A metabolic strategy to enhance long-term survival by Phx1 through stationary phase-specific pyruvate decarboxylases in fission yeast

Effect of Δphx1 mutation on stationary phase transcriptome of S. pombe

In order to understand how Phx1 enables long-term survival, stress resistance, and meiotic sporulation in S. pombe, we compared the global mRNA signatures of wild type vs. Δphx1 deletion mutant cells by microarray analysis. Since the phx1⁺ gene gets expressed from late exponential phase, and the viability of Δphx1 cells starts to decrease rapidly from ~3 days after entering stationary phase (~100 h post-inoculation in EMM; [15]), we prepared RNA samples at 80 h culture time when cells still retained full viability. RNAs from four independent cultures were subjected to cDNA synthesis and hybridization. The microarray analysis revealed that transcripts from 56 genes were decreased by more than 2-fold and expression levels of 97 genes were increased by more than 2-fold in the Δphx1 mutant compared with wild type (Supplementary Table S1, S2). We summarized the affected genes with functional grouping by generic GO-term finder (http://go.princeton.edu/cgi-bin/GOTermFinder) in Table 1 and Table 2 for genes positively and negatively affected by Phx1, respectively.

The genes positively affected by Phx1 (or repressed in the Δphx1 mutant) include those known or predicted to function in thiamine biosynthesis, carbohydrate metabolism, stress responses, transport, RNA metabolic process, and non-coding RNA (Table 1, Table S1). Several genes have functions related to thiamine metabolism. Prominent examples are the nmt1⁺ and nmt2⁺ genes that encode biosynthetic enzymes of pyrimidine and thiazole moiety of thiamine, respectively [17]. The bsu1⁺ (car1⁺) gene encodes a transporter for pyridoxine derivatives and thiamine [18]. Two genes (SPAC13A11.06 and SPAC3G9.11c) encode putative pyruvate decarboxylases that require thiamine pyrophosphate (TPP) as a cofactor. Therefore, it appears that Phx1 contributes to increasing the level of thiamine and TPP-requiring enzyme pyruvate decarboxylase that converts pyruvate to acetaldehyde.

The genes negatively affected by Phx1 (or induced in Δphx1) also involve many genes for carbohydrate and energy metabolism including TCA cycle and pentose phosphate pathway, stress response, conjugation and meiosis, and transport (Table 2, Table S2). Prominently, genes for mitochondrial energy generation such as cytochrome c (cyc1⁺), citrate synthase (cit1⁺), aconitase (aco1⁺), mitochondrial glycerol-3-phosphate dehydro-genase (gut2⁺), and a putative NADH dehydrogenase (SPAC3A11.07) were negatively regulated by Phx1. These results suggest that Phx1 inhibits mitochondrial energy generation. We also noticed that genes predicted for the pentose phosphate pathway such as glucose-6-phosphate dehydrogenase (zwf1⁺, SPAC3C7.13c), glucokinase (SPAC4G9.12), and the gfo/idh family oxidoreductase (SPAC26H5.09c) were induced in Δphx1 mutants, possibly reflecting that the mutant cells are under oxidative stress. This finding coincides with the observation that several genes involved in oxidative stress response are induced in the mutants, such as sod1⁺,encoding CuZnSOD, and srx1⁺, encoding sulfiredoxin. Several transporter genes forhexose (ght3⁺, ght4⁺, ght5⁺, ght8⁺) were also induced in Δphx1 mutants. Considering the previous observation that ght3⁺ and ght4⁺ genes are induced under oxidative and heat stress conditions [19], it is possible that Δphx1 cells experience more oxidative stress than wild type cells.

