The nuclear transcription factor Rtg1p functions as a cytosolic, post-transcriptional regulator in the methylotrophic yeast Pichia pastoris

Rtg1p and Rtg3p are two basic helix–loop–helix, retrograde transcription factors in the budding yeast Saccharomyces cerevisiae. Both factors heterodimerize to activate the transcription of nuclear genes in response to mitochondrial dysfunction and glutamate auxotrophy, but are not well characterized in other yeasts. Here, we demonstrate that the Rtg1p/Rtg3p-mediated retrograde signaling pathway is absent in the methylotrophic yeast Pichia pastoris. We observed that P. pastoris Rtg1p (PpRtg1p) heterodimerizes with S. cerevisiae Rtg3p and functions as a nuclear, retrograde transcription factor in S. cerevisiae, but not in P. pastoris. We noted that P. pastoris Rtg3p lacks a functional leucine zipper and interacts with neither S. cerevisiae Rtg1p (ScRtg1p) nor PpRtg1p. In the absence of an interaction with Rtg3p, PpRtg1p has apparently acquired a novel function as a cytosolic regulator of multiple P. pastoris metabolic pathways, including biosynthesis of glutamate dehydrogenase 2 and phosphoenolpyruvate carboxykinase required for the utilization of glutamate as the sole carbon source. PpRtg1p also had an essential role in methanol metabolism and regulated alcohol oxidase synthesis and was required for the metabolism of ethanol, acetate, and oleic acid, but not of glucose and glycerol. Although PpRtg1p could functionally complement ScRtg1p, ScRtg1p could not complement PpRtg1p, indicating that ScRtg1p is not a functional PpRtg1p homolog. Thus, PpRtg1p functions as a nuclear, retrograde transcription factor in S. cerevisiae and as a cytosolic, post-transcriptional regulator in P. pastoris. We conclude that PpRtg1p is a key component of a signaling pathway that regulates multiple metabolic processes in P. pastoris.

RTG signaling has not been well characterized in yeasts other than S. cerevisiae. In Candida albicans, Rtg3p was shown to be involved in tolerance to cations and antifungal drugs as well as serum-induced filamentation (10). However, its interaction with Rtg1p and its role in glutamate homeostasis have not been investigated. In other yeasts, most studies are focused on the ability of putative RTG homologues to complement corresponding mutations in S. cerevisiae. For example, Rtg2p and Mks1p homologues from Kluyveromyces lactis and Kluyveromyces waltii can complement ⌬rtg2 and ⌬mks1 mutations in S. cerevisiae, and ATP-mediated dissociation of Mks1p from Rtg2p was shown to be conserved among these yeasts as well (6). Putative RTG2 genes from Candida glabrata, Ashbya gossypii, K. lactis, and Vanderwaltozyma polyspora functionally complement S. cerevisiae ⌬rtg2 mutant (11). Pichia pastoris (Komagataella phaffii), a methylotrophic yeast, possesses a potent, methanol-inducible promoter encoding alcohol oxidase I (AOXI) and grows to very high cell densities by virtue of its respiratory metabolism (12,13). It is widely used for the production of several recombinant proteins (14,15). In this study, we demonstrated that Rtg3p is absent in P. pastoris and that Rtg1p functions as a cytosolic regulator rather than a nuclear, retrograde transcription factor. We conclude that the classical mitochondrial retrograde response involving Rtg1p-Rtg3p heterodimer is not functional in respiratory yeasts such as P. pastoris and that Rtg1p has evolved as an Rtg3p-independent regulator of multiple metabolic pathways.

Rtg3p is absent in P. pastoris, and PpRtg1p is a functional homologue of ScRtg1p
The lack of information on mitochondrial retrograde signaling in yeasts other than S. cerevisiae prompted us to investigate this pathway in P. pastoris, a respiratory and methylotrophic yeast of biotechnological importance. BLAST analysis of the P. pastoris genome database using amino acid sequences of ScRtg proteins as the query indicated the existence of putative P. pastoris Rtg (PpRtg) homologues (Table 1). Because RTG signaling culminates in the activation of nuclear genes by the heterodimeric transcription factor Rtg1p-Rtg3p, we focused our attention on these two proteins in this study. The basic helix-loop-helix domains of P. pastoris proteins annotated as PpRtg1p and PpRtg3p share 48 and 44% amino acid identity with those of ScRtg1p and ScRtg3p, respectively ( Fig. 1A and Table 1). PpRtg3p is designated as PpRtgXp in this study (see Table 1) because the leucine zipper essential for heterodimerization with Rtg1p is not conserved in this protein ( Fig. 1B) (16,17). To examine the ability of PpRtg1p and PpRtgXp to heterodimerize with each other as well as with ScRtg homologues, these proteins were expressed as maltosebinding protein (MBP) or GSH S-transferase (GST) fusion proteins in Escherichia coli (Fig. 1C), and protein-protein interactions were studied. GST-Rtg1p bound to GSH beads was incubated with E. coli cell lysates containing MBP-Rtg3p or MBP-RtgXp. After washing, proteins retained on the beads were analyzed by SDS-PAGE and visualized by Coomassie Brilliant Blue R staining. Although PpRtg1p and ScRtg1p interacted with ScRtg3p (Fig. 1D, lanes 4 and 8), PpRtgXp failed to interact with either PpRtg1p or ScRtg1p (Fig. 1D, lanes 3 and 7). Thus, PpRtgXp is unlikely to be a functional homologue of ScRtg3p.
The ability of Rtg1p-Rtg3p heterodimers to bind to radiolabeled S. cerevisiae CIT2 upstream activation sequence containing R box (4) was examined in an electrophoretic mobility shift assay (EMSA). PpRtg1p-ScRtg3p (Fig. 1E, lane 5) but not PpRtgXp-ScRtg1p (Fig. 1E, lane 7) or PpRtgXp-PpRtg1p (Fig.  1E, lane 9) interacted with CIT2 R box. A point mutation within the R box abolished the formation of ScRtg1p-ScRtg3p (Fig. 1E, lane 4) and PpRtg1p-ScRtg3p complexes (Fig. 1E, lane 6). We generated PpRtgXp* by fusing the region containing amino acids 345-486 of ScRtg3p to the C-terminal region of PpRtgXp (Fig. 1F). When expressed as an MBP fusion protein, PpRtgXp* interacted with ScRtg1p as well as PpRtg1p (Fig. 1G, lanes 2 and  5), suggesting that the absence of a dimerization domain may be responsible for the inability of PpRtgXp to dimerize with Rtg1p. PpRtg1p and ScRtg1p were expressed as GFP fusion proteins in S. cerevisiae ⌬rtg1 (⌬Scrtg1), and expression of PpRtg1 GFP and ScRtg1 GFP was confirmed by Western blotting using anti-GFP antibodies ( Fig. 2A). CIT2 mRNA levels were up-regulated in ⌬Scrtg1 expressing PpRtg1 GFP as well as ScRtg1 GFP as evident from quantitative real-time PCR (qPCR) of RNA isolated from S. cerevisiae strains cultured in YNBD Glu Ϫ and YNBD Glu ϩ media (Fig. 2C). PpRtg1 GFP as well as ScRtg1 GFP facilitated the growth of ⌬Scrtg1 in cells cultured in YNBD Glu Ϫ medium

