Genetic and Functional Investigation of Zn₂Cys₆ Transcription Factors RSE2 and RSE3 in Podospora anserina

Abstract

In Podospora anserina, the two zinc cluster proteins RSE2 and RSE3 are essential for the expression of the gene encoding the alternative oxidase (aox) when the mitochondrial electron transport chain is impaired. In parallel, they activated the expression of gluconeogenic genes encoding phosphoenolpyruvate carboxykinase (pck) and fructose-1,6-biphosphatase (fbp). Orthologues of these transcription factors are present in a wide range of filamentous fungi, and no other role than the regulation of these three genes has been evidenced so far. In order to better understand the function and the organization of RSE2 and RSE3, we conducted a saturated genetic screen based on the constitutive expression of the aox gene. We identified 10 independent mutations in 9 positions in rse2 and 11 mutations in 5 positions in rse3. Deletions were generated at some of these positions and the effects analyzed. This analysis suggests the presence of central regulatory domains and a C-terminal activation domain in both proteins. Microarray analysis revealed 598 genes that were differentially expressed in the strains containing gain- or loss-of-function mutations in rse2 or rse3. It showed that in addition to aox, fbp, and pck, RSE2 and RSE3 regulate the expression of genes encoding the alternative NADH dehydrogenase, a Zn₂Cys₆ transcription factor, a flavohemoglobin, and various hydrolases. As a complement to expression data, a metabolome profiling approach revealed that both an rse2 gain-of-function mutation and growth on antimycin result in similar metabolic alterations in amino acids, fatty acids, and α-ketoglutarate pools.

INTRODUCTION

Filamentous fungi are natural scavengers; hence, they have a very flexible metabolism that enables them to live in a variety of environments. They are able to use a wide range of carbon and nitrogen sources, and their biological and ecological capacity to degrade various organic chemicals has been reviewed recently (1, 2). Most filamentous fungi are saprophytes, i.e., they grow on complex compounds that are degraded or used to produce glucose and other carbon sources. Glycolysis and gluconeogenesis are two opposite pathways for glucose metabolism (degradation or biosynthesis of glucose), and the fungi have to adapt their carbon metabolism according to the available carbon source. This adaptation occurs through various mechanisms, including the reprogramming of gene expression (3–6).

Another feature of many filamentous fungi is the presence of an alternative terminal respiratory oxidase, called the alternative oxidase (AOX). AOX is a non-proton-pump enzyme of nuclear origin that is able to receive electrons from reduced ubiquinone and to catalyze the reduction of oxygen to water, bypassing the final steps of the cytochrome-mediated electron transport chain. In plants, the alternative oxidase is present in some tissues or developmental stages and may also be induced in response to a variety of stresses (7–9). In most fungi, it is undetectable, or present at very low levels, under normal growth conditions but becomes expressed when the cytochrome respiratory chain is restricted by inhibitors such as antimycin, azoxystrobin, and cyanide or by mutations (10–13). Under these conditions, the AOX bypasses respiratory complexes III and IV. It is believed to contribute to the maintenance of tricarboxylic acid (TCA) cycle turnover and to act as an overflow for electron transport, preventing the deleterious oxidative stress associated with the increased generation of mitochondrial reactive oxygen species (ROS). Several studies indicate that the AOX is upregulated in oxidative stress conditions (14, 15). It is developmentally regulated in the pathogenic dimorphic fungus Paracoccidioides brasiliensis (15) and plays an important role in the life cycle of the fungal plant pathogen Moniliophthora perniciosa (16). Recently, it has been shown that the AOX is relevant to the virulence of several human-pathogenic fungi, such as Cryptococcus neoformans (17), Aspergillus fumigatus (18), and Paracoccidioides brasiliensis (19). In all of these cases, the AOX is important in the fungal defense against oxidative stress imposed by the host.

In the filamentous ascomycetes Neurospora crassa, Podospora anserina, and Aspergillus nidulans, recent data have shown that two zinc cluster transcription factors, highly conserved in these three species, are required for the expression of the aox gene (12, 20, 21). Surprisingly, in P. anserina and A. nidulans, these transcription factors also have been shown to be responsible for regulating the two key enzymes specific for gluconeogenesis, phosphoenolpyruvate carboxykinase (PCK) and fructose-1,6-biphosphatase (FBP) (20, 21). In P. anserina, RSE2 and RSE3 are also involved in life span control (20). Whether these transcription factors regulate the expression of genes other than aox, fbp, and pck is an open question.

In order to better understand the role of these conserved transcription factors and their impact on a potential metabolic adaptation to an impairment in respiration, we have used a genetic approach and isolated several mutations in P. anserina that lead to an induction in aox expression. All of them are located in rse2 or rse3. Deletions at some positions corresponding to these mutations were constructed. Their effect on the expression of the aox gene was analyzed. The data enabled us to suggest a functional organization of the two transcription factors. In parallel, we have examined transcriptomic changes induced in gain-of-function and knockout (KO) rse2 and rse3 mutants. This analysis revealed that RSE2 and RSE3 modulate the expression of genes other than aox, fbp, and pck. They also regulate the expression of genes encoding NDI1, another alternative respiratory enzyme, a Zn₂Cys₆ transcription factor, a flavohemoglobin, and genes involved in various hydrolytic activities, suggesting that these transcription factors are involved in a complex adjustment to the environment. Additionally, a metabolome analysis revealed that both an rse2 gain-of-function mutation and growth on antimycin lead to similar metabolic alterations in amino acids, fatty acids, and α-ketoglutarate pools, suggesting that the RSE proteins play an important role in the metabolic response to mitochondrial respiration impairment.

MATERIALS AND METHODS

P. anserina strains, growth conditions, genetic analysis, life span measurements, and transformation.

All of the strains used in this study were derived from the s wild-type strain. The long-lived cox5::ble strain has been described previously (22). The rse2^Y326D, rse2^G329S (initially referred to as rse2^Y300D and rse2^G303S, respectively), rse3^G642V, Δrse2, Δrse3, and Δrse2 Δrse3 strains have been described already (20). The ΔKu70 strain provides an efficient method for gene deletion and allelic replacement (23). The nuo19.3 mutant has been described previously (24). Standard culture conditions, media, and genetic methods for P. anserina have been described (25, 26). The M2 medium is a minimal medium in which carbon is supplied as dextrin (0.5%) and nitrogen as urea (0.05%). Acetate medium is a minimal medium in which dextrin is replaced by sodium acetate (1.4%). When necessary, hygromycin, nourseothricin, and antimycin A were added to the medium at 75 μg/ml, 50 μg/ml, and 20 μg/ml, respectively. The germination medium (G) contains ground corn meal (5%) and ammonium acetate (0.5%).

