Redox-dependent Regulation of Gluconeogenesis by a Novel Mechanism Mediated by a Peroxidatic Cysteine of Peroxiredoxin

Peroxiredoxin is an abundant peroxidase, but its non-peroxidase function is also important. In this study, we discovered that Tsa1, a major peroxiredoxin of budding yeast cells, is required for the efficient flux of gluconeogenesis. We found that the suppression of pyruvate kinase (Pyk1) via the interaction with Tsa1 contributes in part to gluconeogenic enhancement. The physical interactions between Pyk1 and Tsa1 were augmented during the shift from glycolysis to gluconeogenesis. Intriguingly, a peroxidatic cysteine in the catalytic center of Tsa1 played an important role in the physical Tsa1-Pyk1 interactions. These interactions are enhanced by exogenous H2O2 and by endogenous reactive oxygen species, which is increased during gluconeogenesis. Only the peroxidatic cysteine, but no other catalytic cysteine of Tsa1, is required for efficient growth during the metabolic shift to obtain maximum yeast growth (biomass). This Tsa1 function is separable from the peroxidase function as an antioxidant. This is the first report to demonstrate that peroxiredoxin has a novel nonperoxidase function as a redox-dependent target modulator and that pyruvate kinase is modulated via an alternative mechanism.


Figure 1. Loss of Tsa1/2 promotes growth suppression in Trp-or Phe-free ethanol medium.
(A) Schematic representation of glucose (yellow arrows) and ethanol (purple arrows) carbon metabolism. Glycolysis metabolites are abbreviated as follows: glucose 6-phosphate (G6P), fructose 6-phosphate (F6P), fructose 1,6-bisphosphate (FBP), dihydroxyacetone phosphate (DHAP), glyceraldehyde 3-phosphate (GAP), 1,2-diphosphateglycerate (DPG), 3-phosphoglycerate (3PG) and phosphoenolpyruvate (PEP). Metabolites in the pentose phosphate pathway (PPP) include 6-phosphogluconate, ribulose 5-phosphate (Ru5P), ribose 5-phosphate (R5P), sedoheptulose 7-phosphate (S7P), erythrose 4-phosphate (E4P) and phosphoribosyl pyrophosphate (PRPP). Trp/Phe synthesis begins with PEP and E4P via chorismate. Utilization of two-carbon compounds such as ethanol and acetate require the glyoxylate cycle and gluconeogenesis (purple arrows). The glyoxylate cycle plays an anaplerotic role in the provision of metabolites to produce essential compounds in the cytoplasm. The key enzymes of the glyoxylate cycle, isocitrate lyase (Icl1) and malate synthase (Mls1), as well as malate dehydrogenase (Mdh2), are upregulated in yeast in response to the ethanol carbon source and are expressed in the cytoplasm (http://www.yeastgenome.org/). Gluconeogenesis uses most of the enzymes that are common to glycolysis, except during the steps responsible for the synthesis of PEP and F6P, which are catalyzed by phosphoenolpyruvate carboxykinase (Pck1) and fructose 1,6-bisphosphatase (Fbp1), respectively. The tricarboxylic acid TCA cycle is indicated by a bright red-colored circle. (B) Growth curves and changes in ethanol levels of wild type (WT, gray and blue lines, respectively) and tsa1/2∆ (yellow and orange lines, mammalian homologue PK PKM2 11 , a high level of PK activity is maintained during glucose metabolism due to high levels of fructose-1,6-bisphosphate (FBP), an allosteric activator of Pyk1 12 . When glucose is exhausted, ATP synthesis depends on the tricarboxylic acid (TCA) cycle, and cells continue to require glucose-6-phosphate (G6P) to synthesize essential metabolites (see Fig. 1A). The reverse pathway (gluconeogenesis) of G6P synthesis begins with oxaloacetate to form phosphoenolpyruvate (PEP), which is a substrate for Pyk1. Pyk1 is catalytically active, even under conditions of glucose and FBP exhaustion 13 . In fact, Pyk1 plays an important role in the synthesis of several amino acids from pyruvate via gluconeogenesis using ethanol as a carbon source 14 . However, Pyk1 may need to be negatively regulated to avoid energy-wasting cycling reactions and enable efficient gluconeogenesis.
After the transition of glycolysis to gluconeogenesis, gluconeogenic metabolism produces catabolites not only for cell growth but also for the establishment of cell integrity during the stationary phase: the accumulation of trehalose and glycogen are required for desiccation tolerance 15 and metabolic storage 16 , respectively.
We demonstrate that the peroxidatic cysteine of Tsa1, which participates in its interaction with Pyk1, has a crucial role in achieving an efficient shift from glycolysis to gluconeogenesis. Our results suggest a novel regulatory mechanism for Pyk1 and an alternative non-antioxidant function of Tsa1 (Prx).

Results
Effect of Tsa1/2 loss on yeast growth in ethanol medium. In the yeast S. cerevisiae, as in many other eukaryotic cells, a high glycolytic flux during exponential growth in glucose results in the production of overflow metabolites, even under aerobic conditions (the Crabtree effect), whereas a low glucose gluconeogenic flux produces assimilation products by consuming overflow metabolites. Ethanol is a significant component of the overflow metabolites of yeast 13 (see also Fig. 1B). The transition from high to low glucose in the late exponential phase of growth is called the "diauxic shift" 17 , and we reasoned that Tsa1 might affect gluconeogenic flux during such a shift 5 . As shown Fig. 1B, we found that the loss of Tsa1/2 (both TSA1 and its orthologue TSA2 were disrupted) did not affect the time-dependent production of ethanol as a metabolite or its consumption. However, as previously indicated [7][8][9] , growth of tsa1/2Δ cells was retarded at the diauxic shift; thus, assimilation of ethanol might be affected by the loss of Tsa1/2. Therefore, we focused our attention on the effect of the loss of Tsa1/2 on yeast growth using ethanol. We initially compared the growth rate of wild type and Tsa1/2-deficient cells in ethanol medium containing amino acids (SEM) and observed a striking growth defect of tsa1/2Δ cells when Trp was omitted from the medium (Fig. 1C). This was also the case in agar medium without Trp and/or Phe ( Fig. 1D; see also Fig. 3B). In contrast, such a growth defect was not observed in glucose medium (SDM -Trp, Fig. 1E).