Post-transcriptional regulation by Rtg1p
( Fig. 2, D and E). Subcellular localization studies using anti-GFP antibodies indicated that PpRtg1p GFP and ScRtg1p GFP localized to the cytosol of ⌬Scrtg1 cultured in YNBD Glu ϩ medium but translocated to the nucleus when cultured under glutamatedeficient conditions (Fig. 2F). These results indicate that PpRtg1p is a functional homologue of ScRtg1p.

Regulation of glutamate utilization pathway of P. pastoris by PpRtg1p
Thus far, Rtg3p-independent function has not been reported for Rtg1p. The absence of Rtg3p in P. pastoris led us to investigate PpRtg1p function in detail. A ⌬Pprtg1 strain was generated, and absence of PpRTG1 mRNA was confirmed by reverse
PpRtg1p and ScRtg1p were expressed as GFP-and FLAGtagged proteins in ⌬Pprtg1, and expression was confirmed by Western blotting using anti-GFP and anti-FLAG antibodies, respectively (Fig. 3, C and D). Both proteins localized to the cytosol of cells cultured in YNBD Glu Ϫ medium (Fig. 3E). Thus, nuclear localization of Rtg1p is observed only in S. cerevisiae but not P. pastoris cultured in YNBD Glu Ϫ medium (compare Figs. 2F and 3E).
P. pastoris can utilize glutamate, aspartate, or proline as the sole source of carbon and nitrogen (18) (Table 2), and growth of ⌬Pprtg1 was severely compromised when cultured in yeast nitrogen base (YNB) Glu ϩ medium containing glutamate as the sole source of carbon (Fig. 4A). When cultured in YNB*D Glu ϩ medium containing glutamate rather than ammonium sulfate as the sole source of nitrogen, ⌬Pprtg1 grew normally (Fig. 4A), indicating that PpRtg1p is essential for the utilization of glutamate as the sole source of carbon but not nitrogen. Because PpRtg1p is a functional homologue of ScRtg1p (Fig. 2, D and E), we examined whether ScRtg1p is a functional homologue of PpRtg1p. Only PpRtg1p GFP but not ScRtg1p FLAG facilitated the growth of ⌬Pprtg1 in YNB Glu ϩ medium (Fig. 4B). To rule out interference from epitope tags, ⌬Pprtg1 strains expressing untagged PpRtg1p and ScRtg1p were constructed (Fig. 4C). Expression of mRNA encoding untagged PpRtg1p and ScRtg1p was confirmed by semiquantitative RT-PCR (Fig. 4D). As observed with epitope-tagged proteins, untagged PpRtg1p but not ScRtg1p facilitated the growth of ⌬Pprtg1 in YNB Glu ϩ medium (Fig. 4E). We conclude that ScRtg1p is not a functional homologue of PpRtg1p.
To understand the mechanism of action of PpRtg1p, GS115 and ⌬Pprtg1 were cultured in YNB Glu ϩ medium, and cell lysates were analyzed by SDS-PAGE followed by Coomassie Brilliant Blue R staining. At least four major proteins, designated as A, B, C, and D, were either undetectable or expressed at reduced levels in ⌬Pprtg1 compared with GS115 (Fig. 5A). Protein bands D and C were identified by MS as phosphoenolpyruvate carboxykinase (PEPCK; accession number XP_002490872.1) and glycerol kinase (GUT1; accession number CCA40348.1), respectively. Bands A and B were suspected to be glutamate dehydrogenase 2 (GDH2) and AOX, respectively, based on their abundance, molecular weight, and results of an earlier study (18). To confirm their differential expression, these proteins were expressed as Myc-or His-tagged proteins in GS115 and ⌬Pprtg1, and lysates of cells cultured in YNB Glu ϩ medium were subjected to Western blotting. All the proteins were expressed at significantly lower levels in ⌬Pprtg1 than in GS115 (Fig. 5B). Among these PpRtg1p-regulated enzymes, PEPCK and GDH2 are essential for growth of cells in YNB Glu ϩ medium (Fig. 5C), suggesting that deficiency of these enzymes may contribute to the growth defect of ⌬Pprtg1 in YNB Glu ϩ medium. Subcellular localization studies using PpRtg1p Myc or PpRtg1p GFP indicate that Rtg1p localizes to the cytosol in cells cultured in YNB Glu ϩ medium (Fig. 5, D and E), suggesting that PpRtg1p is unlikely to function as a nuclear transcription factor. Furthermore, mRNA levels of GDH2 and PEPCK in ⌬Pprtg1 were higher but not lower than those in GS115 as evident from RT-PCR and qPCR analyses

Post-transcriptional regulation by Rtg1p
( Fig. 5, F and G), indicating that PpRtg1p regulates GDH2 and PEPCK synthesis at the post-transcriptional level.

