Nuclear export of tRNA is an essential eukaryotic function, yet the one known yeast tRNA nuclear exporter, Los1, is nonessential. Moreover recent studies have shown that tRNAs can move retrograde from the cytosol to the nucleus by an undefined process. Therefore, additional gene products involved in tRNA nucleus–cytosol dynamics have yet to be identified. Synthetic genetic array (SGA) analysis was employed to identify proteins involved in Los1-independent tRNA transport and in regulating tRNA nucleus–cytosol distribution. These studies uncovered synthetic interactions between los1Δ and pho88Δ involved in inorganic phopshate uptake. Further analysis revealed that inorganic phosphate deprivation causes transient, temperature-dependent nuclear accumulation of mature cytoplasmic tRNA within nuclei via a Mtr10- and retrograde-dependent pathway, providing a novel connection between tRNA subcellular dynamics and phosphate availability.
TRANSPORT of tRNA from its nuclear site of biogenesis to the cytosol is an essential eukaryotic process as tRNA must be available for translation, which occurs in the cytosol. Intracellular tRNA movement in the budding yeast Saccharomyces cerevisiae involves several steps. First, end-matured pre-tRNAs are transported from nucleus to cytosol (“primary tRNA nuclear export”). Once in the cytosol, pre-tRNAs containing introns are spliced by the splicing endonuclease complex located on the cytosolic surface of the mitochondrial outer membrane (Yoshihisa et al. 2003). Numerous nucleoside modification steps also occur in the nucleus and cytosol (Hopper and Phizicky 2003). Mature cytosolic tRNA are then available for translation. Recent work demonstrated that mature tRNA can move from cytosol to nucleus (Shaheen and Hopper 2005; Takano et al. 2005) and vice versa (“re-export”; Whitney et al. 2007). However, many of the gene products involved in the tRNA nucleus-to-cytosol and cytosol-to-nucleus transport pathways have yet to be defined.
Transport of many macromolecules between the nucleus and the cytosol requires the small GTPase, Ran, and its protein-binding partners, the β-importins. Ran is maintained in the GTP-bound state in nuclei by Ran guanosine exchange factor (RanGEF), which is encoded in yeast by PRP20 and resides in the nucleus. β-Importin family members that export cargo from the nucleus to the cytosol bind their cargo only in the presence of RanGTP and then mediate interactions with the nuclear pore complexes to allow the entire complex to move to the cytosol. Once in the cytosol, RanGTP interacts with the Ran GTPase activating protein (RanGAP), encoded by RNA1, which activates GTP hydrolysis, leading to conformational changes that cause the heterotrimeric complexes to dissociate, releasing the cargo in the cytosol (Corbett et al. 1995; Gorlich and Kutay 1999; Weis 2003). For tRNA nuclear export, Los1, a member of the β-importin family, binds tRNA in a RanGTP-dependent manner (Hellmuth et al. 1998). The mammalian homolog of Los1, Exportin-t (Xpo-t), has also been shown to bind end-processed tRNA, with or without introns, as part of a heterotrimeric complex with RanGTP (Arts et al. 1998a,b; Kutay et al. 1998).
Inhibiting the Los1-dependent tRNA transport pathway through a temperature-sensitive mutation of RNA1 (rna1-1) or mutation of LOS1 caused nuclear accumulation of tRNA (Sarkar and Hopper 1998). Similarly, blocking Xpo-t-mediated tRNA transport through microinjection of Xpo-t antibodies in Xenopus oocytes reduced nuclear export of tRNAPhe by 99% (Arts et al. 1998b). Blocking transport through mutation of RNA1, PRP20, LOS1, or some nucleopore genes also caused accumulation of intron-containing pre-tRNA (Hopper et al. 1978, 1980; Kadowaki et al. 1993; Sharma et al. 1996), which indicated that these proteins were involved in the transport of end-matured pre-tRNA to the cytosol.