Trp and Phe are formed from phosphoenolpyruvate (PEP) in the chorismate pathway. Chorismate is derived from PEP and erythrose-4-phosphate (E4P), which is a metabolite of the pentose phosphate pathway (PPP, Fig. 1A). To gain insight into the metabolic alterations caused by a defect in Tsa1/2, we compared metabolites in wild type and tsa1/2Δ cells grown in SEM -Trp ( Fig. 2A,B) and observed in the mutant cells a decrease in metabolites associated with gluconeogenesis (FBP, G6P, glucose-1-phosphate, glycerol-3-phosphate and DHAP) and in PPP (6-phosphogluconate, Ru5P and S7P). In contrast, metabolites of the glyoxylate cycle (citrate, cis-aconitate, and a representative metabolite glyoxylate) were increased, probably as a result of a reduction in PEP. In addition, an increased level of phosphoribosyl pyrophosphate (PRPP) was observed in tsa1/2Δ cells ( Fig. 2A). Because both PRPP and chorismate are required for Trp/Phe synthesis (see Fig. 1A), decreased levels of PEP in tsa1/2Δ cells in SEM -Trp could result in Trp/Phe-dependence for efficient growth and explain why the higher levels of PRPP. These results suggest that a defect in Tsa1/2 results in impaired gluconeogenesis, explaining the dependence on Trp/Phe for efficient growth in SEM.
Pyk1 is a possible regulatory target for Tsa1. Typical Prx enzymes such as Tsa1 have two catalytic cysteine residues (Fig. 3A): a peroxidatic cysteine (Cys48) which reacts directly with H 2 O 2 to form a Cys-sulfenic acid, followed by the formation of a disulfide bond with a resolving cysteine (Cys171) in a homodimer partner molecule 1 . Both cysteines are essential for the peroxidase activity of Tsa1. We addressed whether these Tsa1 cysteines are required for efficient growth in SEM -Trp -Phe. As shown in Fig. 3B, growth was rescued by the wild type (Tsa1 WT ) and the C171T mutant (Tsa1 C171T ) of Tsa1, but not by the C48T mutant (Tsa1 C48T ), showing that although Tsa1-Cys48 has a crucial role in the growth phenotype, the peroxidase activity of Tsa1 is dispensable.
We have previously demonstrated that Cys48 of Tsa1 acts as a peroxide receptor, inducing a transient interaction with the transcription factor Yap1 via a disulfide bond, and that this interaction is enhanced by a resolving cysteine mutation (Tsa1 C171T ) 18 . We sought other putative Tsa1 target proteins that could link to Cys48 of Tsa1 C171T via a disulfide bond using unreduced (1 st ) vs. reduced (2 nd ) two-dimensional SDS-polyacrylamide gel electrophoresis (SDS-PAGE) to separate proteins from immunoprecipitates of HA-tagged Tsa1 C171T (Fig. 3C). We performed peptide mass fingerprinting using MALDI-TOF MS and identified five different proteins (Fig. 3C,D): PI SceI (intein Vde1), Pyk1 (also known as Cdc19; pyruvate kinase), Ssb1/Ssb2 (cytosolic HSC70/HSP70 family), Yol057w (dipeptidyl-peptidase III), and Frd1 (soluble fumarate reductase) (see Figure Supplement 2 for amino acid sequence). The failure to identify Yap1 as a binding partner of Tsa1 in this assay may be due to its respectively) in glucose-rich medium (YPAD). Ethanol was measured using F-Kit Ethanol (J.K. International, Tokyo, Japan). (C) Growth curves of wild type (open diamonds) and tsa1/2∆ (closed square) cells, neither of which are Trp/Phe-auxotrophs, growing in Trp-free ethanol medium (SEM -Trp) and Trp-containing ethanol medium (SEM). The data are presented as the mean + /− standard error of the mean (N = 3). (D,E) Spot assays were performed on wild type (WT) and tsa1/2Δ cells prepared in SDM culture (early stationary phase, 3 days). A series of one-fifth dilutions of 5 × 10 4 cells were spotted on agar plates (SEM -Trp, SEM -Phe, SEM, SDM -Trp, and SDM). See also Figure 18 . Pyk1, which present in the cell at a similar levels to Tsa1 (Fig. 3D), plays an important role in the rate-limiting step of glycolysis 13 . Although the PK activity of Pyk1 is repressed in the absence of the allosteric activator FBP, residual PK activity may need to be repressed to accomplish efficient gluconeogenesis because gluconeogenesis begins with PEP, a substrate of Pyk1 (see Fig. 1A). To examine the genetic interactions of Pyk1 with Tsa1, we measured the effect of PYK1 disruption on the growth retardation of tsa1/2Δ cells in SEM -Trp. As shown in Fig. 3E, disruption of PYK1 resulted in complete suppression of the growth phenotype. Downregulation of PK activity by an enhanced interaction between Tsa1 and Pyk1 during gluconeogenesis. Next, we investigated the effect of Tsa1 on Pyk1 activity and the possible role of the physical and biochemical interaction between Tsa1 and Pyk1. We observed that the PK activity in tsa1/2Δ whole cell extracts grown exponentially in ethanol medium was 119% (+ /− 1.19; p < 0.02) of the activity in wild type cell extracts. In addition, the PK activity of homogenously purified Pyk1 was decreased by the addition of recombinant Tsa1 (Fig. 4A). These results support the hypothesis that the PK activity of Pyk1 is downregulated in the presence of Tsa1/2, and suggest that the metabolic alterations observed in tsa1/2-deficient cells is due to higher Pyk1 activity during gluconeogenesis when grown on ethanol-containing medium.
