|Home | About | Journals | Submit | Contact Us | Français|
Lipid sensing mechanisms at the endoplasmic reticulum (ER) coordinate an array of biosynthetic pathways. A major phospholipid regulatory circuit in yeast is controlled by Scs2p, an ER membrane protein that binds the transcriptional repressor protein Opi1p. Cells grown in the absence of inositol sequester Scs2p–Opi1p at the ER and derepress target genes including INO1. We recently reported that Yet1p and Yet3p, the yeast homologues of BAP29 and BAP31, are required for normal growth in the absence of inositol. Here we show that the Yet1p–Yet3p complex acts in derepression of INO1 through physical association with Scs2p–Opi1p. Yet complex binding to Scs2p–Opi1p was enhanced by inositol starvation, although the interaction between Scs2p and Opi1p was not influenced by YET1 or YET3 deletion. Interestingly, live-cell imaging analysis indicated that Opi1p does not efficiently relocalize to the ER during inositol starvation in yet3Δ cells. Together our data demonstrate that a physical association between the Yet complex and Scs2p–Opi1p is required for proper localization of the Opi1p repressor to ER membranes and subsequent INO1 derepression.
The endoplasmic reticulum (ER) houses numerous enzymes critical for the biosynthesis of secretory proteins and lipids. In addition to these enzymatic pathways, the ER also contains proteins that monitor and regulate biosynthetic processes through the alteration of transcriptional programs. Several elegant mechanisms have been defined, explaining how cells sense and adjust transcriptional profiles to maintain protein and lipid homeostasis (Ire1p/Hac1p, Cox et al., 1993 ; SCAP-SREBP, Wang et al., 1994 ; Spt23p, Hoppe et al., 2000 ; and Scs2p/Opi1p, Loewen et al., 2004 ). In general, ER-localized sensor proteins alter their conformational or oligomeric status in response to changes in protein or lipid composition, which subsequently influences the activation state of associated transcription factors.
In the yeast Saccharomyces cerevisiae, the production of enzymes responsible for phospholipid biosynthesis depends on the interplay between the transcriptional activator Ino2p-Ino4p and its cognate repressor Opi1p (reviewed by Carman and Henry, 2007 ). When inositol is supplied in the culture medium, phospholipid biosynthesis genes containing an inositol-sensitive upstream activation sequence (UASINO) promoter element are repressed through the association of Opi1p with Ino2p (Wagner et al., 2001 ; Jesch et al., 2005 ). Inositol starvation apparently releases Opi1p from Ino2p, thus allowing expression of UASINO-controlled genes (Henry and Patton-Vogt, 1998 ; Gardenour et al., 2004 ; Heyken et al., 2005 ). Current evidence suggests that Opi1p is physically sequestered on the surface of ER membranes during inositol starvation through interaction with Scs2p and accumulated phosphatidic acid (Loewen et al., 2003 , 2004 ). The conversion of phosphatidic acid to phosphatidyl inositol increases with the availability of inositol, which is thought to release Opi1p from ER membranes followed by translocation into the nucleus to repress UASINO genes.
The deletion of SCS2 results in attenuated growth in the absence of inositol, presumably by preventing ER sequestration of Opi1p (Kagiwada et al., 1998 ; Loewen et al., 2003 ). INO1, which encodes the rate-limiting inositol 1-phosphate synthase enzyme, is the major target of Ino2p-Ino4p activation and Opi1p repression. Highlighting the central role of INO1 in the response to inositol starvation, overexpression of INO1 is sufficient to rescue the inositol starvation growth defect of a scs2Δ mutant (Kagiwada et al., 1998 ). An ire1Δ mutant, which also displays inositol starvation growth defects similar to a scs2Δ mutant, can similarly be rescued by overexpression INO1 (Cox et al., 1997 ), suggesting that an inability to properly derepress INO1 is a major reason for inositol auxotrophy in these mutants (Cox et al., 1997 ; Kagiwada and Zen, 2003 ).
Previous studies indicated that yet1Δ and yet3Δ mutants are inositol auxotrophs (Hillenmeyer et al., 2008 ; Wilson and Barlowe, 2010 ). YET1 and YET3 encode sequence homologues of the mammalian BAP31 protein (Toikkanen et al., 2006 ), which has reported roles in substrate-specific ER export, retention, and quality control (e.g., Lambert et al., 2001 ; Ladasky et al., 2006 ; Wang et al., 2008 ). Yet1p and Yet3p form a heteromeric complex that is important for interaction with the ER translocation machinery and for inositol prototrophy (Wilson and Barlowe, 2010 ). In this study, we present evidence showing that the inositol-related growth defects in yet1Δ and yet3Δ mutants are caused by an inability to fully derepress INO1 during inositol starvation. Interestingly, the Yet1p–Yet3p complex interacts with Scs2p and Opi1p in a manner that is regulated by inositol availability and requires both Yet1p and Yet3p. Together our data demonstrate that the Yet1p–Yet3p complex facilitates Scs2p-mediated sequestration of Opi1p.