PpRtg1p is essential for the utilization of methanol, ethanol, acetate, and oleic acid
AOX is a methanol-inducible, peroxisomal enzyme whose promoter is widely used for recombinant protein production (12,13). AOX is encoded by AOXI and AOXII of which AOXI accounts for Ͼ90% of AOX protein synthesized during methanol metabolism (12,13). AOX is essential for the utilization of methanol and fatty acids as the sole source of carbon. To date, post-transcriptional regulation of AOXI has not been reported. Methanol-inducible expression of AOXI is mediated by the zinc finger proteins Mxr1p and Trm1p at the transcriptional level. AOXI mRNA as well as protein levels are down-regulated in ⌬mxr1 and ⌬trm1 cells cultured in YNBM medium (19 -21).
The down-regulation of AOXI Myc in ⌬Pprtg1 cultured in YNB Glu ϩ medium ( Fig. 5B) led us to investigate whether PpRtg1p regulates AOXI expression during methanol metabolism. The results indicate that AOXI Myc protein levels are significantly lower in ⌬Pprtg1 than those in GS115 cultured in YNBM (Fig. 6, A and B). The differential expression of AOX in GS115 and ⌬Pprtg1 can be visualized in Coomassie Brilliant Blue-stained SDS-polyacrylamide gels because it is the most abundant protein in P. pastoris cells cultured in YNBM medium (Fig. 6C). These results were further confirmed by Western blotting of lysates of cells cultured in YNBM medium using anti-AOX antibodies (Fig. 6D). Cells were cultured in YNBM, and immunofluorescence studies were carried out with anti-AOX antibodies. AOX-containing peroxisomes appear as a large vesicle(s) or small puncta in GS115 and ⌬Pprtg1 due to high or low levels of AOX, respectively (Fig. 6E). AOX protein level in ⌬Pprtg1 was lower than that in GS115 but higher than that in ⌬mxr1 (Fig. 6F), and AOXI mRNA levels followed a similar pattern (Fig. 6, G and H) Thus, both AOXI mRNA and protein levels are down-regulated in ⌬Pprtg1 cultured in YNBM. Growth of ⌬Pprtg1 in YNBM was severely impaired, indicating that PpRtg1 has an important role in methanol metabolism (Fig. 7, A and B). The growth defect of ⌬Pprtg1 was rescued by the expression of PpRtg1 GFP but not ScRtg1 FLAG (Fig. 7A). Similar results were observed when untagged PpRtg1p and ScRtg1p were expressed in ⌬Pprtg1 (Fig. 7B). Only PpRtg1 GFP but not ScRtg1 Flag was able to restore AOX protein levels in ⌬Pprtg1 to those of GS115 (Fig. 7C), further confirming that ScRtg1p is not a functional homologue of PpRtg1p. As observed in cells cultured in YNB Glu ϩ medium (Fig. 5, D and E), PpRtg1p localized to the cytosol of cells cultured in YNBM as well (Fig. 7D). PpRtg1p was also essential for the growth of P. pastoris in Figure 3. Analysis of the function and subcellular localization of ScRtg1p and PpRtg1p in P. pastoris cultured in YNBD medium. A, confirmation of the absence of RTG1 mRNA in ⌬Pprtg1 by RT-PCR. PpRTG1-and PGK-specific primers were used for amplification of PpRTG1 and PGK, respectively. DNA molecular weight markers (kb) are shown on the left. B, analysis of growth of GS115 and ⌬Pprtg1 in glutamate-deficient and -sufficient YNBD media. C, schematic diagrams of constructs designed to express PpRtg1p and ScRtg1p in ⌬Pprtg1 as GFP fusion and 3XFLAG-tagged proteins, respectively, from PpRTG1 promoter. Numbers in parentheses indicate molecular mass of proteins (kDa). D, analysis of expression of PpRtg1p GFP and ScRtg1p FLAG by Western blotting using anti-GFP and anti-FLAG antibodies, respectively. Numbers indicate protein molecular mass markers (kDa). E, localization of ScRtg1p Flag and PpRtg1p GFP using anti-FLAG and anti-GFP antibodies, respectively, in P. pastoris strains cultured in YNBD Glu Ϫ medium as indicated. Mouse anti-GFP antibodies, mouse anti-FLAG antibodies, and Alexa Flour 555-conjugated donkey anti-mouse antibodies were used. Hoechst 33342 was used to stain the nucleus.

Post-transcriptional regulation by Rtg1p
YNBE, YNBA, and YNBO but not YNBD and YNBG (Fig. 7E). PpRtg1p predominantly localized to the cytosol irrespective of the carbon source used for culturing the cells (Fig. 7F). Thus, PpRtg1p is a cytosolic protein that regulates multiple metabolic pathways in P. pastoris while retaining its ability to act as a nuclear, retrograde transcription factor in S. cerevisiae (Fig. 8).

Discussion
S. cerevisiae, a Crabtree-positive yeast with a respirofermentative metabolism, utilizes the TCA cycle primarily as a source of ␣-ketoglutarate and other biosynthetic precursors while generating ATP through glycolysis (1,2). In cells with an impaired TCA cycle, the RTG pathway provides a means of ammonium assimilation and a source of glutamate. In contrast, Crabtreenegative yeasts with a respiratory metabolism such as P. pastoris utilize the TCA cycle not only for the generation of biosynthetic intermediates but also for ATP generation via oxidative phosphorylation. Thus, mitochondrial dysfunction in these yeasts results in the deficiency of glutamate as well as ATP, and restoration of glutamate levels alone via the Rtg1p/Rtg3p-mediated retrograde pathway does not appear to be a viable option. Thus, the necessity for glutamate biosynthesis via the RTG pathway in these yeasts is questionable. The fact that deletion of RTG1 does not result in glutamate auxotrophy in P. pastoris led us to examine the function of PpRtg1p. We demonstrate that Rtg3p is absent in P. pastoris and that PpRtg1p has acquired a novel function as a cytosolic regulator of multiple metabolic pathways in P. pastoris while retaining its ability to function as a nuclear, retrograde transcription factor in S. cerevisiae. Interestingly, ScRtg1p does not function as a cytosolic regulator in P. pastoris, indicating that PpRtg1p is a functional homologue of S. cerevisiae Rtg1p, but the latter is not a functional homologue of PpRtg1p. Although this study focused primarily on PpRtg1p, it will be interesting to examine the functions of other proteins of the RTG signaling pathway listed in Table 1. It is pertinent to note that CIT2 encoding peroxisomal citrate synthase, a key protein involved in RTG response, is present only in S. cerevisiae but not P. pastoris (https://www.ncbi.nlm.nih.gov/protein/CCA36579). CIT2 is also absent in several other respiratory yeasts, including Pichia stipitis, C. albicans, Candida utilis, Yarrowia lipolytica, and Kluyveromyces marxianus, which can also utilize amino acids such as glutamate as the sole source of carbon (18,22). It will be interesting to examine the localization and function of Rtg1p of these yeasts.
In this study, we demonstrate that PpRtg1p regulates multiple metabolic pathways. Of these, we have identified the targets of PpRtg1p in at least two metabolic pathways. PpRtg1p regulates the synthesis of PEPCK, GDH2, GUT1, and AOXI during glutamate utilization, whereas AOXI is a target during methanol utilization. Proteins regulated by PpRtg1p during ethanol, acetate, and oleic acid metabolism remain to be identified. PpRgt1p localizes to the cytosol and down-regulates protein but not mRNA levels of PEPCK and GDH2 during glutamate

