Tiered Assembly of the Yeast Far3-7-8-9-10-11 Complex at the Endoplasmic Reticulum*

Background: The Far3-7-8-9-10-11 complex, part of the yeast striatin-interacting phosphatase and kinase (STRIPAK) complex, mediates target of rapamycin complex 2 (TORC2) signaling. Results: The Far3-7-8-9-10-11 complex follows tiered assembly at the endoplasmic reticulum (ER). Conclusion: ER localization of Far9 is required for optimal function in TORC2 signaling. Significance: Our study provides insights into the organization of the yeast STRIPAK complex. Target of rapamycin signaling is a conserved, essential pathway integrating nutritional cues with cell growth and proliferation. The target of rapamycin kinase exists in two distinct complexes, TORC1 and TORC2. It has been reported that protein phosphatase 2A (PP2A) and the Far3-7-8-9-10-11 complex (Far complex) negatively regulate TORC2 signaling in yeast. The Far complex, originally identified as factors required for pheromone-induced cell cycle arrest, and PP2A form the yeast counterpart of the STRIPAK complex, which was first isolated in mammals. The cellular localization of the Far complex has yet to be fully characterized. Here, we show that the Far complex localizes to the endoplasmic reticulum (ER) by analyzing functional GFP-tagged Far proteins in vivo. We found that Far9 and Far10, two homologous proteins each with a tail-anchor domain, localize to the ER in mutant cells lacking the other Far complex components. Far3, Far7, and Far8 form a subcomplex, which is recruited to the ER by Far9/10. The Far3-7-8- complex in turn recruits Far11 to the ER. Finally, we show that the tail-anchor domain of Far9 is required for its optimal function in TORC2 signaling. Our study reveals tiered assembly of the yeast Far complex at the ER and a function for Far complex's ER localization in TORC2 signaling.

Protein phosphorylation plays important roles in many cellular processes. Protein phosphorylation is catalyzed by specific protein kinases, and protein dephosphorylation is carried out by protein phosphatases. Thus, the phosphorylation state of proteins is finely controlled by the opposing activities of protein kinases and phosphatases. Multiple mechanisms exist to finetune the activity of these protein kinases and phosphatases through the regulation of their expression levels, their activity, cellular localization, and availability to substrates, among others. Recently, a large multiprotein complex known as the striatin-interacting phosphatase and kinase (STRIPAK) 2 complex was found in mammals (Table 1). STRIPAK contains PP2A catalytic and scaffolding subunits, striatins, the striatin-associated protein Mob3, two homologous novel proteins STRIP1 and STRIP2, members of the germinal center kinase III family of Ste20 kinases, and the cerebral cavernous malformation 3 protein (1,2). The STRIPAK assembly maintains mutually exclusive interactions with either the CTTNBP2 (cortactin-binding protein 2) proteins or a second subcomplex consisting of sarcolemmal membrane-associated protein (SLMAP) and two related coiled-coil proteins SIKE and FGFR1OP2. The N-terminal region of CCM3 mediates heterodimerization with Ste20 kinases, and its C-terminal domain interacts with striatin (3)(4)(5)(6)(7), which also interacts with the regulatory and catalytic subunits of PP2A, thus bridging a kinase to a phosphatase. This arrangement likely facilitates the regulation of the activity of protein kinases by PP2A (3,8). Both Striatin 3 and STRIP1 localize to the Golgi, and depletion of either results in similar defects suggesting they perform similar functions in regulation of Golgi morphology and mitosis (3,9). The SLMAP gene has several splice variants, encoding tail-anchored membrane proteins that associate with the sarcolemmal membrane in muscle cells, the endoplasmic reticulum and the mitochondrial membrane in nonmuscle cells, and the centrosome and the outer nuclear envelope (9 -13). SLMAP is required for myoblast fusion, centrosome function, and structural arrangement of the excitation-contraction coupling apparatus in cardiomyocytes (9,12,14,15). Much remains to be determined about the role of the STRIPAK components, the mechanism of their regulation, and the identity of their substrates.