Yeast cell survival in the absence of inositol requires the expression of inositol 1-phosphate synthase, Ino1p (Culbertson et al., 1976 ; Donahue and Henry, 1981 ). The removal of genes required for the transcription of INO1 (e.g., ino2Δ), results in strong inhibition of growth in the absence of inositol (Figure 1 and Culbertson and Henry, 1975 ). In contrast, cells harboring mutations in genes involved in the regulated derepression of INO1, such as scs2Δ or ire1Δ, display limited growth when starved of inositol (Figure 1 and Kagiwada et al., 1998 ). Previous studies have indicated that cells deleted for YET1 and/or YET3 are comparably defective for growth in the absence of inositol (Wilson and Barlowe, 2010 ). Thus the findings presented in the following experiments using a yet3Δ mutant should be applicable to a yet1Δ mutant. Figure 1 compares the inositol growth defect of yet3Δ to other inositol starvation growth mutants. As expected, cells lacking INO2 did not grow in the absence of inositol at either 30 or 37°C. Similar to scs2Δ and ire1Δ deletion mutants, cells with a yet3Δ mutation displayed growth attenuation at 30°C, whereas essentially no growth was observed at 37°C. The reason for temperature sensitivity to inositol starvation in these mutants is not entirely clear, although it has been suggested that the cellular inositol requirement may be elevated at higher temperatures (Gaspar et al., 2008 ). Also, higher temperatures have been shown to have an inhibitory effect on INO1 derepression in scs2Δ mutants (Kagiwada and Zen, 2003 ).
The thermosensitive nature of the inositol starvation growth phenotype of the yet1Δ and yet3Δ mutants suggested that, like scs2Δ and ire1Δ mutants, the underlying cause of growth inhibition might be an inability to properly derepress INO1. To test this possibility, we examined levels of Ino1p tagged with the hemagglutinin epitope (Ino1p-HA) in wild-type and yet3Δ mutant cells after various periods of inositol starvation (Figure 2A). We note that prolonged inositol starvation in wild-type strains caused a notable increase in Yet3p protein levels indicative of unfolded protein response (UPR) activation (Cox et al., 1997 ; Wilson and Barlowe, 2010 ). More important, the level of Ino1p-HA in yet3Δ mutant cells was diminished compared with wild-type cells at all time points tested, consistent with a failure to properly derepress INO1. Ino1p-HA levels appeared to be reduced to a greater degree after longer periods of inositol starvation, similar to observations of INO1 mRNA levels in an ire1Δ mutant (Cox et al., 1997 ). To confirm that the observed reduction in Ino1p-HA levels was the result of a failure of the yet3Δ mutant cells to fully derepress INO1, we transformed wild-type, yet3Δ, and scs2Δ mutants with an INO1-LacZ fusion reporter plasmid (pMR1036, Chang et al., 2002 ) and measured reporter expression after inositol starvation. As shown in Figure 2B, both yet3Δ and scs2Δ mutant cells displayed significantly reduced levels of reporter activity compared with wild-type cells after 5 h without inositol supplement. These data demonstrate an INO1 derepression defect in yet3Δ mutant cells, similar to previous studies on ire1Δ, hac1Δ, and scs2Δ mutants (Nikawa, 1994 ; Cox et al., 1997 ; Kagiwada and Zen, 2003 ; Brickner and Walter, 2004 ).
