The yeast transcription factor Stb5 acts as a negative regulator of autophagy by modulating cellular metabolism

ABSTRACT Macroautophagy/autophagy is a highly conserved pathway of cellular degradation and recycling that maintains cell health during homeostatic conditions and facilitates survival during stress. Aberrant cellular autophagy contributes to the pathogenesis of human diseases such as cancer, neurodegeneration, and cardiovascular, metabolic and lysosomal storage disorders. Despite decades of research, there remain unanswered questions as to how autophagy modulates cellular metabolism, and, conversely, how cellular metabolism affects autophagy activity. Here, we have identified the yeast metabolic transcription factor Stb5 as a negative regulator of autophagy. Chromosomal deletion of STB5 in the yeast Saccharomyces cerevisiae enhances autophagy. Loss of Stb5 results in the upregulation of select autophagy-related (ATG) transcripts under nutrient-replete conditions; however, the Stb5-mediated impact on autophagy occurs primarily through its effect on genes involved in NADPH production and the pentose phosphate pathway. This work provides insight into the intersection of Stb5 as a transcription factor that regulates both cellular metabolic responses and autophagy activity. Abbreviations: bp, base pairs; ChIP, chromatin immunoprecipitation; G6PD, glucose-6-phosphate dehydrogenase; GFP, green fluorescent protein; IDR, intrinsically disordered region; NAD, nicotinamide adenine dinucleotide; NADP+, nicotinamide adenine dinucleotide phosphate; NADPH, nicotinamide adenine dinucleotide phosphate (reduced); ORF, open reading frame; PA, protein A; PCR, polymerase chain reaction; PE, phosphatidylethanolamine; PPP, pentose phosphate pathway; prApe1, precursor aminopeptidase I; ROS, reactive oxygen species; RT-qPCR, real-time quantitative PCR; SD, standard deviation; TF, transcription factor; TOR, target of rapamycin; WT, wild-type.


