Expression control of the AMPK regulatory subunit and its functional significance in yeast ER stress response

AMP-activated protein kinase (AMPK) is an evolutionarily conserved heterotrimeric kinase complex consisting of a catalytic subunit, α, and two regulatory subunits, β and γ. Previously, we demonstrated that Snf1, the Saccharomyces cerevisiae ortholog of AMPK, negatively regulates the unfolded protein response (UPR) pathway and the Hog1 MAP kinase pathway in ER stress response. However, it remains unclear how the alternate three β subunits, Sip1, Sip2, and Gal83, of the Snf1 complex participate in ER stress response. Here, we show that Gal83 plays a major role in Snf1-mediated downregulation of the UPR and Hog1 pathways. Gal83 is the most abundant β subunit in the normal state and further induced by ER stress. This induction is mediated via activation of the GAL83 promoter by the UPR. When expressed under the control of the GAL83 promoter, Sip2 exhibits potent functional activity equivalent to Gal83. Our results suggest that the functional significance of the β subunit of Snf1 AMPK in ER stress response is defined by modulation of the expression level through regulation of the promoter activity.

remained unclear which β subunit is important for the negative regulation of the UPR and Hog1 pathways. In this study, we found that Gal83 plays a major role in Snf1-mediated downregulation of the UPR and Hog1 during ER stress response. Among the β subunits, Gal83 is the most abundant under normal conditions, and its expression is further induced by ER stress in a manner dependent on the UPR. When SIP2 was expressed from the GAL83 promoter, loss of Gal83 could be effectively complemented. These results suggest that the functional significance of Gal83 as the AMPK β subunit in ER stress response is defined by its promoter.

Results
Gal83 is the principal β subunit of the Snf1 complex in ER stress response. To investigate which β subunits are involved in regulation of the UPR and Hog1 pathways, we employed reg1 mutation which causes Snf1 hyperactivation. The kinase activity of Snf1 is regulated through phosphorylation of Thr-210 located in its kinase domain 20,21 : Snf1 is phosphorylated and activated by three upstream kinases, Sak1, Tos3, and Elm1 [22][23][24] ; Snf1 inactivation is mediated by the Reg1-Glc7 protein phosphatase 1 complex 25,26 . Reg1 is the regulatory subunit that guides Glc7 catalytic subunit toward Snf1 27,28 . Previously, we showed that Snf1 hyperactivation caused by reg1 mutation leads to rapid downregulation of the UPR activity 19 . The reverse transcription-PCR (RT-PCR) analysis revealed that a large fraction of HAC1 mRNA remained the unspliced form (HAC1 u ) in wild-type cells under unstressed conditions (Fig. 1a). Exposure to dithiothreitol (DTT), which causes ER stress by blocking disulfide bond formation in the ER, induced HAC1 mRNA splicing. The amount of the spliced form of HAC1 mRNA The mean of HAC1 s /(HAC1 u + HAC1 s ) at 3 hours after DTT addition with SEM (n = 4) is shown in (b). *P < 0.05 and **P < 0.01 as determined by Tukey's test. Original data are presented in Supplementary Fig. 4. (c) Hog1 activation after DTT treatment. Wild-type (WT) and indicated mutant strains were grown at 25 °C until exponential phase and treated with 4 mM dithiothreitol (DTT) for 3 hours. Extracts prepared from each cell were immunoblotted with anti-phospho-p38 (P-Hog1) and anti-Hog1 antibodies. Original data are presented in Supplementary Fig. 5. (d) Loss of Gal83 has the strongest influence on ER stress sensitivity. Wild-type (WT) and indicated mutant strains were spotted onto YPD medium lacking or containing 0.5 μ g/ml tunicamycin (TM) and incubated at 25 °C.