The inositol starvation–related growth defects of INO1 derepression mutants (ire1Δ and scs2Δ) can be suppressed by overexpression of INO1 (Cox et al., 1997 ; Kagiwada et al., 1998 ). Additionally, the inositol auxotrophy of a hac1Δ mutant was reported to be suppressed by a multicopy plasmid bearing SCS2 (Nikawa et al., 1995 ). Thus we tested whether overexpression of these genes could suppress the inositol growth defect of a yet3Δ mutant. Strikingly, a 2μ INO1 plasmid suppressed the inositol growth defect of yet3Δ mutant cells as efficiently as this plasmid suppressed scs2Δ mutant cells at both 30 and 37°C (Figure 3A). This finding provides further evidence that the inositol auxotrophy in yet3Δ mutants is the result of an inability to properly derepress INO1. Notably, yet3Δ and scs2Δ mutant cells were able to activate an UPR during inositol starvation similar in magnitude to wild-type cells (unpublished data), suggesting that unlike Ire1p, Yet3p and Scs2p are not involved in UPR signaling. Interestingly, a 2μ SCS2 plasmid had no detectable influence on the growth of yet3Δ cells at 30°C but clearly enhanced growth at 37°C. Multicopy SCS2 similarly suppressed yet1Δ and the double yet1Δ yet3Δ mutant strains (unpublished data), suggesting that elevated levels of Scs2p partially bypass YET1 and YET3 function during inositol starvation. Reciprocal experiments with multicopy YET1 and/or YET3 showed no detectable effect on the inositol growth defect of scs2Δ or ire1Δ mutants (unpublished data). These suppression results also correlated with readouts from the INO1-LacZ reporter plasmid in the presence of inositol where multicopy YET1/YET3 produced mild increases in reporter activity relative to induction by multicopy Scs2p (Supplemental Figure 1).
To further explore the genetic relationship between YET3 and SCS2, we next examined the growth of a yet3Δ scs2Δ double mutant on medium lacking inositol. Even though the yet3Δ scs2Δ mutant displayed no apparent growth defect in the presence of inositol, essentially no growth was observed in the absence of inositol (Figure 3B). This result suggests that the Yet proteins and Scs2p can partially function (with respect to INO1 derepression) in the absence each other, but removal of both components results in a severe defect in INO1 derepression.
Opi1p sequestration on ER membranes during inositol starvation is associated with INO1 derepression (Loewen et al., 2004 ). Additionally, Opi1p has been reported to remain partially bound to INO1-associated chromatin during inositol starvation in hac1Δ and scs2Δ mutants, which have defects in INO1 derepression (Brickner and Walter, 2004 ). Given our evidence that INO1 was not properly derepressed in a yet3Δ mutant, we next asked whether green fluorescent protein fused to Opi1p (GFP-Opi1p) as previously described (Loewen et al., 2003 ) was localized to the ER during inositol starvation in this mutant. As shown in Figure 4, GFP-Opi1p localized to the nuclear rim in wild-type cells when inositol was present in the medium, consistent with previous reports (Loewen et al., 2003 , 2004 ). A similar GFP-Opi1p localization pattern was observed in yet3Δ mutant cells grown with inositol supplement. In striking contrast, ER localization of GFP-Opi1p was not observed in yet3Δ mutant cells after inositol starvation as was observed in wild-type cells. Instead the majority of GFP-Opi1p appeared to remain in the nucleus of inositol-starved yet3Δ mutant cells (Figure 4) as also observed when Opi1p was tagged with the c-Myc epitope (Opi1p-MYC) in Supplemental Figure 2, thus correlating deficient INO1 expression with a failure to properly sequester Opi1p in this mutant.
Scs2p function in the derepression of INO1 appears to be through binding and sequestration of the Opi1p repressor, as highlighted by aberrant Opi1p occupancy at INO1-associated chromatin and the lack of INO1 derepression during inositol starvation in scs2Δ cells (Kagiwada and Zen, 2003 ; Loewen et al., 2003 ; Brickner and Walter, 2004 ). Given the phenotypic similarities between the yet1Δ, yet3Δ, and scs2Δ mutants reported earlier in addition to the ER localization of Yet1p, Yet3p, and Scs2p (Kagiwada et al., 1998 ; Toikkanen et al., 2006 ; Wilson and Barlowe, 2010 ), we next examined whether the Yet proteins physically interact with Scs2p and Opi1p. To test for associations, we constructed a strain that expresses chromosomally tagged Scs2p-HA and Opi1p-MYC and performed immunoprecipitation experiments from digitonin solubilized semi-intact cell membranes. As shown in Figure 5A, immunoprecipitation of Scs2p-HA recovered Opi1p-MYC, as expected, but also coimmunoprecipitated Yet1p and Yet3p. Similarly, Opi1p-MYC immunoprecipitations yielded Scs2p-HA and both Yet1p and Yet3p. Interestingly, Sec61p, which we previously showed to be associated with Yet1p and Yet3p (Wilson and Barlowe, 2010 ), was absent from these immunoprecipitations, suggesting that the associations of Yet1p and Yet3p with the Sec complex and with Scs2p–Opi1p are mutually exclusive.
