The accelerated catabolism of proline, serine, and alanine, together with the reduced biosynthesis of arginine, provides a shortcut for ATP regeneration via the TCA cycle. Thus, efficient energy metabolism can be maintained under the inhibitor stress. Apparently, enriched genetic background by aforementioned genes and a well-maintained redox balance through the reprogrammed expression responses involved in numerous pathways of the tolerant yeast strain are accountable for the acquired yeast tolerance and the detoxification of the inhibitors.

Integrated multiple gene interactions

Yeast exhibit an accelerated glucose conversion rate once they are recovered from the furfural and/or HMF challenges compared with what would normally occur without the inhibitors (Taherzadeh et al. 2000; Liu et al. 2004, 2005). The inhibition of glucose phosphorylation, together with repression of PFK1, PFK2, PYK2, and CDC19 seemed responsible for the delayed glycolysis inhibited by furfural and HMF treatment (Liu et al. 2009). Such a delayed biological process in yeast can also be attributed to a lack of ATP, NAD(P)H, and intermediate metabolites necessary to support cell growth and reproduction (Wahlbom and Hahn-Hägerdal 2002; Fisk et al. 2006; Liu 2006). For the tolerant yeast, in addition to numerous induced expressions, gene transcription levels of PGK1, ENO1, ENO2, PYK2, CDC19, PDA1, and PDB1 encoding varied enzymes for pyruvate metabolisms did not show repressed effect in response to the inhibitor challenge at the early stage. This allowed a smooth flow of the central metabolic pathways. Since the tolerant yeast is able to in situ detoxify the aldehyde inhibitors, with the significant reduction in the concentration of the inhibitory aldehydes, more NADPH thus generated could shift from detoxification to accelerate biosynthesis processes and cell growth. In the meantime, alcohol dehydrogenase is favored for the conversion of acetaldehyde to ethanol with sufficient NADH supply, which contributes to the accelerated glucose consumption.

It should be pointed out that many genes that initially were repressed but were able to recover to their normal functional levels after the inhibitor challenges are necessary components in these globally integrated interactions under the stress. The functions of these genes allowed the tolerant yeast to maintain balanced biological processes to complete ethanol fermentation. In the absence of such reprogrammed transcription dynamics at the genome level, continued inhibition and repression by furfural and HMF, as demonstrated by a wild-type strain, led to loss of cell function and eventual death.

Global response

Gene expression profiling overview

Earlier studies on laboratory yeast strain response to environmental stimuli suggested a core set of genes are involved with the environmental stress response (Gasch et al. 2000; Gasch and Werner-Washburne 2002). For industrial ethanologenic yeast, 100 to 400 genes showed differential transcription expression response to furfural and HMF individually or in combination (Liu and Slininger 2005; Liu 2006; Liu and Slininger 2006; Liu et al. 2009; Ma and Liu 2010; Li and Yuan 2010). Unlike the laboratory strain showing mostly transient responses, ethanologenic yeast displays relatively persistent expression dynamics indicating different mechanisms may possibly be involved. Most studies on yeast response were characterized using a wild-type strain and differentially expressed genes distributed in a wide range of functional categories. Classic observations on stress-related high-osmolarity glycerol pathway, heat-shock protein genes, and transcription factors Msn2p/Msn4p are commonly observed (Lin et al. 2009a, b; Li and Yuan 2010; Ma and Liu 2010). Tolerant yeast responded to aldehyde inhibitors differently than a wild-type, although there are some overlappings during the early time point responses. In the remaining sections of this review, an emphasis is given mainly to potential candidate genes in order to address global integration and interactions of the tolerance yeast.