Numerous mutations have been identified that affect tRNA nucleus–cytosol distribution without causing accumulation of intron-containing pre-tRNA. These include temperature-sensitive mutations of methionyl- (mes1-1), isoleucyl- (ils1-1), and tyrosyl- (tys1-1) aminoacyl-tRNA synthetases (Sarkar et al. 1999), 3′ tRNA CCA-nucleotidyl transferase (cca1-1; Grosshans et al. 2000; Feng and Hopper 2002), and the tef1Δ tef2-1 double mutation of the eukaryotic elongation factor 1A (Grosshans et al. 2000). Lack of intron-containing pre-tRNA accumulation indicates that tRNA nuclear accumulation observed in these mutants occurs via accrual of fully processed, mature tRNA from the cytosol, and not by inhibition of the primary tRNA nuclear export pathway that exports pre-tRNA to the cytosol.
One genetic approach for defining parallel pathways that accomplish essential cellular functions is the identification of new mutations that have synthetic interactions with known mutations that impair or block one of the pathways. As tRNA nuclear export is an essential process and LOS1 is a known nonessential nuclear exporter of tRNA (Hurt et al. 1987), this approach should be useful for the identification of genes that encode proteins participating in Los1-independent transport of end-matured pre-tRNA and mature tRNA and, possibly, negative regulators of tRNA retrograde transport.
Previous work employing standard synthetic lethal methodology identified several synthetic interactions with los1Δ. The genes identified affected a variety of cellular processes, including tRNA aminoacylation (Simos et al. 1996a), tRNA modification (Simos et al. 1996a), Pol III transcription (Simos et al. 1996b), and translation (Hellmuth et al. 1998; Grosshans et al. 2000). Synthetic genetic array (SGA) analysis (Tong et al. 2001) allows extensive, large-scale, systematic searches for genetic interactions. This approach was utilized with a collection of promoter replacement alleles of essential genes and los1Δ (Davierwala et al. 2005). Depletion of Taf3, a subunit of the Pol II transcription initiation complex, and depletion of Pop5, a subunit of both RNase MRP (pre-rRNA cleavage) and RNase P (pre-tRNA 5′-end processing), were found to have synthetic interactions with los1Δ (Davierwala et al. 2005). None of the previously reported studies have identified tRNA nuclear exporters that function in parallel to Los1 or regulators of tRNA nucleus–cytosol distribution.
Here, we describe the results of SGA screens of los1Δ using the yeast MATa deletion collection. Novel synthetic interactions were observed between los1Δ and snt309Δ, aro7Δ, rpl12aΔ, rpl13bΔ, lrp1Δ, gtr1Δ, gtr2Δ, or pho88Δ. Of these verified candidates, gtr1Δ, gtr2Δ, and pho88Δ caused tRNA nuclear accumulation. As gtr1Δ, gtr2Δ, and pho88Δ have been previously shown to influence inorganic phosphate (Pi) uptake (Bun-Ya et al. 1992; Yompakdee et al. 1996; Lagerstedt et al. 2005), we examined the affect of Pi starvation upon tRNA nucleus–cytosol distribution. Transient, temperature-dependent, tRNA nuclear accumulation was observed in wild-type cells grown on synthetic media lacking Pi for 1–2 hr at 23° or 30°. Northern and heterokaryon analyses of Pi-deprived cells and of pho88Δ cells indicated that the observed nuclear accumulation was due to accrual of mature tRNA, which entered the nucleus via retrograde transport. This work provides a novel connection between Pi availability and tRNA nucleus–cytosol distribution.