Introduction
Macroautophagy/autophagy is a dynamic pathway of cellular degradation and recycling that is highly conserved from yeast to humans. Basal autophagy is low, but is upregulated during stress conditions, such as nutrient deprivation and pathogen infection. In the cell, autophagy serves both metabolic and quality control functions [1]. Classically, autophagy degrades cellular components, thereby enabling the cell to recycle the necessary macromolecules to ensure survival during nutrientlimiting conditions. The hallmark morphological feature of autophagy is the double-membrane autophagosome. During the early stages of autophagy in yeast, induction and nucleation of the phagophore (the autophagosome precursor) occurs at a distinct perivacuolar site -the phagophore assembly site (PAS). Expansion of the phagophore membrane, closure, and maturation of the vesicle culminates in the formation of the autophagosome. Flux through a late stage in the pathway results in the fusion of the autophagosome with the vacuole (in yeast) or the lysosome (in mammalian cells). Within the vacuole, the autophagic cargo is degraded, and the resulting biomolecules are transported to the cytosol for reuse as macromolecular building blocks. Despite its significance and obvious connections, the intimate relationship between cellular metabolism and autophagy (including how individual metabolites affect autophagy) is not well understood.
Two major metabolic triggers of autophagy occur through the depletion of intracellular amino acid levels and the generation of reactive oxygen species (ROS). However, at a biochemical level, the manner in which autophagy contributes to individual metabolic pathways and the mechanisms by which unique metabolites affect autophagy activity are still largely open to investigation. Furthermore, whether there are differences in cellular metabolism that could be the result of the type of autophagy and the particular stimulus activating autophagy is still largely unknown. Under nutrient-limiting conditions, the overall cellular levels of glucose and amino acids decrease, leading to a decline in downstream metabolism and the corresponding metabolic intermediates [2]. However, during starvation, the cell is also required to scavenge for metabolites for use in anabolic reactions to maintain overall homeostasis. The most wellstudied example of the interplay between autophagy and cellular metabolism comes from studies investigating cancer metabolism. Host autophagy maintains circulating amino acid pools that promote tumor growth [3,4]. In fact, certain types of cancers have been described as having an addiction to autophagy (reviewed in [5]).
Furthermore, metabolites such as reduced nicotinamide adenine dinucleotide phosphate (NADPH) contribute to cancer cell survival [6]. Increased NADPH levels in pancreatic ductal carcinoma support cancer progression [7]. NADPH is a major source of reducing equivalents in the cell, providing electrons for redox and anabolic reactions [8] (and reviewed in [6]). The pentose phosphate pathway (PPP) is the major source of NADPH generation in yeast [9]. The oxidative branch of the PPP diverges from glycolysis at the first committed step and consumes glucose-6-phosphate as a primary substrate to generate metabolites such as ribulose 5-phosphate and NADPH [10,11]. The PPP is a central player in cellular metabolism, thereby generating precursors for the biosynthesis of nucleotides, amino acids, and coenzymes [10,11]. The non-oxidative branch of the PPP is key for redox homeostasis [11]. In addition, the PPP is crucial in cancer cell survivalnot only for the generation of pentose phosphates to support increased nucleic acid synthesis, but also for providing NADPH to fulfill metabolic needs [10]. Furthermore, adequate NADPH levels are necessary for maintaining cancer cell survival under stress conditions [10]. However, the genetic and metabolic regulation of the PPP and the interaction between the PPP and autophagy is not well defined.
Interestingly, the transcription factor (TF) Stb5 binds to and directly regulates genes in the PPP and in NADPH production [12][13][14]. Stb5 can function both as a transcriptional activator and a repressor to modulate the cellular response to oxidative stress and multidrug resistance , and stb5∆ (EDA249) cells were grown to mid-log phase in YPD (+N) and then starved for nitrogen (-N) for 2.5 h. The Pho8Δ60 activity was measured and normalized to the activity of starved WT cells, which was set at 100%. (C) WT (WLY176) and stb5∆ (EDA123) cells transformed with a pRS426 (2µ) plasmid encoding GFP-Atg8 under the CUP1 promoter were grown to mid-log phase in rich selective medium and then starved for nitrogen (SD-N) for 0, 0.5, 1, or 2 h. Cells were collected at the indicated time points and protein extracts were analyzed by SDS-PAGE and blotted with anti-GFP or anti-Pgk1 (loading control) antibodies. The blot shown is representative of at least 3 independent experiments. (D) Densitometry of blots represented in (C). The relative percent processed GFP-Atg8 was calculated by determining the ratio of free GFP: total GFP (sum of full-length GFP-Atg8 and free GFP). Results shown are relative to the level of free GFP in the stb5∆ strain during starvation (2 h SD-N), which was set to 100%. (E) WT (WLY176) and stb5∆ (EDA123) cells were grown to mid-log phase in YPD (+N) and then starved for nitrogen for 2 h (-N). Protein extracts were analyzed by SDS-PAGE and blotted with anti-Atg8 or anti-Pgk1 (loading control) antisera (S.E. denotes short exposure; L.E. denotes long exposure). The blot shown is representative of 5 independent experiments. (F) Densitometry of blots represented in (E). The ratio of lipidated Atg8 (Atg8-PE) to the unlipidated form of Atg8 (Atg8) was quantified and normalized to starved WT cells, which was set to 1 (n = 5). In (B) and (D), results are shown as the mean of at least 3 independent experiments ± standard deviation (SD). For (B), (D), and (F), p values are as follows: *p < 0.05; ***p < 0.001; ****p < 0.0001. Also see Fig. S1 and Table S1. [12][13][14]. Stb5 is a C6 zinc cluster (also known as Zn[II] 2 Cys 6 ) DNA-binding protein [15,16]. The C6 zinc cluster DNAbinding domain contains six cysteines in the arrangement CX 2 CX 6 CX 6 CX 2 CX 6 C, where X is any amino acid, which complexes two Zn 2+ ions (reviewed in [16]). The DNAbinding domain of Stb5 is located at the N terminus (Cys22 through Cys49; Figure 1A). Ume6 is a DNA-binding protein that plays in important role in autophagy and also possesses the C6 zinc cluster motif at its C terminus [16,17]. In addition, Stb5 is also predicted to have two N-terminal intrinsically disordered regions (IDRs) located at Val81 to Ala100 and Asn155 to Asn249 [18] ( Figure 1A). Stb5 recognizes various motifs on target promoters, including 5'-CGGNStTAta-3' [19], 5'-CGGNStTAta-3' [14], or 5'-CGGNSNTA-3' [12], where N is A, C, T, or G and S is C or G, with lower case letters less preferred (reviewed in [20]). Furthermore, loss of STB5 decreases NADPH levels under nitrogen-limitation conditions [13] and nitrogen starvation is a potent stimulus for autophagy induction in yeast [21].
To address the gap in our understanding of the interplay between autophagy and cellular metabolic pathways, we investigated the role of the metabolic TF Stb5 in autophagy [22]. In this study, we find that chromosomal deletion of STB5 in yeast enhances autophagy activity, supporting a role for Stb5 as a negative autophagy regulator. Atg1 is a fundamental regulator of autophagy activity [23], and Stb5 protein expression decreases significantly in an Atg1dependent manner when cells are starved for nitrogen. Although we observed that select autophagy-related (ATG) transcripts are upregulated in stb5Δ cells under nutrientreplete conditions, the primary effect on autophagy occurs through Stb5-mediated transcription of genes involved in the generation of cellular NADPH pools. In addition, transcriptional targets of Stb5, including YMR315w, ALD6, and ZWF1 negatively regulate autophagy. Altogether, our data support a model in which autophagy is stimulated when cellular levels of the metabolite NADPH decrease; this mechanism may serve to provide macromolecular building blocks for cellular anabolism during nutrient-limiting conditions.