Many genes whose expression is correlated with different phases of meiotic differentiation were affected by Phx1. Among the previously reported meiosis-correlated genes [20], 21 genes were repressed in the Δphx1 mutant and 40 genes were induced (Table S1). It is likely that the mis-regulation of these genes lie behind the sporulation-deficient phenotype of the Δphx1/Δphx1 diploid mutant [15]. Inspection of the affected meiotic genes revealed that Phx1 primarily up-regulated middle and late meiotic genes, functioning during meiotic divisions and sporulation (16 out of 21 genes; Table S1), and down-regulated genes of early meiotic genes, functioning during nitrogen starvation and meiotic prophase (27 out of 40 genes; Table S2). It can be hypothesized that Phx1 is a major regulator that activates later phases of meiotic gene expression (such as meiotic divisions I, II, and spore formation)and represses genes of earlier meiotic stages (such as starvation response, pheromone sensing, conjugation, S phase and DNA recombination). How Phx1 works in concert with other known regulators of late meiotic stages, such as Rsv1, Rsv2, Atf21 and Atf31 [21], will be an interesting topic to investigate in the future.

Phx1 functions to maintain the thiamine pool during stationary phase

We further examined whether thiamine-metabolic genes are indeed regulated by Phx1. Particularly, the possibility of feedback inhibition of thiamine biosynthetic (nmt1⁺ and nmt2⁺) and transport (bsu1⁺) genes by thiamine in the Δphx1 mutant needed to be evaluated. For this purpose, we monitored the mRNA levels of nmt1⁺, nmt2⁺, and bsu1⁺ genes along with the thi9⁺ gene, encoding a thiamine transporter [18], which is thiamine-repressible but not affected by Phx1. Transcripts from the tnr3⁺ gene encoding thiamine diphosphokinase [22], which is not repressible by thiamine, were monitored in parallel as a control.

Results from qRT-PCR of RNAs from wild-type and Δphx1 mutant cells demonstrated that thenmt1⁺, nmt2⁺, and bsu1⁺ transcripts decreased in the Δphx1 mutant in the absence of thiamine, whereas the thi9⁺ gene was not, as observed in microarray analysis (Fig. 1, panel A1). Added thiamine repressed expression of all genes, except tnr3⁺ in the wild type (Fig. 1, panel A2). The thi9⁺ gene was also repressed by thiamine in the Δphx1 mutant (Fig. 1, panel A2). Therefore, the decrease in the expression of three Phx1-dependent genes (nmt1⁺, nmt2⁺, bsu1⁺) is not due to feedback regulation by increased thiamine in the Δphx1 mutant, but suggests that Phx1 positively regulates the three genes for thiamine supply.

We then determined the intracellular thiamine pool. The amounts of thiamine, thiamine monophosphate (TMP), and thiamine pyrophosphate (TPP) were measured by HPLC in cell-free extracts obtained at different growth phases. As shown in Fig. 1B, the total thiamine pool decreased as cells progressed into stationary phase. The decrease, however, was more pronounced in the Δphx1 mutant, where the total thiamines were reduced to about 60% of wild type level during stationary phase (80 h culture). The majority of the thiamine pool consisted of TPP, which is the active form as an enzyme co-factor, consistent with previous data [17]. The Δphx1 mutant contained a much lower amount of TPP, only about 30% of the wild-type level during the stationary phase (80 h; Fig. 1C). The ratio of TPP to thiamine was maintained from 8.4 to 5.5 in the wild type from exponential (12 h) to late stationary phase (80 h) cells, whereas the ratio changed from 9.4 to 0.8 in the mutant during the same period. The amount of TMP was below our detection limit. Overall, these measurements demonstrated that Phx1 is needed to maintain the level of thiamine pool, especially TPP, during stationary phase.