MATERIALS AND METHODS
Strains and media:
Yeast strains used for these studies are listed in Table 1. All plasmid transformations were made by lithium acetate (LiOAc) transformation (Schiestl and Gietz 1989). Deletions were accomplished by LiOAc transformation of the parent yeast strain with a PCR product cassette that replaced the endogenous gene. Sequences for the oligonucleotides used are listed in supplemental Table S1 at http://www.genetics.org/supplemental/. The los1∷natMX4 cassette was generated by PCR amplification using oligonucleotide primers RLH007A and RLH007B and a natR-MX4 template (p4339; Tong et al. 2001). LOS1KO1B was derived from the transformation of Y3656 with the los1∷natMX4 cassette. The deletion of LOS1 was verified using the primers RLH010A and RLH010B. The pho88∷natMX4 cassette was generated by PCR amplification using oligonucleotide primers RLH017A and RLH017B and a natR-MX4 template (Goldstein and McCusker 1999). 8MS88ΔN2C was derived from the transformation of MS739 with the pho88∷natMX4 cassette. BY88ΔF1 was derived from the transformation of BY4741 with the pho88∷natMX4 cassette. The pho88∷hphMX4 cassette was generated by PCR amplification using the oligonucleotide primers RLH017A and RLH017B and the hphR-MX4 template (Goldstein and McCusker 1999). BY88ΔD1 and BY88ΔG4 were derived from independent transformations of BY4741 with the pho88∷hphMX4 cassette. 8MS88ΔH2D was derived from the transformation of MS739 with the pho88∷hphMX4 cassette. YD05055 (los1∷kanMX) was transformed with YCpLOS1 (Hurt et al. 1987) and then with the pho88∷hphMX4 cassette to form strain los1Δ pho88Δ + YCpLOS1. The deletion of PHO88 was verified using primers RLH017C, RLH017D, and RLH017E. The strain los1Δ pho88Δ + YCpLOS1 was grown on nonselective media to allow plasmid loss from which los1Δ pho88Δ was derived.
Yeast strains were maintained on YEPD [yeast extract, trypticase peptone, dextrose, supplemented with adenine (0.04 g/liter), uracil (0.04 g/liter)] medium; SC (synthetic complete) defined medium; Difco yeast nitrogen base without amino acids (6.7 g/liter; Becton Dickinson, Sparks, MD) supplemented with amino acids, dextrose (2%), adenine (0.04 g/liter), and uracil (0.04 g/liter). SC–Pi liquid medium is synthetic defined medium lacking K2PO4 [the source of Pi in SC; yeast nitrogen base lacking amino acids and K2PO4 (5.7 g/liter; ForMedium, Norwich, UK) supplemented with amino acids, uracil, and 1.0 g/liter KCl]. Solid YEPD medium supplemented with 2 g/liter KH2PO4 (Fisher) was utilized for the “wild-type” heterokaryon fluorescence in situ hybridization (FISH) analysis. To select for natMX cassette integration, the strains were grown on YEPD + clonNAT (100 mg/liter; Werner BioAgents, Jena, Germany) solid medium. To select for hphMX cassette integration, the strains were grown on YEPD + hygromycin B (300 mg/liter; Calbiochem, La Jolla, CA) solid medium.
SGA was performed and scored as previously described (Tong et al. 2001) using LOS1KO1B as the bait strain. The final replica plates were grown at 22° or 37°. Tetrad dissection was done by standard procedures.
Serial dilutions were made of the indicated yeast cultures. Aliquots (5 μl) of each dilution were spotted onto each plate of the indicated solid medium. The plates were incubated for 2–3 days at the indicated temperature. Images of the plates were captured using Eagle Eye II (Stratagene, La Jolla, CA), digital camera, or scanner (ScanMaker 8700 by Microtek, Carson, CA). Adobe Photoshop 5.0 was used for image assembly.
Fluorescence in situ hybridization:
FISH was performed as previously described (Sarkar and Hopper 1998) with the modifications detailed in Stanford et al. (2004). Heterokaryons for FISH were mated and grown as described in Shaheen and Hopper (2005) except the solid YEPD medium used for the “wild-type” mating was medium supplemented with 2 g/liter KH2PO4 (Fisher) to prevent premature induction of the PHO pathway, as low levels of PHO5 expression have been observed when cells are grown in YEPD medium (Yoshida et al. 1989a,b). Oligonucleotides used as probes are listed in supplemental Table S1 at http://www.genetics.org/supplemental/. Each slide contained positive and negative controls for tRNA nuclear accumulation. When adjectives describing the relative amount of tRNA nuclear accumulation are used, the probes were hybridized under the same conditions. All critical experiments were independently viewed and scored by at least two people, one of whom was unaware of the experimental details. A Nikon Microphot-FX microscope was used to observe fluorescence and a Sensys charge-coupled device camera (Photometrics, Tucson AZ) using QED software (QED Imaging, Pittsburgh) was used to capture the images. Adobe Photoshop 5.0 was used for image assembly.