The metabolic transcription factor Stb5 is a negative regulator of autophagy
Our lab previously conducted a broad screen to search for DNA-binding proteins regulating ATG gene expression and autophagy in yeast [24]. Following up on our previous work [24], we examined potential transcriptional regulators of autophagy and tested whether the yeast metabolic transcription factor Stb5 could be involved in modulating autophagy ( Figure 1). To determine whether Stb5 plays a role in nonselective autophagy, we tested stb5 null cells using the modified vacuolar phosphatase (Pho8Δ60) assay ( Figure 1B). Vacuolar Pho8∆60 activity is a quantitative enzymatic measurement of autophagy activity and autophagic body capacity/ autophagosome volume [25,26], although some minimal degree of basal phosphatase activity can be observed in cells under nutrient-rich conditions [27]. In yeast, nitrogen starvation is a potent stimulus for autophagy induction [21]. Here, we examined the cells at 2.5 h of nitrogen starvation, which is the shortest time point that we found to both produce a robust level of Pho8∆60 activity and reduce background activity. Shorter time points of starvation (1-2 h) lead to higher background levels and a less robust signal-to-noise ratio [27]. As expected, when cells were starved for nitrogen, robust autophagy activity was observed in the wild-type (WT) strain and little activity in the negative control atg8∆ strain (~17%; Figure 1B). Autophagy activity was enhanced in the stb5∆ strain above the level seen in WT under basal and starvation conditions (~10% and > 15% above WT, respectively; Figure 1B).
We also assayed autophagy flux using the GFP-Atg8 processing assay ( Figure 1C,D). Atg8 is required for autophagy [28] and associates with the inner and outer membranes of both the phagophore and mature autophagosome, being subsequently removed from the outer surface and recycled [29]. Autophagic flux culminates in the fusion of the autophagosome with the vacuole. Following autophagosome-vacuole fusion, the GFP-Atg8 that was present on the inner autophagosome membrane is rapidly hydrolyzed while the GFP moiety remains relatively stable within the vacuole. The GFP-Atg8 chimera thus provides a method to assess nonselective autophagy based on the release of free GFP; increasing amounts of free GFP correspond to a greater degree of flux [21,30,31]. This assay also reflects the surface area of the inner autophagosome vesicle (referred to as the autophagic body) that is released following fusion with the vacuole [25]. Here, we used a GFP-Atg8 plasmid expressed under the control of the copper promoter because transcriptional regulators of autophagy may influence ATG8/Atg8 expression levels [17,32]. The stb5∆ strain showed higher levels of free GFP (and thus higher autophagy activity) than the WT strain at 1 and 2 h of nitrogen starvation when using the GFP-Atg8 processing assay ( Figure 1C,D).
Multiple assays are recommended to assess autophagy activity [31]. Thus, we also examined the lipidation status of Atg8 in the stb5∆ strain ( Figure 1E,F). Atg8 exists in two species -a non-lipidated soluble species and a lipidated phosphatidylethanolamine (PE)-conjugated membrane-associated species [31]. When starved for nitrogen, cells lacking STB5 exhibited a greater ratio of lipidated Atg8-PE compared to WT cells ( Figure 1E,F). Finally, we examined cell viability during prolonged nitrogen starvation. Autophagy-deficient cells display reduced viability under these conditions [33,34] as was seen for example with an atg10∆ strain (Fig. S1A); it is presumed that excessive autophagy would result in a similar phenotype [35]. stb5∆ cells demonstrated decreased survival under prolonged nitrogen starvation (Fig. S1A), consistent with previous reports [13], suggesting that prolonged autophagy in the absence of STB5 is detrimental. Taken together, our results indicate that Stb5 functions as a negative regulator of autophagy.
Finally, we also examined whether the loss of Stb5 had an impact on selective autophagy by testing the processing of precursor aminopeptidase I (prApe1) in a vac8∆ strain background ( Fig. S1B and C). The cytoplasm-to-vacuole targeting (Cvt) pathway is a biosynthetic mechanism of selective autophagy that traffics cargo, such as the precursor form of the resident vacuolar hydrolase Ape1, to the vacuole independent of the secretory pathway. Precursor Ape1 trafficking by the Cvt pathway occurs under nutrient-rich conditions and requires its receptor, Atg19 [36], and the scaffold protein Atg11 [37]. In the vac8∆ mutant, the Cvt pathway is defective under nutrient-rich conditions, and prApe1 can only traffic to the vacuole and undergo processing when cells are starved for nitrogen in a selective process that still requires both Atg11 and Atg19 [38]. Precursor Ape1 processing can be monitored by SDS-PAGE and western blotting; prApe1 (61 kDa) is proteolytically processed to Ape1 (50 kDa) within the vacuole. When assayed for prApe1 processing, stb5∆ cells did not demonstrate any significant difference in selective autophagy activity compared to WT cells when starved for nitrogen ( Fig.  S1B and C), supporting the hypothesis that the loss of STB5 negatively regulates only nonselective autophagy.

Stb5 protein expression decreases with nitrogen starvation and autophagy induction
During nutrient-replete conditions, factors that negatively regulate autophagy are typically inactivated following autophagy induction (reviewed in ref [39]). To determine how Stb5 protein levels could be affected during nitrogen starvation and autophagy induction, STB5 was chromosomally tagged at its C terminus with protein A (PA), and endogenous Stb5 fusion protein levels were assessed by western blot (Figure 2A,B). Stb5-PA protein levels significantly decreased with nitrogen starvation in WT cells (~30% by 2 h; Figure 2A, B), further supporting its role as a negative autophagy regulator. To investigate whether autophagy is involved in the downregulation of Stb5 during nitrogen starvation, we examined the expression of the Stb5-PA fusion protein in a strain lacking the autophagy regulator ATG1 ( Figure 2C,D). Atg1 is a serine/threonine kinase that functions as a central regulator of autophagy [23]. However, Stb5 fusion protein levels were stabilized in the absence of ATG1 ( Figure 2C,D), suggesting that Stb5 is downregulated in an Atg1-dependent manner. In addition, we examined Stb5-PA fusion protein expression in a strain lacking the vacuolar protease Pep4/proteinase A (Fig.  S2A,B). In the absence of PEP4, Stb5-PA fusion protein levels were elevated under both nitrogen-replete and starved conditions compared to WT cells (Fig. S2A,S2B), suggesting that Stb5 downregulation was dependent on vacuolar Pep4 in addition to Atg1.
We also investigated Stb5-PA fusion protein expression in a strain lacking ATG4 ( Fig. S2C and D). Atg4 is a cysteine protease required for Atg8 proteolytic processing, which is necessary for subsequent lipidation and autophagosome biogenesis [34,40]. As expected, Stb5-PA fusion protein levels decreased in WT cells; however, Stb5-PA fusion protein levels also decreased in the absence of ATG4 ( Fig. S2C and D), indicating that Atg4 has no role in the downregulation of Stb5 under starvation conditions. We also examined STB5 mRNA levels under both growing and nitrogen-starvation conditions in WT cells with realtime quantitative PCR (RT-qPCR; Fig. S2E). In contrast to our western blot results, we observed a significant increase (>2-fold) of STB5 mRNA levels at 1 h of nitrogen starvation (Fig. S2E), suggesting that STB5 may be regulated at multiple levels under these conditions.