Phx1 is needed for the synthesis of stationary phase-specific pyruvate decarboxylases

The microarray analysis revealed that transcripts from two genes (SPAC13A11.06 and SPAC3G9.11c) encoding putative pyruvate decarboxylases (PDCs) were significantly lowered in Δphx1 mutant (Table 1, Table S1). The PDC enzyme utilizes thiamine pyrophosphate (TPP) as a cofactor, and constitutes a distinct subgroup of the TPP-dependent enzyme family [23]. There are four PDC-like proteins predicted from the genome of S. pombe. Since the gene name pdc1 has been already used for mRNA decapping scaffold protein (SPAC4F10.01) [24], we propose to name the paralogs as pdc101(SPAC1F8.07c), pdc102(SPAC186.09), pdc201(SPAC3G9.11c), and pdc202(SPAC13A11.06). Phylogenetic relatedness of PDC-like proteins in S. pombe as well as in other representative fungi is presented in Fig. 2A.

The Pdc201 and Pdc202 cluster together, along with three reported PDCs of S. cerevisiae (ScPdc1, 5, 6; [25], two PDCs of basidiomycota (U. maydis, C. neoformans), PdcA of Aspergillus fumigatus and Ilv-5 of Neurospora crassa, and more distantly with ScAro10 (YDR380w) and AfPdcB (Afu6g00750). ScAro10 was reported to be involved in the degradation of aromatic amino acids as a phenylpyruvate decarboxylase [26]. Pdc101 and Pdc102 clustered closely in a separate clade, along with PDCs of N. crassa and A. fumigatus and four PDC proteins from plant kingdom such as Arabidopsis thaliana and Chlamydomonas reinhardtii (Fig. 2A).

We monitored the expression profiles of four pdc genes during growth and stationary phases. Each mRNA level was determined by qRT-PCR, and presented as a normalized value relative to the control act1⁺ mRNA (Fig. 2B). The results demonstrate that pdc101⁺ is the predominant gene in the exponential phase, whose expression decreases during stationary phase. Therefore, Pdc101 is most likely the primary pyruvate decarboxylase that supports exponential growth. Expression of pdc201⁺ and pdc202⁺ genes increased markedly during stationary phase, by about 500-fold and 50-fold, respectively. The pdc102⁺ gene produced hardly detectable mRNA in both conditions. Expression of pdc201⁺ andpdc202⁺ genes was dependent on Phx1, whereas that of pdc101⁺was not (Fig. 2B). Taken together, we found that pdc201⁺ and pdc202⁺ genes are induced highly during the stationary phase and their induction is mediated by the positive regulator Phx1.

We then determined whether Phx1 indeed controls the level of PDC enzyme activity. Results in Fig. 2C demonstrated that PDC enzyme activity increased during stationary phase in wild type, but decreased in the mutant. PDC activity at the exponential phase was not affected by phx1 mutation. Therefore, consistent with the transcript regulation, Phx1 indeed controls the level of stationary phase-specific PDC enzymes. Since Phx1 controls the synthesis of both PDC apo-proteins and its cofactor TPP,we investigated the effect of adding thiamine to a stationary phase cell culture. Results in Fig. 2D show that added thiamine increased activities of PDCs in the wild type and the mutant, suggesting that TPP can be a limiting factor for PDC activity during stationary phase. However, thiamine addition did not fully elevate the PDC activity to wild type level,confirming that the phx1 mutation indeed significantly decreased the level of PDC proteins during stationary phase. Therefore, our results revealed that Phx1 positively regulates the supply of thiamine pyrophosphate as well as the production of Pdc201 and Pdc202 proteins, leading to increase in PDC enzyme activity during stationary phase. PDC diverts pyruvate from being used in mitochondria for the TCA cycle and respiration to the direction of ethanol fermentation.We hypothesize that this metabolic shift could be a mechanism behind the cellular strategy to maintain viability during stationary phase.