(HAC1 s ) peaked 1.5 to 3 hr after DTT addition and gradually decreased thereafter (Fig. 1a). In reg1 mutant cells, promotion of HAC1 mRNA splicing by DTT treatment was apparently normal; however, HAC1 s was decreased rapidly within 3 hr of DTT addition (Fig. 1a). This reg1 defect could be significantly restored by snf1 mutation, while cells harboring snf1 single mutation only exhibited a mild defect in HAC1 mRNA splicing under unstressed conditions 19 . Thus, alteration of the kinetics of HAC1 mRNA splicing caused by reg1 mutation was expected to be highly sensitive to reduction of Snf1 function. Indeed, loss of all three β subunits clearly suppressed rapid downregulation of HAC1 mRNA splicing observed in reg1 mutant cells ( Fig. 1a and b). To further explore the role of the β subunits in the UPR regulation, we examined HAC1 mRNA splicing in single and double mutant cells of the β subunits in a reg1 mutant background ( Fig. 1a and b). The gal83 mutation slightly delayed downregulation of HAC1 mRNA splicing, but neither sip1 nor sip2 mutation did. The gal83 sip1 and gal83 sip2 double mutations significantly delayed downregulation of HAC1 mRNA splicing, while sip1 sip2 double mutation did not. These results suggest that Gal83 is the most important β subunit for Snf1-mediated regulation of the UPR activation in response to DTT. To test whether similar effects were observed using different types of the ER stressor, we next monitored HAC1 mRNA splicing in cells exposed to tunicamycin, which causes ER stress by inhibition of N-linked glycosylation (Supplementary Fig. 1a and b). In wild-type cells, the high amount of HAC1 s was kept over 7.5 hr of tunicamycin addition. In contrast, in reg1 mutant cells, a gradual decrease in HAC1 s was detected within 7.5 hr after tunicamycin addition. Thus, tunicamycin-induced activation of the UPR was sustained long-term compared to DTT; however, reg1 mutation caused a significant decline in the UPR activity, similar to the case when cells were treated with DTT. Furthermore, this reg1 phenotype could be suppressed by gal83 sip1 and gal83 sip2 double mutations, but not by sip1 sip2 double mutation. Taken together, these results suggest that Gal83 acts as the main β subunit of the Snf1 complex in the UPR regulation.
We previously showed that Snf1 is involved in negative regulation of the UPR pathway by using W303 derivative strains 19 . To investigate whether the inhibitory effect of the Snf1 complex on the UPR is restricted to the W303 background, we employed BY4741 derivatives as budding yeast cells harboring a different genetic background. We first compared the kinetics of HAC1 mRNA splicing between wild-type and reg1 mutant cells, and found that an accelerated decline in the UPR activity was occurred in reg1 mutant cells (Supplementary Fig. 2a  and b). Next, we tested double mutant cells of the β subunits in a reg1 mutant background. The gal83 sip1 and gal83 sip2 double mutations modestly delayed downregulation of HAC1 mRNA splicing, while sip1 sip2 double mutation did not (Supplementary Fig. 2a and b). Thus, both in the W303 and BY4741 backgrounds, reg1 mutation downregulates the UPR activity, and this reg1 defect could be restored by loss of Gal83 in combination with Sip1 or Sip2. Therefore, it is suggested that the mechanism by which Snf1 negatively regulates the UPR is widely used in budding yeast cells with different genetic backgrounds. We hereafter mainly described the results from experiments using W303 derivatives, although similar results were obtained from all experiments we have tried using BY4741 derivatives and they were shown in Supplementary Fig. 2.
Next, we examined which β subunits are involved in regulation of the Hog1 pathway. To monitor Hog1 activity, we used anti-phospho-p38 antibodies that recognize the phosphorylated form of mammalian p38 MAPK. As shown previously 19 , western blot analysis with anti-phospho-p38 antibodies strongly detected the activated Hog1 in wild-type cells treated with DTT (Fig. 1c). However, activated Hog1 level was significantly decreased in reg1 mutant cells (Fig. 1c). We have also demonstrated that snf1 mutation could completely restore the reduction of Hog1 activity caused by reg1 mutation 19 . To investigate the role of the β subunits in regulation of Hog1 activity, we first detected activated Hog1 in the reg1 sip1 sip2 gal83 quadruple mutant cells. Loss of all three β subunits clearly suppressed the reduction of Hog1 activity caused by reg1 mutation (Fig. 1c). Therefore, we compared the effects of sip1, sip2 and gal83 mutations on reg1-caused reduction of Hog1 activity (Fig. 1c). The gal83 mutation slightly restored Hog1 activity. On the other hand, neither sip1 nor sip2 single mutation had an obvious effect on Hog1 activity. We also found that the activated Hog1 levels were increased in the following order: sip1 sip2 < sip1 gal83 < sip2 gal83 < sip1 sip2 gal83. Similar observations were seen when cells were exposed to tunicamycin ( Supplementary Fig. 1c). These results suggest that Gal83 is the most important β subunit for Snf1-mediated regulation of the Hog1 pathway.