Life spans were measured on M2 medium on three subcultures derived from two to five independent spores exhibiting a given genotype as previously described (20). Transformation of P. anserina was carried out as described previously (27) on protoplasts obtained by incubation with 40 mg/ml glucanex (Laffort).

Quantitative RT-PCR experiments.

For RNA extraction, the different strains were grown on cellophane sheets (Bio-Rad) overlaid on petri dishes containing M2 medium for 24 h at 27°C. The cellophane sheets were then transferred to petri dishes containing M2 (control) or M2 supplemented with antimycin A or acetate medium for incubation during 6 h at 27°C. Mycelia were harvested, flash frozen in liquid nitrogen, and ground in a Fastprep apparatus (MP Bio) with glass beads and the RLT buffer provided with the RNA extraction kit (RNeasy plant kit [Qiagen]). For quantitative reverse transcription-PCR (RT-qPCR) analysis, 2 μg of freshly extracted total RNA was reverse transcribed with oligo(dT)₁₅ using Superscript II reverse transcriptase (Invitrogen). The RT-qPCR amplification mixture (10 μl) was examined with the LightCycler 480 SYBR green I master kit (Roche) on 2 μl of a 1/10 dilution of the reverse transcription reaction. Reactions were run on and read in a LightCycler 480 real-time PCR system (Roche). Data were analyzed with software using the “second derivative” method of quantification. The cycling conditions comprised 5 min at 95°C and 40 cycles at 95°C for 10 s, 60°C for 20 s, and 72°C for 20 s. For a given gene, each plate contained cDNA samples representing all of the conditions examined and a standard curve of three serial dilution points (1/4 dilutions). All measures were performed with three biological replicates for each condition. For microarray validation, the biological replicates used for RT-qPCR were different from those used in microarray experiments. The gene-specific primers used are listed in Table 1. For all genes, a non-reverse-transcribed control was performed on each biological replicate under the same conditions as the reverse-transcribed samples to control the absence of genomic DNA contamination. The reference gene used was Pa_3_5110 (gene gpd, encoding the glyceraldehyde phosphate dehydrogenase), as previously described (20).

TABLE 1.

Primers used for quantitative RT-PCR experiments

Gene no.	Gene name	Primers (5′ to 3′)
Pa_3_5110	gpd	CACCGAGGACGAGATTGTCT
		TCAGGGAGATACCAGCCTTG
Pa_3_1710	aox	GAGGAAATGTTGGCAGTGGT
		GATGTCTGTTCCCCATCGAC
Pa_4_9360	fbp	CACCGGTGACTTTACGCTCC
		GGAGAATTGGAGGGCGTGGC
Pa_4_3160	pck	ACCAAACCATCCGACATGC
		GGTCTTGTTTACTGTGTTGA
Pa_5_1700	fhb	TACGAGTTTGGAGACATGCG
		GAGTCACCCAGGAACAAACC
Pa_7_1820	ndi1	CTTATTCCTCCACGACATCGC
		GGAATTGGGGGACGATGACG
Pa_4_3860		AGGAGGGGGATGGCGATG
		CCTGCGCCTTCACATGATG
Pa_6_4030		GTTTGGTTCCCTGGGAGTC
		GGCTCGTTGCTGGAACTTC
Pa_7_1770		CGGCAAGGGCCATTTCGTC
		GTAATTCATCGGGGCGTGG
Pa_7_2610		GGCGGATGAACTCAAGGGG
		AGGCCCATCATGCCGACG

Construction of P. anserina strains carrying mutated rse2 and rse3 alleles.

The mutations rse2^Δ223, rse2^Δ329, rse2^Δ452, rse2^Δ454-504, rse3^Δ251, rse3^Δ438, rse3^Δ640-645, and rse3^Δ646-696 were introduced into P. anserina by gene replacement according to a previously described protocol (23; also see Fig. S1 in the supplemental material). The mutation rse2^Δ324-346 was introduced ectopically as described in Fig. S1. The primers used are listed in Table S1.

Microarray analyses.

The construction and optimization of gene expression microarrays for P. anserina has been described previously (28). Briefly, 1-μg aliquots of total RNA were amplified and Cy labeled with Agilent's low-RNA input fluorescent linear amplification (LRILAK) plus kit and the two-color RNA spike-in kit (Agilent). Four biological replicates labeled with Cy-3 for each of the different experimental conditions were compared to a common reference labeled with Cy-5 in indirect comparisons. The common reference was obtained by mixing RNA extracted from the different conditions as indicated in reference 28. Microarrays were scanned using the Agilent DNA microarray scanner (Agilent) at a resolution of 5 μm using the extended dynamic range (XDR) feature. A moderated t test with adjustment of P values (29) was computed to measure the significance of each difference of expression. Genes were considered differentially transcribed if their absolute fold change of up- or downregulation was greater than 2 with a P value of <0.005.

Metabolite extraction, identification, and quantification.

The strains were grown on cellophane sheets (Bio-Rad) overlaid on petri dishes containing M2 medium (M2) or M2 supplemented with antimycin A (M2+AA) for 48 h at 27°C. Each experiment was performed in triplicate. Around 100-mg aliquots of mycelium were harvested for each sample and were lyophilized. Fifteen mg of dry material was ground to a fine powder and extracted into 80% methanol-20% water containing two internal standards (α-aminobutyrate and ribitol), both at 100 μM as described in reference 30. Gas chromatography coupled with time-of-flight mass spectrometry (MS) analysis was carried out as previously described (30). Metabolite derivatives were identified by comparison of the fragmentation pattern to MS databases using a match cutoff criterion of 750/1,000 and by retention index (RI) using an alkane series as standards. Because automated peak integration was occasionally erroneous, integration was verified manually for each compound in all analyses. This enabled detection of 97 different metabolites. Integrated peak areas were obtained after deconvolution by the LECO PEGASUS III ChromaTOF software and quantified using the appropriate software option. These data were then normalized to the internal standard (ribitol) peak area for each injection. Fold change was calculated as the ratio of the testing condition value (the wild-type strain grown on M2+AA or the rse2^Y326D mutant grown on M2) to the control value (the wild-type strain grown on M2). The level of significance was set at P < 0.05 (t test).

Microarray data accession number.

All microarray data are MIAME compliant. The raw data have been deposited in the MIAME-compliant Gene Expression Omnibus database and are accessible through the GEO series accession number GSE51360.

RESULTS

Isolation of novel mutations within the rse2 and rse3 genes.