To address the interaction between Tsa1 and Pyk1, we performed a pull-down experiment. The level of Tsa1 co-immunoprecipitated with HA-tagged Pyk1 (HA-Pyk1) was higher in cells growing exponentially in ethanol medium compared with glucose medium (Fig. 4B, the level of Tsa1 bound to Pyk1 was increased by a factor of 4∼ 7). Interestingly, when Tsa1 Cys48 was mutated, the Pyk1-Tsa1 interaction in cells grown in ethanol medium was significantly decreased (approximately 3-fold, p = 0.0009) (Fig. 4C). The level of Pyk1 protein considerably decreased in cells grown in ethanol medium, whereas the level of Tsa1 reciprocally increased such that the molecular ratio of Tsa1 to Pyk1 varied from 1.3 during exponential phase growth in glucose medium to more than 14 in ethanol medium (Fig. 4B, WCE). Despite the increase in the Tsa1/Pyk1 ratio, our results suggest that the direct interaction between Pyk1 and Tsa1 was enhanced by a Tsa1 Cys48-dependent mechanism when grown on ethanol-containing medium.
To determine whether the interaction of Tsa1 and Pyk1 is direct, we performed a pull-down experiment using purified recombinant proteins. rHA-Pyk1 or rPyk1 (Pyk1 without HA-tag, used as a control for a nonspecific interaction) were mixed with rTsa1. As shown in Fig. 4D, a pull-down experiment using HA-specific beads revealed that both disulfide dimers and monomers of rTsa1 were directly bound to rHA-Pyk1 (lane 1).

Induction of the physical interaction between Tsa1 and Pyk1 in the diauxic shift in glucose medium.
After glucose is exhausted in the diauxic shift in SDM (medium containing glucose as a carbon source), cells begin to use overflow metabolites such as ethanol for gluconeogenesis. Therefore, we predicted that the interaction between Pyk1 and Tsa1 would be enhanced in such glucose-exhausted conditions, resulting in the downregulation of Pyk1 activity. To test this hypothesis, we examined PK activity during the early stationary phase in glucose medium and observed a greater decrease in wild type cells than in tsa1/2Δ cells (Fig. 5A). Next, we observed that the level of Tsa1 bound to Pyk1 also increased during the early stationary phase in glucose medium (Fig. 5B), and repeated experiments indicated that this increase was approximately two-fold in the diauxic shift (48 h) and four-fold in the early stationary phase (Fig. 5C, 60-72 h, 3 days). As observed in cells grown in ethanol medium, we observed a significant shift in the ratio of the levels between Tsa1 and Pyk1, demonstrating an alteration from 1.3 in the exponential and diauxic shift to more than 12 in early stationary phase (72 h) (Fig. 5D).
To examine the role of Tsa1 cysteines in the interactions between Pyk1 and Tsa1, we expressed HA-Pyk1 as well as cysteine mutants of Tsa1 in tsa1/2Δ cells and performed co-IP experiments (Fig. 5E). Mutation of Tsa1 Cys48 decreased the interaction with Pyk1 (compare lane 6 with lanes 7 and 9 in the lower panel), whereas mutation of the resolving cysteine (C171T) substantially enhanced the interaction with Pyk1 (lane 8). It is likely that Tsa1 Cys48-sulfenic acid is relatively stable after the loss of the resolving Cys (Tsa1 C171T ) and that this form of Tsa1 can preferentially attack the Cys of other proteins. The level of a slowly migrating complex of Pyk1 (indicated as Pyk1-Tsa1, 110 K; upper panel, lane 3), which is presumably a Tsa1 C171T -HA-Pyk1 disulfide-bonded complex, was much less than the level of monomer Pyk1 (indicated as HA-Pyk1). Thus, the slowly migrating complexes of Tsa1 (lane 3 of lower panel, "#") may consist mainly of mixed disulfide complexes of non-Pyk1 proteins linked with Tsa1 that can physically interact with HA-Pyk1.