We previously found that reg1 mutant cells exhibit hypersensitivity to ER stress and this reg1 phenotype is completely suppressed by snf1 mutation 19 . To ask whether Gal83 is actually important for regulation of ER stress response, we examined growth of yeast cells on medium containing tunicamycin (Fig. 1d). Similar to snf1 mutation, loss of all three β subunits suppressed the ER stress sensitive phenotype observed in reg1 mutant cells. The reg1 hypersensitivity to ER stress was partially suppressed by gal83 mutation and to a much lesser extent by sip2 mutation. In contrast, sip1 single mutation failed to suppress the ER stress sensitive phenotype of reg1 mutants. ER stress tolerance caused by gal83 mutation was significantly enhanced by sip2 mutation and to a lesser extent by sip1 mutation. Thus, Gal83 has the strongest influence on ER stress sensitivity among three β subunits. This finding is consistent with the results showing that Gal83 acts as the major β subunit for Snf1-mediated regulation of the UPR and Hog1 pathways.
The relative functional significance of the β subunits is modulated by Snf1 activation level. Next, we investigated whether the relative contribution of the Snf1 β subunits to the UPR activity seen in the reg1 mutants could be observed in cells harboring the wild-type REG1 gene. Previously, we found that snf1 mutation elevates HAC1 mRNA splicing under unstressed conditions, but does not cause a significant change in its kinetics during ER stress response 19 . Therefore, we compared HAC1 mRNA splicing in double mutant cells of the β subunits under unstressed conditions (Fig. 2a). Intriguingly, the gal83 sip1 double mutation increased the level of HAC1 s , while sip1 sip2 and gal83 sip2 double mutations did not. We also found that the level of HAC1 s in the gal83 sip1 sip2 triple mutant cells was higher than that in the gal83 sip1 double mutant cells. These results suggest that Gal83 and Sip1 have greater potential to regulate the UPR than Sip2 in wild-type backgrounds, and that the relative contribution of three β subunits is slightly different between wild-type and reg1 mutant backgrounds.
Previously, we showed that the sip1 sip2 gal83 triple mutant cells were resistant to tunicamycin, although none of their single mutants exhibited the obvious tunicamycin-resistant phenotype 19 . To further clarify involvement of Snf1 β subunits in ER stress response, we tested sip1 sip2, gal83 sip1 and gal83 sip2 double mutants for growth on medium containing tunicamycin (Fig. 2b). The gal83 sip1 and gal83 sip2 double mutants exhibited the tunicamycin-resistant phenotype, while sip1 sip2 double mutants did not. This indicates that Gal83 plays a major role in Snf1-mediated ER stress response. Furthermore, gal83 sip1 mutant cells were more resistant to tunicamycin than gal83 sip2 mutant cells. This observation suggests that, in contrast to a reg1 mutant background, Sip1 is more important for the function in Snf1-mediated ER stress response than Sip2 in a wild-type background.