Induced expression

The inhibitor-induced expression consists of only a small portion of genes responding to the challenge at the genome level. However, many of these genes have multiple functions. Some notable function categories involve cytoplasm, nucleus, membrane, mitochondrion, cellular protein catabolic process, transport, response to stress, amino acid and derivative metabolic process, hydrolase activity, peptide activity, oxidareductase activity, protein binding, protein fate, cellular transport, and several groups of unknown functions (Tables 2 and 3). At least seven transcription factor genes, YAP1, YAP5, YAP6, PDR1, PDR3, RPN4, and HSF1, were identified as key regulators for the induced expression response in yeast adaptation to HMF challenge (Song et al. 2009; Ma and Liu 2010). Most of these regulatory genes displayed greater than twofold increase of mRNA abundance after challenges by furfural and HMF. Protein binding motif analysis revealed that each of these transcription factor genes harbors multiple protein binding sites for Pdr3p, Yap1p, Yap5p, Yap6p, Rpn4p, and Hsf1p. For example, DNA binding motifs of Pdr1/3p are present in promoter regions of PDR3, YAP5, YAP6, and RPN4 (Ma and Liu 2010; Fig. 4). DNA binding sites of Yap1p and Hsf1p exist in all five transcription factor genes except for PDR1 having one Yap1p site and PDR3, two Hsf1p sites. Most transcription factor genes have multiple binding sites for multiple transcription factors. For example, RPN4 has 13 binding sites of four transcription factors, and PDR3 has six sites for two. These observations suggest potential interactions involving multiple transcription factors exist for inhibitor tolerance. High expression of RPN4 by HMF treatment was suggested to be regulated by Yap1p, Pdr1p, Pdr3p, and Hsf1p based on ChIP-chip assay data, genome expression, and microarray assays of transcription factor mutations (Lee et al. 2002; Harbison et al. 2004; Hahn et al. 2006; Larochelle et al. 2006; Workman et al. 2006; Salin et al. 2008; Ma and Liu 2010). Numerous studies also demonstrated positive feedback of enhanced expression of RPN4 to its regulators of Yap1p and Pdr1p (Harbison et al. 2004; Haugen et al. 2004; Salin et al. 2008). In addition, DNA binding motif of a transcription factors’ own is present in its promoter region, such as PDR3, YAP1, and HSF1 (Fig. 4). These suggest a possible self-regulated expression interaction involved in yeast tolerance response as well as co-regulation and interactions of multiple transcription factors under the stressed condition.

Table 3 Protein functional categories for significantly induced genes by HMF during the lag phase in S. cerevisiae
Full size table
Fig. 4
figure 4

DNA binding sites in the promoter regions from −1,000 to −1 for seven selective transcription factor genes YAP1, YAP5, YAP6, PDR1, PDR3, RPN4, and HSF1 of S. cerevisiae showing overlapping and multiple binding sites that indicate gene co-regulation roles of key transcription factor genes

Full size image

Repressed response

Most differentially expressed genes show repressed response to inhibitor challenges regardless of the treatment methods used. The difference lies that repressed genes in certain categories are able to recover over time while others remain repressed as demonstrated by comparative transcription dynamic analyses (Liu et al. 2009; Ma and Liu 2010). The importance of repressed genes is often neglected in contrast to overwhelmingly emphasized attention to the induced genes. In fact, many “overlooked” genes play necessary roles in yeast adaptation as they are able to recover and function under stress. As mentioned, the lack of such functional genes can result in non-viable biological processes including ethanol fermentation. Under certain conditions, down-regulated expression could be efficient means of energy utilization for economic pathway development (Ma and Liu 2010). The repressed genes are mainly involved in the functional categories of ribosome biogenesis, amino acid and derivative metabolic process, RNA synthesis, RNA metabolic process, transport, transcriptional and translation controls, mitochondrial, and others (Ma and Liu 2010; Li and Yuan 2010). For many repressed genes, at least five important regulatory genes including ARG80, ARG81, GCN4, RAP1, and FHL1 were found to be involved in the significantly down-regulated expression. For example, ARG1, ARG3, ARG4, ARG5, ARG6, ARG7, and ARG8 involved in arginine biosynthesis repressed by HMF were regulated by the transcription factor genes ARG80 and ARG81 as well as GCN4 (De Rijcke et al. 1992; Natarajan et al. 2001; Ma and Liu 2010). In addition to regulation of arginine biosynthesis, GCN4 regulates expression of many other genes related to amino acid biosynthesis such as a number of genes involved in biosynthesis of histidine, leucine, and lysine (Natarajan et al. 2001; Ma and Liu 2010). Among the many genes repressed by HMF, a large number of genes are involved in ribosome biogenesis and protein translation processes, which were predicted to be regulated by transcription factor genes RAP1 and FHL1.