Requirement of Tsa1 Cys48 for the H 2 O 2 -induced interaction with Pyk1. The result demonstrating
that the H 2 O 2 -reactive cysteine of Tsa1, Cys48, is required for an efficient interaction with Pyk1 in cells growing putative target proteins (X) that can react with Cys48-SOH may be trapped using HAT-tagged Tsa1 C171T . Tsa1 C171T binding proteins were purified from yeast cells (BY4742 tsa1Δ ) expressing HAT-tagged Tsa1 C171T . Cells were grown in SDM -Leu and were treated with 0.5 mM H 2 O 2 for 5 min. Proteins in the purified fraction were fractionated by two-dimensional gel electrophoresis. The first dimension (horizontal) and the second dimension (vertical) were performed under non-reduced and reduced conditions, respectively (left panel). Non-specific proteins that interacted with HAT-tagged Tsa1 C48T, C171T mutants were also fractionated (right panel). A gradient gel (10% to 16%) was used for the second dimension. (D) Amino acid length, molecular weight (http://www.yeastgenome.org/), and protein abundance 32  in ethanol medium, implies a role of metabolically produced H 2 O 2 in the Pyk1-Tsa1 interaction. In fact, the ROS levels in cells growing exponentially in ethanol medium were significantly higher than those growing in glucose medium, and a further increase in ROS levels was observed in early stationary phase in glucose medium (Fig. 6A). The correlation between cellular ROS levels and the Pyk1-Tsa1 interaction and the requirement of Tsa1 Cys48 for an efficient interaction between Tsa1 and Pyk1 suggests that the oxidation of Cys48 by H 2 O 2 might trigger the interaction between Pyk1 and Tsa1. Therefore, we next examined H 2 O 2 -dependent interactions and the role of the identified cysteine residues on such interactions. Stationary phase yeast cells are extremely resistant to H 2 O 2 19 . In fact, the cells from early stationary phase were resistant to a 15-min treatment with 1 mM H 2 O 2 , but they were slightly sensitive to 3 mM and 10 mM H 2 O 2 ( Figure Supplement 4). Thus, we used these condition for the Pyk1-Tsa1 interaction. As shown in Fig. 6B, a dramatic increase in the interaction between Pyk1 and Tsa1 WT occurred in early stationary phase cells in response to 1 mM and 3 mM H 2 O 2 (compare lane 7 with lanes 8 and 9 in the lower panel). The maximum interaction was increased by a factor of 5 (1 mM H 2 O 2 ; lower panel, lane 8) and was equivalent to the level of interaction between Tsa1 C171T and Pyk1 under unstressed conditions (Fig. 5E, lower panel, lane 8). The interaction between Pyk1 and slowly migrating bands containing Tsa1 (> 100 K) as well as monomer and disulfide-linked dimer Tsa1 were also enhanced by H 2 O 2 (Fig. 6B, lower panel, lanes 2 and 3). As predicted, the interaction between Tsa1 and Pyk1 under the steady state and the H 2 O 2 -induced conditions were considerably decreased when Cys48 in Tsa1 was mutated (lower panel, lanes 10, 11 and 12), confirming the critical role of Tsa1 Cys48 in the physical interaction between Pyk1 and Tsa1. The slowly migrating complexes of Tsa1 (lower panel, lanes 2 and 3), which were only present in the anti-Tsa1 blot but not the anti-Pyk1 blot, may consist of mixed disulfide complexes of non-Pyk1 proteins linked to Tsa1 that can physically interact with HA-Pyk1.
It is possible that Pyk1 Cys reacts with Tsa1-Cys48-SOH to form a transient disulfide bond, followed by the formation of a disulfide bond within Pyk1 by a disulfide bond exchange reaction. To evaluate this possibility, we performed a PEG-maleimide assay in which the number of oxidized cysteine residues can be estimated by mobility analysis via SDS-PAGE 4 . As shown in Fig. 6C, the oxidation status of the cysteine residues of Pyk1 WT was not altered in response to 1 mM H 2 O 2 , despite the repression of PK activity in the lysate by H 2 O 2 treatment (Fig. 6D). Pyk1 oxidation status was also not affected by the loss of Tsa1/2. In addition, Pyk1 protein levels were not affected by the loss of Tsa1/2. Accordingly, it is possible that the redox status of Pyk1 was unaffected by the H 2 O 2 -induced Pyk1-Tsa1 interaction, and the disulfide bond between Pyk1 and Tsa1 was only observed in the context of the Tsa1 C171T mutation (see Fig. 5E). Taken together, these results depict a unique model in which Tsa1 can downregulate Pyk1 activity due to the direct physical interaction between Tsa1 and Pyk1. While a low level of Tsa1-Pyk1 interaction is constitutive, this interaction is largely induced in response to H 2 O 2 via a Tsa1 Cys48-dependent mechanism.
Role of Tsa1-Cys48 oxidation on the interaction with Pyk1. Peroxiredoxin (Tsa1) forms dimers and decamers under physiological conditions 2 . The decamer and higher multimeric forms are induced under oxidative stress or heat stress via high H 2 O 2 -induced hyperoxidation of the peroxidatic cysteine (Cys48-thiol) to Cys48-sulfinic acid (-SOOH, hyperoxidation) 20 . Interestingly, multimerization of Tsa1 is associated with a peroxidase-to-chaperone functional switch, which decreases the levels of proteins that are aggregated by heat stress, and this phenomenon is required for heat shock resistance 20 . To examine whether the Pyk1-Tsa1 interaction depends on the H 2 O 2 concentration and hyperoxidation of Tsa1, we treated early stationary phase yeast cells with a higher but tolerable concentration of H 2 O 2 (see above; 1 mM, 3 mM or 10 mM) for 5 min. As shown in Fig. 7A, a significant level of Tsa1 was bound to Pyk1 in 1 mM H 2 O 2 (lane 2, IP, anti-Tsa1), but the binding was inhibited in 3 mM and 10 mM H 2 O 2 (lanes 3 and 4 of IP, anti-Tsa1). The Pyk1-bound Tsa1 was unreactive to anti-Prx-SO 2 /SO 3 antibody (lanes 2 to 4, IP, anti-Prx-SO 2 /SO 3 ). In contrast, the hyperoxidation level of Tsa1  5). (C) The redox status of Pyk1 WT and Pyk1 CA in yeast lysates after treatment with H 2 O 2 or DTT. Lysates were prepared from early stationary phase cells (3 days), and the lysates (10 μ g) were treated with 1 mM H 2 O 2 or 50 mM DTT for several minutes. PEGylated maleimide assays were performed as previously described 4,18 using 5 mM PEGylated maleimide (MW 2 kDa). Pyk1 migrates at 65 kDa (see Fig. 4), whereas Pyk1 with seven attached PEGylated maleimide molecules migrates at 90 kDa. The difference in mobility between Pyk1 WT and Pyk1 CA is due to the C121A/C174A substitution. Pyk1 CA is used as a control for Pyk1 with two oxidized cysteines. (D) PK activity was determined in the lysate prepared from PYK1 WT and tsa1/2Δ cells. Lysates were prepared from early stationary phase cells (3 days) and treated with 1 mM H 2 O 2 for a few minutes, and then the PK activity was determined. "**" indicates p values of 0.0085. The data are expressed as the mean + /− the standard error of the mean (N = 3).  Cys48 in whole cell extracts was increased in 3 mM and 10 mM H 2 O 2 (lanes 3 and 4, WCE, anti-Prx-SO 2 /SO 3 ). Thus, it is possible that the hyperoxidation of Tsa1 was not responsible for the binding to Pyk1. Furthermore, the Tsa1 C48D mutation, which can mimic hyperoxidation and enhance chaperone activity 21 , failed to enhance the binding of Tsa1 to Pyk1 (Fig. 7B).