Post-transcriptional regulation by Rtg1p
utilization. During methanol metabolism, in addition to a decrease in AOX protein, AOXI mRNA levels are also affected in ⌬Pprtg1 cultured in YNBM. The exact mechanism by which PpRtg1p regulates glutamate and methanol metabolism remains to be investigated. This study demonstrates, for the first time, regulation of methanol-inducible expression of AOXI by a cytosolic protein in P. pastoris. To date, strategies aimed at the development of versatile, AOXI-based expression systems have focused on manipulating the transcriptional regulatory circuits (14,(23)(24)(25)(26)(27). The identification of Rtg1p-dependent, post-transcriptional regulatory circuit could lead to the development of novel synthetic biology tools for genetic manipulation of AOXI expression. Finally, this study demonstrates that the Rtg1p/Rtg3-mediated RTG signaling pathway studied extensively in S. cerevisiae may not be conserved in P. pastoris and related yeasts.

Experimental procedures Media and culture conditions
S. cerevisiae BY4741 and ⌬rtg1 strains were obtained from Euroscarf, Frankfurt, Germany. P. pastoris GS115 was a kind gift from James Cregg and has been described (19). P. pastoris KM71 strain was purchased from Thermo Fisher Scientific. Yeast strains were cultured at 30°C in an orbital shaker at 180 rpm overnight in YPD (1.0% yeast extract, 2.0% peptone, 2.0% glucose), washed with sterile water, and shifted to different minimal media containing 0.17% YNB without amino acids and with 0.5% ammonium sulfate supplemented with 2.0% glucose (YNBD), 2.0% glycerol (YNBG), 1.0% glutamate (YNB Glu ϩ ), 1.0% methanol (YNBM), 1.0% ethanol (YNBE), or 0.5% oleic acid (YNBO). For solid medium, agar was added to a final concentration of 2%. In some experiments, YNB medium without ammonium sulfate (YNB*) was used. For the growth of S. cerevisiae BY4741 strain, YNBD medium was supplemented with histidine (1.8 mg.ml), leucine (38 mg/ml), methionine (7.6 mg/ml), and uracil (7.6 mg/ml). Where indicated, glutamate was added at a final concentration of 0.2% to YNBD medium. For growth kinetics, colonies were first cultured overnight in YPD, then washed with sterile water, and shifted to different media with initial A 600 of ϳ0.1 per ml of media. Aliquots of cells were removed at regular intervals, and A 600 was measured. For spot assays, colonies were first cultured overnight in YPD, washed, and resuspended in sterile water to an A 600 of 1 per ml. Serial dilutions of 1:10 until an A 600 of 10 Ϫ4 per ml were made, and 2 l from each dilution was then spotted on solid medium.
E. coli DH5␣ and BL21 (DE3) strains were used for the isolation of recombinant plasmids and expression of recombinant proteins, respectively. Bacterial and yeast transformations were done using CaCl 2 and an electroporation method (Gene Pulser, Bio-Rad), respectively, according to the manufacturer's instructions.

Post-transcriptional regulation by Rtg1p
Oligonucleotides were purchased form Sigma-Aldrich. Anti-FLAG tag and anti-Myc tag antibodies were purchased from Sigma-Aldrich and Merck Millipore, respectively. Mouse anti-GFP and mouse anti-His tag antibodies were purchased from Santa Cruz Biotechnology Inc. Donkey anti-mouse Alexa Flour 555/488 antibodies were purchased from Thermo Fisher Scientific. Anti-AOXI antibodies were generated by injecting AOXI purified from P. pastoris cell extracts into mice. Anti-phosphoglycerate kinase (PGK) antibodies were obtained by immunizing rabbits with recombinant, histidine-tagged P. pastoris PGK. Nucleotide sequence of primers used in qPCRs and RT-PCRs will be provided on request. Restriction enzymes, Taq DNA polymerase, and T4 DNA ligase were purchased from New England Biolabs (Frankfurt, Germany).

qPCR and RT-PCR
Total RNA was isolated from yeast cells using an RNA isolation kit (Promega), and qPCR was carried out essentially as described (21) using a StepOnePlus Real-Time PCR system (Thermo Fisher Scientific). For semiquantitative RT-PCR, cDNA was prepared using a RevertAid First Strand cDNA Synthesis kit (Thermo Fisher Scientific), and PCR was carried out for 20, 25, and 30 cycles in a 2720 thermal cycler (Thermo Fisher). PCR products were resolved on 1% agarose gels and visualized by ethidium bromide staining (0.05 g/ml) using a UV transilluminator.