In both wild-type and reg1 mutant backgrounds, Gal83 acts as the most important β subunit of the Snf1 complex in ER stress response. However, the degree to which Sip1 and Sip2 participate in ER stress response seemed to vary between wild-type and reg1 mutant backgrounds: Sip1 is more and less important than Sip2 in wild-type and reg1 mutant backgrounds, respectively. To elucidate the mechanism by which functional significance of Sip1 and Sip2 is altered, we compared the mRNA levels of SIP1 and SIP2 in wild-type and reg1 mutant cells by a quantitative real-time RT-PCR (qRT-PCR). The SIP1 mRNA level was only modestly reduced by reg1 mutation (Fig. 2c); in contrast, the SIP2 mRNA level was significantly increased by reg1 mutation (Fig. 2d). Furthermore, the increase in SIP2 mRNA caused by reg1 mutation was clearly inhibited by loss of Snf1 (Fig. 2d). These results Wild-type (WT) and indicated mutant strains were grown at 25 °C until exponential phase, and total RNAs prepared from each strain were subjected to RT-PCR of HAC1. Positions of unspliced HAC1 (HAC1 u ) and spliced HAC1 (HAC1 s ) are indicated. The data show the mean of HAC1 s / (HAC1 u + HAC1 s ) with SEM (n = 3). *P < 0.05 and **P < 0.01 as determined by Tukey's test. Original data are presented in Supplementary Fig. 6. (b) ER stress sensitivity of the cells lacking two β subunits. Wild-type (WT) and indicated mutant strains were spotted onto YPD medium lacking or containing 1.5 μ g/ml tunicamycin (TM) and incubated at 25 °C. (c,d) The mRNA levels of SIP1 (c) and SIP2 (d). Wild-type (WT) and indicated mutant strains were grown at 25 °C until exponential phase, and total RNAs were prepared from each strain. The mRNA levels were quantified by qRT-PCR analysis, and relative mRNA levels were calculated using ACT1 mRNA. The data show mean ± SEM (n = 4). *P < 0.05 and **P < 0.01 as determined by Tukey's test. suggest that expression changes of SIP1 and SIP2 mRNAs caused by reg1 mutation-mediated Snf1 activation contribute to enhance the relative functional significance of Sip2 in reg1 mutant cells.
Gal83 expression is higher than those of Sip1 and Sip2, and induced by ER stress. In order to elucidate why Gal83 is the most important for Snf1 to negatively regulate ER stress response, we compared the expression levels of Sip1, Sip2, and Gal83. We generated yeast strains carrying the carboxyl-terminally GFP-tagged genes and quantitated their expression by western blot analysis with anti-GFP antibodies ( Fig. 3a and b). In the normal state, the protein abundance of Gal83 was higher than those of Sip1 and Sip2. This observation is consistent with the previous report 17 . Intriguingly, Gal83, but neither Sip1 nor Sip2, was increased following exposure to DTT ( Fig. 3a and b). Induction of Gal83 was also observed when cells were treated with tunicamycin ( Supplementary Fig. 1d). To investigate how the expression level of Gal83 is upregulated by ER stress, we quantitated GAL83 mRNA by qRT-PCR. We found that the GAL83 mRNA level is transiently increased by ER stress: the amount of GAL83 mRNA peaked 1.5 hr after DTT treatment and decreased thereafter (Figs 3c and 4c). Similar induction was observed when cells were exposed to tunicamycin and in strains harboring the BY4741 background ( Supplementary Figs 1e and 2c). Next, we examined whether increased expression of GAL83 is due to its transcriptional activation by ER stress. To address this, we generated a P GAL83 -GFP reporter, consisting of the 5′ upstream region of the GAL83 gene to drive GFP expression (Fig. 3d), and monitored the amount of GFP mRNA by qRT-PCR. GFP expression from the P GAL83 -GFP reporter was increased after treatment with DTT and tunicamycin ( Fig. 3e and Supplementary Fig. 1f), suggesting that the GAL83 promoter is activated by ER stress. We also tested the possibility that Gal83 is stabilized by ER stress, which consequently contributes to upregulation of the protein level of Gal83. Cells expressing Gal83-GFP were treated with or without DTT, and the protein level of Gal83-GFP was examined following cycloheximide treatment. However, the stability of Gal83-GFP was apparently unaffected by DTT (Fig. 3f). Based on the findings that the induction level of the P GAL83 -GFP reporter was comparable with that of Gal83-GFP protein ( Fig. 3b and e), Gal83 expression is induced by ER stress through transcriptional activation of the GAL83 gene.