Prior to this study, two alleles in rse2 and one allele in rse3 had been identified as being responsible for constitutive expression of the AOX. These two genes encode two Gal4p-type C₆ zinc cluster proteins. In order to find potential additional genes controlling the regulation of the AOX, we used a screening strategy based on the isolation of mutations able to improve the phenotype of the respiratory cox5::ble mutant (20). This mutant presents alterations in germinating and growing mycelium, a reduction of the pigmentation and growth rate, female sterility, and a spectacular increase in life span. All of these characteristics are partially suppressed by overexpression of the AOX (10, 20). Of 40 cox5::ble mutant cultures, 23 independent fast-growing sectors appeared spontaneously. These revertant sectors were selected and crossed with the wild-type strain. In all cases, two types of cox5::ble ascospores were obtained: those giving rise to the original mutant phenotype (cox5::ble genotype) and those giving rise to an improved fast-growing mycelium (cox5::ble, su genotype), consistent with the presence of extragenic suppressors in all revertants. For each of the 23 strains carrying a suppressor mutation, rse2 and rse3 were sequenced. The results showed that eight and ten strains carried a mutation in rse2 or rse3, respectively, and five did not harbor mutations in these two genes.

Analysis of the five latter strains showed only a weak suppression of the cox5::ble mutation. The pigmentation of the mycelium was restored and the growth rate was slightly increased (about 40%), but the germination rate of the cox5::ble mutant was not improved. RT-qPCR experiments indicated that the expression of the aox gene was not induced in the strains bearing these undetermined suppressor mutations in a cox5⁺ context. Thus, we can conclude that these mutations did not suppress the cox5::ble mutation via overexpression of the AOX. These mutants will be described elsewhere.

The positions of the newly identified and the previously published rse2 and rse3 mutations within the RSE2 and RSE3 protein sequences are shown in Fig. 1. Analysis of the rse2 sequence revealed that our first annotation was incorrect and that exon 5 was truncated by 78 nucleotides. The previously described rse2^Y300D and rse2^G303S mutations correspond to rse2^Y326D and rse2^G329S alleles, respectively, in the correct annotation. All of the new alleles are substitutions, except one in rse3 that contains an 18-bp duplication of nucleotides causing a repeat of the amino acids 640 to 645 (rse3^Dup640-645). Some positions were modified by independent mutations: E251 (E251K and E251G), A438 (A438T and A438V), and V626 (V626M and V626L) in rse3. The mutations G634V in rse3 and T449I in rse2 also were isolated twice independently. Therefore, the screen is practically saturated and has not identified genes others than rse2 and rse3 that are able to activate the expression of AOX.

FIG 1.

Localization of the mutations in the RSE2 and RSE3 proteins and effects of the mutations in a cox5⁺ and cox5::ble context. Schematic representation of the RSE2 (A) and RSE3 (B) proteins. The Zn₂Cys₆ zinc cluster DNA-binding motif is indicated (DBD). The mutations screened as gain of function of the aox expression are indicated by a black line for substitutions and by a white triangle for the 18-bp duplication causing a repeat of the amino acids 640 to 645 inside the RSE3 protein. An asterisk means that the substitution was found independently in two mutants. The deletions created in each protein are shown by a black triangle. For each protein, the last line represents the C-terminal truncated protein. In a cox5⁺ context, the ability of the strains to grow on a medium containing 20 μg/ml antimycin A (AA) is shown by the following scoring system: +, growth after a delay; ++, growth without delay; −, no growth; ε, slow growth, slower than that of the wild type, after a delay. In a cox5::ble context, the growth rate (in mm/day) and the mycelium aspect of the different strains is indicated. The rse2^Δ324-346 allele was integrated ectopically in a Δrse2 strain (see the text).

Functional analysis of the transcription factors RSE2 and RSE3.

The 18 new rse2 or rse3 mutants analyzed in this study shared similar phenotypes. Like the previously described mutants rse2^Y326D, rse2^G329S, and rse3^G642V, they constitutively expressed the AOX at a high level, as shown by their ability to partially suppress the cox5::ble mutation defects, and to grow without delay on medium containing antimycin in a cox5⁺ context (Fig. 1). Antimycin is an inhibitor of the respiratory complex III, and this leads to an induction of the AOX. In the presence of antimycin, the wild-type strain grows after a delay necessary for the induction, whereas strains displaying a deregulated constitutive expression of the AOX grow without a delay (10, 20). All of the rse2 and rse3 substitutions and duplication indicated in Fig. 1 were dominant when recovered in a heterokaryotic mycelium containing cox5::ble rse2⁺ rse3⁺ nuclei associated with cox5::ble nuclei carrying any of these rse2 or rse3 mutations. Thus, they correspond to gain-of-function mutations.

The mutations obtained in rse2 and rse3 are located between amino acids 223 and 491 in the RSE2 protein (which corresponds to 56% of the protein) and between amino acids 251 to 642 in the RSE3 protein (which corresponds to 64% of the protein). The functional organization of a prototypic Zn₂Cys₆ protein consists of an N-terminal Zn₂Cys₆ DNA-binding motif and a transcriptional activation region frequently located in the C terminus. The mechanisms required to convert Zn₂Cys₆ proteins from an inactive to an active state apparently are not conserved. Inhibitory regions thought to play a role in regulating the transcriptional activity have been characterized in several of these proteins (31–33). Due to the relatively high frequency of suppressor mutations obtained, we hypothesized that at least some of these gain-of-function mutations correspond to the loss of an inhibitory function.

To test this possibility, we first created an RSE2 protein carrying an internal deletion of 23 amino acids (Δ324-346), which removes a region that covers four substitution sites (Fig. 1). The plasmid containing the mutant allele rse2^Δ324-346 and a hygromycin resistance cassette was used to transform the Δrse2 strain, leading to ectopic integrations. Transformants resistant to hygromycin were selected and first tested on antimycin. Our rationale was that the ability to grow on this medium would mean that the rse2^Δ324-346 transgene has retained its functionality, since the recipient strain was Δrse2. Among 10 transformants able to grow on antimycin, two were crossed with the cox5::ble strain, leading to four genotypes containing the cox5::ble mutation: cox5::ble Δrse2 spores that were unable to germinate, as already described (20), cox5::ble rse2⁺ spores that displayed a characteristic growth rate of 2.5 ± 0.3 mm/day, and cox5::ble Δrse2 rse2^Δ324-346 and cox5::ble rse2⁺ rse2^Δ324-346 spores that displayed a growth rate of 3.5 ± 0.1 mm/day. Viability of the cox5::ble Δrse2 rse2^Δ324-346 strain confirmed that the RSE2(Δ324-346) protein is functional. Its improved growth rate indicates that the deletion renders the protein constitutively more active, which suggests that the region between amino acids 324 and 346 displays a slight inhibitory function.