Orthologs of mammalian STRIPAK components have been reported to exist in many eukaryotes ( Table 1). The Drosophila STRIPAK complex has been reported to be involved in Hippo signaling by mediating phosphorylation of the Hippo kinase and the transcriptional activator Yorkie (2). In Neurospora crassa, orthologs of STRIPAK complex components are required for hyphal fusion (16,17). In the ascomycete Sordaria macrospora, the STRIPAK complex is required for sexual development and vegetative hyphal fusion (18,19). In the fission yeast Schizosaccharomyces pombe, components of the STRIPAK complex localize to the mitotic spindle pole body in early mitosis and are required for the establishment of asymmetry of the septation initiation network, a conserved signaling pathway that is required for cytokinesis and mitotic transitions (9,20).
In yeast, Far3, Far7, Far8, Far9, Far10, and Far11 have been reported to form a complex (22,24,25). However, it is unclear how these proteins assemble together to form the final complex, and identification of the cellular component of this complex could potentially provide insights into the mechanism of its function. Cellular localization of subsets of the Far complex components has been reported in three different studies; however, the results were not consistent (23,27,35). A genomewide study on the localization of yeast proteins C-terminally tagged with green fluorescent protein (GFP) found that Far3, Far7, and Far8 localize to the ER (35). In that study, Far9 was shown to be localized in the cytoplasm; Far10 localization was ambiguous, and there were no data on Far11. In another study, Beilharz et al. (27) showed that N-terminal GFP-tagged Far9 and Far10 localize to the ER and clusters within the bounds of the ER, respectively. In the third reported study on the localization of Far proteins with a C-terminal fluorescent tag, Far11 was reported to co-localize with Chc1, a late-Golgi protein, Far3 with Cop1, an early Golgi protein, and Far9 with Sec13, an ER-to-Golgi protein that is located on ER-derived transport vesicles (23). To gain insights into how the Far proteins assemble into a complex and address the inconsistency in their cellular localization, we constructed functional GFP-tagged Far proteins and analyzed their localization in various far mutants. Our data show that all of the Far proteins localize in a tiered fashion at the endoplasmic reticulum and ER localization of Far9 is required for its optimal function in TORC2 signaling.

EXPERIMENTAL PROCEDURES
Strains, Plasmids, Growth Media, and Growth Conditions-Yeast strains and plasmids used in this study are listed in Tables  2 and 3, respectively. Yeast cells were grown in SD (0.67% yeast nitrogen base plus 2% dextrose), YNBcasD (SD medium plus 1% casamino acids), or YPD (1% yeast extract, 2% peptone, 2% dextrose) medium at temperatures as indicated in the text and in the figure legends. When necessary, amino acids, adenine, and/or uracil were added to the growth medium at standard concentrations to cover auxotrophic requirements (36).
Cellular Extract Preparation and Co-immunoprecipitation-Total cellular protein extracts were prepared by disrupting yeast cells in extraction buffer (1.85 N NaOH, 7.5% ␤-mercaptoethanol) followed by precipitation with trichloroacetic acid as described (37). For co-immunoprecipitation experiments, cellular lysates were prepared in IP buffer (50 mM Tris-HCl, pH 7.6, 150 mM NaCl, 0.5% Triton X-100, and protease inhibitors). Cell extracts (ϳ3 mg of proteins) were incubated at 4°C for 1 h with anti-Myc antibody (9E10, Roche Applied Science) or anti-HA antibody (12CA5, Roche Applied Science) as indicated, after which 30 l of a 50% slurry of protein G-Sepharose (Roche Applied Science) was added to each sample, and the samples were further incubated at 4°C for 2 h. Washed immunoprecipitates bound to the Sepharose beads were released by boiling in 1ϫ SDS-PAGE loading buffer. The released immune complexes were analyzed by SDS-PAGE and immunoblotting. Myc-, HA-, and GFP-tagged proteins were probed with anti-Myc antibody 9E10, high affinity anti-HA antibody 3F10 (Roche Applied Science), and anti-GFP antibody B-2 (Santa Cruz Biotechnology, Inc), respectively. Chemiluminescence images of Western blots were captured using the Bio-Rad Chemi-Doc photo documentation system (Bio-Rad) and processed using Bio-Rad Quantity One software.