Subcellular localization studies
Yeast cells were treated with 1% formaldehyde for 1 h and resuspended in buffer containing 40 mM K 2 HPO 4 (pH 7.4), 1.2 M sorbitol, 50 mM MgCl 2 , and 30 mM ␤-mercaptoethanol. Zymolyase (G Biosciences, St. Louis, MO) was added (7 l/100 l) and incubated at 37°C for 1 h. Cells were centrifuged at 1500 rpm for 10 min in a microcentrifuge (Hettich, Tuttlingen, Germany) at room temperature, resuspended in phosphatebuffered saline (PBS), spread evenly onto glass coverslips, and then air-dried. Coverslips were incubated in blocking buffer (PBS containing 0.05% Tween 20 and 2% BSA) for 1 h followed by incubation in blocking buffer containing primary and secondary antibodies at room temperature for 1-3 h. Coverslips were washed with PBS and then treated with Hoechst 33342 (Sigma-Aldrich; 1 g/ml) or 4Ј,6-diamidino-2-phenylindole (DAPI) (1 g/ml) for 7 min. After washing with PBS, coverslips were air-dried, and cells were visualized using a fluorescence microscope (Leica DMLA) or confocal microscope (Zeiss LSM 880 with Airyscan) with appropriate filters.

Post-transcriptional regulation by Rtg1p
For visualization of GFP-tagged proteins by live-cell imaging, yeast cells (3 l) were placed on a glass slide layered with 1% agarose, and the cells were allowed to settle for 1 min. A coverslip was placed on top of the cell suspension, the edge of the coverslip was sealed with nail polish, and cells were visualized using a confocal microscope as mentioned above.

EMSA
EMSA was carried out essentially as described (20). Briefly, oligonucleotides radiolabeled at their 5Ј-ends using [␥-32 P] ATP and T4 polynucleotide kinase were incubated with recombinant proteins in EMSA buffer (50 mM Tris (pH 8.0), 250 mM NaCl, 0.25% Nonidet P-40, 5 mM DTT, and 30% glycerol). The reaction mixture was electrophoresed on a 5% polyacrylamide gel at 4°C for 6 h. The gel was dried and exposed to a phosphorimaging screen overnight. DNA and DNA-protein complexes were visualized using a Typhoon FLA 9000 laser scanner.

Western blotting
Yeast cells were resuspended in a lysis buffer containing 20 mM Tris (pH 8.0), 400 mM NaCl, 10 mM MgCl 2 , 10 mM EDTA (pH 8.0), 10% glycerol, 7 mM ␤-mercaptoethanol, and protease inhibitor mixture (cOmplete, ULTRA, Mini, EDTA-free, EASYpack). Chilled, 0.5-mm glass beads (Biospec Products) were added (1 g/g of cells) and vortexed 10 times for 1 min, each time keeping the cells on ice for 1 min between vortexings and using the highest setting of the vortex mixer. Cell debris was removed by centrifugation at 13,000 rpm in a microcentrifuge (Hettich). Proteins were estimated using Bradford reagent (Bio-Rad) and resolved by SDS-PAGE, electroblotted onto a 0.22-m polyvinylidene difluoride membrane using transfer buffer (39 mM glycine, 48 mM Tris (pH 8.0), and 20% methanol). The membrane was blocked overnight in 5% nonfat milk

Statistical analysis
Statistical tests, including Student's t test and one-way analysis of variance followed by Tukey's multiple comparison test, were carried out using GraphPad Prism 5 software. Data are presented as mean Ϯ S.D. A p value summary is indicated on the bar of each figure: *, p Ͻ 0.05; **, p Ͻ 0.005; ***, p Ͻ 0.0005; ns, not significant.

Mass spectrometry
Protein bands of interest were excised precisely from the SDS-polyacrylamide gel and subjected to in-gel trypsin digestion using a Trypsin Profile IGD kit (Sigma-Aldrich). MALDI-TOF was performed in an HCT Ultra PTM Discovery System (ETD II, Bruker Daltonics) with an 1100 series HPLC (Agilent). For identification of proteins, the obtained "peak list" was searched against the NCBI nonredundant database using Mascot protein mass fingerprint software.

Expression of recombinant S. cerevisiae and P. pastoris Rtg proteins in E. coli and study of protein-protein interactions
For the construction of pGEX-PpRTG1 plasmid, PpRTG1 gene was amplified from GS115 genomic DNA by PCR using the primer pair 5Ј-CGCGGATCCATGGATAGTAATCAAT-GGCCC-3Ј and 5Ј-AAGGAAAAAAGCGGCCGCTCAATC-TGCTAGTGGTCCAAC-3Ј (BamHI and NotI restriction sites are underlined). The PCR product was digested with BamHI and NotI and cloned into pGEX4T1 vector.
pGEX-ScRTG1 plasmid was constructed by the PCR amplification of ScRTG1 gene from S. cerevisiae BY4741 genomic DNA using the primer pair 5Ј-CGCGGATCCATGAGCAGC-ATTCCAGCTGGC-3Ј and 5Ј-AAGGAAAAAAGCGGCCG-CTTAGCTACCATTACCGTAC-3Ј and cloning the PCR product into pGEX4T1 vector at the BamHI and NotI sites. Restriction sites in the primers are underlined.
pMAL-ScRTG3 plasmid was constructed by the amplification of ScRTG3 gene from S. cerevisiae BY4741 genomic DNA using the primer pair 5Ј-CGCGGATCCATGATGAACAATA-ACGAAAGTG-3Ј and 5Ј-AAAACTGCAGCTACCCCGAAC-CAAATTC-3Ј (BamHI and PstI restriction sites are underlined). The PCR product was digested with BamHI and PstI and cloned into pMAL-c2X vector.
pMAL-PpRTGX plasmid expressing PpRtgX as an MBP fusion protein was constructed by amplifying PpRTGX gene from P. pastoris GS115 genomic DNA by PCR using the primer pair 5Ј-CGCGGATCCAATGGCACAAGATGCTCCCT-TTG-3Ј and 5Ј-AAAACTGCAGTTAGTTCGGAACGTCA-AGG-3Ј (BamHI and PstI restriction sites are underlined). The PCR product was digested with BamHI and PstI and cloned into pMAL-c2X vector.
For the study of protein-protein interactions, E. coli cells were suspended in a buffer containing 50 mM Tris (pH 8.0), 300 mM NaCl, 10% glycerol, 10 mM ␤-Mercaptoethanol, and 1 mM PMSF and subjected to sonication in a Vibracell 750 (Sonics and Materials Inc., Newtown, CT). Cell lysates containing GST-Rtg1p were incubated with GSH-agarose beads at 4°C for 1 h followed by brief centrifugation and washing with a buffer A containing 50 mM Tris (pH 8.0) and 150 mM NaCl. GST-Rtg1p bound to GSH-agarose beads was incubated with E. coli cell lysates containing MBP-Rtg3p, MBP-RtgXp, or MBP-RtgXp* for 1 h. After centrifugation and washing twice with buffer A, proteins bound to GSH beads were resolved on an SDS-polyacrylamide gel and visualized by Coomassie Brilliant Blue R staining.