To examine in more detail the role of the regions of RSE2 and RSE3 in which gain-of-function mutations were obtained, we have introduced site-specific deletions of individual codons. We chose to delete codons corresponding to residues E223, G329, and F452 in RSE2 and to residues E251 and A438 in RSE3. We also deleted the motif RDGKLE in RSE3 (residues 640 to 645) whose duplication results in a gain of function. The activation domains of the Zn₂Cys₆ transcription factors, in contrast to the DNA-binding domains, are less understood, and their structure seems not to be highly conserved. For several of these transcription factors, they have been localized to a C-terminal region of about 100 amino acids (31). To investigate whether this is also the case for RSE2 and RSE3, the C termini of the proteins were deleted. Having already deleted residues 640 to 645 in RSE3, we decided to remove the last 51 amino acids of the protein from residue 646 [RSE3(Δ646-696)] and to delete the last 51 amino acids of RSE2 [RSE2(Δ454-504)]. The mutated alleles rse2^Δ223, rse2^Δ329, rse2^Δ452 rse2^Δ454-504, rse3^Δ251, rse3^Δ438, rse3^Δ640-645, and rse3^Δ646-696 were introduced at the corresponding locus in the wild-type strain. The primary transformants were crossed with the cox5::ble mutant, and monokaryotic cox5⁺ and cox5::ble spores bearing a mutated rse2 or rse3 allele (identified in the progeny by resistance to nourseothricin) were obtained. The activity of the mutated proteins was revealed by testing their ability to confer viability on a cox5::ble strain and to allow a cox5⁺ strain to grow on medium containing antimycin. The inducible or constitutive nature of the activity of the mutated proteins was revealed in a cox5⁺ context depending on the ability of the strains to grow after or without a delay on antimycin. Results are shown in Fig. 1. All of the mutations conferring a constitutive activity to RSE2 and RSE3 showed a suppressor effect of the cox5::ble phenotype.

The deletion of residue 223 in the protein RSE2, like the deletion of residue 251 and of the RDGKLE (residues 640 to 645) motif in the protein RSE3, led to the same effect as the E223K substitution in RSE2, the E251K/G substitutions, and the duplication of the motif in RSE3, i.e., the ability of the cox5⁺ strain to grow on antimycin without delay and a strong improvement of the cox5::ble phenotype. The deletion of residue 329 in RSE2 resulted in a weaker effect than that of the G329S substitution but also caused an activation of the protein. The cox5⁺ rse2^Δ329 strain grew with a slight delay on antimycin, and the growth rate of the cox5::ble rse2^Δ329 strain was 4.0 ± 0.2 mm/day; that of the cox5::ble rse2⁺ strain was 2.5 ± 0.3 mm/day. Since the deletion of residues E223 and G329 in RSE2 and of residues 251 and 640 to 645 in RSE3 led, like their substitution or duplication, to constitutive hyperactive proteins, we propose that these residues belong to regions that inhibit the transcription in the absence of the inducer and therefore to regions involved in receiving the inducer signal. In contrast, the deletion of the 51 C-terminal amino acids in the two proteins (rse2^Δ454-504 and rse3^Δ646-696) abolished the function of these proteins. The cox5⁺ strains were unable to grow or grew very poorly on antimycin, and the cox5::ble ascospores displayed a very long germination delay (2 to 3 weeks instead of 4 days for cox5::ble rse2⁺ rse3⁺ ascospores) and produced very tiny thalli that were unable to grow further. These data are in agreement with a C-terminal localization of the activation domain, as is seen in many Zn₂Cys₆ proteins. The deletion of the residues F452 in RSE2 (rse2^Δ452) and A438 in RSE3 (rse3^Δ438) also resulted in a loss of function. This could mean that these residues participate in the activation of the protein or that the deletions generated nonfunctional proteins. A summary of the properties of the different mutations and of the regions that they identify is given in Table 2.

TABLE 2.

Properties of different rse2 and rse3 mutations and of the regions that they identify^a

Mutation	Activity (plus inducer)	Nature of activity	Possible consequence of the mutation
None	+	I
rse2Δ223	+++	C	Loss of function of a region involved in the inhibition of the transcription in the absence of inducer
rse2E223K	+++	C	Loss of function of a region involved in the inhibition of the transcription in the absence of inducer
rse2Δ329	++	C	Loss of function of a region involved in the inhibition of the transcription in the absence of inducer
rse2G329S	+++	C	Loss of function of a region involved in the inhibition of the transcription in the absence of inducer
rse2Δ452	−		Loss of function of an activation region or generation of a nonfunctional protein
rse2F452L	+++	C	Gain of function of an activation region
rse2Δ454-504	−		Loss of function of an activation region
rse3Δ251	+++	C	Loss of function of a region involved in the inhibition of transcription in the absence of inducer
rse3E251K/G	+++	C	Loss of function of a region involved in the inhibition of the transcription in the absence of inducer
rse3Δ438	−		Loss of function of an activation region or generation of a non-functional protein
rse3A438T/V	+++	C	Gain of function of an activation region
rse3Δ640-645	+++	C	Loss of function of a region involved in the inhibition of the transcription in the absence of inducer
rse3Dup640-645	+++	C	Loss of function of a region involved in the inhibition of the transcription in the absence of inducer
rse3Δ646-696	−		Loss of function of an activation region

^aThe different mutations are indicated. The transcriptional activity of the RSE proteins in the presence of an inducer (loss of function of the respiratory complex IV) is represented by + (activity of the wild-type proteins), ++ (weaky increased activity), +++ (hyperactivity), and − (absence of activity). It was evaluated from the growth rate of the cox5::ble strain carrying each mutation. The inducible (I) or constitutive (C) nature of this activity was evaluated from the capacity of the cox5⁺ strain carrying each mutation to grow with, or without, delay on antimycin (Fig. 1). The function of the region that they identify was deduced from these properties.

We previously showed that the life span of the rse2^Y326D and rse3^G642V mutants was slightly decreased, whereas that of the Δrse2, Δrse3, and Δrse2 Δrse3 mutants significantly increased (20). The life span of several of the new mutants was determined. Results are given in Table 3. The gain-of-function mutations of rse2 and rse3 did not significantly modify the life span. In contrast, in accordance with our previous results, the three loss-of-function strains examined (rse2^Δ452, rse2^Δ454-504, and rse3^Δ646-696 strains) had an extended longevity (about 15 cm versus 9 cm for the wild type). To complete this phenotypic analysis, we looked at the fertility of the KO strains and found that the fertility of the Δrse3 and Δrse2 Δrse3 strains is greatly diminished, whereas it is not modified in the Δrse2 strain. Homozygous crosses for Δrse3 were not sterile but produced 4 to 5 times fewer mature perithecia than wild-type homozygous crosses.