Stb5 negatively regulates the expression of select ATG mRNAs under nutrient-rich conditions
As Stb5 functions in dual capacities to regulate cellular transcription as both a repressor and an activator in yeast [12][13][14] and functions as an autophagy repressor under nutrient-rich conditions, we investigated whether the loss of STB5 altered ATG mRNA levels ( Figure 3A,B). In the absence of STB5 under nutrient-replete conditions, select ATG mRNAs including ATG8, ATG10, and ATG41 were upregulated ( Figure 3A). When cells were starved for nitrogen, only ATG10 levels remained higher in stb5∆ cells compared to WT, and ATG29 levels were lower in stb5∆ cells ( Figure 3B), although in both instances, the differences were relatively small. These results suggest that Stb5 plays a negative role in autophagy regulation primarily under nutrientrich conditions, when autophagy is kept at a low basal level. Atg41 is required for nonselective autophagy, regulates the frequency of autophagosome formation in yeast, and may function as a component of the Atg9 complex that delivers membranes to the expanding phagophore during autophagy induction [25]. As ATG41 was significantly upregulated in the absence of STB5, we next examined whether the observed changes in mRNA levels corresponded with increases at the protein level (Fig. S3). We used a strain expressing ATG41 chromosomally tagged with PA at its C terminus (Fig. S3). As anticipated, we observed a significant upregulation of Atg41-PA with starvation in the WT strain (Fig. S3), consistent with previous work [25]. In stb5∆ cells, Atg41-PA protein levels were higher under nutrient-rich conditions compared to WT cells, consistent with our RT-qPCR results ( Figure 3A and Fig. S3).

Stb5 does not directly bind to the promoters of ATG genes
The Stb5 protein is 743 amino acids in length and has a DNAbinding domain located at residues Cys22 to Cys49 ( Figure 1A). Considering that Stb5 functions as a transcription factor [12][13][14] and that we observed alterations in the expression of select ATG mRNAs in the absence of STB5 (Figure 3), we investigated whether Stb5 directly binds to the promoters of ATG genes (Fig. S4). Using the freely available online yeast transcriptional repository Yeastract (www.yeastract.com), we found that ATG10 has 2 predicted Stb5 consensus sites within its promoter region [41]. To test whether Stb5 directly binds within the ATG10 promoter, we performed chromatin immunoprecipitation (ChIP) experiments using the aforementioned Stb5-PA strain with primers targeted to the two predicted Stb5 binding sites (Fig. S4A). As negative controls, we used primers designed to amplify the ALG9 gene and the open reading frame (ORF) of ATG10; as a positive control we used primers for SNQ2 ( [42]; Fig. S4A). However, we did not find any enrichment of Stb5 binding to either of the 2 predicted sites on the ATG10 promoter under nutrient-rich or nitrogen-starved conditions (Fig. S4A). In addition to the two consensus cites noted above, there are multiple Yeastract predicted sites for Stb5 binding along the ATG8 promoter that lie within regions −209 to −75 upstream of the ATG +1 start site. Using ChIP, we found no evidence for Stb5 binding at the predicted sites within the ATG8 promoter region during nutrient-rich conditions (Fig. S4B). Additionally, we found no ChIP-based evidence that Stb5 bound to the promoters of ATG29 or ATG41 at the predicted sites (data not shown). Thus, we conclude that Stb5 may not directly bind to the promoters of ATG genes.
As we found no evidence for the direct binding of Stb5 to ATG promoters (Fig. S4 and data not shown), we considered whether Stb5 may heterodimerize with another DNA-binding protein to modulate ATG gene expression (Fig. S5). Stb5 can form heterodimers with Pdr1 to differentially regulate cellular transcriptional activity [43]. PDR1 is a multidrug resistance gene and zinc finger transcription factor [44,45]. Accordingly, we examined whether ATG gene expression could be affected in a pdr1∆ strain under either nutrient-rich or starvation conditions (Fig. S5). Another zinc finger transcription factor -Pdr3 -binds to consensus sites similar to Stb5 and Pdr1 [43,46] (reviewed in [20]), and we also examined ATG mRNA levels in the absence of PDR3 (Fig. S5). However, we found no evidence that Pdr1 or Pdr3 significantly altered mRNA levels of ATG genes comparable to what we observed in stb5Δ cells (Figure 3 and Fig. S5), supporting the conclusion that PDR1 and PDR3 do not significantly affect the ATG mRNA transcripts examined here.