Pdc201 and Pdc202 contributes to long-term survival of S. pombe

Whether PDC contributes to long-term survival of S. pombe was assessed by gene disruption and overexpression. We created Δpdc201, Δpdc202, and Δpdc102 mutants and compared their long-term viability in EMM with the wild type (972) and the Δphx1 mutant (JY01). We were not able to obtain a Δpdc101 mutant, consistent with the prediction that this gene is essential [27]. The results in Fig. 3A demonstrated that the Δpdc201 and Δpdc202 mutations reduced cell viability in the stationary phase, even though not as pronounced as the effect of Δphx1 mutation. The Δpdc102 mutation did not affect cell viability significantly. The double mutation of Δpdc201Δpdc202 did not cause further reduction in viability (supplementary Fig. S1). We then examined the effect of overproduction by cloning the pdc201⁺ or pdc202⁺ gene behind the adh1⁺ promoter on pAEP1, a pREP1-based vector [16], and introduced them to the wild type (JH43) or Δphx1 (ESX5). Results in Fig. 3B and 3C demonstrated that overproduction of either Pdc201 or Pdc202 increased the long-term viability of both the wild type and Δphx1 mutant. These observations support the proposal that the mechanism behind the positive action of Phx1 in extending stationary cell survival relies heavily on the increase in pyruvate decarboxylases.

Phx1 decreases mitochondrial respiration and ROS production, and increases ethanol fermentation upon entering into stationary phase

Based on the increased synthesis and activity of Pdc201 and Pdc202 enzymes during stationary phase, we hypothesized that mitochondrial respiration may decrease as S. pombe cells enter into stationary phase. We therefore estimated oxygen consumption rate in the wild type and the Δphx1 mutant grown in minimal media. Fig. 4A demonstrated that the oxygen consumption in the wild type cells peaks at around 40 h after inoculation, when cells enter the stationary phase, and decreases steadily thereafter. On the other hand, the Δphx1 mutant maintained the high rate of respiration longer, for about 20 hours after cells entered stationary phase. When we measured glucose consumption during the growth of wild type and Δphx1 mutant cells, we found that glucose was exhausted at around 40 h, and there was no difference between wild type and Δphx1 mutant in consuming glucose (data not shown). This implies that the Δphx1mutation does not affect sugar metabolism, and possibly energy metabolism, during the exponential growth.

Since mitochondrial respiration is considered to be a major source of reactive oxygen species (ROS) whose accumulation damages cell components and hence could curtail lifespan, we examined the amount of intracellularROSproduction by a redox-sensitive fluorescent dye: 2′,7′-dichlorofluorescin diacetate (DCFH-DA). Fig. 4B demonstrated the level of DCFH-oxidizable ROS in the wild type and Δphx1cells grown for different lengths of time in minimal media as in Fig. 4A. The fluorescence of individual cells was measured by flow cytometry, and the intensity from 10,000 cells was compared relative to the value at 12 h culture. The results indicated that theΔphx1 mutant accumulated higher amount of ROS than the wild type during stationary phase (60, 72 h culture time), when Δphx1 mutant cells still retain full viability in minimal media [15]. This coincides with the prolonged respiration observed in the mutant. We further examined whether the Δphx1cells experience more oxidative stress by monitoring the level of protein oxidation. Immunodetection of carbonylated proteins (Fig. 4C) demonstrated that proteins are more oxidized in Δphx1 at late stationary phase. These results support a proposal that the actual mechanism by which Phx1 contributes to extend cell viability is through reducing ROS-mediated oxidative damage by inhibiting respiration in stationary phase.

Extended lifespan of Δpka1, Δsck2, or Δpyp1 mutants was compromised by Δphx1 mutation