To further characterize the association between the Yet proteins and Scs2p–Opi1p, we incubated ER microsomes (Wuestehube and Schekman, 1992 ) with the cross-linking reagent dithiobis[succinimidylpropionate] (DSP) and then performed denaturing to recover Scs2p-HA or Opi1p-MYC (Figure 5B). Both Yet1p and Yet3p were associated with Scs2p-HA and Opi1p-MYC in a cross-linker–dependent manner, consistent with our native immunoprecipitation results. We note that Scs2p-HA was not efficiently cross-linked to Opi1p-MYC under these conditions, presumably because primary amino groups in these proteins were not positioned to generate a cross-linked product. Given that we did not detect the Scs2p–Opi1p association using DSP, these results provide evidence that the Yet complex is proximally associated with both Scs2p and Opi1p. However, these results do not exclude the possibility that additional components are required for association of Yet1p–Yet3p with Scs2p–Opi1p.
We next examined the influence of inositol starvation on the interactions between Scs2p–Opi1p and the Yet proteins. For this experiment, cells were grown to log phase in the presence of inositol, washed, and then cultured for 3 h in the presence or absence of inositol. As shown in Figure 5C, immunoprecipitation of Scs2p-HA or Opi1p-MYC revealed that inositol starvation increased the amount of Yet1p and Yet3p coassociated with Scs2p-HA and Opi1p-MYC, suggesting that association of the Yet1p–Yet3p complex with Scs2p and Opi1p was stimulated by this condition. Notably, an increase in Scs2p–Opi1p association was not detected in these experiments as predicted from the model of INO1 derepression (Loewen et al., 2004 ). However, control experiments to assess protein association after detergent solubilization indicated that a mixture of digitonin-solubilized SCS2-HA opi1Δ semi-intact cells with OPI1-MYC scs2Δ semi-intact cells produced postsolubilization complexes in which Opi1p-MYC was efficiently coimmunoprecipitated with Scs2p-HA (Supplemental Figure 3). In contrast, a similar mixing experiment with SCS2-HA yet3Δ and scs2Δ YET3 digitonin-solubilized semi-intact cells showed that postsolubilization association between Scs2p-HA and Yet3p was negligible (Supplemental Figure 3).
Scs2p-HA membrane association and ER localization in yet3Δ mutant cells were indistinguishable from that of wild-type cells (unpublished data). In addition, the Scs2p–Opi1p interaction does not appear to depend on YET1 (Figure 6A) or YET3 (unpublished data), at least under conditions where membranes have been detergent solubilized. In contrast, Yet3p coassociation with Scs2p-HA or Opi1p-MYC was disrupted in yet1Δ mutant cells (Figure 6A). Similar results for Yet1p were obtained from Scs2p-HA immunoprecipitation experiments performed in a yet3Δ strain (unpublished data). These data suggest that, whereas the Yet complex does not influence the level of Scs2p–Opi1p coassociation, both members of the Yet complex are required for interaction with Scs2p–Opi1p. Furthermore these results correlate inositol prototrophy with an association between the Yet complex and Scs2p–Opi1p.
The protein domains required for the Scs2p–Opi1p interaction are well documented (Loewen et al., 2003 ; Kaiser et al., 2005 ; Loewen and Levine, 2005 ). Scs2p belongs to the conserved VAMP-associated protein (VAP) family in which the major sperm protein domain (N-terminal) in the cytoplasmic region of Scs2p interacts with the double phenylalanine in an acidic tract (FFAT) motif–containing domain in Opi1p (Kaiser et al., 2005 ; Loewen and Levine, 2005 ). The observation that both Scs2p-HA and Opi1p-MYC coimmunoprecipitate Yet1p and Yet3p suggests that the Yet proteins interact with a complex of Scs2p and Opi1p. To test the individual importance of Scs2p and Opi1p for Yet complex association, we immunoprecipitated Scs2p-HA from opi1Δ cells and Opi1p-MYC from scs2Δ cells and then examined the levels of associated Yet1p and Yet3p (Figure 6B). Interestingly, deletion of either OPI1 or SCS2 decreased the level of associated Yet1p and Yet3p with the remaining Scs2p or Opi1p partner. Taken together with the cross-linking data shown in Figure 5B, these results support a model in which the Yet complex most efficiently associates with a complex of Scs2p and Opi1p. We noted that the amount of Yet complex associated with Opi1p-MYC in the scs2Δ mutant was higher than the amount associated with Scs2p-HA in the opi1Δ mutant. We reasoned that the SCS2 homologue, SCS22 (Loewen and Levine, 2005 ), might contribute to Yet complex association with Opi1p-MYC in the absence of Scs2p. However, immunoprecipitation of Opi1p-MYC in the scs2Δ scs22Δ double mutant background did not reduce the amount of Yet complex coassociated with Opi1p-MYC compared with the single scs2Δ mutant (Supplemental Figure 4). These observations indicate that the Yet complex can inefficiently associate with Opi1p in the absence of both Scs2p and Scs22p.