Small RNAs were extracted from yeast cultures grown to densities similar to those used for FISH as previously described (Hopper et al. 1980). Samples (10 μg of RNA) were electrophoretically separated at 4° in 10% polyacrylamide gel containing 8 m urea and 1× TBE (0.09 m Tris, 0.09 m borate, 0.001 m ethylenediaminetetraacetic acid). RNAs were electrophoretically transferred onto Hybond N+ membranes (Amersham Pharmacia) using a Hoefer TE42 Transphor apparatus (Hoefer Scientific) filled with 1× TAE buffer (40 mm Tris, 20 mm acetate, 1 mm EDTA). Membranes were prehybridized at 37° in 4× SSC (1× is 0.15 m NaCl, 0.015 m sodium citrate), 2.5× Denhart's solution [1× is 0.02% Ficoll (type 400), 0.02% polyvinylpyrrolidone, 0.02% bovine serum albumin], 50 μg/ml single-stranded salmon sperm DNA, and 0.05% sodium dodecyl sulfate. Probe hybridization was at 37° in the same buffer with the addition of oligonucleotide probes (supplemental Table S1 at http://www.genetics.org/supplemental/) that were 5′ end labeled with [32P]ATP using T4 polynucleotide kinases (Promega, Madison, WI). After hybridization, membranes were UV crosslinked, washed in 2× SSC, and used to expose Kodak BioMax MS film (Eastman Kodak, Rochester, NY).
Acid phosphatase assay:
Acid phoshatase activity was determined as previously described (Byrne et al. 2004) except that yeast cultures were grown to early log phase and the volume of cells was increased to compensate for the reduced number of cells.
SGA analysis of los1Δ uncovers eight novel interactions:
SGA was performed twice using LOS1KO1B (los1∷natR) as bait. All of the incubation steps for the first screen were conducted at 22°, the standard temperature for SGA. The final replica plates of the second screen were incubated at 37°, a temperature that exaggerates the los1Δ phenotype in some strains (Hopper et al. 1980). The first screen provided 18 novel candidate interactions, whereas the second screen uncovered 118 candidates. As only six candidates appeared in both screens—deletions of ARC1 (YGL105W), LDB16 (YCL005W), LRP1 (YHR081W), RCY1 (YJL204C), WHI3 (YNL197C), and RPL13B (YMR142C)—the total number of candidates was 131 (supplemental Table S2 at http://www.genetics.org/supplemental/). Candidates of the first screen did not necessarily overlap with the candidates found in the second screen, possibly due to changes in the phenotype of individual deletion strains at different temperatures, manipulation errors, false positives, or false negatives. ARC1, a known synthetic lethal interaction (Simos et al. 1996b), was found in both screens, indicating that the assay was functioning as predicted.
Tetrad analysis and growth assays of selected haploid progeny were utilized to detect and verify synthetic lethal, slow, or enhanced growth interactions. Tetrad analysis of 97 candidate interactions revealed eight bona fide los1Δ interactions (Table 2); only three of these appeared as candidates in both searches. Synthetic enhanced growth interaction was observed between los1Δ and snt309Δ (YPR101W). Synthetic slow-growth interactions were observed between los1Δ and aro7Δ (YPR060C), rpl12a (YEL054C), rpl13bΔ (YMR142C), lrp1Δ (YHR081W), gtr1Δ (YML121W), or gtr2Δ (YGR163W).
A synthetic lethal interaction between los1Δ and pho88Δ (YBR106W) was documented upon tetrad dissection at 23° by the absence of nonparental ditypes and the absence of los1Δ pho88Δ containing progeny in tetratype asci as predicted by two-gene synthetic lethality (Figure 1A). However, we were able construct a viable los1Δ pho88Δ strain via direct mutagenesis (see materials and methods). The los1Δ pho88Δ strain exhibited slower growth at 23° than strains containing either individual mutation (Figure 1B). The slow growth vs. synthetic lethal interactions observed between los1Δ and pho88Δ are addressed in the discussion.