Genomic adaptation

YAP family- and YAP1-regulated oxidoreductase activity networks

The lag phase for cell growth in response to inhibitor challenges has been used as a measure of strain tolerance and to study the mechanisms of genomic adaptation (Liu et al. 2004; Liu 2006; Ma and Liu 2010). Recently, 365 candidate genes were identified as involved in yeast adaptation and tolerance to HMF (Ma and Liu 2010). The interventional networks and interplays are complex and comprehensive. However, at least three significant components are recognized by some key regulators. First, numerous functional encoding genes such as ARI1, ADH6, ADH7, and OYE3, as well as gene interactions involved in the biotransformation and inhibitor detoxification, are the direct driving force to reduce the HMF damage in cells. The yeast activator protein (YAP) family contains eight transcription factors with a b-ZIP protein at the DNA binding domain (Rodrigues-Pousada et al. 2010). Transcription factor Yap1p, the major oxidative stress regulator, acts as a sensor for oxidative molecules and activates the transcription response of anti-oxidant genes by recognizing Yap1p response elements (YRE), 5’-TKACTMA-3’, in the promoter region (Harbison et al. 2004; Fernandes et al. 1997; Dubacq et al. 2006). Under HMF challenged conditions, YAP1 displayed consistently higher induced abundance of at least two- to threefold increase during the lag phase (Ma and Liu 2010). There are at least 41 HMF-induced genes possessing the YRE sequence in their promoter region. Many genes were confirmed to be regulated directly by YAP1 or indirectly through YAP5 and YAP6 (Fig. 5). Most YAP1-regulated genes were classified in the functional categories of redox metabolism, amino acid metabolism, stress response, DNA repair, and others (Table 2). For example, the highly induced oxidoreductase genes ADH7, GRE2, and OYE3 were found as regulons of YAP1 (Lee et al. 2002; Haugen et al. 2004; Dubacq et al. 2006; Ma and Liu 2010). A recently characterized new aldehyde reductase gene, ARI1, was found to be regulated by Yap6p which is a regulon of YAP1 (Harbison et al. 2004; Liu and Moon 2009; Ma and Liu 2010). ADH7 and GRE2, two confirmed HMF-detoxification genes encoding reductase activities, were co-regulated by Yap5p and Yap6p (Harbison et al. 2004; Workman et al. 2006; Ma and Liu 2010). A few enzyme encoding genes, for example, ALD4 and GRE2, were also co-regulated by Pdr1p. In addition, YAP1 and other YAP gene family members were shown to co-regulate numerous genes in a wide range of functional categories such as PDR, heat-shock protein, chaperones, and amino acid metabolism (Fig. 5, Table 2). Multiple functions of a gene are commonly observed in yeast tolerance and co-regulation of numerous genes can be a reflection of the multi-functions of such genes.

Fig. 5
figure 5

A schematic diagram showing key gene regulatory elements involved in tolerance and in situ detoxification of lignocellulose hydrolysate inhibitors for S. cerevisiae. Important transcription factor genes and major functional gene categories are highlighted. See text for detailed descriptions

Full size image

Single YAP gene deletion mutations are able to grow normally without HMF treatment. However, in the presence of 15 mM HMF, mutations Δyap1, Δyap4, Δyap5, and Δyap6 showed delayed growth compared with their parental strain (Ma and Liu 2010). Mutant Δyap1 displayed a 4-day-long lag phase and high sensitivity indicating a profound defect function affected by the YAP1 gene. The deletion mutation of YAP1 also showed increased sensitivity and decreased reduction activity toward coniferyl aldehyde (Sundstrom et al. 2010). This evidence supports the significant role of the YAP gene family in adaptation and tolerance to HMF. Thus, YAP1 regulated networks involving the functional reductase enzymes as described in a previous section is an important component for yeast tolerance and in situ detoxification of aldehyde inhibitors such as furfural, HMF, and coniferyl aldehyde. Excellent comprehensive reviews on Yap1p regulations involved in yeast stress response are available (Herrero et al. 2008; Rodrigues-Pousada et al. 2010).