We next attempted to identify the regulator of GAL83 expression. Previous studies demonstrated that in budding yeast, several signaling pathways, including the UPR, Mpk1, Hog1, and Snf1, become activated in response to ER stress [1][2][3][4][5][6]19 . Under our experimental conditions, the UPR pathway consisting of Ire1 and Hac1 was quickly activated after DTT addition and thereafter downregulated ( Fig. 1a and b). On the other hand, our previous analyses revealed that activation of Hog1 and Snf1 was occurred comparatively late after exposure to DTT and maintained long-term 19 . Furthermore, the activation time course of Mpk1 was similar to those of Hog1 and Snf1 under our experimental conditions (Supplementary Fig. 3). Therefore, we examined whether the UPR pathway is involved in transcriptional activation of the GAL83 gene. We found that induction of Gal83 protein and GAL83 mRNA following exposure to DTT was impaired in hac1 and ire1 mutant cells ( Fig. 4a and b). The hac1 and ire1 mutations inhibited induction of GAL83 mRNA when cells were exposed to tunicamycin and in strains harboring the BY4741 background ( Supplementary Figs 1e and 2c). These results indicate that the UPR pathway induces GAL83 expression during ER stress response.
The reg1 mutation leads to Snf1 hyperactivation and consequent decreased activity of the UPR pathway during ER stress response 19 (Fig. 1a). This observation raised the possibility that GAL83 expression level was reduced by reg1 mutation. To test this possibility, we measured the amount of GAL83 mRNA in reg1 mutant cells. We found that GAL83 mRNA levels were reduced by reg1 mutation (Fig. 4c). Similar result was obtained in reg1 mutant cells harboring the BY4741 background ( Supplementary Fig. 2d). These results suggest that in ER stress response, Snf1 downregulates expression of its regulatory subunit by inhibiting the UPR activity.
Sip2 expressed from the GAL83 promoter compensates for loss of Gal83. Comparison of the expression levels among the β subunits led us to hypothesize that their protein abundance, but not their protein structure, determines their demands for ER stress response mediated by Snf1. To test this hypothesis, we first generated a P SIP2 -GFP reporter, consisting of the 5′ upstream region of the SIP2 gene to drive GFP expression, and compared its activity to express GFP with a P GAL83 -GFP reporter. The qRT-PCR analysis showed that, under unstressed conditions, GFP mRNA level from a P SIP2 -GFP reporter was half of that from a P GAL83 -GFP reporter (Fig. 3g). Thus, the difference in promoter activity was reflected in their protein level ( Fig. 3b and g). We next generated a P GAL83 -SIP2 construct, which expresses SIP2 under the control of the GAL83 promoter (Fig. 5a). To confirm that the expression pattern of SIP2 mRNA from the P GAL83 -SIP2 integration mimics that of GAL83 mRNA, we quantified SIP2 mRNA in the quadruple mutant cells harboring the wild-type SIP2 (P SIP2 -SIP2) or P GAL83 -SIP2 integration by qRT-PCR. As expected, we found that under unstressed conditions, SIP2 mRNA was expressed at a higher level from the P GAL83 -SIP2 integration than from the P SIP2 -SIP2 integration, and that expression of SIP2 mRNA from the P GAL83 -SIP2 integration was increased by DTT and tunicamycin ( Fig. 5b and Supplementary Fig. 1g). Then, we compared the ability of P GAL83 -SIP2 to alter the phenotype caused by reg1 sip1 sip2 gal83 quadruple mutation with those of P GAL83 -GAL83 and P SIP2 -SIP2. The reg1 sip1 sip2 gal83 quadruple mutant cells were resistant to ER stress, while the reg1 single mutant cells were sensitive to ER stress (Fig. 5c). When harboring the P GAL83 -GAL83 integration, the reg1 sip1 sip2 gal83 quadruple mutants exhibited ER stress hypersensitivity, similar to the reg1 single mutants. However, the reg1 sip1 sip2 gal83 quadruple mutants harboring the P SIP2 -SIP2 integration displayed the intermediate phenotype between the reg1 sip1 sip2 gal83 quadruple mutant and the reg1 single mutant. These results indicate that the P GAL83 -GAL83 integration has a stronger activity to complement sip1 sip2 gal83 triple mutations than the P SIP2 -SIP2 integration. The activity of the P GAL83 -SIP2 integration was the same as that of the PGAL83-GAL83 integration. This result suggests that in ER stress response, the difference in the promoter activity between the GAL83 and SIP2 genes is more important than their difference in the protein structure.  