The oxidoreductase Ymr315w negatively regulates nonselective autophagy
Because we found no evidence for the direct regulation of ATG mRNAs by Stb5, we investigated whether potential metabolic targets of Stb5 could modulate autophagy. Ymr315w is an NADP(H) oxidoreductase that functions to supply the cell with NADPH and may be a transcriptional target of Stb5 [14]. NADPH is the reduced form of cellular NADP + and is an essential electron donor and driver of anabolic reactions (reviewed in [47]). NADPH/NADP + serves as a pool of redox cofactors in the cell [48]. To investigate whether Ymr315w could also modulate autophagy, we tested the phenotype of the ymr315w∆ strain using the Pho8Δ60 assay ( Figure 4A). In the ymr315∆ strain, we found that Pho8∆60-dependent phosphatase activity was upregulated (>20%) relative to WT under nitrogen-starvation conditions ( Figure 4A). Thus, our findings support a role for the metabolic oxidoreductase Ymr315w as a negative modulator of autophagy.

Stb5 positively regulates YMR315w expression
Because we had observed that cells lacking YMR315w exhibited higher levels of autophagy activity ( Figure 4A) and Stb5 may be a direct transcriptional regulator of YMR315w [14], we investigated whether Stb5 could modulate YMR315w mRNA expression ( Figure 4B). During nutrient-rich conditions, YMR315w transcript levels were significantly lower in the stb5∆ strain compared to WT; no significant differences were noted when the strains were starved for nitrogen ( Figure 4B), supporting the idea that Stb5 positively regulates YMR315w under nutrient-rich conditions.
When cells expressing endogenous YMR315w with a C-terminal PA tag were starved for 2 h, we did not observe any significant changes in the expression of the Ymr315-PA fusion protein ( Figure 4C), suggesting that Ymr315w is maintained at a steadystate level independent of nitrogen-limiting conditions. Next, we tested whether changes in YMR315w mRNA expression corresponded to altered protein levels by examining protein extracts from WT and stb5∆ strains expressing the Ymr315w-PA fusion protein under growing conditions ( Figure 4D,E). We observed that expression was significantly lower in stb5∆ cells compared to the WT under nutrient-rich conditions ( Figure 4D,E), consistent with our RT-qPCR results ( Figure 4B).
Previous work indicated that a region in the YMR315w promoter contains a putative Stb5 binding site [14]. However, to our knowledge, direct DNA binding analysis had not been previously conducted. The predicted site was identified by Hector et al. and is located −180 bp upstream of the ATG +1 start site in the YMR315w promoter; the sequence is 5'-CGGAGTTATC-3' [14]. Therefore, we examined, the direct binding of Stb5 to the YMR315w promoter by performing ChIP analysis using primers covering the putative binding site ( Figure 4F). We found enrichment of Stb5 binding to the YMR315w promoter at the predicted site and to the positive control SNQ2 ( Figure 4F). No enrichment was detected when primers were used targeting a non-coding region at chromosome VI (ChrVI), SLD3 (a gene with no known connection to Stb5), or the ORF of YMR315w ( Figure 4F). Taken all together, our results indicate that Stb5 directly modulates the positive expression of the metabolic oxidoreductase YMR315w ( Figure 4B-F).

Ald6 is a negative autophagy regulator and is transcriptionally targeted by Stb5
Ald6 is a cytosolic Mg 2+ and NADPH-dependent acetaldehyde dehydrogenase that facilitates the reaction converting acetaldehyde to acetate (acetaldehyde + NADP + → acetate + NADPH) [49], and is an important source of NADPH generation in the cell [50]. Previous work showed that Ald6 protein expression decreases in an Atg7-dependent manner during autophagy induction [51]. Furthermore, cell survival is enhanced in the absence of ALD6 when cells are starved for nitrogen [51]. The authors suggested that Ald6 "enzymatic activity may be disadvantageous for survival during nitrogen starvation" [51]. Interestingly, ALD6 is a potential transcriptional target of Stb5 [12] and an important source of NADPH in the cell [50]. As previous work indicated that Ald6 protein levels decrease during autophagy [51] and ALD6 may be a transcriptional target of Stb5 [12], we hypothesized that Ald6 could function as a negative regulator of autophagy. To investigate this possibility, we tested the phenotype of the ald6∆ strain using the Pho8Δ60 assay. In the ald6∆ strain, we found that Pho8∆60-dependent phosphatase activity was upregulated (~15%) relative to WT under starvation conditions ( Figure 5A). We further validated our results using the GFP-Atg8 processing assay in WT and ald6Δ cells. When ald6Δ cells were starved for nitrogen, we observed enhanced GFP-Atg8 processing as indicated by the release of free GFP compared to WT ( Figure 5B,C), supporting the idea that Ald6 functions as a negative autophagy regulator.
Furthermore, we explored whether Stb5 functions as a transcriptional regulator of Ald6 by assaying for ALD6 expression in WT and stb5Δ strains by RT-qPCR analysis ( Figure 5D). In the WT strain, ALD6 transcript levels decreased dramatically during nitrogen starvation ( Figure 5D), consistent with what has previously been observed for Ald6 protein levels [51]. In cells lacking STB5, ALD6 expression was significantly lower in both nutrient-replete and starved conditions ( Figure 5D), supporting the view that Stb5 positively modulates the expression of ALD6. Furthermore, we confirmed the direct binding of Stb5 to the ALD6 promoter by ChIP, again using ChrVI and SNQ2 as negative and positive controls, respectively ( Figure 5E). Together, our results indicate that Ald6 -which serves to generate NADPH in the cellnegatively regulates autophagy activity and is a direct transcriptional target of Stb5.