Fluorescence Microscopy-Fluorescence of GFP-and RFPtagged proteins was analyzed in live cells grown in SD medium    pRS418-FAR7-FAR3-HA, expressing nontagged Far7 from its own promoter and Far3 from its own promoter with a 3ϫHA tag at the C terminus This study Fig. 5C pTP673 pRS418-FAR3-FAR7-HA, expressing nontagged Far3 from its own promoter and Far7 from its own promoter with a 3ϫHA tag at the C terminus This study Fig. 5C pZL2762 pRS416-FAR11-HA, expressing Far11 from its own promoter with a 3ϫHA tag at the C terminus 24 Fig. 6  pTP554 pRS416-MKS1p-GFP-FAR9⌬C, expressing Far9⌬C from the MKS1 promoter with a GFP tag at the N terminus This study Fig. 7, B and D to the mid-logarithmic growth phase by fluorescence microscopy. Cells were concentrated by centrifugation at 5000 ϫ g for 2 min, and fluorescence images were immediately captured using a Nikon Eclipse E800 microscope equipped with an HBO 100 W/2 mercury arc lamp, a Nikon Plan Fluor ϫ100 objective lens, a Photometrics Coolsnap fx CCD camera, and a Nikon B-2E/C filter set (excitation light wavelengths 465-495 nm, emission light wavelengths 515-555 nm, dichromatic mirror cut-on wavelength 505 nm) for GFP images and a Y-2E/C filter set (excitation light wavelengths 540 -580 nm, emission light wavelengths 600 -660 nm, dichromatic mirror cut-on wavelength 595 nm) for RFP images. Digital images were acquired using the Metamorph Imaging Software and processed using ImageJ (National Institutes of Health) and Adobe Photoshop software.
Pheromone Response Halo Assay-Sensitivity to the mating pheromone ␣-factor was assayed by standard plate halo assays as described previously (29,38). Briefly, 2 g of ␣-factor was applied to a sterile 6-mm filter paper disc placed onto a lawn of 1 ϫ 10 6 cells spread on a YNBcasD plate. Halo formation was documented after 3 days of cell growth.
Generation of Spheroplasts, Membrane Isolation, and Extraction-Cells expressing epitope-tagged Far proteins were grown at 30°C overnight to a concentration of A 600 1.5-2.0. Cells were converted into spheroplasts as described by Jazwinski (39). Briefly, cell pellets were prepared and resuspended to a concentration of A 600 5 in reduction buffer (1 M sorbitol, 50 mM Tris-HCl, pH 7.6, 10 mM dithiothreitol (DTT)) and incubated for 10 min at room temperature. Cells were again harvested, and pellets were resuspended to a concentration of A 600 25 in spheroplasting buffer (1 M sorbitol, 50 mM Tris-HCl, pH 7.6, 1 mM DTT). Spheroplasts were prepared by treatment with Zymolyase 20T, rinsed in 1 M sorbitol solution, suspended in lysis buffer (50 mM Tris-HCl, pH 7.6, 200 mM sorbitol, 1 mM EDTA, 1 mM DTT, and protease inhibitors), and homogenized using a glass homogenizer. Cell debris was sedimented two times by centrifugation at 2000 ϫ g for 5 min, and the remaining supernatant was centrifuged at 100,000 ϫ g for 15 min to generate a crude membrane fraction. The resulting membrane pellet was resuspended in lysis buffer. Aliquots of the membrane resuspension were treated with an equal volume of either H 2 O, 200 mM Na 2 CO 3 , or 2% Triton X-100 for 60 min at 4°C with occasional gentle agitation. Membrane pellets were re-isolated by ultracentrifugation at 200,000 ϫ g for 30 min. 10% trichloroacetic acid and 0.015% deoxycholic acid were used to precipitate proteins from supernatant fractions. Equivalent amounts of proteins from the supernatant and pellet fractions were analyzed using Western blotting and enhanced chemiluminescence.