Supplementary Fig. 7. (c) The mRNA levels of GAL83 after DTT treatment. Wild-type (WT) strains were grown at 25 °C until exponential phase and treated with 4 mM dithiothreitol (DTT) for the indicated time. The mRNA levels were quantified by qRT-PCR analysis, and relative mRNA levels were calculated using ACT1 mRNA. The data show mean ± SEM (n = 3). **P < 0.01 as determined by Student's t-test. (d) Schematic representation of the structure of P GAL83 -GFP and P SIP2 -GFP. (e) Effects of DTT treatment on expression of P GAL83 -GFP reporter. Wildtype (WT) cells harboring the integration which expresses GFP under the control of GAL83 promoter were grown at 25 °C until exponential phase and treated with 4 mM dithiothreitol (DTT) for the indicated time. The GFP mRNA levels were quantified by qRT-PCR analysis, and relative mRNA levels were calculated using ACT1 mRNA. The data show mean ± SEM (n = 3). *P < 0.05 as determined by Student's t-test. (f) Stability of Gal83. Cells expressing Gal83-GFP were treated with or without 4 mM dithiothreitol (DTT) for 1.5 hours, and then treated with 0.2 mg/ml cycloheximide (CHX) for the indicated time. The triangle indicates the point when DTT was added. Extracts prepared from each cell were immunoblotted with anti-GFP and anti-Hog1 antibodies. Hog1 was used as a loading control, since it has been reported to be a stable protein 35 . (g) Comparison between GFP expression levels from P GAL83 -GFP and P SIP2 -GFP reporters. Wild-type (WT) cells harboring the integration which expresses GFP under the control of GAL83 or SIP2 promoters were grown at 25 °C until exponential phase. The GFP mRNA levels were quantified by qRT-PCR analysis, and relative mRNA levels were calculated using ACT1 mRNA. The data show mean ± SEM (n = 3). **P < 0.01 as determined by Student's t-test.

Discussion
In budding yeast, the UPR signaling pathway, composed of Ire1 ER transmembrane sensor and Hac1 transcription factor, plays a pivotal role in ER stress response 1,2 . We previously demonstrated that the budding yeast ortholog of AMPK, Snf1, acts as a negative regulator of the UPR 19 . Snf1 is also involved in downregulation of the Hog1 MAPK during ER stress response. Similar to mammalian AMPK, Snf1 forms a heterotrimeric complex with two regulatory subunits, β and γ 7 . Budding yeast expresses three β subunits, and one of them is incorporated into each Snf1 complex 7 . However, it has remained unclear which β subunit functions in ER stress response mediated by Snf1. Here, we revealed that Gal83 makes the greatest contribution to the regulation of UPR and Hog1 among three β subunits. Consistently, loss of Gal83 caused stronger resistance to ER stress than those of Sip1 or Sip2. These indicate that Gal83 is the principal β subunit in ER stress response.  