Zwf1 negatively modulates autophagy independent of Stb5
Zwf1 is also known as glucose-6-phosphate dehydrogenase (G6PD) and functions as a major source for generating  Table S2.
NADPH from NADP + [9,50]. Zwf1 catalyzes the initial step of the pentose phosphate pathway, which yields precursors for the synthesis of nucleotides and some amino acids in addition to producing NADPH [50,52]. Zwf1 is highly conserved from yeast to human; the human ortholog can complement the yeast null strain [53]. Previous work has suggested that Stb5 may transcriptionally regulate ZWF1 [12]. Thus, we examined whether loss of ZWF1 affected autophagy activity (Fig. S6). In the zwf1∆ strain, we found that Pho8∆60-dependent phosphatase activity was enhanced (>15%) relative to WT under starvation conditions, suggesting that Zwf1 functions as a negative modulator of autophagy (Fig. S6A). In WT cells, we noted that ZWF1 mRNA levels were upregulated when cells were starved (Fig. S6B). However, we noted no additional changes in mRNA levels in the absence of STB5, suggesting that the transcriptional regulation of ZWF1 was independent of Stb5 during autophagy (Fig. S6B). However, we observed that Stb5 bound the ZWF1 promoter under nutrient-rich conditions (Fig. S6C) but did not appear to actively influence transcription. Taken together, our data indicate that Zwf1 negatively modulates autophagy activity, but this occurs independently of transcriptional regulation by Stb5.

NADP(H) levels increase with nitrogen starvation and autophagy induction
The molecular cross-talk between the PPP and autophagy is not well defined; this prompted us to investigate how , and ald6∆ (EDA267) cells were grown to mid-log phase in YPD (+N) and then starved for nitrogen (-N) for 2.5 h. The Pho8Δ60 activity was measured and normalized to the activity of starved WT cells, which was set at 100% (n = 4). (B) WT (WLY176) and ald6∆ (EDA266) cells transformed with a pRS426 (2µ) plasmid encoding GFP-Atg8 under the CUP1 promoter were grown to mid-log phase in rich selective medium and then starved for nitrogen for 0 (+N) or 2 h (-N). Cells were collected at the indicated time points and protein extracts were analyzed by SDS-PAGE and blotted with anti-GFP or anti-Pgk1 (loading control) antibodies. A representative blot is shown (n = 3). (C) Densitometry of blots in (B). The relative percent processed GFP-Atg8 was calculated by determining the ratio of free GFP:total GFP (sum of full-length GFP-Atg8 and free GFP). Results shown are relative to the level of free GFP in the ald6∆ strain during starvation (-N), which was set to 100% (n = 3). (D) WT (WLY176) and stb5∆ (EDA123) cells were grown to mid-log phase in YPD (+N) and then nitrogen starved (-N) for 1 h. Total RNA was extracted, and RT-qPCR was performed. Results shown are relative to the level of ALD6 mRNA expression in WT cells under rich conditions (+N), which was set to 1. The geometric mean of UBC6 and SLD3 was used to quantify relative expression levels (n = 4). (E) Stb5-PA binds the ALD6 promoter. Cells (EDA216) endogenously expressing Stb5-PA or mock (WLY176) were analyzed by ChIP under nutrient-replete (+N) conditions. ChIP was performed on a large noncoding region located at 260 kb on chromosome VI (ChrVI) as a negative control, the ALD6 promoter, and the SNQ2 promoter as a positive control. Results were normalized to input DNA, calibrated to the ChrVI PCR product, and presented as fold-enrichment of Stb5 binding (n = 3). In (A) and (C-E), results shown are the mean ± SD (*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001). Also see Table S2. Figure 6). When WT yeast cells were starved for nitrogen, we observed a significant increase in NADP(H) levels compared to cells under nutrient-replete conditions (~3.5-fold and ~ 2-fold in Figure 6A and B, respectively, which may reflect strain-specific differences). We also observed a significant decrease in NADP(H) levels in stb5∆ cells under both nutrient-rich and nitrogen-starved conditions ( Figure 6B), consistent with what has been observed by others under nitrogen-limited conditions [13]. As a control, NADP(H) levels were lower in the zwf1∆ strain under nutrient-replete conditions, whereas the deletion of ATG8 had no effect ( Figure 6B). Taken together, these findings indicate that NADP(H) levels increase when cells are stimulated to undergo autophagy, thereby providing essential biomolecular building blocks to sustain cellular metabolism under nutrientlimited conditions. In the absence of STB5, cellular NADP(H) levels decreased under both nutrient-rich and nitrogenstarved conditions compared to WT cells, consistent with the role of Stb5 as a transcription factor that regulates both cellular metabolic responses and autophagy activity.