RESULTS
GFP-tagged Far3, Far7, Far8, Far9, Far10, and Far11 Localize on the Endoplasmic Reticulum-Elucidating the cellular localization of the Far complex thus far has not been straightforward. Several studies report inconsistent data with the components of the Far complex localizing to different cellular compartments from one study to the next (23,27,35). Furthermore, none of the studies comprehensively analyzed the intracellular localization of all components of the complex. In all cases, fluorescent tags were fused to each Far protein to localize each component. One explanation for the differential localization of the Far complex components is that some of the fusion proteins may not have been functional as the functionality of the fusions of the previous studies was not indicated. Another factor that could affect the localization of a fusion protein is the placement of the fluorescent tag. For instance, Far9 and Far10 contain a hydrophobic tail-anchor domain; therefore, the addition of a fluorescent protein at the C-terminal end of Far9 and Far10 could affect their cellular localization. Indeed, in the genome-wide cellular localization study on yeast proteins and in the study by Lisa-Santamaria et al. (23), Far9 and Far10 were tagged with a C-terminal fluorescent protein, and in the study by Beilharz et al. (27), Far9 and Far10 carried an N-terminal GFP tag (35). Therefore, we generated N-terminal GFP-tagged Far9 and Far10 fusion constructs and C-terminal GFP-tagged Far3, Far7, Far8, and Far11 constructs on centromeric plasmids and determined their functionality and cellular localization. Expression of GFP-tagged Far3, Far7, Far8, and Far11 fusion proteins was under the control of their respective endogenous promoters. Expression of GFP-tagged Far9 and Far10 was under the control of a stronger but still relatively weak promoter of MKS1 (40) due to our initial observations that N-terminal GFP-tagged Far9 and Far10 under the control of their endogenous promoters did not yield enough signal to determine their cellular localization. To minimize potential interference of GFP with the functionality and thus localization of Far proteins, we also introduced a 10-alanine linker between the Far proteins and the GFP tag.
We first sought to determine the functionality of our GFPtagged Far proteins. A quantitative assay is best suited for determining the function of epitope-tagged proteins. However, expression of FUS1-lacZ, a pheromone-responsive reporter gene, is not affected by the far mutations in our study despite the role of these Far proteins in mediating cell cycle arrest in response to the mating pheromone and thus is not useful in the determination of the functionality of GFP-tagged Far fusion proteins (22). Furthermore, we conducted a quantitative mating assay to assess whether far mutations affect mating efficiency, and we found that they reduced the mating efficiency marginally by ϳ10% (data not shown), which is not sensitive enough to provide quantitative information on the functionality of GFP-tagged Far proteins. Therefore, to assess the functionality of GFP-tagged Far proteins, we used a pheromone response halo assay (Fig. 1A). Wild-type mating type a (MATa) cells normally arrest cell growth around a paper disc infused with ␣-factor and create a cell-free zone in the shape of a halo. We tested far3, far7, far8, far9, far10, and far11 single deletion mutants and found that they became resistant to ␣-factor-induced cell cycle arrest, which was manifested by increased cell growth around the disc containing ␣-factor, consistent with previous results (22). After these far⌬ mutants were transformed with centromeric plasmids encoding respective wildtype FAR genes tagged with GFP, the resultant transformants were as sensitive to ␣-factor as wild-type cells, indicating that GFP-tagged Far3, Far7, Far8, Far9, Far10, and Far11 proteins were all functional.