PDR family and PDR1/3 involved cellular transport interactions

The second significant element for yeast tolerance and the in situ detoxification is the PDR gene family-centered functions that are regulated by Pdr1/3p as well as other co-regulator genes such as YAP1 and HSF1 (Fig. 5). The PDR genes encode plasma membrane proteins and function as transporters of ATP-binding cassette (ABC) proteins. These genes mediate membrane translocation of ions, and a wide range of substrates and often exhibit multiple functions in response to a large variety of unrelated chemical stresses (Mamnun et al. 2002; Moye-Rowley 2003; Jungwirth and Kuchler 2006; MacPherson et al. 2006). Many genes of the PDR family displayed consistent expressions of three- to 30-fold increases induced by furfural and HMF treatment (Liu et al. 2007; Song et al. 2009; Alriksson et al 2010; Ma and Liu 2010). Gene products of these increased transcripts are characterized in a broad range of protein categories such as drug/toxin transport for TPO1 and TPO4, transport ATPase for RSB1, and ABC transporters for PDR15 (Tomitori et al. 2001; Teixeira and Sá-Correia 2002; Ma and Liu 2010; Table 3). SNQ2, YOR1, PDR5, and PDR12 encoding proteins shared functions of all these three categories. These genes are considered as a core set candidate genes promoting cellular survival and adaptation to the inhibitor stress. In addition, many PDR proteins have functions as ATP binding and chemical resistance agent.

Most of these genes have the pleiotropic drug response element (PDRE) in their promoter regions. HMF-induced transcription factor genes PDR1 and PDR3 regulate gene expression under a large variety of unrelated chemical stress conditions by binding to the PDRE of target genes (Mamnun et al. 2002; Moye-Rowley 2003; Jungwirth and Kuchler 2006; MacPherson et al. 2006). Both Pdr1p and Pdr3p recognize CGG triplets oriented in opposite directions to form an inverted repeat and able to form homodimers or heterodimers to activate target gene expression (Mamnun et al. 2002; Hellauer et al. 1996). Many induced genes regulated by Pdr1p and/or Pdr3p in this group are involved in export of both xenobiotic compounds and endogenous toxic metabolites using ABC transporters (Pdr5p, Pdr15p, Snq2p, and Yor1p), lipid composition of the plasma membrane (Rsb1p and Ict1p), export of polyamines by polyamine transporters (Tpo1p and Tpo4p), DNA repairing (Mag1p and Ddi1p), and other functions (Katzmann et al. 1995; Mahé et al. 1996; Wolfger et al. 1997; De Risi et al. 2000; Onda et al. 2004; Alenquer et al. 2006; Salin et al. 2008; Ma and Liu 2010). At least eight genes induced by HMF were regulated by both Pdr1p and Pdr3p. These two regulators also recognize and activate other subsets of genes. For example, Pdr3p participates in certain processes that do not involve Pdr1p, such as regulating DNA damage-inducible genes MAG1 and DDI1 (Zhu and Xiao 2004). Similarly, certain genes are only regulated by Pdr1p, such as RSB1, ADH7, and PRE3 (Lee et al. 2002; Harbison et al. 2004; Kihara and Igarashi 2004). The PDR3 promoter contains two PDREs that can be autoregulated by itself in addition to being a regulon of Pdr1p (Delahodde et al. 1995; De Risi et al. 2000). PDR1 and PDR3 also demonstrated regulatory connections with a broad range of functional category genes as well as most active regulatory genes.