To further elucidate the significance of sulfenic acid formation of Tsa1-Cys48, we deleted the carboxy-terminal YF-motif 2 to generate a hyperoxidation-resistant mutant of Tsa1 (Tsa1∆ C). In fact, the level of hyperoxidation was decreased by the deletion mutation of Tsa1 (Figure Supplement 4B). Interestingly, in response to 1 mM H 2 O 2 and especially to 10 mM H 2 O 2 , the Tsa1∆ C efficiently formed complexes of proteins with a variety of molecular sizes of approximately 25-34 kDa and 100 kDa in the IP fraction of HA-Pyk1 (Fig. 7C, lanes 11 and 12; indicated as an asterisk and "#", respectively). These Tsa1∆ C complexes were sensitive to DTT treatment (compare lanes  11 and 12 with lanes 23 and 24, respectively). To explore the correlation between the Tsa1-Pyk1 interaction and yeast growth in SEM -Trp, we examined whether Tsa1∆ C could suppress the growth phenotype. We found that the growth of cells expressing Tsa1∆ C on SEM -Trp was similar to that of cells expressing Tsa1 C171T ( Figure  Supplement 4C). Thus, the enhanced interaction between Tsa1 and Pyk1 correlated to the Trp-requirement phenotype.
These results suggest that the H 2 O 2 -induced efficient interaction between Tsa1 and Pyk1 requires sulfenic acid (-SOH) formation of Tsa1-Cys48 but not hyperoxidation of Tsa1-Cys48. In addition to the observation that Pyk1-bound Tsa1 C171T efficiently formed mixed-disulfide complexes (see Fig. 5E), the above results suggested that the Tsa1-Cys48-SOH may induce mixed-disulfide formation with various proteins to enhance the direct physical interaction between Tsa1 and Pyk1 in a non-redox manner. Tsa1 functions as a peroxidase at low levels of H 2 O 2 and as a hyperoxidation-induced molecular chaperone at high levels of H 2 O 2 . Our results suggested an alternative function of Tsa1 as a redox-dependent target modulator (Fig. 7D).
Requirement of Cys48 but not the peroxidase activity of Tsa1 for efficient growth during the diauxic shift and biomass in glucose medium. Tsa1 is required for efficient growth during the diauxic shift in rich glucose medium 5,7-9 (see Fig. 1B). Because PYK1 is essential for vegetative growth in glucose medium, it is not possible to perform genetic analyses to estimate the role of Pyk1 in phenotype. Therefore, we examined the importance of the peroxidatic cysteine (Cys48) on phenotype in glucose medium. As shown in Fig. 8A, the growth retardation due to Tsa1/2 loss during the diauxic shift was almost completely suppressed by expressing Tsa1 C171T ; however, this did not occur in response to Tsa1 C48T expression. Trehalose and glycogen biosynthesis are induced during the diauxic shift via gluconeogenesis 17 . The levels of trehalose and glycogen were decreased by the loss of Tsa1/2; however, these levels were recovered by the expression of Tsa1 WT and Tsa1 C171T but not by Tsa1 C48T (Fig. 8B). These results suggested that Tsa1 Cys48 is essential for the regulation of the gluconeogenic flux during the diauxic shift and yeast growth. As expected 1,2 , yeast cells carrying either Tsa1 C48T or Tsa1 C171T were sensitive to H 2 O 2 (Fig. 8C). Collectively, these data provide evidence for a role of the nonperoxidase function of Tsa1 in gluconeogenesis.

Discussion
Using ethanol medium as a diauxic shift-mimicking condition, we assumed an important role for Tsa1 Cys48 and identified Pyk1 as a regulatory target for Tsa1. Because the loss of Tsa1/2 did not affect growth in ethanol medium when Trp and Phe were supplied (Fig. 1C), Tsa1 might control not only Pyk1 but also additional Tsa1 target protein(s) for efficient growth in glucose medium during the diauxic shift to ensure maximum biomass. Nevertheless, the improved understanding of the Tsa1-dependent regulatory mechanism of Pyk1 provides evidence for an alternative role of the peroxidatic cysteine Cys48 of Tsa1 in efficient gluconeogenesis. This finding is further supported by the ability of Tsa1 C171T expression to almost completely recover the growth retardation and to almost restore the levels of trehalose/glycogen during the diauxic shift in glucose medium (Fig. 8).