We next examined the cellular localization of the GFPtagged Far proteins in their respective far deletion mutant strains. Fig. 1B, left column, shows that all exhibited localization patterns suggestive of ER localization. To support this assumption, in cells expressing GFP-tagged Far proteins, we co-expressed the C-terminal RFP-tagged Shr3, an ER-localized chaperone for packaging amino acid permeases into COPII-coated transport vesicles (41). Fig. 1B shows that Shr3-RFP and each of the six Far-GFP fusion proteins co-localize, indicating that the Far complex localizes to the ER. We expressed GFP-tagged Far proteins in respective far deletion mutant strains of two other strain backgrounds, BY4741, which is derived from S288c, and SY2227 (22), and we found that in these strains, GFP-tagged Far proteins also localized to the ER (data not shown), indicating that ER localization of the Far complex is not strain-dependent.
Far9 and Far10 Localize to the ER Independently of the Other Far Proteins-To better understand how the Far complex is organized on the ER, we sought to determine which Far protein(s) establishes a foothold on the ER. To that end, we analyzed the cellular localization of individual GFP-tagged Far proteins in a far3/7/8/9/10/11⌬ sextuple mutant strain. Fig. 2 shows that Far3-GFP, Far7-GFP, Far8-GFP, and Far11-GFP, GFP-FAR9, pTP179; GFP-FAR10, pTP203; FAR11-GFP, pZL2564) were grown on YNBcasD medium in the presence of a paper filter disc containing ␣-factor as described under "Experimental Procedures." B, co-localization of GFP-tagged Far proteins with ER-localized Shr3-RFP. far⌬ mutant cells (far3⌬, TPY157; far7⌬, TPY147; far8⌬, TPY213; far9⌬, TPY357; far10⌬, TPY151; far11⌬, TPY116) co-expressing respective GFP-tagged Far proteins as described for A, and RFP-tagged Shr3 (pTP201) were grown in SD medium and observed by fluorescence microscopy. GFP and RFP fluorescence images were captured and processed using the same parameters for each channel. Vacuolar autofluorescence in the RFP channel was sometimes observed. when expressed individually in the sextuple mutant strain, did not exhibit ER localization, indicating that they were unable to localize to the ER in the absence of other Far complex components. In contrast, GFP-Far9 and GFP-Far10 fusion proteins could still localize to the ER in the sextuple mutant, indicating that Far9 and Far10 establish ER localization for the complex. Far9 and Far10 contain a tail-anchor domain at their C termini. Therefore, it is not surprising that they may function as the ER anchor for the Far complex. Furthermore, the localization of Far9 and Far10 on the ER when expressed individually in the sextuple mutant suggests that these two homologous proteins do not require hetero-oligomerization for ER recruitment.
Tiered Assembly of the Components of the Far Complex at the ER-To further characterize the organization of the Far complex on the ER, we sought to determine the order in which the rest of the complex localizes to the ER. To achieve this, we determined whether ER localization of GFP-tagged Far3, Far7, Far8, or Far11 could be altered by the absence of just one of the other five Far complex components. Accordingly, we characterized the cellular localization of Far3-GFP in far3⌬ far7⌬, far3⌬ far8⌬, far3⌬ far9⌬, far3⌬ far10⌬, and far3⌬ far11⌬ double deletion mutant cells. Far3-GFP localization at the ER was abolished by far7⌬, far8⌬, far9⌬, and far10⌬ mutations but still showed normal ER localization in a far11⌬ mutant, indicating that ER localization of Far3-GFP requires Far7, Far8, Far9, and Far10 but not Far11 (Fig. 3A). Using the same strategy, we determined the cellular localization of Far7-GFP, Far8-GFP, and Far11-GFP in respective double deletion mutant cells. Likewise, ER localization of Far7-GFP and Far8-GFP was abolished by all respective far⌬ mutations except a far11⌬ mutation (Fig. 3, B and C). Interestingly, ER localization of Far11-GFP was disrupted by the deletion of any of the other five Far complex components (Fig. 3D), suggesting that Far11 is the most peripheral component of this complex at the ER. These data also suggest that Far3, Far7, Far8, and Far11 are peripheral membrane proteins because their ER localization requires Far9 and Far10. This possibility was supported by our initial observation early in our studies that the localization of GFP-tagged Far3, Far7, Far8, and Far11 became more cytoplasmic when they were expressed in wild-type cells in comparison with the respective deletion mutant cells, suggesting that these four GFP-tagged Far proteins compete with their nontagged counterparts for Far9/10-dependent ER localization (data not shown). The interdependence of Far3, Far7, and Far8 for ER localization also suggests that they might form a subcomplex before their ER recruitment.