Discussion
Here we present data supporting a model for the yeast metabolic transcription factor Stb5 as a negative regulator of autophagy (Figure 7). Chromosomal deletion of STB5 in yeast resulted in a significant upregulation of autophagy when examined by multiple assays. We also found that Stb5 was well expressed during nutrient-rich conditions and its expression significantly decreased during nitrogen starvation (and was dependent on Atg1 and Pep4), further supporting the hypothesis that Stb5 functions as a negative modulator of autophagy. In contrast, STB5 transcript levels increased during starvation, suggesting that multiple modes of regulation exist in the cell to potentially maintain Stb5 and NADPH levels for anabolic metabolism during conditions in which external nutrients are depleted. Previous work indicates that Stb5 may autoregulate by binding its own promoter [13].
Larochelle et al. identified genes bound by Stb5 through genome-wide analysis with ChIP-chip [12]. These targets included genes involved in the pentose phosphate pathway, NADPH production, and oxidative stress resistance [12]. A study by Ouyang et al. examined WT and stb5 null cells by transcriptome-wide analysis using ChIP-exo under 4 different environmental conditions (ethanol limited, glucose limited [both aerobic and anaerobic] and nitrogen limited) [13] but did not include nutrient-rich conditions (YPD) which is the primary basis for our observations. Here, DNA binding targets of Stb5 were identified, including those involved in NADPH generation and the PPP, confirming and expanding on the repertoire of Stb5 transcriptional targets previously described by Larochelle and colleagues [12].
Our data further indicate that, in the absence of STB5, key ATG transcripts -ATG8 and ATG41 -are upregulated. During the induction of canonical autophagy, ATG8/Atg8 and ATG41/Atg41 levels increase dramatically [25,54]. However, we found no evidence for the direct binding of Stb5 to predicted sites within the promoter regions of either ATG8 or ATG41. However, we cannot exclude the possibility that Stb5 may bind to sites within the promoter other than what we have tested here. We also found no evidence for the involvement of Pdr1 (which can function as a heterodimer with Stb5 to modulate transcriptional activity [12,42]) in the regulation of ATG8 or ATG41 by Stb5. This is consistent with previous work indicating that Stb5 binds promoters in the absence of Pdr1 when cultured in rich medium [12]. However, it is also possible that Stb5 may exert direct effects on ATG promoters -either through as-yet unidentified heterodimeric partners (as has been suggested by others [12]) or on ATG transcripts that we have not examined in this study. Our results suggest that the observed upregulation in ATG8 and ATG41 levels are secondary consequences of the loss of STB5 and are not due to the direct transcriptional control of these genes by Stb5. Instead, the altered expression of ATG8 and ATG41 are likely the result of enhanced cell stress in the absence of STB5, and, subsequently, NADPH limitation, and are related to autophagy induction. Taken together, our results support the hypothesis that Stb5 is a regulator of autophagy and metabolic activity in yeast by directly regulating certain transcriptional targets -YMR315w and ALD6 -that modulate cellular NADPH levels. Chromosomal deletion of either YMR315w or ALD6 enhanced starvation-induced autophagy, indicating that these transcriptional targets of Stb5 are also negative autophagy regulators. Our data also indicate a role for the direct regulation of these genes by Stb5 as evidenced by our ChIP experiments. Ymr315w is a conserved (from yeast to humans) NADP(H)-specific oxidoreductase that is regulated in response to NADPH limitation [14]. In the presence of an intact PPP, Ymr315w positively regulates NADPH levels in the cell [14]. Thus, both Stb5 and Ymr315w are positive regulators of cellular NADPH levels [14]. Furthermore, the exogenous addition (leading to enhanced expression) of STB5 and YMR315w increases NADPH levels [14].
Ald6 is a conserved cytosolic Mg 2+ -activated aldehyde dehydrogenase [49]. The reaction catalyzed by Ald6 converts acetaldehyde to acetate using NADP + as a coenzyme [13]. Acetate production is the second major source of NADPH in yeast (reviewed [52]). The loss of STB5 also reduces acetate levels in the cell [52]. Previous work from the Ohsumi lab showed that Ald6 is a preferred substrate for autophagic degradation during nitrogen starvation [51]. Cells lacking both ALD6 and ATG7 demonstrate enhanced viability over atg7∆ cells during an extended time course of nitrogen starvation [51]; Atg7 is an E1-like enzyme that is required for the conjugation of Atg12-Atg5 during phagophore expansion [55,56]. The authors concluded that Ald6 May be detrimental for cell survival during nitrogen starvation [51]. Stb5 is essential for acetaldehyde tolerance in yeast [57], and Ald6 and Zwf1 may have overlapping roles in NADPH production [58]. Strains lacking both ALD6 and ZWF1 are not viable [50]. One difficulty in analyzing the effect of Stb5 is that it functions as a negative regulator during nutrient-rich conditions. In general, autophagy is maintained at a low basal level under these conditions through the action of various negative regulators including Ume6 [17], Pho23 [59], Rph1 [33], Dcp2 [35], Dhh1 [35], Xrn1 [27], and, of course, the target of rapamycin (TOR) [60]. Hence, removing any one of these negative factors (except for TOR) is insufficient to allow autophagy induction on its own. Instead, it is necessary to measure effects on individual genes and/or proteins. For example, deletion of UME6 results in an increase in ATG8 transcripts and Atg8 protein; that increase does not result in an elevated level of autophagy under nutrient-rich conditions [17]. When autophagy is monitored following a shift to starvation conditions, it is possible to see an elevated level in the ume6∆ strain [17]. A similar situation exists with the elimination of Stb5 and its downstream targets -it is only possible to see an effect on autophagic flux, which requires terminal assays, after shifting to starvation conditions, although we can clearly detect an effect on individual genes in nutrientrich conditions.
In this study, we also found that Zwf1 (glucose-6-phosphate dehydrogenase, G6PD), a regulator of cellular metabolism, the PPP, and NADPH generation also negatively regulates starvation-induced autophagy independent of Stb5. In the absence of ZWF1, cellular NADPH pools decrease, consistent with what has been observed by others [14]. Although we found that Stb5 directly binds to the promoter of ZWF1, we found no evidence for altered ZWF1 transcript levels in the absence of STB5. This is consistent with previous observations, in which Stb5 binds to the ZWF1 promoter under different environmental conditions, but ZWF1 expression is unchanged in the absence of STB5 [13]. ZWF1 functions in a housekeeping role [9]; this could provide an explanation as to why Stb5 is bound, but not mediating transcriptional activity under the conditions examined here.
NADPH/NADP + is the preferred cofactor for anabolic reactions (reviewed in [48]). Here, we found that cellular NADP(H) levels were significantly upregulated when cells were starved for nitrogen. NADP(H) levels increase when cells are stimulated to undergo autophagy, thereby providing essential biomolecular building blocks to sustain cellular metabolism under nutrient-limited conditions. Under nutrient-rich conditions, NADPH levels are lower in stb5∆ cells compared to WT. The cell may stimulate autophagy in response to decreased NADPH levels as indicated by the enhancement of autophagy activity in the stb5∆ strain above WT levels as observed in the Pho8Δ60 assay.
However, under nitrogen-limited conditions, stb5∆ cells generate less NADPH compared to WT, consistent with what has been observed by others [13]. These results indicate that depleted NADPH levels in the cell could upregulate autophagy as we observed in stb5∆ cells, suggesting that autophagy is triggered as a mechanism to restore and maintain cellular pools of this critical metabolite. Our work herein supports the conclusion that the loss of NADPH-generating genes (STB5, YMR315w, ALD6, and ZWF1) enhances autophagy. NADPH limitation may serve as an intracellular sensor to alert the cell to promote autophagy induction. Thus, autophagy activation may provide a mechanism to regenerate NADPH, thereby sustaining NADPH pools for anabolic reactions to continue essential metabolism under nutrient-limited conditions.
While this manuscript was in revision, Kataura and colleagues identified an evolutionarily conserved role for autophagy in the maintenance of cellular nicotinamide adenine dinucleotide (NAD) levels, providing a link between autophagy and NAD metabolism [61].