We demonstrate that after the diauxic sift, a significant level of Tsa1 is associated with Pyk1, with eight molecules of Tsa1 C171T estimated to be associated with each tetramer of Pyk1 (Fig. 5E) if all of the Pyk1 is completely associated. This association corresponds to the Tsa1-Pyk1 binding in cells growing in ethanol medium (Fig. 4B) as well as to the early stationary phase in response to H 2 O 2 (Fig. 6B). The interaction results in the downregulation of PK activity in cells during gluconeogenesis. Our metabolomic analysis indicated that the level of FBP, an allosteric activator in cells in ethanol medium and cells in the early stationary phase in glucose medium (gluconeogenesis), was 4% or 5% of that in cells growing exponentially in glucose medium (glycolysis, see Table Supplement 1). Pyk1 is catalytically active, even in the presence of such a low level of FPB 10,13 . In addition, Pyk1 plays an important role in the synthesis of several amino acids from pyruvate via gluconeogenesis using ethanol as a carbon source 14 . Thus, we speculate that Pyk1 must be moderately downregulated to avoid energy-wasting cycling reactions and to achieve efficient gluconeogenesis. The results presented herein provide a mechanistic understanding of Pyk1 modulation under these circumstances. We show that downregulation of the PK activity of Pyk1 is governed by the physical interaction between Pyk1 and Tsa1 and that this downregulation is induced by two mechanisms. First, while the Pyk1 protein level is constitutively high from the exponential phase to the diauxic shift (48 h), it decreases to approximately 13% in the early stationary phase (post-diauxic shift; see Fig. 5D). Second, we demonstrate that the peroxidatic cysteine (Cys48) of Tsa1 plays an important role in the ROS-induced physical interaction of Tsa1-Pyk1. The H 2 O 2 -reactive catalytic center of Tsa1-Cys48 is required for proper metabolic flow (Fig. 3B) and for Tsa1 binding to Pyk1 in ethanol medium (Fig. 4B). The correlation between intracellular ROS levels (Fig. 6A) and the level of Tsa1 bound to Pyk1 in ethanol medium is consistent with the idea that ROS (H 2 O 2 )-induced oxidation of Tsa1-Cys48 is responsible for the physical interaction of Pyk1-Tsa1. In the early stationary phase in glucose medium, Tsa1-Cys48 is also essential to ensure a sufficient Pyk1-Tsa1 interaction, especially in response to H 2 O 2 (Fig. 6B). The slowly migrating Tsa1-containing bands found in the Pyk1-IP fraction (Fig. 6B) (Tsa1-Cys48). (A) Growth curve of tsa1/2Δ cells carrying pRS315 empty vector (EV) or pRS315-TSA1 (WT) or pRS315-TSA1 C48T (C48T) or pRS315-TSA1 C171T (C171T). The cells were cultured in SDM -Leu. (B) The levels of trehalose and glycogen in these Tsa1 mutant-expressing cells. The cells from the above cultures were collected (10 OD 600 units) at 36 h, and trehalose and glycogen levels were evaluated as previously described 33 . The data are expressed as the mean + /− the standard error of the mean (N = 3). "*" and "**" indicate p values < 0.02 and < 0.01, respectively. Some p values greater than 0.03 are also indicated. (C) The above cells were spotted on SDM-Leu ager containing the indicated concentration of H 2 O 2 , and incubated for two days. (D) Model for Tsa1 as a H 2 O 2 receptor and target modulator. Tsa1-Cys48 in the Tsa1 tetramer is oxidized to sulfenic acid and forms mixed disulfide followed by interactions with Pyk1. Pyk1 activity is suppressed in the absence of the allosteric activator FBP. It is possible that an unknown Tsa1 target(s) may be similarly regulated in a Tsa1-dependent manner to ensure gluconeogenesis. mixed disulfide bond between Tsa1-Cys48 with other proteins may induce conformational alterations in the Tsa1 tetramer to trigger physical interactions between Tsa1 and Pyk1 (Fig. 8D). Yeast cells need to adapt to changes in the availability of carbon sources and oxygen, which can result in mitochondrial ROS generation 22 . The mammalian enzyme PKM2 is regulated by the inhibition of tetramer (active form) formation mediated by several mechanisms 23 , such as serine availability, PEP levels, and tyrosine phosphorylation. However, this is not the case with Pyk1 since it is constitutively present as a tetramer 12 . Moderate changes in Pyk1 activity in the presence of low levels of the allosteric activator FBP (under low glucose) are likely to be crucial for controlling gluconeogenesis, and metabolites from PEP in yeast cells. Our results suggest that Tsa1 can act as a redox sensor and a metabolic modulator during this adaptive program: the glycolysis-to-gluconeogenesis switch (Fig. 8D).
Additional factors might also be regulated similarly by Tsa1-Cys48-dependent mechanisms to enhance trehalose/glycogen levels during the diauxic shift in glucose medium (Fig. 8). The Tsa1-interacting protein Ssb1/Ssb2 (cytosolic Hsc70/Hsp70 family) and Frd1 are additional candidate Tsa1 target proteins that may be correlated to phenotype. Hsc70 is a multi-functional chaperone that is present at a level comparable to Tsa1. Recently, it has been shown that the lifespan is extended by the enhanced physical interaction between Tsa1 and Hsp70 24 . This interaction is induced by hyperoxidation of Tsa1 by H 2 O 2 , and enhances the recovery of aggregated proteins during aging. In addition, Frd1 is a cytosolic soluble fumarate reductase 25 that may affect gluconeogenesis. We examined whether Frd1 could affect gluconeogenesis. However, we found that neither the disruption of FRD1 nor the overexpression of Frd1 affected the observed phenotype (our unpublished observation). Thus, the identification of additional Tsa1 target(s) is necessary to complete our understanding of the role of Tsa1-Cys48 in gluconeogenesis.
Tsa1 is a major yeast Prx that performs not only thioredoxin-dependent peroxidase activities but also chaperone functions 20 , both of which are conserved in mammalian systems 26 . Our results suggest a novel function of Prx that involves redox regulation of the peroxidatic cysteine. Consequently, this is the first report to demonstrate that Tsa1 has a dual function as an H 2 O 2 receptor and as a modulator of target proteins by direct physical interactions.
Spot assays were performed as follows: a series of one-fifth dilutions of 5 × 10 4 of the appropriate cells prepared under the indicated conditions were spotted on an agar plate (2% Bacto agar, BD Difco), and the agar plate was incubated at 30 °C for 2 days (SDM), 3 days (SDM containing H 2 O 2 ), or 4-5 days (SEM). Amino acids and unspecified reagents were purchased from Nacalai Tesque, Kyoto, Japan.