TABLE 3.

Life spans of rse2 and rse3 mutants^a

Strain	Life span (cm ± SD)
Wild type	8.6 ± 1.9
rse2E223K	8.9 ± 1.5
rse2G329S	7.6 ± 1.1
rse2F452L	9.0 ± 0.9
rse2Δ452	17.3 ± 3.7
rse2Δ454-504	15.4 ± 4.7
rse3E251K	7.1 ± 0.7
rse3Δ251	9.0 ± 1.4
rse3Δ640-645	7.7 ± 1.1
rse3Δ646-696	12.6 ± 1.9

^aLife spans were measured on M2 medium at 27°C in race tubes on three subcultures derived from two to five independent spores. The mean life span is given.

Identification of genes regulated by the RSE2 and RSE3 transcription factors.

In a previous study, we showed that RSE2 and RSE3 activated the expression of the aox gene and the gluconeogenic genes fbp and pck (20). The connection between expression of the aox, fbp, and pck genes may not be obvious. It is observed when the wild-type strain is grown for 6 h on medium containing antimycin. Figure 2A shows an increase of the expression of the aox, fbp, and pck genes of about 700-, 2.5-, and 7-fold, respectively, under these conditions. It is also observed when the wild-type strain is grown on medium containing acetate as the sole source of carbon with an increase in the expression of the aox, fbp, and pck genes of about 3-, 8-, and 80-fold, respectively (Fig. 2B). Therefore, it seems that there is a correlated expression of the aox, fbp, and pck genes in both respiratory deficiency and gluconeogenic conditions. However, the induction level of the aox gene on one hand and of the gluconeogenic genes on the other hand is quite different on antimycin and acetate, suggesting that these genes respond to different environmental signals according to the growth medium. We have previously shown that strains deleted for one or the other of the transcription factors (Δrse2 rse3⁺ and rse2⁺Δrse3 strains) were unable to grow on medium containing antimycin (20). The Δrse2 rse3⁺ and rse2⁺ Δrse3 mutants also grew very poorly on acetate and exhibited an extremely spindly mycelium and extremely reduced growth rate. Interestingly, the Δrse2 Δrse3 double deletion mutant showed no growth at all on acetate, revealing a synergistic effect between RSE2 and RSE3 for the activation of the gluconeogenesis pathway.

FIG 2.

RT-qPCR analysis of the aox, fbp, and pck transcript levels under different culture conditions. Relative abundance of aox, fbp, and pck transcripts in the wild-type strain grown under normal conditions (gray) or in the presence of antimycin A (A) or acetate (B) during 6 h (black). Levels are represented in relative units; the normalized amount of each transcript in the wild-type strain grown under standard conditions is set to 1. Mean values ± standard deviations are shown.

In order to identify more exhaustively the genes whose expression is controlled by RSE2 and RSE3, transcriptome profiles were determined in the single rse2^Y326D and rse3^G642V and in the double rse2^Y326D rse3^G642V mutants, as well as in the single Δrse2 and Δrse3 and in the double Δrse2 Δrse3 mutants. RNA was extracted from cultures after 48 h and prepared for hybridization to whole-genome gene expression microarrays developed for P. anserina (28) (see Materials and Methods). Analysis of transcription patterns of the wild type and the six mutants revealed 598 genes to be significantly differentially expressed 2-fold or more in at least one mutant strain compared to the wild type (see Data Set S1 in the supplemental material). Given the finding that both RSE2 and RSE3 are required to grow on antimycin and acetate and that both DNA-binding domains of the two N. crassa (Aod2 and Aod5) and the two A. nidulans (AcuM and AcuK) orthologues are required to bind to a sequence element in the aox and the acuF (pck) promoters (12, 21), it is reasonable to assume that these transcription factors work as heterodimers. Thus, we hypothesized that the genes regulated by RSE2 and RSE3 will show similar regulation in the three gain-of-function or the three KO mutants. The binding motif (AIM) of the two N. crassa orthologues is CGG(N)₇CGG. It is conserved in the aox promoter of closely related species, notably in P. anserina (11), so we hypothesized that the direct targets of RSE2 and RSE3 will present an AIM motif in their 5′ region. Therefore, we focused on two groups of genes: those that were up- or downregulated in the three rse2^Y326D, rse3^G642V, and rse2^Y326D rse3^G642V strains (Fig. 3, clusters A and C) and those that were up- or downregulated in the three KO strains (Fig. 3, clusters B and D). The presence or absence of one or more copies of the AIM motif in the 1,000 bp upstream of the ATG of the different genes and their annotated function are shown in Fig. 3. Cluster A (Fig. 3) contains 18 genes whose expression is increased in the three gain-of-function mutants. It contains the aox, fbp, and pck genes as expected, as well as ndi1, which encodes an alternative NADH dehydrogenase (24), and genes that encode putative glutamate dehydrogenase (GDH; Pa_2_6020), glutamate synthase (GOGAT; Pa_3_6960), flavohemoglobin (FHB; Pa_5_1700), phenylalanine ammonia-lyase (Pa_5_1980), α/β hydrolase fold protein (Pa_6_4030), haloacid dehalogenase (Pa_2_1020), two putative epoxide hydrolases (Pa_7_1770 and Pa_4_3860), a putative Zn₂Cys₆ transcription factor (Pa_7_2610), and 5 genes of unknown function (34). It is worth noting that about 70% of the genes of cluster A (12 out of 18) contain one or more copies of the AIM motif in their upstream sequence (Fig. 3), whereas this percentage is about 6% for the genes whose expression did not change in any of the rse mutants. Therefore, genes with an upstream AIM motif are clearly overrepresented in cluster A relative to their percentage in the genome. In a previous study (20), this motif was not identified for pck and fbp; this may have been due to a problem in the genome sequence data at the time. Cluster D (Fig. 3) contains 14 genes that showed decreased expression levels in the 3 KO mutants. Half of them encode proteins without functional annotation, and the other genes are involved in secondary metabolism (one polyketide synthase [PKS], Pa_1_11870, one hybrid PKS-NRPS, Pa_6_10100) and in carbohydrate metabolism (two glycoside hydrolases, Pa_1_11790 [family 15] and Pa_1_1010 [family 11], and one pyruvate decarboxylase, Pa_5_4580). Except for a few genes, including the PKS-NRPS gene, the majority of the genes of this cluster do not contain an upstream AIM motif. The genes of clusters A (upregulated in the three gain-of-function mutants) and D (downregulated in the three KO mutants) are potential targets of these transcription factors acting as activators. In contrast, the RSEs seem to be negative regulators of genes of clusters B (8 genes) and C (4 genes), which showed increased expression levels in the three KO mutants and decreased expression levels in the three gain-of-function mutants, respectively. Clusters B and C are predominantly enriched in orphan and nonannotated genes. Only two genes present a functional annotation: Pa_4_2420, which encodes a putative glycoside hydrolase (cluster B), and the glucose-repressible gene grg1 (Pa_2_11900; cluster C). It has been shown in N. crassa that the orthologue (grg-1 or ccg-1) is an abundantly expressed gene displaying complex regulation that involves clock control, light control, and metabolic control (35). It is intriguing that the expression of grg1 decreased in the rse2^Y326Dand rse3^G642V mutant strains that display a decreased life span (20), whereas its expression has been shown to be reduced in the long-lived mutant grisea and increased during senescence (36). It should be noted that in the three KO mutants, we failed to identify up- or downregulated gene(s) that could easily account for the increased longevity of the Δrse2, Δrse3, and Δrse2 Δrse3 mutants (20).