Gene deletion mutation assays of Δpdr1 displayed reduced transcriptional abundance for many genes including PDR5, PDR10, PDR15, YOR1, SNQ2, ICT1, GRE2, TPO1, YMR102C, and YGR035C compared with its parental strain (Ma and Liu 2010). The mutation Δpdr3 appeared to have a similar regulatory effect but at a less degree except for a clear positive effect on PGA3. These results confirmed the influence of PDR 1 and PDR3 on the expression of their potential regulons. It is likely that ABC transporters play a key role to export excessive toxic compounds such as furfural and HMF, and endogenous toxic metabolites from intracellular environment brought about by the inhibitor damage. As mentioned above, the shortcut of the TCA cycle could provide energy for the pumping of HMF and toxic metabolites by ABC transporters under the stress.

RSB1 and ICT1 are involved in phospholipid synthesis and transportation for membrane structure and functions that are responsible for yeast tolerance to organic solvents (Miura et al. 2000; Ghosh et al. 2008). It is possible that the induction of these PDR genes prevents the fast influx of HMF into cytoplasm and important organelles by membrane remodeling, thus, increasing the cell’s tolerance to HMF. MAG1 encodes a 3-methyladenine (3MeA) DNA glycosylase (Chen et al. 1990), which acts in the first step of a multistage base excision repair pathway for the removal of lethal lesions such as 3MeA and protects yeast cells from killing by DNA-alkylating agents (Fu et al. 2008). DDI1, located immediately upstream of MAG1 and transcribed in an opposite direction, encodes an ubiquitin-related protein and is involved in a DNA-damage cell-cycle checkpoint (Clarke et al. 2001). Regulatory interactions of PDR gene family are complex, and many genes appeared to be regulated by multiple transcription factor genes involving PDR1, PDR3, YAP1, and HSF1. Regulatory roles of PDR1 and PDR3 to HMF challenge were suggested by computational modeling (Song and Liu 2007; Song et al. 2009).

Protein modification interplays mediated by RPN4, HSF1, and co-regulators

Another important component of the yeast tolerance and its capability for in situ detoxification involves degradation of damaged proteins and protein modifications such as SHP1 and SSA4, regulated by transcription factor genes RPN4 and HSF1 as well as interplays with other closely related regulator genes such as YAP1 and PDR1 (Fig. 5). Chemical stress causes damage to protein conformation leading to protein unfolding and aggregation (Goldberg 2003). Small heat-shock proteins, acting as chaperones, assist in folding or refolding nascent or proteins and enzymes to maintain a functional conformation (Burnie et al. 2006). For example, HSP26 and SSA4 encoding chaperones were significantly induced to counteract the furfural–HMF complex damage to proteins. The deletion mutation of SSA4 displayed a significant longer lag phase under the HMF challenge, indicating its important role in adaptation and tolerance to HMF (Ma and Liu 2010). While the presence of chaperones contributes protein protection, prolonged inhibitor stress may result in irreversible protein damages. Misfolded or damaged proteins, especially aggregated proteins, are highly toxic to cells (Goldberg 2003). Degradation of misfolded and damaged proteins by the ubiquitin-mediated proteasome pathway plays an important role in maintaining normal cell function and viability (Goldberg 2003; Wang et al. 2008; 2010). Denatured proteins are targeted via the covalent attachment of ubiquitin to a lysine side chain, and polyubiquitinated proteins are finally delivered to proteasome to be degraded. Strains with deletion mutations of these genes are sensitive to HMF such as OTU1 and SHP1. It was suggested that the degradation of proteins by the ubiquitin-mediated proteasome pathway has regulatory roles on cell cycle, metabolic adaptations, gene regulation, development, and differentiation (Glickman and Ciechanover 2002).