In experiments with yet1Δ yet3Δ mutants, we frequently observed a small number of colonies that grew up on plates lacking inositol at 37°C (unpublished data). The occurrence of such suppressor colonies was not specific to the yet1Δ yet3Δ strain or to the strain background used (BY4742), because they also appeared in a yet3Δ mutant and in a yet1Δ mutant in the FY833 background. Similar suppressor colonies were observed in scs2Δ and ire1Δ mutants at a comparable frequency (roughly 1:5000 colony-forming units after 1 wk in the absence of inositol at 37°C), suggesting that this phenomenon was not specific to cells lacking YET1 and/or YET3. Importantly, comparable colony formation was not observed in an ino1Δ mutant, signifying a dependence on INO1. One such suppressed strain was isolated in a yet3Δ background (referred to as yet3Δsup) and growth was compared with the wild-type and parental yet3Δ strain (Figure 7A). Notably, the growth of yet3Δsup was more robust than that of the wild type in the absence of inositol at 37°C, whereas the parental yet3Δ strain showed essentially no growth under this condition. A constitutive loss of INO1 repression that is characteristic of overproduction of inositol mutants (Opi−) could explain this phenotype (Greenberg et al., 1982a ; 1982b ). To examine the regulation of INO1 in the yet3Δsup isolate, the expression of Ino1p-HA was monitored in the presence and absence of inositol. Indeed, as shown in Figure 7B, the yet3Δsup isolate expressed elevated levels of Ino1p-HA in both inositol-replete and -starvation conditions, demonstrating the predominant characteristic of an Opi− mutant.
To further characterize this yet3Δsup isolate, we crossed it to the parental yet3Δ mutant strain and found that the resulting diploid was sensitive to inositol starvation (Figure 7C). Tetrad dissection of this heterozygous diploid showed that the Opi− trait segregated 2:2. Together these observations provide evidence that the Opi− trait was most likely a recessive mutation. Growth of the spores from one such tetrad in the presence or absence of inositol is shown in Figure 7C. The most likely cause of the Opi− phenotype was a loss of function mutation in OPI1. To test this hypothesis, we sequenced the promoter and open reading frame of OPI1 in the yet3Δsup strain and the parental yet3Δ strain and found that yet3Δsup, but not the parent, contained a nonsense mutation (cytosine to thymine at position +871) in the glutamine repeat region of the open reading frame. Interestingly, premature stops in this region have previously been found to result in the Opi− phenotype (White et al., 1991 ). Thus the suppression of the inositol growth phenotype observed in the yet3Δsup strain indicates that, like scs2Δ and ire1Δ mutants (Cox et al., 1997 ; Brickner and Walter, 2004 ), elimination of functional Opi1p is sufficient to rescue the inositol-starvation growth defect of the yet3Δ mutant.
In this study, we present evidence demonstrating that Yet1p and Yet3p are required for derepression of INO1. Cells harboring yet1Δ and yet3Δ mutations displayed comparable inositol-starvation growth phenotypes (Wilson and Barlowe, 2010 ) that were similar in magnitude to that of the known inositol derepression mutants scs2Δ and ire1Δ (Cox et al., 1997 ; Kagiwada and Zen, 2003 ). INO1 expression during inositol starvation was decreased in a yet3Δ mutant, as assessed by Ino1p-HA immunoblot and by an INO1-LacZ transcriptional reporter assay. Consistent with a defect in INO1 derepression, we found that a significant amount of Opi1p remained in the nucleus during inositol starvation in a yet3Δ mutant. Furthermore the Yet1p–Yet3p complex interacted with Scs2p and Opi1p in manner that was stimulated by inositol starvation. Disruption of the Yet1p–Yet3p complex by yet1Δ mutation prevented Yet3p from interacting with Scs2p and Opi1p, connecting the inositol auxotrophy of a yet1Δ mutant to the loss of Yet complex association with Scs2p–Opi1p (similar results were obtained for Yet1p in a yet3Δ mutant, unpublished data). Together with our previous findings (Wilson and Barlowe, 2010 ), these results indicate that the Yet1p–Yet3p complex directly regulates Scs2p–Opi1p-mediated INO1 derepression.