Yeast metabolomics analysis. Yeast metabolites were extracted from cell cultures (10 OD 600 units) as previously described 28 , and metabolomic analysis was performed using capillary electrophoresis mass spectrometry 29 . Measurements of intracellular ROS by flow cytometry. Dihydroethidium (DHE; Wako pure chemical, Osaka, Japan) was dissolved in DMSO (1.9 mg/ml). Yeast cell cultures (0.5 ml x OD 600nm = 1) were mixed with 5 μ l of DHE stock solution and incubated for 10 min at room temperature. After the cells were washed with medium, the fluorescence of each group of cells (10,000 cells) was measured with a FACSCalibur (Becton, Dickinson and Company, Franklin Lakes, NJ, USA) using FL3. We determined a threshold level for each fluorescent probe by measuring the background fluorescence of unlabeled cells and comparing the average fluorescence values, which were obtained using appropriate conditions (laser levels and detector gain) that had been determined for each fluorescent probe.
Manipulation of genes and yeast strain construction. We manipulated a parental S. cerevisiae strain BY4742 (MATα his3Δ 1 leu2Δ 0 lys2Δ 0 ura3Δ 0) (EUROSCARF, Frankfurt, Germany) derived from S288C to construct the strains listed in Table S2. Tsa1 and Tsa2 are homologous Prxs that are present in the cytoplasm and have similar peroxidase activity. Although TSA1 is the most abundant Prx (more than 90% of the total Prx 7 ), TSA2 expression (approximately 1% of the total Prx) is induced in a tsa1Δ genetic background and tends to compensate for the lost cellular antioxidant activity 8 . Thus, we examined tsa1Δ tsa2Δ double knockout cells (abbreviated hereafter as tsa1/2Δ and tsa1/2Δ for genotype and strain, respectively). The parental strain BY4742 was used as wild type cells.
To construct pyk1::kanMX, we replaced a coding region of PYK1 with the kanMX gene from pUG6 30 . These DNA fragments first form as plasmid DNA and are then PCR-amplified. After transfection of the yeast cells, appropriate transformants were selected on agar plates containing G418 (kanMX) and hygromycin B (hphMX) 31 . To construct the Cys121/Cys174 mutant of PYK1 (PYK1 CA ), two codons, "TGT" at Cys121 and Cys174, were mutated to "GCT" (Ala) using a PCR-based method. Other mutations were introduced in a similar manner. We introduced DNA fragments encoding a duplicated hemagglutinin epitope HA 2 -tag Scientific RepoRts | 6:33536 | DOI: 10.1038/srep33536 (MAAMYPYDVPDYAGSYPYDVPDYAGSRV) at the 5′ end of the coding sequence of Pyk1 (pRS313-HA-PYK1). Next, we used a basic yeast gene replacement method as described elsewhere 27 . To replace the genomic allele of PYK1 with a PYK1 cysteine mutant, we first introduced the URA3 gene into the NdeI site at -285 of the PYK1 coding region, and the resulting PYK1-pro::URA3 allele was replaced at the corresponding genomic allele. We then again replaced the genomic locus with a DNA fragment containing PYK1 WT or PYK1 CA . We confirmed all of the genomic replacements by PCR. The list of these strains is provided in Table Supplement 2. The TSA1 gene (− 764 to + 343 from the Tsa1 ORF) was cloned into pRS315 (CEN-ARS origin, LEU3). Next, codon 48 "TGT" (Cys) was mutated to "GAC" (Asp). We also deleted a C-terminal region corresponding to codon 180 to 196 to prepare pRS315-TSA1Δ C. Other Tsa1 mutants (Tsa1 C48T , Tsa1 C171T , Tsa1 C48,171T ) have been previously described 18 . Affinity purification and identification of Tsa1 C171T binding protein. To decrease non-specific binding during the purification of the Tsa1 protein and its binding proteins from yeast cells, we replaced the polyhistidine (His)-tag region in the Tsa1 expression plasmids 18 with a HAT-tag (histidine affinity tag; amino acid sequence "KDHLIHNVHKEHAHAHNK") to produce pRS-315-HAT-tag-TSA1 C171T and pRS-315-HAT-tag-TSA1 C48T . We treated an exponential-phase culture (0.6 OD 600 ) of tsa1Δ cells carrying a pRS-315-HAT-tag-TSA1 C171T SDM -Leu with 0.5 mM H 2 O 2 in fresh SDM -Leu for 5 min. The cells were washed with 10 mM N-ethylmaleimide (NEM) and suspended in Buffer L (50 mM NaH 2 PO 4 , pH 8.0, 300 mM NaCl, 0.2% Triton X-100, 50 mM NEM, 1 mM phenylmethanesulfonyl fluoride, 0.5 mM N-p-tosyl-L-phenylalanine chloromethyl ketone, 0.05 mM N α -tosyl-L-lysine chloromethyl ketone hydrochloride, 2 μ g/ml pepstatin A, 5 μ g/ml aprotinin and 3 μ g/ml leupeptin). The cells were then frozen and disrupted in liquid nitrogen by grinding in a mortar placed on dry ice. The supernatant lysate was collected after centrifugation (15,000 × g, 10 min), and purification was performed using TALON ® Metal Affinity Resins (Takara Clontech Laboratories Inc. CA, USA). After incubation of 1 ml of the resins with the lysate, the resins were washed two times with Buffer W1 (50 mM NaH 2 PO 4 , pH 8.0, 300 mM NaCl, 2% Triton X-100 and 10 mM imidazole), eluted with 1 ml of elution buffer (50 mM NaH 2 PO 4 , pH 8.0, 300 mM NaCl, and 150 mM imidazole), and concentrated using Vivaspin 500 devices (30,000 MWCO PES, Vivaproducts, Littleton, MA USA). The proteins were quantified with a DC Protein Assay kit (Bio-Rad Laboratories, Hercules, CA) with BSA as the standard. Tsa1-bound proteins (100 μ g) in sample buffer (50 mM Tris-HCl, pH 6.8, 2% SDS, 0.1% bromophenol blue, and 10% glycerol) were separated by two-dimensional gel electrophoresis (non-reduced and reduced in the first and second dimensions, respectively). Polyacrylamide gel electrophoresis was conducted using SDS-PAGE (10%, with an acrylamide to bisacrylamide ratio of 29:1). We used a disk gel (3.5-mm inner diameter; 14 cm and 2 cm for the running gel and stacking gel, respectively) for the first dimension of electrophoresis (2 mA/disk) and a slab gel (1-mm thick, and 165 mm and 30 mm for the running and stacking gels, respectively) for the second dimension. After the first dimension of electrophoresis, the disk gel was soaked in a buffer containing 50 mM Tris-HCl, pH 6.8, 2% SDS, 0.1% BPB, 10% glycerol and 5% 2-mercaptoethanol for 60 min prior to the second dimension of electrophoresis. The slab gels were stained using a Silver Stain II Kit (Wako Pure Chemical) using 0.2 g/ml sodium thiosulfate as the developer. Proteins in the spots were carboxymethylated and trypsinized. The resulting peptides were analyzed by TOF-MS/MS using a MALDI-Qq-TOF MS/MS QSTAR Pulsar i (Applied Biosystems, ThermoFisher Scientific).