FIG 3.

Clustering of genes similarly up- or downregulated in the three gain-of-function or the three loss-of-function rse2 and rse3 mutants. The 6 columns indicate the fold change in gene expression between the wild-type strain and rse2^Y326D (1), rse3^G642V (2), rse2^Y326D rse3^G642V (3), Δrse2 (4), Δrse3 (5), and Δrse2 Δrse3 (6). The color panel indicates relatively decreased (red), increased (green), and unchanged (black) expression levels. Cutoffs of 2-fold changes and statistical significance of P < 0.05 were used on the transcriptomic data. Genes are ordered in 4 clusters: A, upregulated in the 3 gain-of-function mutants (yellow bar); B, upregulated in the 3 KO mutants (green bar); C, downregulated in the 3 gain-of-function mutants (blue bar); D, downregulated in the 3 KO mutants (red bar). For each gene, the number, the presence (+) or absence (−) of the binding motif CGG(N)₇CGG in the 1,000 bp upstream ATG, and the gene name or function are given on the right. HP, hypothetical protein.

We focused on genes of cluster A that are upregulated in the 3 gain-of-function mutants and investigated whether, as observed for the aox, fbp, and pck genes, the transcription of the other annotated genes of this cluster is activated in the presence of antimycin and acetate and depends on RSE proteins. Therefore, the expression of ndi1, gdh, gogat, fhb, Pa_5_1980 (phenylalanine ammonia-lyase), Pa_6_4030, Pa_2_1020, Pa_7_1770 and Pa_4_3860 (hydrolases), and Pa_7_2610 (transcription factor) was tested in the wild-type and in the Δrse2 Δrse3 double mutant strains in the presence of antimycin or acetate (6 h). We found that gdh, gogat, and Pa_5_1980 were not activated in the presence of antimycin or acetate in either strain, suggesting that these genes are false positives. For some unknown reason, we were not able to amplify the Pa_2_1020 hydrolase mRNA. Results of RT-qPCR experiments for the other genes are shown in Fig. 4. The expression of all of these genes was activated in the wild-type strain grown on antimycin (Fig. 4A). The strongest inductions were observed for fhb (about 250×), Pa_7_2610, encoding a Zn₂Cys₆ transcription factor (about 80×), and Pa_4_3860, encoding a putative epoxide hydrolase (about 50×). All of these genes were also activated on acetate (Fig. 4B) but to a much lower level than on antimycin. The induction factor is about 8 for the gene encoding the transcription factor, 3 for the three hydrolases, and 2 for fhb and ndi1. The antimycin and acetate induction depends on RSE2 and/or RSE3 for all of the genes tested, since it was abolished in the Δrse2 Δrse3 strain (Fig. 4). It is worth noting that among the nine genes that were induced on antimycin in an RSE-dependent manner, eight contained an upstream AIM motif; only one, fhb, did not. However, while in vivo data indicate that the expression of the aox, fbp, and pck genes is completely dependent on functional RSEs and that strains deleted for these transcription factors cannot grow on antimycin or acetate, it may be possible that, for the expression of at least some of the other genes, other regulators take precedence over the RSEs. For instance, ndi1 can be expressed without RSE2. This was shown by associating the Δrse2 mutation with the loss-of-function mutation of complex I, nuo19.3. We have previously shown that the expression of NDI1 is essential to overcome lethality due to the absence of complex I (24); however, we found that the double mutant Δrse2 nuo19.3 is viable, indicating that ndi1 is still expressed.

FIG 4.

Relative abundance of Pa_5_1700, Pa_7_2610, Pa_4_3860, Pa_7_1820, Pa_6_4030, and Pa_7_1770 transcripts in the wild-type strain and the Δrse2 Δrse3 strain grown under different conditions. (A) Total RNAs were extracted from cultures of the wild-type strain (wt) and the Δrse2 Δrse3 mutant grown under standard conditions (gray) or in the presence of antimycin A during 6 h (black). (B) Total RNAs were extracted as described for panel A from cultures grown under standard conditions (gray) or after being transferred for 6 h on acetate (black). Levels are represented in relative units, and the normalized amount of each transcript in the wild-type strain grown under standard conditions is set to 1. Mean values ± standard deviations are shown.

Analysis of the effect of antimycin and of the rse2^Y326D mutation by metabolite profiling.

In order to investigate whether the transcriptomic changes observed in a mutant strain with constitutive activation of RSE2 and RSE3 were correlated with metabolic changes, we examined a pool of metabolites in the rse2^Y326D mutant and the reference strain. We also monitored the changes in metabolite amount in the wild-type strain in response to antimycin (see Data Set S1 in the supplemental material). Ninety-seven metabolites representing different pathways, mainly of primary metabolism, were detected and quantified. Most of them were significantly affected by antimycin treatment (Fig. 5A) and in the rse2^Y326D mutant (Fig. 5B). When these changes were mapped onto a schematic metabolic network, it could be seen that they were not randomly distributed but instead occurred in specific regions of the network (Fig. 5). After 48 h of antimycin treatment of the wild-type strain, there was a pronounced decrease in α-ketoglutarate that indicates a perturbation of the TCA cycle. On the other hand, there was a strong increase in the amounts of numerous amino acids and fatty acids. The level of arginine and glutamine increased by 9- and 15-fold, respectively. The level of fatty acids (linoleic, linolenic, palmitoleic, and palmitic) increased by 2- to 111-fold. These data suggest an increase in the flux through cataplerotic pathways that remove TCA cycle intermediates to convert them to amino acids and fatty acids. Increased levels of mannitol (9-fold), glycerol-2-P, and glycerol-3-P (2- to 3-fold) also were found, suggesting a higher rate of flux through pathways downstream of glycolysis. Remarkably, in the rse2^Y326D mutant, many similarities can be observed, in particular a decrease in α-ketoglutarate and an increase in amino acids, fatty acids, glycerol-2-P, and glycerol-3-P. It is worth noting that the transcripts encoding enzymes involved in the TCA cycle, fatty acid, and amino acid biosynthetic pathways were not significantly changed in the rse2^Y326D mutant. This strongly suggests that the metabolic changes occurring in the rse2^Y326D mutant were not the direct result of transcriptional changes.