To confirm that Far3, Far7, Far8, and Far11 are peripherally associated with the ER membrane, we conducted membrane extraction experiments using carbonate or Triton X-100. Alkaline carbonate extraction removes peripheral membrane proteins but not integral membrane proteins from membrane fractions. In contrast, Triton X-100 treatment disrupts membranes and releases both peripheral and integral membrane proteins. A crude membrane fraction was isolated and treated with 100 mM carbonate, 1% Triton X-100, or water. Fig. 4 shows that after water treatment, Far3, Far7, Far8, and Far11 fractionated in both the supernatant and the pellet, whereas Far10 fractionated only in the pellet. Furthermore, 100 mM carbonate treatment released most of Far3, Far7, and Far8, all of Far11, yet very little of Far10 into the supernatant fraction. However, 1% Tri- ton X-100 treatment efficiently released Far10 into the supernatant fraction. Surprisingly, for unknown reasons a small portion of Far3, Far7, and Far8 was resistant to Triton X-100 extraction. Together, these data indicate that Far3, Far7, Far8, and Far11 are peripheral membrane proteins, whereas Far10 is an integral membrane protein. Based on the efficiency of carbonate extraction of Far proteins from membranes and the fluorescence microscopy data, the Far complex appears to assemble at the ER in the spatial order of Far9/10, Far3/7/8, and then Far11.
Far8 has been reported to interact with Far3 and Far7 by yeast two-hybrid and co-immunoprecipitation analyses in wild-type strains (22,25). Although we carried out the interaction analysis between Far8 and Far3 or Far7 in sextuple mutant cells, which were not used in previous studies, the failure to detect their interactions in such cells was still surprising because these three proteins appear to require each other for ER localization as shown in Fig. 3. One possibility is that Far8 may only bind to the Far3-Far7 complex. To test this hypothesis, we first confirmed whether Far8 interacts with Far3 or Far7 in respective double deletion mutant cells. We generated a far3⌬ far8⌬ double mutant carrying plasmids encoding FAR3-HA and FAR8-GFP. Fig. 5B shows that Far8-GFP was co-immunoprecipitated specifically with Far3-HA (compare lanes 1 and 2). Similarly, we found that Far8-GFP specifically interacts with Far7-HA in a far7⌬ far8⌬ double mutant (Fig. 5B,  lanes 3 and 4). We next examined whether Far8 interacts with the Far3-Far7 complex in the sextuple mutant. We co-expressed Far3-HA, Far8-GFP, and nontagged Far7 in far3/7/8/ 9/10/11⌬ sextuple mutant cells and found that expression of Far7 in the sextuple mutant was sufficient for Far3-HA to interact with Far8-GFP (Fig. 5C, compare lanes 1 and 2). Similarly, reintroduction of Far3 into sextuple mutant cells co-expressing Far7-HA and Far8-GFP also enabled an interaction between Far7 and Far8 (Fig. 5C, compare lanes 3 and  4). These data together indicate that Far3, Far7, and Far8 form a subcomplex in the absence of the other components of the Far complex.