SGA analysis uncovered genes influencing tRNA nucleus–cytosol distribution:
The subcellular distributions of tRNA between the nucleus and the cytosol of selected synthetic interactors were evaluated using fluorescence in situ hybridization (Sarkar and Hopper 1998). Gene deletions that had synthetic slow-growth phenotypes were expected to cause tRNA nuclear accumulation if the growth defect was due to impairment of tRNA export. Nuclear accumulation of both tRNAMet and tRNATyr, as indicated by the increased FITC signal that colocalizes with the DAPI-stained DNA, was observed in los1Δ (Figure 2A, column 2), pho88Δ (Figure 2A, column 5), gtr1Δ (supplemental Figure S1 at http://www.genetics.org/supplemental/ and data not shown), and gtr2Δ (supplemental Figure S1 at http://www.genetics.org/supplemental/ and data not shown) strains (Table 2). Several independent pho88Δ strains were generated, and all exhibited tRNA nuclear accumulation (Figure 2B, columns 2 and 4; also see supplemental Figure S1 at http://www.genetics.org/supplemental/). In contrast, arc1Δ cells exhibited nuclear accumulation of tRNAMet while the cellular distribution of tRNATyr was similar to wild-type cells (Figure 2A, column 3). Deletion of arc1Δ influencing the nucleus–cytosol distribution of tRNAMet, but not tRNATyr, is consistent with the function of Arc1 as a cofactor of the methionyl- and glutamyl-tRNA synthetases (Simos et al. 1996a). However, not all deletions that had synthetic slow-growth interactions with los1Δ caused tRNA nuclear accumulation. For example, wild-type, rpl13bΔ, and rpl12aΔ cells had a similar cellular distribution of tRNAMet and tRNATyr (Figure 2A, columns 1 and 4, and supplemental Figure S1 at http://www.genetics.org/supplemental/, respectively; Table 2).
Candidate deletions that caused tRNA nuclear accumulation had been implicated in uptake of Pi. Gtr1 and Gtr2 formed a small GTPase complex (Nakashima et al. 1999) that negatively regulated the Ran cycle (Nakashima et al. 1996) and was required for efficient Pi uptake in response to Pi starvation (Bun-Ya et al. 1992; Lagerstedt et al. 2005). Deletion of PHO88 caused impaired Pi uptake and slowed growth (Yompakdee et al. 1996). PHO88 was selected for further study because it had strong synthetic interactions with los1Δ, its deletion caused tRNA nuclear accumulation, and it has been shown to impact the regulation of the PHO pathway (Yompakdee et al. 1996).
Pi deprivation of wild-type cells caused tRNA nuclear accumulation:
We obtained a plasmid encoding an N-terminal GST-tagged Pho88 (Martzen et al. 1999) and found that GST-Pho88 was located in the endoplasmic reticulum (ER; data not shown). This location is consistent with the results of the genomewide protein localization study that employed C-terminal GFP tags (Huh et al. 2003) and the observed membrane association of Pho88 (Yompakdee et al. 1996). As Pho88 was located in the ER, it was unlikely to directly bind and transport tRNA from the nucleus to the cytosol. Given that deletion of PHO88 caused a defect in Pi uptake from media (Yompakdee et al. 1996), it was possible that the low intracellular Pi levels caused the tRNA nuclear accumulation. This possibility had precedent as amino acid starvation has been shown to cause tRNA nuclear accumulation (Shaheen and Hopper 2005). If pho88Δ cells redistributed tRNA in response to low intracellular Pi levels, then wild-type cells that cannot import Pi due to lack of substrate should also have tRNA nuclear accumulation.
To address this prediction, wild-type cells (strain BY4742) were grown on synthetic complete media lacking KH2PO4 (SC–Pi) for various lengths of time. This approach allowed a controlled starvation period and minimized the risk of suppressor appearance. The cellular distributions of tRNA were then determined by FISH. As expected, tRNAHis was evenly distributed in wild-type cells grown in fresh SC media for 2 hr (Figure 3A, column 1). An even distribution of tRNAHis was also visible in cells grown for 30 min in SC–Pi (data not shown). Cells grown at 23° in SC–Pi for 1 hr displayed nuclear accumulation of tRNAHis (Figure 3A, column 2, row 1) as did cells grown at 30° in SC–Pi for 1 hr (Figure 3A, column 2, row 3). Thus, cells redistributed tRNA in response to Pi removal.