We note that, whereas 2μ overexpression of SCS2 partially rescued yet3Δ (Figure 3A), similar overexpression of YET1 and/or YET3 had no detectable suppressive effect on an scs2Δ mutant (unpublished data). These results are consistent with a model in which the Yet complex confers inositol prototrophy through Scs2p. In other words, excess Yet complex provides no benefit in the absence of SCS2. It seems paradoxical then that an scs2Δ mutant displays a more modest growth phenotype than a yet3Δ mutant at 30°C and that the yet3Δ scs2Δ double mutant exhibits a clear synthetic defect in the absence of inositol. However, yeast contains a homologue of SCS2, known as SCS22, that has been reported to play a minor role in phospholipid metabolism because an scs2Δ scs22Δ double mutant showed a synthetic growth defect in the absence of inositol (Loewen and Levine, 2005 ). Thus one explanation for this paradox is that although Scs22p does not notably influence the association of the Yet complex with Opi1p in the absence of Scs2p (Supplemental Figure 4), it may possess a basal level of derepression activity that requires the Yet complex but is not significantly enhanced by its overexpression.
The observation that Yet1p–Yet3p interactions with Scs2p and Opi1p were increased by inositol starvation suggests that this association facilitates Scs2p-mediated INO1 derepression. It does not seem to be simply a matter of modulating the association between Scs2p and Opi1p because the yet1Δ mutation did not detectably affect this interaction, at least under the detergent solubilization conditions we tested. However, the observation that Opi1p was not correctly localized to the ER in a yet3Δ mutant during inositol starvation suggests a role for the Yet1p–Yet3p complex in this process. Our favored model is that the Yet proteins bind to and sequester Scs2p–Opi1p complexes at ER membranes during inositol starvation, thus preventing diffusion and transport to the inner nuclear membrane (INM) and subsequent Opi1p-mediated repression of UASINO genes. Previous studies have shown that, in the absence of inositol, Opi1p binds to Scs2p and an increased pool of ER-localized phosphatidic acid to remain outside the nucleus (Loewen et al., 2004 ). While this study was under review, two genome-wide screens of the yeast deletion collection identified >200 genes that influence growth rates in the absence of inositol, including yet1Δ and yet3Δ (Villa-Garcia et al., 2010 ; Young et al., 2010 ). Interestingly, several of these gene deletions cause reductions in intracellular pH, which decreases binding of Opi1p to phosphatidic acid and prevents efficient sequestration of Opi1p (Young et al., 2010 ). The Yet1p–Yet3p complex does not appear to influence intracellular pH but may act to further stabilize the association between Opi1p-Scs2p and the pool of ER-localized phosphatidic acid to ensure stringent derepression during inositol starvation.
Alternatively, the Yet complex could play a more active role in the movement of Scs2p–Opi1p. The mechanism by which Opi1p exits the nucleus upon inositol starvation has not been described. ER localization of Opi1p does not depend on the major nuclear export receptors (Loewen et al., 2004 ), and it has been suggested that Opi1p binds Scs2p on the INM (Brickner and Walter, 2004 ). One possibility is that Opi1p exits the nucleus in complex with Scs2p. If this is the case, Yet1p–Yet3p may somehow facilitate the INM to ER relocalization of Scs2p–Opi1p during inositol starvation. These models are consistent with observations in mammalian systems showing that the Yet protein homologue BAP31 influences localization of specific substrate molecules (e.g., Annaert et al., 1997 ; Paquet et al., 2004 ; Szczesna-Skorupa and Kemper, 2006 ). A final possibility is that the Yet1p–Yet3p complex facilitates Scs2p-mediated nuclear membrane recruitment of INO1, a requisite part of its activation (Brickner and Walter, 2004 ; Brickner et al., 2007 ). Remarkably, artificial recruitment of the INO1 locus to the nuclear membrane bypassed the SCS2 requirement in the inositol-starvation response (Brickner and Walter, 2004 ). In contrast, this strategy was not sufficient to overcome the inositol requirement of a hac1Δ strain.