Interaction between Far9 and Far11 Requires the Far3-7-8-Subcomplex-ER localization of Far11-GFP was disrupted by deletion of any of the other Far complex components as shown in Fig. 3D. These findings along with the findings that Far3, Far7, and Far8 form a subcomplex suggest that the Far3-7-8subcomplex may bridge the interaction of Far11 and Far9/10 at the ER. To test this possibility, interaction between Myc-tagged Far9 and HA-tagged Far11 was analyzed in far9⌬ far11⌬ double mutant cells without (WT) or with an additional mutation of far3⌬, far7⌬, or far8⌬. Far9-myc was immunoprecipitated from cell lysates, and immunoprecipitates were probed with anti-HA antibody to detect Far11-HA via Western blotting. Fig.  6 shows that Far11-HA was co-immunoprecipitated with Far9myc, and deletion of FAR3, FAR7, or FAR8 greatly reduced their interaction. These data indicate that ER recruitment of Far11 by Far9 requires the Far3-7-8-subcomplex.
ER Localization of Far9 Is Required for Its Optimal Function in TORC2 Signaling-Yeast Far9 and Far10 and their human and fly orthologs all contain a tail-anchor domain and a forkhead-associated domain (Fig. 7A). Tail-anchored proteins utilize the tail-anchor domain for membrane association (42). The finding that Far9 and Far10 are able to localize to the ER in the absence of the other Far complex components prompted us to determine the role of Far9's tail-anchor in the ER localization of Far9. Accordingly, we constructed a GFP-tagged C-terminal truncation mutant of Far9, GFP-Far9⌬C, and examined its location in far9⌬ mutant cells. Unlike GFP-tagged wild-type Far9, GFP-Far9⌬C localized diffusely in the cytoplasm (Fig. 7B), indicating that the tail-anchor domain of Far9 is required for its ER localization. Recently, we proposed that the Far complex antagonizes TORC2 signaling by showing that a far9⌬ mutation or loss of other Far complex components are able to bypass a tor2 temperature-sensitive (tor2-21) mutation (24). We sought to test whether ER localization of the Far complex is required for its function. To that end, we introduced a far9⌬C mutation, which results in the synthesis of Far9 without the tail-anchor domain, at the genomic FAR9 locus in tor2-21 mutant cells. We then tested whether the far9⌬C mutation could bypass the tor2-21 mutation and found that far9⌬C was able to partially mimic far9⌬ in suppressing the growth defect of tor2-21 mutant cells grown at 36 and 37°C (Fig. 7C). To exclude the possibility that the removal of the tail-anchor domain of Far9 may reduce the steady-state level of Far9 by reducing its stability, which could explain the partial suppression of the tor2-21 growth defect at high temperatures, we examined the levels of GFP-tagged Far9 and Far9⌬C in tor2-21 far9⌬ cells and found that Far9⌬C was expressed to similar levels as full-length Far9 (Fig. 7D). Together, these data suggest that the suppression of the tor2-21 mutation by far9⌬C results from the loss of ER localization of Far9 and that ER localization of Far9 is required for its optimal function.

DISCUSSION
In this report, we systematically examined the cellular localization of functional GFP-tagged Far3, Far7, Far8, Far9, Far10, and Far11 and found that they all localized to the ER. Our findings are partly consistent with two previous studies (27,35) yet disagree with another (23). Inconsistencies in the localization of the Far proteins may be explained by several possibilities. Cytoplasmic localization of Far9 in the genome-wide study on yeast protein localization is most likely to be due to the tagging with GFP at its C terminus, which is expected to interfere with tailanchor domain-dependent ER membrane insertion of Far9. The clustering effect of Far10 in the study by Beilharz et al. (27) may result from a higher level of overexpression of Far10 from the relatively strong MET25 promoter than in our current study (44,45). Our data disagree with the study by Lisa-Santamaria et al. (23) on the localization of Far3, Far9, and Far11. In that study, the authors added a C-terminal CFP tag to Far9. Furthermore, the functionality of the fusion proteins in the aforementioned study was not reported. If the fusion proteins were not functional, this could present another possibility that accounts for the localization differences.