Supplementary Fig. 8. (b) Effects of the hac1 and ire1 mutations on DTT-induced upregulation of GAL83 mRNA. Wild-type (WT) and hac1 and ire1 mutant strains were grown at 25 °C until exponential phase and treated with 4 mM dithiothreitol (DTT) for the indicated time. The mRNA levels were quantified by qRT-PCR analysis, and relative mRNA levels were calculated using ACT1 mRNA. The data show mean ± SEM (n = 3). **P < 0.01 as determined by Tukey's test. (c) Effects of the reg1 mutation on ER stress-induced upregulation of GAL83 mRNA. Wild-type (WT) and reg1 mutant strains were grown at 25 °C until exponential phase and treated with 4 mM dithiothreitol (DTT) for the indicated time. The mRNA levels were quantified by qRT-PCR analysis, and relative mRNA levels were calculated using ACT1 mRNA. The data show mean ± SEM (n = 3). *P < 0.05 and **P < 0.01 as determined by Scientific RepoRts | 7:46713 | DOI: 10.1038/srep46713 Our analyses utilizing a highly sensitive reg1 mutant background showed the relative contribution of three β subunits to the UPR regulation is identical to that to Hog1 regulation. This implies that the UPR and Hog1 pathways are regulated through a similar mechanism involving Snf1. However, the demands of Snf1 activity for the UPR and Hog1 pathways are likely to differ from each other. The decreased HAC1 mRNA splicing caused by reg1 mutation could be suppressed by either sip1 gal83 or sip2 gal83 double mutations at the level similar to the sip1 sip2 gal83 triple mutations. In contrast, sip1 gal83 and sip2 gal83 double mutations lead to weak suppression of decreased Hog1 activity caused by reg1 mutation, compared to the sip1 sip2 gal83 triple mutations. Therefore, it is suggested that Snf1-mediated regulation of the UPR and Hog1 pathways requires the relatively high and low activities of Snf1, respectively.
Previous studies have characterized the various functional differences of the β subunits 7,14-17 . For instance, only Sip2 has been implicated in intrinsic aging 29 . However, there appear to be little functional differences in ER stress response, as upregulation of Sip2 expression level using GAL83 promoter could effectively complement loss of Gal83. Rather, their abundance controlled by the promoter activity may be critical to define their importance in ER stress response. Consistent with previous studies 15 , we observed that Gal83 is the most abundant in normal conditions among three β subunits. Thus, the greatest contribution of Gal83 to Snf1-mediated ER stress response is consistent with the highest expression level of Gal83. However, there may be the difference in activity between Sip1 and Sip2. In wild-type cells, Sip1 was less abundant than Sip2; nevertheless, the relative contribution of Sip1 to ER stress response was greater than that of Sip2. Therefore, it is likely that, in the case of ER stress response, Sip1 has a higher activity per molecule than Sip2. What is regulated in ER stress response by the β subunits? The β subunits are believed to function in determination of substrate preferences and subcellular localizations of the Snf1 complex 7,14-17 . However, these might not be the case for ER stress response, as their expression levels seem to be a critical determinant. Therefore, it should be further elucidated how the expression levels of three β subunits control the function of the Snf1 complex in ER stress response.
Previous reports showed that a shift from fermentable to nonfermentable carbon sources upregulates the expression level of Sip2 15 . However, it remains unclear how Sip2 expression is modulated by carbon sources. A previous study using mammalian cells has revealed that the β 1 subunit of AMPK is induced by cold stress and chemotherapeutic drug 30 ; however, induction mechanism remains unclear. Therefore, it has yet to be elucidated how environmental changes alter the expression levels of the β subunits. In this study, we showed that The SIP2 mRNA levels were quantified by qRT-PCR analysis, and relative mRNA levels were calculated using ACT1 mRNA. The data show mean ± SEM (n = 3). *P < 0.05 as determined by Tukey's test. (c) Effects of the P GAL83 -GAL83, P SIP2 -SIP2 and P GAL83 -SIP2 integrations on ER stress sensitivity. Wild-type (WT) and reg1 single and reg1 sip1 sip2 gal83 quadruple mutant strains harboring the integration of P GAL83 -GAL83, P SIP2 -SIP2, or P GAL83 -SIP2 were spotted onto YPD medium lacking or containing 0.5 μ g/ml tunicamycin (TM) and incubated at 25 °C.