The Yet complex is required for Scs2p–Opi1p function although the Yet1p and Yet3p proteins appear to possess distinct functional properties. A similar relationship was reported for the mammalian homologues BAP29 and BAP31, which perform overlapping but nonidentical functions (Ladasky et al., 2006 ; Abe et al., 2009 ). We observed that inositol starvation has a more significant effect on Yet1p association with Scs2p and Opi1p (clearly apparent for Opi1p-MYC coimmunoprecipitation; Figure 5C) when compared with Yet3p association with Scs2p and Opi1p. Moreover, 2μ overexpression of YET1 in the presence of inositol elevated basal INO1 promoter activity, whereas 2μ overexpression of YET3 reduced this promoter activity (Supplemental Figure 1). Taken together, these results suggest that increasing the relative amount of Yet1p in association with Yet3p positively influences the ability of the Yet complex to act in INO1 derepression.
We previously reported genetic and physical interactions between the Yet1p–Yet3p complex and the Sec translocation complex (Wilson and Barlowe, 2010 ). More specifically, yet1Δ and yet3Δ mutants displayed synthetic growth defects in the absence of inositol when combined with sec63–1, sec61–2, or sec71Δ mutations. To test whether these genetic relationships are shared with other mutations in the INO1 derepression pathway, we examined the growth of scs2Δ sec63–1 and scs2Δ sec71Δ double mutant strains (Supplemental Figure 5). Interestingly, both scs2Δ sec63–1 and scs2Δ sec71Δ mutant strains showed synthetic growth defects in the absence of inositol. These findings indicate some type of a connection between Scs2p-mediated INO1 derepression and the ER translocation apparatus. We also considered the possibility that a general reduction in phosphatidyl inositol (PI) levels was not well tolerated by translocation-defective cells because PI levels are known to be diminished when cells are starved for inositol (Chang et al., 2002 ) and are presumably reduced further in yetΔ and scs2Δ mutants. However, this possibility seems unlikely because the ire1Δ mutation, which also attenuates INO1 derepression (Cox et al., 1997 ), did not display synthetic growth defects in the absence of inositol when combined with sec71Δ (Supplemental Figure 5). These results support a specific genetic connection between YET1, YET3, and SCS2 with components of the ER translocation apparatus.
In this report, we demonstrate a direct role for the yeast BAP31 homologues in regulation of Scs2p–Opi1p-mediated derepression of INO1. Scs2p belongs to a conserved family of ER-localized VAPs, which interact with a variety of intracellular proteins often through binding to a FFAT motif as found in Opi1p (Lev et al., 2008 ). Many FFAT motif–containing proteins, including oxysterol binding proteins and ceramide transport proteins, functionally interact with VAPs to regulate lipid synthesis and transport in animal cells (Wyles et al., 2002 ; Kawano et al., 2006 ). Interestingly, a mutation in one of the human VAP genes produces late-onset familial motor neuron disease (Nishimura et al., 2004 ). On the basis of sequence conservation and preserved ER localization for the BAP31 and VAP families, we speculate that Yet1p–Yet3p regulation of Scs2p–Opi1p function in yeast represents a conserved regulatory module that controls lipid synthesis and/or transport in other eukaryotic species. Further mechanistic dissection of this pathway in yeast should contribute to our general understanding of BAP31- and VAP-controlled processes with potential connections to human disease.
Yeast strains used in this study are listed in Table 1. All C-terminal epitope tagging was achieved using described methods (Longtine et al., 1998 ), and all yeast transformations were performed using the lithium acetate technique (Ito et al., 1983 ). Deletion mutants containing the natMX4 cassette were created with p4339 (Tong et al., 2001 ). Yeast cells were grown at 30°C in 1% yeast extract, 1% peptone, and 2% dextrose (YPD) medium unless otherwise noted. For plasmid selection, yeast cells were grown in 0.67% yeast nitrogen base without amino acids, 2% dextrose, and requisite supplements (YMD). For inositol-starvation growth assays, strains were grown overnight in YMD. After washing with sterile water, strains were plated on YMD with or without 75 μM inositol and grown at indicated temperatures. For experiments shown in Figures 2, ,4,4, ,5C,5C, and and7B,7B, starter cultures were grown in 0.67% yeast nitrogen base (without inositol) and complete supplement (MP Biomedicals, Solon, OH), 2% dextrose and 75 μM inositol (CSMD). After washes with sterile water, cells were resuspended and grown for the indicated time periods in either CSMD without inositol or CSMD with 75 μM inositol. All defined growth media lacked choline.