Unexpectedly, tRNA nuclear accumulation in response to Pi deprivation demonstrated temperature- and time-dependent components. While FISH is primarily a qualitative technique, the tRNAHis nuclear accumulation in cells grown at 23° reproducibly appeared to be more intense than in cells grown at 30° (Figure 3A, columns 2 and 3, rows 1 and 3). In contrast, cells grown at 37° in SC–Pi did not exhibit nuclear accumulation of tRNAHis (Figure 3A, columns 1–4, row 5). When cells were grown at 23° or 30° for 2 hr in SC–Pi, tRNA nuclear accumulation appeared diminished in comparison to the 1-hr time point. Moreover, cells grown for 3.5 or 12 hr in SC–Pi had tRNAHis nucleus–cytosol distributions similar to cells grown in fresh SC. The redistribution of tRNATyr and tRNAMet in response to growth in SC–Pi mirrored the demonstrated redistribution of tRNAHis (data not shown). The data indicated that denying wild-type cells readily available sources of Pi caused temperature-dependent, transient tRNA nuclear accumulation, in contrast to pho88Δ cells, which constitutively accumulated tRNA in the nucleus.
Activation of the PHO pathway is a well-studied response to Pi deprivation (Carroll and O'Shea 2002). The subcellular location and activity of Pho4p, the master transcription factor of the Pi starvation response, is regulated through phosphorylation at four sites. Pho4 is cytosolic, inactive, and highly phosphorylated when cells are grown in media rich in Pi as the Pho80/Pho85 cyclin–CDK complex is fully active and maintains Pho4 in a highly phosphorlyated state (Carroll and O'Shea 2002). When cells are grown in low Pi medium, the Pho80/Pho85 cyclin–CDK complex is partially inhibited by Pho81, causing Pho4 to be only partially phosphorylated. In this partially phosphorylated state, Pho4 is nuclear and activates the transcription of genes such as Pho84, a high-affinity Pi transporter (50% of maximal expression). Partially phosphorylated Pho4 has a minimal effect upon the transcription of other phosphate-responsive genes, including the acid phosphatase (rAPase), encoded by PHO5, which is expressed at 10% of maximum levels (Springer et al. 2003). Full inhibition of Pho80/Pho85 by Pho81 in response to Pi deprivation causes Pho4 to become fully dephosphorylated and competent in activating transcription of all its target genes, including Pho5 (Springer et al. 2003).
To ascertain when tRNA nuclear accumulation occurs with respect to the PHO pathway-mediated gene expression, we determined the enzymatic activity of secreted Pho5. As the amount of the p-nitrophenylphosphate converted to p-nitrophenolate is proportional to the amount of Pho5 (Toh-e et al. 1973), determining production of p-nitrophenolate provides an indirect measurement of Pho5 expression. The rAPase activity was assayed (Byrne et al. 2004) using intact, early log-phase wild-type cells (strain BY4742) grown in SC–Pi, which duplicated the growth conditions of 23° cell cultures used for FISH analysis. Cells grown for 2 hr in SC–Pi had 145% of basal activity, whereas cells grown for 3.5 hr in SC–Pi had 278% of basal activity (Table 3). After 10 hr of growth in SC–Pi, cells had 1100% of basal activity (Table 3). The rAPase activity indicates that Pho5 has low levels of expression at the time points when tRNA nuclear accumulation occurs (1–2 hr of Pi deprivation).
Nuclear accumulation of tRNA appears to be an early response to Pi deprivation as it occurs slightly after Pho4p translocation from the cytosol to the nucleus (Kaffman et al. 1998) in contrast to the later, strong induction of Pho5p expression (Table 3). To address the possibility that the tRNA nuclear accumulation in response to Pi deprivation could be dependent on the PHO pathway, the subcellular tRNA distributions in Pi-deprived and replete pho4Δ or pho81Δ cells were determined. Unstarved pho4Δ cells grown at 23° in fresh SC have a slight accumulation of tRNATyr (Figure 3B, column 1). As pho4Δ and pho81Δ cells grown at 23° in SC–Pi for 1.5 hr also exhibited nuclear accumulation of tRNA (Figure 3B, columns 2 and data not shown), an intact PHO pathway is not required for tRNA nuclear accumulation induced by Pi deprivation. In addition, pho4Δ pho88Δ cells displayed tRNA nuclear accumulation that is similar to the accumulation within pho88Δ cells (Figure 3B, columns 4 and 3, respectively). Therefore, Pi deprivation-induced tRNA nuclear accumulation is an early, signal-mediated response to Pi deprivation that does not require the PHO pathway.