transcription of the GAL83 gene is activated rapidly and transiently by ER stress. In yeast ER stress response, several signaling pathways, including the UPR, Mpk1, Hog1, and Snf1, become activated [1][2][3][4][5][6]19 . ER stress induced-activation of Mpk1, Hog1, and Snf1 was maintained long-term. In contrast, the UPR activity was increased rapidly after ER stress treatment and gradually decreased thereafter. In accord with the rapid and transient activation of the UPR, expression of the well-known UPR target genes, such as ERO1 and KAR2, was induced rapidly and transiently following exposure to ER stress 19,31 . Similar expression pattern was seen in the GAL83 gene. Furthermore, induction of GAL83 was impaired in hac1 and ire1 mutant cells. These observations suggest that the GAL83 gene is directly controlled by the UPR during ER stress response. Consistent with our previous finding that Snf1 negatively regulates the UPR 19 , we found here that the expression level of GAL83 was downregulated in Snf1-hypreactivated cells. This suggests that Snf1 negatively regulates itself through transcriptional inhibition of its regulatory subunit Gal83. On the other hand, it is also possible that the UPR negatively regulates itself through potentiating Snf1 function, based on the observation that Gal83 expression was induced in a manner dependent on the UPR. Taken together, the UPR and Snf1 may form a feedback loop to modulate the signal mediating ER stress response (Fig. 6). Since failure of the UPR to be downregulated properly results in hypersensitivity to ER stress 31,32 , it may be anticipated that the defect in Snf1-mediated feedback inhibition of the UPR causes ER stress sensitive phenotype. However, cells deleted for components of the Snf1 complex did in fact display resistance to ER stress. Why does loss of Snf1 function cause ER stress resistance phenotype? We have previously shown that in snf1 mutant cells, the basal activity of the UPR is increased compared with wild-type cells, but attenuation of the UPR activity is apparently normal 19 . This observation suggests that upregulated, but controllable, UPR activity possibly contributes to ER stress resistant phenotype observed in the snf1 mutants, and further indicates that Snf1 plays an auxiliary role in downregulation of the UPR. Previous studies also revealed that the changes of Ire1 phosphorylation state lead to attenuation of the UPR 31,32 . Therefore, it is possible that Snf1 participates in modulating the phosphorylation state of Ire1. Thus, further analyses should be needed to reveal the physiological importance of the feedback regulation between the UPR and Snf1, and will provide valuable insights into the mechanism to finely tune the UPR during ER stress response.

Materials and Methods
Strains. Strains used in this study are listed in Table 1 and Supplementary Table 1. Yeast strains harboring the complete gene deletions and carboxyl-terminally GFP-tagged genes were generated by a PCR-based method as described previously 33 . BY4741 and its mutant derivatives, reg1, hac1, and ire1, were obtained from Open Biosystems. All other strains were constructed by a PCR-based method and verified by PCR to confirm that replacement had occurred at the expected locus. Standard procedures were followed for yeast manipulations 34 . Plasmids. Plasmids used in this study are described in Table 2. In-Fusion cloning kits (Takara) was used to construct plasmids. The P GAL83 -GFP and P SIP2 -GFP were constructed as follows. The DNA fragment encoding GFP followed by the ADH1 terminator (T ADH1 ) was obtained by PCR using the pFA6a-GFP vector 33 as a template. The GFP-T ADH1 DNA fragment was fused to 698-bp and 690-bp genomic fragments containing 5′ upstream sequences of the GAL83 and SIP2 genes, respectively, yielding the P GAL83 -GFP and P SIP2 -GFP plasmids. The P GAL83 -SIP2 was constructed as follows. The coding region of the SIP2 gene together with a 537-bp 3′ downstream sequence was amplified by PCR using genomic DNA as a template. The SIP2 DNA fragment was fused to a 698-bp genomic fragment containing 5′ upstream sequences of the GAL83 gene, yielding the P GAL83 -SIP2 plasmid. Schemes detailing construction of plasmids and primer sequences are available on request.
Stress sensitivity. Assays for tunicamycin toxicity were carried out as follows. Cells were grown to exponential phase, and cultures were adjusted to an optical density of 0.5. Cell cultures were then serially diluted 5-fold, spotted onto normal plates or plates containing the indicated concentrations of tunicamycin, followed by incubation at 25 °C for 3 days (for plates lacking tunicamycin) or more than 5 days (for plates containing tunicamycin).