Whole cell extract preparation, co-immunoprecipitation of Pyk1 and immunoblotting. We successfully identified the Pyk1-Tsa1 complex using co-IP experiments. pyk1∆ tsa1/2∆ cells carrying an expression vector for HA-Pyk1 and wild type Tsa1 or Tsa1 mutant were collected by centrifugation at 5,000 × g for 1 min and then suspended in IP buffer (100 mM 2-morpholinoethanesulfonic acid, pH 7.0, 200 mM KCl, 15 mM MgCl 2 , 5% glycerol, 0.15% NP-40, 50 mM N-ethylmaleimide (NEM)) containing 2 mM phenylmethanesulfonyl fluoride, 1 μ g/ml leupeptin and 1 μ g/ml pepstatin A. NEM was added to block free Cys-SH to prevent artificial thiol-disulfide exchange reactions. The cells were then frozen in liquid nitrogen and disrupted by shaking at 2,000 rpm for 30 sec with a Multi-beads shocker (Yasui Kikai Corporation, Osaka, Japan). The whole cell extract (WCE; 400 μ g) was mixed with anti-HA antibody beads (Wako Pure Chemical) or with anti-HA Affinity Matrix (Roche Diagnostics, Basel, Germany) at 4 °C for 3 h. Bound proteins were eluted from the beads with 90 μ l of sample buffer (50 mM Tris-HCl, pH 6.8, 2% SDS, 0.1% bromophenol blue, and 10% glycerol) with or without 50 mM DTT. The immunoprecipitates were analyzed by SDS-PAGE (10% and 15% acrylamide gel for Pyk1 and Tsa1, respectively) and immunoblotted using anti-Pyk1 rabbit polyclonal antibodies, anti-Tsa1 rabbit polyclonal antibodies 18 , and anti-Prx-SO 2 /SO 3 antibody (LabFrontier, Seoul, Korea). Chemiluminescent images of the immunoblot were obtained using Immobilon Western Chemiluminescent HRP Substrate (Millipore, Billerica, MA, USA) and the VersaDoc ™ imaging system (Bio-Rad). Anti-Pyk1 polyclonal antibody was prepared by immunizing a rabbit with bacterially expressed HIS-Pyk1 protein as previously described 4 . Quick-CBB (Wako pure chemical) was used for Coomassie brilliant blue staining.
Determination of pyruvate kinase activity. We prepared WCE as described above except that NEM and NP40 were omitted, and then determined the pyruvate kinase (PK) activity in the WCE as previously described 12 . FBP was omitted from the reaction mixture. Briefly, 800 μ M PEP was added to a reaction mixture (150 μ l) containing 100 mM 2-morpholinoethanesulfonic acid, pH 7.0, 200 mM KCl, 15 mM MgCl 2 , 5% glycerol, 150 μ M NADH, 2.4 mM ADP, 25 U of lactate dehydrogenase (Toyobo, Osaka, Japan) and 20 μ g of WCE. NADH oxidation was determined at 30-s intervals by measuring the OD 340 using a spectrophotometer (Varioskan Flash, Thermo Scientific, MA, USA).
Interaction between recombinant Pyk1 and Tsa1 proteins. HA-PYK1 and TSA1 were cloned into pET15b (Millipore), and the recombinant proteins were isolated from E. coli. Recombinant HIS-HA-Pyk1 (20 μ g) and recombinant HIS-Tsa1 (23 μ g) were mixed in 24 mM KH 2 PO 4 /K 2 HPO 4 (pH 7.0) and 10 mM MgSO 4 (50 μ l) with or without 10 mM DTT and then incubated for 10 min. The reaction mixtures were then mixed with IP buffer containing 50 mM NEM (600 μ l) that included anti-HA antibody beads and incubated for 1 h with rotation. The beads were washed three times with IP buffer and then boiled in sample buffer with or without 50 mM DTT. HIS-HA-Pyk1 was detected using anti-HIS-Tsa1 antibody (Fig. 4D) because antibody against the HIS tag had been generated during rabbit immunization rabbit with HIS-Tsa1 4 .