At least 14 ubiquitin-related and proteasome genes (PRE1, PRE3, PRE6, PRE7, PRE10, PUP3, RPN9, RPN12, ECM29, RPT2, RPT3, RPT4, SHP1, and OTU1) for protein degradation were identified in relationship to HMF adaptation (Ma and Liu 2010). These genes encoding enzymes for degradation of damaged proteins maintain cell viability and functions under the inhibitor stress. The induction of these genes was predicted to be under the control of the transcription factor Rpn4p by binding to the proteasome-associated control element (PACE, 5’- GGTGGCAAA-3’), and the PACE was found in the promoter region of most ubiquitin-related and proteasome genes induced by HMF (Mannhaupt et al. 1999; Ma and Liu 2010). The expression of RPN4 was persistently enhanced over time during the lag phase. Rpn4p levels are regulated by the 26S proteasome via a negative feedback control mechanism (Xie and Varshavsky 2001). It is also required for regulation of genes involved in DNA repair and other cellular processes, such as DNA damage-inducible genes MAG1 and DDI1 (Harbison et al. 2004; Zhu and Xiao 2004). Interestingly, Rpn4p is a feedback regulator of YAP1 and PDR1 (Salin et al. 2008). This was further demonstrated by the comparative performance of the deletion mutation response to HMF. While it was able to grow and establish a culture normally without HMF challenge, the strain harboring Δrpn4 failed to recover in the presence of HMF (Ma and Liu 2010). These results confirmed the vital role of RPN4 involvement in yeast tolerance. The enhanced expression of HSF1 by HMF was consistent and statistically significantly greater. The up-regulated HSP26 and SSA4 for protein folding and refolding have been reported to be regulated by Hsf1p (Harbison et al. 2004; Ferguson et al. 2005; Ma and Liu 2010). Regulator gene HSF1 is an essential gene and a positive regulator of other transcription factor genes RPN4, PDR3, YAP5, and YAP6 (Lee et al. 2002; Harbison et al. 2004; Hahn et al. 2006; Workman et al. 2006). Therefore, the significant roles of HSF1 involved in the complex co-regulation networks for the yeast tolerance cannot be underestimated.

Conclusion and perspectives

It is clear that yeast tolerance and in situ detoxification of lignocellulosic hydrolysate inhibitors such as aldehydes, involve complex interplays of many genes at multiple levels at the genome scale. Functional reduction enzymes, largely involved in oxidoreductase activities, are the direct driving force in biotransformation of aldehyde inhibitors reducing the inhibitory damages. This group of genes and their interactions are regulated by members of the yeast activator protein gene family that is led by YAP1. These activities are closely related to the center metabolic pathways and ethanol fermentation. Tolerant yeast can be obtained with enhanced genetic background and reprogrammed pathways to overcome furfural–HMF stress. Identification of the inhibitor functional group and the use of structure–function strategy led to a better understanding of yeast tolerance and detoxification. Numerous members of the PDR gene family, showing consistent high levels of transcription abundance under the inhibitor stress, are considered as tolerance candidate genes. They are actively involved in exporting xenobiotic products and endogenous toxic metabolites and regulated mainly by PDR1 and PDR3. These function-specific and multifunctional cellular transporters and ATP binding agents located at cell wall and nucleus membranes are critical for cell survival and adaptation in the presence of the inhibitors. Another necessary component of the yeast tolerance involves genes functioning in protein folding, modification, and destination that is essential to reduce degraded protein toxicity and restore protein functions. Such genes are regulated by RPN4, HSF1, and other co-regulators. Furthermore, all regulators rolling these three basic components are co-regulatory and interactive. However, important elements of yeast tolerance are not limited to these outlined above. As indicated by recent transcriptome and proteomic studies, general stress response and several additional significant functional categories are recognized such as DNA repairing, oxidative stress, osmotic, and salt stress (Lin et al. 2009a, b; Ma and Liu 2010). While characterization and annotation of individual gene functions are necessary, identification of responsible functional categories and their interplays is of more importance from a global point of view. Time-course studies and temporal dynamic approaches reveal relevant and informative insight into a life response and should be used more widely for yeast tolerance mechanism studies. The snap-shot kind of method needs be limited and avoided. As demonstrated by transcription factor gene-linked regulatory interactions using systems biology approaches (Ma and Liu 2010), identification of major regulatory networks backboned with key regulators will further our understanding of the tolerance mechanisms in depth. Fortunately, having many advanced tools available in genomics, proteomics, metabolomics, biological engineering, and chemical engineering, more detailed understanding of molecular mechanisms and interplays of yeast tolerance at genome level is expected.