Yeast strains, media, and cell culture
Yeast strains used in this study are listed in Table S1. Yeast cells were grown in YPD (1% yeast extract, 2% peptone, and 2% glucose) or synthetic minimal medium (SMD; 0.67% yeast nitrogen base and 2% glucose, supplemented with the appropriate auxotrophic amino acids and vitamins). Autophagy was induced by shifting mid-log phase cells from rich medium to nitrogen starvation medium (SD-N; 0.17% yeast nitrogen base without ammonium sulfate or amino acids and 2% glucose) for the indicated times. Gene deletions and chromosome tagging were performed using standard methods [62,63].

Antibodies and inhibitors
Antisera to Ape1 [64] and Atg8 [65] have been described previously. Some experiments used a commercial Atg8 antibody purchased from Santa Cruz Biotechnology (sc -373963). Pgk1 antibody was a generous gift from Dr. Jeremy Thorner (University of California, Berkeley) or purchased from Invitrogen (22C5D8). Antibody to GFP was purchased from Clontech (JL-8; 63281). A commercial antibody that recognizes the PA tag was purchased from Jackson Immunoresearch (323-005-024). Dpm1 monoclonal antibody was purchased from Invitrogen (A6429).

Chromatin immunoprecipitation
ChIP was performed as previously described with modifications [25,32,66]. Yeast cells were cultured in SMD or starved (SD-N) as indicated. Formaldehyde (1% final) was added for DNA-protein cross-linking. Samples were isolated, harvested, and chromatin was isolated. DNA was sheared by sonication, and the sheared chromatin was immunoprecipitated on IgG Sepharose 6 Fast Flow beads (Fisher Scientific, 45-000-173) or saved as input. The protein-DNA complex was eluted from the beads, and reverse cross-linking was performed. ChIP and input samples were purified (Zymo Research, D5201) and analyzed by RT-qPCR. Primer sequences are included in Table S2.

RNA and real-time quantitative PCR (RT-qPCR)
Unless otherwise noted, yeast cells were cultured in YPD to mid-log phase and then shifted to SD-N (1 h) for autophagy induction. Cells (1 OD 600 unit) were then collected, and the pellets were flash frozen in liquid nitrogen. Total RNA was extracted using an RNA extraction kit (Clontech, 740955.250). Reverse transcription was carried out using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems/ Thermo Fisher Scientific, 4368814). For each sample, 1 µg RNA was used for cDNA synthesis. RT-qPCR was performed using the Power SYBR Green PCR Master Mix (Applied Biosystems/Thermo Fisher Scientific, 4367659) in a CFX Connect (Bio-Rad, 1855201) real-time PCR machine. For all RT-qPCR experiments, melt curves were run after the PCR cycles to verify primer specificity. Relative gene expression was calculated using the 2 −ΔΔCT method [67] and normalized as indicated. Primer sequences are included in Table S2.

NADP(H) quantification
Enzymatic cycling assays to determine NADP(H) were performed according to the NADP/NADPH quantification kit (BioVision; Abcam, K347) with the following modifications: Yeast cells were cultured in YPD to mid-log phase and then shifted to SD-N (2 h) for autophagy induction. Cells (1 OD 600 unit) were collected, and the pellets were flash frozen in liquid nitrogen. Cell pellets were resuspended in 300 µL of Extraction Buffer (provided in the kit) and processed for NADP(H) measurements as previously described [14]. Colorimetric detection was performed using the SmartReader 96 microplate absorbance reader (Accuris, MR9600). The values were normalized to the protein content of each sample determined by the Pierce BCA Protein Assay kit (Thermo Scientific, PI23227).

Statistical analysis
The two-tailed Student's t test was used to determine statistical significance unless otherwise indicated. For all figures, p values are as follows: *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. A p value < 0.05 was considered significant.