FIG 5.

Changes in the levels of metabolites in a wild-type strain grown on antimycin A (A) and in the rse2^Y326D strain (B). (A) The changes in metabolite contents in the wild-type strain grown on M2 supplemented with antimycin A for 48 h were calculated by dividing the metabolite level on antimycin A by that under standard conditions. (B) The changes in metabolite contents in the rse2^Y326D strain were calculated by dividing the metabolite level in the rse2^Y326D mutant by that in the wild-type strain. The level of significance was set at P < 0.05 (t test). The metabolites with gray characters were undetectable. Red box, increase of more than 2-fold; orange box, increase of less than 2-fold; dark blue box, reduction of more than 2-fold; blue box, reduction of less than 2-fold; white box, no change. Abbreviations: 3-PGA, 3-phosphoglycerate; A-CoA, acetyl-coenzyme A; α-KG, alpha-ketoglutarate; Ala, alanine; Arg, arginine; Asn, asparagine; Asp, aspartate; b-Ala, beta-alanine; Cys, cysteine; E A, ethanolamine; Fru, fructose; Fru 1,6-bP, fructose-1,6-biphosphate; Fru 6P, fructose 6-phosphate; GAL 3P, glyceraldehyde-3-phosphate; Glc, glucose; Glc 6P, glucose-6-phosphate; Gln, glutamine; Glu, glutamate; Gly, glycine; Gly 2P, glycerol-2-phosphate; Gly 3P, glycerol-3-phosphate; H-Ser, homoserine; Ile, isoleucine; Leu, leucine; Lys, lysine; Met, methionine; OAA, oxaloacetate; Orn, ornithine; PEP, phosphoenolpyruvate; Phe, phenylalanine; Pro, proline; Put, putrescine; Ser, serine; Thr, threonine; Tyr, tyrosine; Ura, uracil; Val, valine.

DISCUSSION

The RSE2 and RSE3 transcription factors of P. anserina have previously been shown to activate genes encoding the alternative oxidase and gluconeogenic enzymes. The results reported here strongly suggest that they are the main, if not the only, transcriptional regulators of the aox gene. A screening procedure to isolate mutations with increased expression of the aox gene based on the isolation of cox5::ble suppressors was undertaken. Of the 23 suppressors analyzed, eight were located in rse2 and 10 in rse3, which, given the three previously described mutations (20), correspond to 21 independent mutational events obtained in these two genes. Some of these mutations were recovered independently several times, in particular in rse3, indicating that the screen was practically saturated. No new gene controlling the expression of the AOX was identified in this study. At first sight, five mutations outside gene rse2 or rse3 did not modify the expression of the aox gene. Characterization of these mutations is likely to give new insights into the physiology of the cox5::ble respiratory-deficient mutant.

The 10 mutations in rse2 identified nine positions of amino acid substitutions distributed essentially in the second half of the protein, and the 11 mutations in rse3 identified five positions with a strong biased distribution in the C terminus of the protein. It is worth noting that simply changing one amino acid at these positions converts RSE2 and RSE3 into constitutive hyperactive proteins. Most of the mutations affect residues conserved in their orthologues (see Fig. S2 in the supplemental material). This conservation is an additional argument that supports a functional significance for these positions. The RSE2 and RSE3 proteins and their orthologues in N. crassa and A. nidulans lack a “middle homology” region that is thought to play a role in regulating the transcriptional activity (31–33). However, the deletion of residues 223 and 329 in RSE2 and of residue 251 in RSE3 increased the activity of the proteins, and this strongly suggests that the central region of RSE2 and RSE3 also contains sequences that inhibit the transcriptional activity in the absence of the inducer. The mutations obtained from this region (rse2^E223K, rse2^G329S, and rse3^E251K/G) then would cause the loss of function of a domain involved in the response to the inducer signal. In contrast, deletion of the 51 C-terminal amino acids in both proteins results in the loss of function of the proteins. A similar result was observed for the A. nidulans AcuM and AcuK orthologues (21; also see Fig. S2). These data are in agreement with a C-terminal location of the activation domain for these proteins, as is the case for many zinc cluster proteins (31–33). Several pieces of data suggest that there are differences in the functional organization of RSE2 and RSE3. First, whereas the gain-of-function mutations in RSE2 are distributed throughout the terminal half of the protein, the gain-of-function mutations in RSE3 are clustered in a short segment of the protein located about 50 amino acids from the C terminus. Second, the deletion of amino acids 640 to 645 in RSE3 led to the constitutive activation of the protein, suggesting that in addition to the region around residue 251, the C-terminal region around residue 640 is also involved in receiving the inducer signal. Another possibility is that the mutations clustered in the C-terminal region of RSE3 increase the activity of the activation domain by changing intramolecular interactions or interactions with auxiliary protein(s), resulting in increased transcriptional activity. It must be noticed, however, that no partner able to modulate the activity of RSE2 and RSE3 has been found in our screen.

Whether RSE2 and RSE3 are regulators of genes other than the aox, fbp, and pck genes is an open question. We show here that 598 genes, i.e., about 5% of the genes, undergo a transcriptional change in response to gain- or loss-of-function mutations in genes rse2 and rse3, and that 44 genes appeared to respond in a similar manner in the three gain-of-function or the three KO strains. The microarray data, supplemented by RT-qPCR experiments, indicate that RSE2 and RSE3 modulate the expression not only of AOX, FBP, and PCK but also of a transcription factor, of another alternative respiratory enzyme, NDI1, of a flavohemoglobin, of various hydrolytic enzymes, and possibly of a PKS and an NRPS/PKS hybrid supposed to produce diverse secondary metabolites. In view of these results, we propose that the RSEs regulate genes that contribute to various adaptative strategies and defense mechanisms that could be crucial in the natural life style of P. anserina. Importantly, the RSEs modulate the expression of a transcription factor and of at least one gene (fhb) that does not contain an AIM upstream motif, which implies that these transcription factors belong to a regulatory cascade.