Genomic DNA from BY4742 was used as the template for the construction of plasmids. For SCS2/pRS426, SCS2 including 479 nucleotides upstream of the open reading frame (ORF) and 298 nucleotides downstream of the ORF were cloned into BamHI and EcoRI sites in the polylinker of pRS426 (Christianson et al., 1992 ). For INO1/pRS426, INO1 including 504 nucleotides upstream of the ORF and 337 nucleotides downstream of the ORF were cloned into BamHI and EcoRI sites in the polylinker of pRS426; pMR1036 (Chang et al., 2002 ) and pTL211 (Loewen et al., 2003 ) have been described.
Antiserum directed against Yet1p (Toikkanen et al., 2006 ), Yet3p (Wilson and Barlowe, 2010 ), Sec13p (Salama et al., 1993 ), Kar2p (Brodsky et al., 1993 ), and Sec61p (Stirling et al., 1992 ) have been described. The sheep anti–mouse and donkey anti–rabbit secondary horseradish peroxidase–linked antibodies were from GE Healthcare (Piscataway, NJ), and monoclonal antibodies against MYC (9E10) and HA (HA.11) were from Covance (Princeton, NJ).
Lysates (Figures 2A and and7B)7B) were derived from ~0.5–1.0 OD600 equivalents of yeast cells and prepared using the bead-beat method in JR lysis buffer (20 mM HEPES pH 7.4, 0.1 M sorbitol, 50 mM potassium acetate, 2 mM EDTA, 1 mM phenylmethylsulfonyl fluoride). Lysates were then centrifuged for three minutes at 1000 × g at 4°C to create a low-speed supernatant. This low-speed supernatant was mixed 2:1 with 5× reducing sample buffer, heated for 6 min at 75°C, and resolved by SDS–PAGE. After transfer to nitrocellulose membranes and antibody incubations, immunoblots were developed with SuperSignal West Pico chemiluminescent substrate (Pierce Chemical, Rockford, IL) and visualized by charge-coupled device (CCD) camera (UVP BioImaging, Upland, CA). Native immunoprecipitation experiments were performed with digitonin-solubilized semi-intact cell membranes as detailed previously (Wilson and Barlowe, 2010 ). For cross-linking experiments, 0.5 mM DSP (Pierce Chemical), or an equivalent volume of dimethyl sulfoxide (vehicle) was incubated for 15 min at 20°C with ~0.6 A280 units of microsomal membranes (Wuestehube and Schekman, 1992 ) prepared from log phase cells grown for 3 h in CSMD without inositol prior to harvest. After quenching with glycine, SDS was added to 1% and samples were heated to 75°C for 3 min. Samples were clarified by centrifugation and 100 μl of the supernatant fluid was diluted 10-fold with 15 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1% Triton X-100, and 2 mM sodium azide and used in immunoprecipitations.
Cells transformed with pMR1036 were grown to log phase in CSMD-leucine with 75 μM inositol, washed in sterile water, resuspended in CSMD-leucine without 75 μM inositol, and grown for 5 h. β-Galactosidase assays were performed according to the manufacturer’s instructions on β-galactosidase assays in yeast (Thermo-Fischer Scientific, Rockford, IL).
Stationary cultures grown in CSMD with 75 μM inositol were washed twice with CSMD without inositol and then diluted into either CSMD with 75 μM inositol or CSMD without inositol and grown for 14 h to an OD600 of 0.25–0.5. At this point, images were acquired with a Hamamatsu Orca R2 cooled CCD camera mounted on an Eclipse Ti Nikon microscope. Optical sections (0.2 μm) in the z axis were collected in GFP and differential interference contrast (DIC) channels with a 100× 1.4 NA objective. Iterative restoration and contrast enhancement were performed using Volocity software (Perkin Elmer, Waltham, MA).
We thank C. Bentivoglio for technical assistance, S. Henry for pMR1036, T. Levine for pTL211, and M. Makarow for Yet1p antiserum. We also thank D. Compton and W. Wickner for the use of their microscopes. We are grateful to J. Merritt, J. Scarcelli, A. Lorente, C. Hodge, V. Starai, C. Stroupe, W. Wickner, L. Myers, J. Brickner, and C. Cole for helpful advice. This work was supported by NIH grant GM52549.
This article was published online ahead of print in MBoC in Press (http://www.molbiolcell.org/cgi/doi/10.1091/mbc.E10-07-0559) on March 3, 2011.
Contributions of authors: J.D.W. wrote the manuscript, designed experiments, and generated data, except where noted; S.L.T. performed microscopy in Figure 4; C.B. provided advice in planning experiments and interpretation of data and edited the manuscript.