Generation of S. cerevisiae ⌬Sc-ScRTG1 GFP and ⌬Sc-PpRTG1 GFP
ScRTG1 along with ϳ1.0-kb promoter was amplified by PCR from S. cerevisiae BY4741 genomic DNA using the primer pair 5Ј-CGCGGATCCGTACTTGGGATTTCCCATAGC-3Ј and 5Ј-CTCCTTTACTAGTCAGATCTACCATGCTACCATTA-CCGTACTCACCATACC-3Ј. The gene encoding GFP was amplified from pREP41GFP vector (28) using the primer pair 5Ј-GGTATGGTGAGTACGGTAATGGTAGCATGGTAGA-TCTGACTAGTAAAGGA-3Ј and 5Ј-CCGCTCGAGTTAGT-GGTGGTGGCTAGCTTT-3Ј (BamHI and XhoI sites are underlined). The individual PCR products were purified and used as templates in a final PCR with the primers containing BamHI and XhoI sites to obtain P RTG1 ScRTG1-GFP.

Generation of P. pastoris ⌬Pprtg1 strain
The ⌬Pprtg1 strain in which the coding region of RTG1 was replaced by a Zeocin expression cassette was generated as follows. The 1.08-kb RTG1 promoter was amplified from P. pastoris GS115 genomic DNA by PCR using the primer pair 5Ј-GATTTCTGCGATTGGAGCGATTTC-3Ј and 5Ј-CTAT-GGTGTGTGGGGGATCCGCATGGAGTCAACAGGAGTA-GTGAAG-3Ј (962 to 984 bp of pGAPZA in reverse complement (in italics), Ϫ43 to Ϫ20 bp of RTG1 in reverse complement). In the second PCR, 1.2 kb of the Zeocin expression cassette was amplified by PCR using the primer pair 5Ј-CTTCACTACTCCTGTTGACTCCATGCGGATCCCCC-ACACACCATAG-3Ј (Ϫ43 to Ϫ20 bp of RTG1 (italics), ϩ962 to ϩ984 bp of pGAPZA vector) and 5Ј-GAGAAGAATGTC-CATGAATGTATCCCTCACATGTTGGTCTCCAGCTT-G-3Ј (ϩ743 to ϩ767 bp in reverse complement of 3Ј-flanking region of RTG1 (in italics), ϩ2137 to ϩ2159 bp in reverse complement of pGAPZA). In the third PCR, 974 bp of the 3Ј-flanking region of RTG1 was amplified using the primer pair 5Ј-CAAGCTGGAGACCAACATGTGAGGGATACATTCAT-GGACATTCTTCTC-3Ј (ϩ2137 to ϩ2159 bp of pGAPZA, ϩ743 to ϩ767 bp of 3Ј-flanking region of RTG1) and 5Ј-ATA-CATCTGCTCATGAGTTCATCC3Ј (ϩ1693 to ϩ1716 bp in the reverse complement of 3Ј-flanking region of RTG1). All three PCR products were purified and used as templates in the final PCR along with the primer pair 5Ј-GATTTCTGCGATT-GGAGCGATTTC-3Ј and 5Ј-ATACATCTGCTCATGA-GTTCATCC-3Ј to obtain a 3.234-kb product consisting of the Zeocin expression cassette flanked by 1.080 kb of RTG1 promoter and 974 bp of 3Ј-flanking region of RTG1 that was transformed into P. pastoris GS115 strain. Zeocin-resistant transformants that did not express RTG1 mRNA as evident from RT-PCR were selected and designated as ⌬Pprtg1.

Generation of P. pastoris ⌬Pp-PpRTG1 GFP and ⌬Pp-ScRTG1 FLAG strains
Expression cassettes comprising genes encoding PpRtg1p and ScRtg1p along with ϳ1 kb of PpRTG1 promoter were cloned into pIB3 vector (Addgene, 25452) and expressed in ⌬Pprtg1 as GFP-tagged and FLAG-tagged proteins, respectively. PpRTG1 (1.741 kb) along with its ϳ1-kb promoter was amplified from GS115 genomic DNA using the primer pair 5Ј-CCGCTCGAGGATTTCTGCGATTGGAGCG-3Ј and 5Ј-CCTTTACTAGTCAGATCTACCATATCTGCTAGTGGT-CCAACACCAATG-3Ј (XhoI site is underlined), and a 714bp coding region of GFP was amplified from the vector pREP41GFP vector (28) using the primer pair 5Ј-CATTGGT-GTTGGACCACTAGCAGATATGGTAGATCTGACTAGT-AAAGG-3Ј and 5Ј-CCCAAGCTTCTAGTGGTGGTGGCTA-GCTTTG-3Ј (HindIII site is underlined). Both PCR products were purified and used as templates in the final PCR along with the primer pair 5Ј-CCGCTCGAGGATTTCTGCGATTGGA-GCG-3Ј and 5Ј-CCCAAGCTTCTAGTGGTGGTGGCTA-GCT-3Ј. The PCR product was cloned into pIB3 vector and transformed into E. coli DH5␣ competent cells. Recombinant plasmid containing P RTG -PpRTG1-GFP construct was linearized using SalI and transformed into ⌬Pprtg1 by electroporation. Recombinant clones were selected by plating on YNBD His Ϫ plates, and clones expressing GFP-tagged Rtg1p were confirmed by Western blotting using anti-GFP antibody.