In S. pombe, it has been demonstrated that nutrient signaling via serine/threonine kinases Pka1 and Sck2 has strong pro-aging effects, and deletion of either the pka1 or sck2 gene extends chronological lifespan [4]. We examined whether Phx1 has any genetic relation with Pka1 or Sck2, by examining long-term survival of Δsck2Δphx1 and Δpka1Δphx1 double deletion strains. The Δsck2 and Δpka1 mutants grown in complex media demonstrated extended long-term survival as previously reported [4, 6]. Introduction of the Δphx1 mutation to these long-lived strains compromised their survival significantly (Fig. 5A). The stronger effect of the phx1 mutation in the Δsck2 mutant background suggests that in the absence of the Tor kinase, Phx1 may contribute more to cell survival than in the absence of protein kinase A. Constitutive activation of the Sty1 MAP kinase pathway of S. pombe has also been reported to extend chronological lifespan [6]. We examined whether the extended lifespan of a Δpyp1 mutant, where Sty1 is constitutively activated by absence of inactivating phosphatase, is affected by the phx1 mutation. Fig. 5B demonstrated that the prolonged survival phenotype of Δpyp1 was greatly compromised by Δphx1, since the double mutant showed a similarly reduced survival as the Δphx1 single mutant. Thus, Phx1 could be involved in mediating longevity in association with all three major pathways known to be involved in chronological lifespan.

Strains and culture media

S. pombe strains and plasmids we used in this study are listed in supplementary Table S3. The Δpdc102, Δpdc201, Δpdc202, Δpka1, and Δsck2 deletion strains were obtained from the Bioneer mutant library version 2.0 (Bioneer Corporation, Korea), and verified by PCR for correct deletions. Through mating of Bioneer auxotrophic strains with 972 (h^-), we isolated prototrophic Δpdc201(JY02), Δpdc202 (JY03), Δpdc102 (JY04), Δpka1(JY05) and Δsck2 (JY06) mutants, following spore selection and confirmation by PCR. The prototrophic Δpka1Δphx1 (JY07) and Δsck2Δphx1 (JY08) double deletion mutants were obtained by transforming Δpka1::kmx4 or Δsck2::kmx4 cassette to prototrophic Δphx1 (JY01) strain, following marker selection and confirmation by PCR. The Δpdc201Δpdc202 (JY11) double mutant was obtained by introducing Δpdc202::natMX6 cassette to the prototrophic Δpdc201 (JY02) strain. All recombinant plasmids were confirmed by nucleotide sequencing. Growth and maintenance of S. pombe strains were generally done in Edinburgh minimal medium (EMM, 2% glucose) or in yeast extract medium (YE, 3% glucose) with appropriate supplements [15]. For conjugation and sporulation, malt extract medium (ME, 3% malt extract) was used. For liquid culture, cells were inoculated at initial OD₆₀₀ of 0.02 and were routinely grown in EMM to OD₆₀₀ of ~1 for exponential phase (~18 h of culture time) or OD₆₀₀ of ~10 for stationary phase (~50 h of culture time) at 30°C with shaking at 180 rpm.

Microarray analysis

We used DNA microarrays displaying probes for >99% of all known and predicted genes of S. pombe spotted in duplicate onto glass slides [39]. For RNA preparation, WT(JH43) and Δphx1 (ESX5) cells were grown in EMM for 80 hours after inoculation at initial OD₆₀₀ of 0.02. RNA extraction, hybridization and initial data processing and normalization were performed as previously described [39]. Four independent biological experiments were performed, including two dye swaps. The data were visualized and analyzed using GeneSpring (Agilent), and were deposited in ArrayExpress (E-MTAB-2285). The significance of overlaps between different gene lists was calculated in GeneSpring using a standard Fisher's exact test, and P values were adjusted with a Bonferroni multiple testing correction. Cut-off values of 2-fold change of the average value from four biological repeats. Gene annotations were downloaded from PomBase (http://www.pombase.org/).

Quantitative real-time PCR (qRT-PCR)

Each RNA sample (1 μg/μl) was reverse-transcribed into cDNA using RevertAid™ Reverse Transcriptase kit (Fermentas). Each PCR was performed with SYBR Green/ROX qPCR master mix (Fermentas) and act1⁺ mRNA-specific primers were used for an internal control. Triplicate PCRs for gene-specific primer pairs of each gene were carried out according to manufacturer's instruction in a qRT-PCR machine(Stratagene MX3000P QPCR system, Agilent Technologies) with analysis software MXpro (Agilent Technologies). Real-time PCR data were analyzed by the comparative C_T method [40] to calculate fold changes.