Flavohemoglobins are widely distributed in bacteria and fungi and are known to function in nitric oxide (NO) detoxification, converting the NO radical to nitrate (37, 38). The P. anserina genome shows two highly conserved putative FHBs (Pa_5_1700 and Pa_1_3460), but only one (Pa_5_1700, encoded by fhb) responds to impaired respiration and depends on RSE2-RSE3. Its cellular location (mitochondrial or cytosolic) is unknown. FHB of Saccharomyces cerevisiae (YHb) has been proposed to be activated in the presence of antimycin and to protect against oxidative stress (39), but this function has been questioned (40). In aspergilli, FHBs have been biochemically characterized as NO dioxygenases (41) with an unexpected function in promoting oxidative stress (42). In P. anserina, deletion of fhb confers hypersensitivity to the NO donor, DETA NONOate (data not shown), demonstrating that this flavohemoglobin plays a role in the genetic response to nitrosative stress. The RSEs are also activators of various hydrolytic enzymes. They activate the expression of several members of the α/β hydrolase superfamily (Pa_4_3860, Pa_1_1770, and Pa_6-4020), including two epoxide hydrolases, and also seem to activate two members of the glycoside hydrolase family (Pa_1_11790 and Pa_1_1010). Epoxide hydrolases are responsible for the detoxification of xenobiotics by catalyzing the degradation of epoxides that frequently arise from oxidative metabolism of endogenous and xenobiotic compounds via chemical and enzymatic oxidation processes, including the cytochrome P450 monooxygenase (43).

The nature of the signal(s) responsible for the activation of the RSEs and the mechanism(s) of this activation are not currently understood. Our results indicate that the expression of the genes encoding the AOX, NDI1, FHB, the transcription factor Pa_7_2610, and the 3 hydrolases Pa_6_4030, Pa_4_3860, and Pa_1_1770 is increased much more on antimycin than on acetate, whereas the opposite is observed for genes encoding FBP and PCK. These observations suggest that the inducer-mediated activation of RSE2 and RSE3 is different on acetate and antimycin, or that the activity of these transcription factors depends on interactions with other regulators that themselves depend on the growth conditions. Furthermore, the existence of genes downregulated in the KOs and not upregulated in the strains that constitutively express the aox gene (cluster D) also suggests that the activation of the RSEs requires different signals according to the genes. In A. nidulans, it has been shown that AcuK and AcuM are permanently present in the nucleus and proposed that the inducer is malate both on gluconeogenic carbon sources and on antimycin. The activation would occur via the binding of malate to one or both of the potential PAS domains of AcuK and AcuM (21). These domains are known to be involved in protein-protein interactions and to be able to bind a variety of ligands (44, 45). Here, our metabolome data do not show significant accumulation of malate following a 48-h antimycin treatment, and the basis of the activation of these transcription factors remains to be further explored.

A comparison of genes that showed increased or decreased relative expression in the different rse mutants revealed a larger proportion of genes with decreased expression in the KO mutants and a larger proportion of genes with increased expression in the gain-of-function mutants. These data are consistent with the role of RSE2/RSE3 functioning in transcriptional activation. However, the RSEs can also act to reduce the expression of some genes. Note that most of these genes do not contain an AIM motif; thus, the regulation is probably indirect.

RSE2 and RSE3 are each essential for growth on antimycin and acetate, and their orthologues in N. crassa and A. nidulans have been shown to bind to DNA as a heterodimer by their N-terminal sequences (11, 21). These results strongly support a model in which heterodimerization of the two proteins is essential for gene activation. However, it appeared that in the absence of RSE3 but not of RSE2, fertility of the strains is compromised. This observation suggests that although RSE2 and RSE3 can function as a heterodimer, they have distinct functions. Although our microarray analysis did not clearly reveal sets of genes whose expression depends only on RSE2 or RSE3, the reduced fertility of Δrse3 mutants suggests that RSE3 alone is able to operate in the form of a homodimer or complexed with an additional unidentified protein(s). Such a situation has been described for some zinc cluster family members. For instance, the Oaf1 and Pip2 proteins and the Rsc3 and Rsc30 proteins in S. cerevisiae have been shown to interact as heterodimers, but one of the proteins is also able to operate in the absence of the other one (46, 47). Additional work is needed to clarify this point.

Analysis of the metabolic responses to antimycin treatment revealed a profound decrease in α-ketoglutarate level, suggesting a perturbation of the TCA cycle and an increase in the levels of amino acids and fatty acids, mannitol, Gly-2P, and Gly-3P. These metabolites are linked to downstream TCA or glycolytic intermediates. The primary role of the TCA cycle is the oxidation of acetyl-coenzyme A to carbon dioxide. However, the TCA cycle also functions in biosynthetic processes in which intermediates are removed from the cycle to be converted to glucose (gluconeogenesis), fatty acids, and amino acids. If TCA intermediates are removed from the cycle, they must be replaced to permit the cycle to function. This is performed by reactions named anaplerotic, whereas reactions linked to biosynthetic processes that remove intermediates from the cycle are called cataplerotic (48). Therefore, it seems that in response to antimycin, there is a relative increase in the flux through cataplerotic pathways that remove TCA cycle intermediates and through pathways that pump glycolytic intermediates. A possible explanation for this observation is that when the respiratory chain is impaired, it is important to slow down the TCA cycle to maintain mitochondrial NADH homeostasis. We suggest that using TCA intermediates for amino acid and fatty acid biosynthesis constitutes an efficient way to reduce the overload on the TCA cycle and to allow the reoxidation of reducing equivalents, at the same time reducing the production of ROS. Interestingly, the metabolite profile of the rse2^Y326D mutant partly overlapped with that of the wild type grown in the presence of antimycin. This suggests that the RSEs play an important role in the metabolic response to mitochondrial respiration impairment. However, it is worth noting that there is no direct correlation between these metabolic changes and the abundance of relevant transcripts in the rse2^Y326D mutant. It is possible that the t statistic test used for the analysis of the transcriptomic data gives a robust measure of significant differences but misses more subtle changes. The lack of correlation between metabolism and transcripts has been much discussed (49, 50). The metabolic control is mainly performed at posttranslational levels. The activation of the RSEs may, for instance, modify this posttranslational regulation. Additional experiments, in particular the analysis of the changes in the transcriptome following antimycin treatment, are required to understand which role gene expression plays in metabolic reprogramming in response to mitochondrial impairment.