Generation of GS-PpRTG1 GFP -OE
P. pastoris GS115 expressing PpRtg1p as a GFP fusion protein from GAPDH promoter (GS-PpRTG1 GFP -OE) was generated as follows. PpRTG1 gene (741 bp) was amplified from GS115 genomic DNA using primers 5Ј-CCGCTCGAGATGG-ATAGTAATCAATGGCCC-3Ј and 5Ј-CCTTTACTAGTCA-GATCTACCATATCTGCTAGTGGTCCAACACCAATG-3Ј (XhoI site is underlined in the forward primer). GFP was amplified by PCR from pREP41GFP vector (28) using the primer pair 5Ј-CATTGGTGTTGGACCACTAGCAGATATGGTAGAT-CTGACTAGTAAAGG-3Ј and 5Ј-ATAAGAATGCGGCCG-CCTAGTGGTGGTGGCTAGCTTTG-3Ј (NotI site is underlined in the reverse primer). Both PCR products were purified and used as templates in the final PCR along with the primer pair 5Ј-CCGCTCGAGATGGATAGTAATCA-ATGGCCC-3Ј and 5Ј-ATAAGAATGCGGCCGCCTAGTG GTGGTGGCTAGCTTTG-3Ј. Following XhoI and NotI digestion, the PCR product was cloned into pGAPZA vector and transformed into E. coli DH5␣ competent cells. Recombinant plasmid containing P GAPDH PpRTG1-GFP was linearized with AvrII and transformed into GS115 by electroporation, and Zeocin-resistant colonies were selected on Zeocin-containing YPD plates. PpRtg1-GFP expression was confirmed by Western blotting using anti-GFP antibody.
Generation of ⌬Pp-PpRTG1* and ⌬Pp-ScRTG1* strains expressing untagged ScRtg1p and PpRtg1p, respectively Expression cassettes comprising genes encoding PpRtg1p and ScRtg1p along with ϳ1 kb of PpRTG1 promoter were cloned into pIB3 vector (Addgene, 25452) and expressed in ⌬Pprtg1 (as untagged proteins). The PpRTG1 expression cassette (1.741 kb) was amplified from GS115 genomic DNA using the primer pair 5Ј-CCGCTCGAGGATTTCTGCGATTGGA-GCG-3Ј and 5Ј-CCCAAGCTTTCAATCTGCTAGTGGTCC-AACACCAATG-3Ј (XhoI and HindIII sites are underlined) and cloned into pIB3 vector. The recombinant plasmid (P RTG -PpRTG1) was linearized with SalI and transformed into ⌬Pprtg1 by electroporation. Recombinant clones were selected by plating on YNBD His Ϫ plates, and expression of PpRTG1 in the recombinant clones was confirmed by RT-PCR.
For the generation of ⌬Pp-ScRTG1*, ϳ1-kb PpRTG1 promoter encoding ScRTG1 gene was generated by a series of PCRs using overlapping primers. First, PpRTG1 promoter (1.0 kb) was amplified from P. pastoris genomic DNA isolated from GS115 and BY4741, respectively, using overlapping primer pairs 5Ј-CCGCTCGAGGATTTCTGCGATTGGAGCG-3Ј and 5Ј-GCCAGCTGGAATGCTGCTCATGAAGGTAGTCT-TTCGTAGATGGAGTC-3Ј. ScRTG1 gene was amplified from S. cerevisiae genomic DNA using the primer pair 5Ј-GACTCC-ATCTACGAAAGACTACCTTCATGAGCAGCATTCCAG-CTGGC-3Ј and 5Ј-CCCAAGCTTCTAGCTACCATTACCG-TACTCACC-3Ј. PCR products from these two reactions were purified and used as templates in the final PCR along with the primer pair 5Ј-CCGCTCGAGGATTTCTGCGATTGGA-GCG-3Ј and 5Ј-CCCAAGCTTCTAGCTACCATTACCGTA-CTCACC-3Ј (XhoI and HindIII sites are underlined). The PCR product P RTG ScRTG1 consisting of ScRTG1 downstream of PpRTG1 promoter was linearized using SalI and transformed into ⌬Pprtg1 by electroporation. Recombinant clones were selected by plating on YNBD His Ϫ plates, and expression of ScRTG1 in the recombinant clones was confirmed by RT-PCR.

Generation of P. pastoris GS-⌬PEPCK and GS-⌬GUT1 strains
P. pastoris ⌬PEPCK strain in which the coding region of PEPCK was replaced by a zeocin expression cassette was generated as follows. PEPCK promoter (Ϫ1000 bp) was amplified by PCR using the primer pair 5Ј-CTACATCGGAACCAATT-TTGAATTCGCCG-3Ј (Ϫ1000 to Ϫ971 bp of PEPCK) and 5Ј-CTATGGTGTGTGGGGGATCCGCATGTGGATAATA-ATCTATGAGAAGG-3Ј (962 to 985 bp of pGAPZA in reverse complement (in italics), Ϫ24 to Ϫ1 bp of PEPCK in reverse complement). In the second PCR, 1.2 kb of zeocin expression cassette was amplified by PCR using the primer pair 5Ј-CCTT-CTCATAGATTATTATCCACATGCGGATCCCCCACACA-CCATAG-3Ј (Ϫ24 to Ϫ1 bp of PEPCK (in italics), ϩ962 to ϩ985 bp of pGAPZA vector) and 5Ј-CAAATAATAT-TATCCTG TTGCAATGCTCACATGTTGGTCTCCAGC-3Ј (ϩ1675 to ϩ1698 bp in reverse complement of 3Ј flanking region of PEPCK [in italics], ϩ2137 to ϩ2159 bp in reverse complement of pGAPZA). In the third PCR, ϳ1 kb of the 3Ј-flanking region of PEPCK was amplified using the primer pair 5Ј-GCTGGAGACCAACATGTGAGCATTGCAACAGG-ATAATATTATTTG-3Ј (ϩ2137 to ϩ2159 bp of pGAPZA (in italics), ϩ1675 to ϩ1698 bp of 3Ј-flanking region of PEPCK) and 5Ј-CAATTCCATAAAGTATATCTTGCTGACCAGT-ATC-3Ј (ϩ2640 to ϩ2674 bp in the reverse complement of 3Ј-flanking region of PEPCK). All three PCR products were purified and used as templates in the final PCR along with the primer pair 5Ј-CTACATCGGAACCAATTTTGAATTCG-CCG-3Ј and 5Ј-CAATTCCATAAAGTATATCTTGCTGAC-CAGTATC-3Ј to obtain an ϳ3.2-kb product consisting of Zeocin expression cassette flanked by 1 kb of PEPCK promoter and 1 kb of 3Ј-flanking region of PEPCK that was transformed into P. pastoris GS115 strain. Zeocin-resistant colonies were selected, and deletion of PEPCK was confirmed by the absence of PEPCK gene in the genomic DNA using PCR.
P. pastoris GS-⌬GUT1 strain in which 1 kb from the 1.8-kb GUT1 coding region was replaced by a zeocin expression cassette was generated as follows. First, ϳ1 kb of GUT1 promoter was amplified by PCR using the primer pair 5Ј-CCGCTCGAG-GGGAAACTAACAATAGTGTTGCCTTG-3Ј (Ϫ973 to Ϫ947 bp of GUT1 with XhoI site (underlined)) and 5Ј-CTATGGTG-TGTGGGGGATCCGCATATAGTAGATATATCTGTGGT-ATAG-3Ј (962 to 985 bp of pGAPZA in reverse complement (in italics), Ϫ25 to Ϫ1 bp of GUT1 in reverse complement). The 1.2-kb Zeocin expression cassette was amplified by PCR using