ER/nuclear envelope localization of proteins associated with the STRIPAK complex is likely to be a conserved feature, which is expected to be mediated by the tail-anchor domain of Far9/ Far10 orthologs in different species (Table 1). Apart from the ER localization of human and S. cerevisiae STRIPAK components, two components of the fission yeast S. pombe STRIPAK complex, Csc2 and Csc3 (S. cerevisiae Far11 and Far8 orthologs, respectively), also associate with the nuclear envelope/ER. Conversely, Csc1, the S. cerevisiae Far9/10 ortholog, was not found to localize to the nuclear envelope (20). However, this could be due to the addition of GFP at its C terminus, which interferes with its tail-anchor domain-dependent ER localization. ER localization of the STRIPAK components appears to be important for their function in human cells (11,14,15). Likewise, we show that, in yeast, ER localization of the Far complex is required for its optimal function in TORC2 signaling. Why is ER localization of the yeast Far complex required for its func- tion in TORC2 signaling? We and others have shown that the yeast Far complex and PP2A negatively regulate TORC2 signaling. Both TORC2 and its substrates localize to the plasma membrane (30,32,33). The association with the Far complex could bring PP2A to the periplasmic region to facilitate dephosphorylation of Slm1 and other TORC2 substrates located at the plasma membrane.
Human SLMAP and the S. pombe STRIPAK complex also localize to the centrosome and the spindle pole body, respectively (9,12,20). Consistent with their localization, the STRIPAK complex from these two species has been reported to play roles in mitosis (9,12,20). Our failure to detect localization of the yeast Far complex components and even the Far9 truncation mutant without its tail-anchor domain to the spindle pole body is consistent with the hypothesis of Frost et al. (9) that the function of the STRIPAK complex may have been "repurposed" and the S. cerevisiae STRIPAK complex has lost its function in mitosis.
Apart from identifying the cellular localization of the Far complex, we also determined the spatial order by which the Far complex organizes itself on the ER; Far9/10 establish a foothold on the ER utilizing a tail-anchor domain, and Far3/7/8 form a subcomplex that bridges Far11 to Far9/10 at the ER (Fig. 8). Previously, we reported that mutations in genes encoding the Far complex components suppress TORC2 deficiency in the following order: far11⌬ Ͼ far8/9⌬ Ͼ far3/7⌬ Ͼ far10⌬ (24). It is interesting to note that Far11, which is the most peripheral component of the complex, is also the most important among the six Far proteins in TORC2 signaling. Although this result is surprising, it may help understand why ER localization of the Far complex is not absolutely required for its function in TORC2 signaling.
We found that Far3 and Far7 are able to form a complex in the absence of the other four Far proteins. Interestingly, the mutant phenotypes of far3⌬ and far7⌬ on TORC2 signaling are most similar (24). Orthologs of Far3 and Far7 are only found in a restricted set of fungal species and lack apparent orthologs in animals (9). However, yeast Far3/Far7 and human SIKE/ FGFR1OP2 are all relatively small proteins predicted to have coiled-coil domains (22,25,46). Therefore, it is possible that Far3/Far7 may be the functional or structural counterparts of human SIKE/FGFR1OP2 (Table 1). Interestingly, the STRIPAK complex in S. pombe contains a novel protein of 166 residues, Csc4, which is also predicted to have a coiled-coil domain (20). It is possible that these small coiled-coil domain proteins may play the same structural role in the STRIPAK complex in different species as a result of divergent evolution.
Human Striatin 3 and its Sordaria macrospora ortholog PRO11 have been reported to contain an N-terminal coiled-coil region, and both proteins are critical components in the organization of their respective STRIPAK complexes (3,18). Like human striatins, yeast Far8 was also predicted to contain an N-terminal coiled-coil domain (22,25). Interestingly, almost all components of the Far complex are predicted to contain coiledcoil domains (22,25). These coiled-coil domains may mediate protein-protein interactions and provide a structural framework for the organization of the Far complex. Further research will be conducted to uncover the role of Far8 in the yeast STRIPAK complex and how the Far complex interfaces with PP2A.