Analysis of thiamine and thiamine phosphates by HPLC

Cells were grown in liquid EMM at 30°C and harvested at different time points, followed by washing with distilled water. Thiamine and thiamine phosphates were extracted and determined by HPLC as described by Schweingruber AM et al. [17]. The thiamine mixture was injected into an HPLC column (a PRP-1 main column, 10 μm, 250X4.1 mm, Hamilton no.79427) in Varian Prostar HPLC system equipped with a fluorescence detector (excitation at 365 nm and emission at 430 nm). Thiamine and its derivatives were eluted with steep gradients of solvent A (8.5 mM sodium phosphate buffer, pH 8.5) and solvent B (methanol, HLPC-grade, Merck). Thiamine phosphates were eluted in the order of triphosphate, diphosphate and monophosphate at 10% B, followed by thiamine after steep gradient of 50% B. As standards, thiamine monophosphate (TMP) and thiamine pyrophosphate (TPP) (Sigma) were monitored in parallel.

Enzyme activity assay

Pyruvate decarboxylase activity in cell extracts (20 μg total protein) was determined as described previously [41]. The rate of disappearance of absorption at 340 nm by NADH was monitored on Simadzu UV-1650pc spectrophotometer. The enzyme activity was presented as U/mg protein, where 1 unit will convert 1.0 μmole of pyruvate to acetaldehyde per minute at pH 6.0 at 25°C.

Long-term survival assay

To measure cell viability during stationary phase, cells were inoculated to liquid EMM or YE media at initial OD₆₀₀ of 0.02 and grown for about 40 h or 30 h until OD₆₀₀ reached the maximum cell density of about 9 or 12, respectively. From this time point (day 0), aliquots were taken each day and plated on complex YES medium, followed by incubation at 30°C for 2 days for colony counting. The average % viabilities with standard deviations were assessed from three independent experiments.

Measurement of oxygen consumption

The respiration rate was measured polarographically at 30°C by using an oxygen electrode probe in YSI5300A Biological Oxygen Monitor System (Yellow Spring Instrument). After collection of 0.5 ml of the cells grown in EMM at each time point and at indicated OD₆₀₀ point by centrifugation at 3000 g, cells were suspended in 3 ml of 50 mM potassium phosphate (pH 6.5) with 0.1 M glucose. Cell suspension was introduced into the sample chamber and the amount of oxygen consumed was recorded for 15 minutes.

Measurement of intracellular ROS level

Intracellular ROS levels were determined by using redox-sensitive fluorescent probe 2′, 7′-dichlorofluorescin diacetate (DCFH-DA, Sigma).To each aliquot of 0.5 ml cultured cells, DCFH-DA (50 mM in DMSO) was added to 50 μM (final), and incubated for 1 h at 30°C in the dark. The amount of ROS-oxidized DCF was analyzed by flow cytometry, using FACS Canto (Becton Dickinson) at low flow and monitoring in FITC-A channel (detecting green fluorescence).A total of 10,000 live cells for each sample were analyzed for DCF - dependent fluorescence.

Measurement of protein carbonylation

Protein carbonylation was detected by OxyBlot^™ protein oxidation detection kit as recommended by the manufacturer (Millipore). Cells were grown in EMM, and harvested at each time point during stationary phase. Protein extracts were prepared after vortexing harvested cells with glass beads in lysis buffer containing either 1-2% β-mercaptoethanol or 50 mM DTT. Following protein quantification by Bradford assay, each 5 μl aliquot containing 20 μg of protein was processed for DNP derivatization, electrophoresed on 10% SDS-PAGE, and immunodetected by anti-DNP antibody. As a loading control, a parallel electro-transferred membrane was stained with Ponceau S solution. The anti-DNP immunoblots and Ponceau S stained blots were scanned and quantified by chemiluminescence imaging system (Fusion solo, Vilber Lourmat) with Multi Gauge (Fuji) program.