Post-transcriptional regulation by Rtg1p
the primer pair 5Ј-CTATACCACAGATATATCTACTATAT-GCGGATCCCCCACACACCATAG-3Ј (Ϫ25 to Ϫ1 bp of GUT1 promoter (in italics), ϩ962 to ϩ985 bp of pGAPZA) and 5Ј-CTGTCCAACCAAAGAGGCAGATTGGTGCTCACATG-TTGGTCTCCAGCTTGC-3Ј (ϩ966 to ϩ991 bp in reverse complement of 3Ј-coding region of GUT1 (in italics), ϩ2131 to ϩ2159 bp in reverse complement of pGAPZA). In the third PCR, ϳ800 bp of the 3Ј-coding region of GUT1 was amplified using the primer pair 5Ј-GCAAGCTGGAGACCAACATGTG-AGCACCAATCTGCCTCTTTGGTTGGACAG-3Ј (ϩ2131 to ϩ2159 bp of pGAPZA (in italics), ϩ966 to ϩ991 bp of coding region of GUT1) and 5Ј-CCCAAGCTTAGCAGTGTCCTTA-AGC-3Ј (ϩ1882 to ϩ1897 bp in the reverse complement of 3Ј-coding region of GUT1 with HindIII site (underlined)). All three PCR products were purified and used as templates in the final PCR along with the primer pair 5Ј-CCGCTCGAGGGG-AAACTAACAATAGTGTTGCCTTG-3Ј and 5Ј-CCCAAGC-TTAGCAGTGTCCTTAAGC-3Ј to obtain an ϳ3-kb product consisting of Zeocin expression cassette flanked by 1 kb of GUT1 promoter and ϳ800 bp of 3Ј-flanking region of GUT1 that was transformed into GS115. Zeocin-resistant colonies were selected, and deletion of GUT1 was confirmed by the absence of 1 kb of GUT1 gene in genomic DNA using PCR. Expression cassettes comprising genes encoding PEPCK and GUT1 along with ϳ1 kb of their promoters were cloned into pIB3 vector (Addgene) and expressed in P. pastoris GS115 and P. pastoris ⌬rtg1 (⌬Pprtg1) as Myc-tagged proteins. AOXI was cloned with ϳ1 kb of its promoter in pGAPBA vector as a Myctagged protein, whereas GDH2 was cloned with 545 bp of its promoter in pIB3 vector as a His-tagged protein. PEPCK and GUT1 were amplified from GS115 genomic DNA using primer pairs 5Ј-GGGGTACCCACCCACCCTCAAGTGC-3Ј and 5Ј-CCCAAGCTTCTACAGGTCTTCTTCAGAGATCAGTTT-CTGTTCCAACTGAGGGCCGGCCTG-3Ј (KpnI and HindIII sites are underlined) and 5Ј-CCGCTCGAGGGGAAACTAA-CAATAGTGTTGCCTTG-3Ј and 5Ј-CCCAAGCTTCTACA-GGTCTTCTTCAGAGATCAGTTTCTGTTCAGCAGTGT-CCTTAAGC-3Ј (XhoI and HindIII sites are underlined), respectively. GDH2 was amplified from GS115 genomic DNA using primer pairs 5Ј-CCGGAATTCCTCTCATGTTCGGA-TAATTCCAGCGGCTTTC-3Ј and 5Ј-CCGCTCGAGCT-AATGATGATGATGATGATGCAATCCCCGAGACTTG-TAC-3Ј (EcoRI and XhoI sites are underlined). PCR products were cloned into pIB3 vector and transformed into E. coli DH5␣ competent cells. Recombinant plasmids containing P PEPCK PEPCK-Myc and P GUT1 GUT1-Myc were linearized using BsrGI and SalI, respectively, and transformed into GS115 and ⌬Pprtg1 by electroporation. Recombinant plasmid containing P GDH2 GDH2-His was linearized using StuI and transformed by electroporation into GS115 and ⌬Pprtg1. Recombinant clones were selected by plating on YNBD His Ϫ plates, and clones expressing Myc-tagged PEPCK and GUT1 and Histagged GDH2 were confirmed by Western blotting using anti-Myc and anti-His antibodies.

Generation of P. pastoris GS-GDH2
AOXI was amplified from GS115 genomic DNA by the primer pair 5Ј-CCGCTCGAGATGTTGGTATTGTGAAAT-AGACGCAGATCG-3Ј and 5Ј-ATATGCGGCCGCTGAATC-TAGCAAGACCGGTCTTCTCG-3Ј (XhoI and NotI restriction sites are underlined). PCR products were cloned into respective sites of pGAPBA vector and transformed into E. coli DH5␣ competent cells. The recombinant plasmid was linearized with AvrII and transformed into GS115 and ⌬Pprtg1 by electroporation. Recombinants expressing Myc-tagged Table 3 List of yeast strains used in this study