Mature tRNA accumulated within nuclei in response to Pi deprivation enter via the retrograde transport pathway:
The current model of tRNA nucleus–cytosol transport has two modes for tRNA nuclear egress: primary nuclear export of newly synthesized tRNAs and re-export of mature tRNA. Blocking the primary nuclear export of tRNAs by mutation of LOS1, or any other members of the Ran pathway, causes accumulation of intron-containing pre-tRNAs. If Pi deprivation or deletion of PHO88 caused a block in the primary tRNA nuclear export pathway, then accumulation of intron-containing pre-tRNAs should have occurred. To determine whether tRNA nuclear accumulation observed in response to Pi deprivation and in pho88Δ cells occurs due to an inhibition of the primary export pathway or the re-export pathway, tRNA from wild-type cells grown in SC–Pi and pho88Δ cells were analyzed by Northern blot analysis. Depriving wild-type cells of Pi did not cause accumulation of intron-containing pre-tRNAs (Figure 4A). The independently derived pho88Δ strains did not exhibit accumulation of intron-containing pre-tRNAs (Figure 4). Therefore, mature tRNA had accumulated within nuclei of pho88Δ cells and Pi-deprived wild-type cells.
Since Northern analysis indicated that the tRNA nuclear accumulation observed in response to Pi deprivation was not due to impaired export of newly synthesized tRNA, tRNA nuclear accumulation was likely to have occurred through the tRNA retrograde pathway, similar to amino-acid-starvation-induced tRNA redistribution (Shaheen and Hopper 2005). To verify that tRNA retrograde nuclear accumulation occurs during Pi deprivation and when Pho88 is absent, the distribution of tRNAGlu-D was determined in “wild-type” heterokaryons grown in SC–Pi and in heterokaryons without Pho88. Heterokaryons were generated by mating yeast cells, in which one of the mating types carries a plasmid-expressing tRNAGlu-D from Dictyostelium discoideum and a kar1-1 mutation (strains MS739 + tRNAGlu-D or 8MS88ΔH2D + tRNAGlu-D), with KAR1 yeast cells (strain BY4741 or BY88ΔD1). The kar1-1 mutation blocked nuclear fusion upon mating and resulted in cells with a shared cytosol and two separate nuclei, where only one nucleus carried the plasmid. If mature tRNA, in response to Pi deprivation, had entered nuclei through retrograde tRNA transport, then all nuclei in a heterokaryon would have accumulated tRNAGlu-D when grown in SC–Pi. If it is independent of retrograde tRNA transport, then only nuclei that carried the plasmid-expressing tRNAGlu-D should have accumulated tRNAGlu-D when grown in SC–Pi.
Heterokaryons grown at 23° in SC–Pi for 1 or 2 hr, or in fresh SC for 2 hr, were subjected to FISH analysis. To quantify the data, heterokaryons were scored by determining if all, some, or none of the nuclei accumulate tRNAGlu-D. Nearly every heterokaryon grown in SC had an even distribution of tRNAGlu-D throughout the cell (Figure 5A, column 1; Table 4). Of the heterokaryons grown in SC–Pi for 1 hr, 83.5% accumulated tRNAGlu-D in all nuclei (Figure 5A, column 2; Table 4), 14.4% accumulated tRNAGlu-D in some nuclei, and 2.1% had no nuclear accumulation of tRNAGlu-D. The majority of heterokaryons grown in SC–Pi for 2 hr still accumulated tRNAGlu-D in all nuclei (72.7%; Figure 5A, column 3); however, as previously observed in haploid cells (Figure 3), the accumulation was not as prominent as the 1-hr time point and there was an increase in cells that have no nuclear accumulation (17.2 vs. 2.1%). The reduction of the total number of cells exhibiting tRNA nuclear accumulation after 2 hr of growth in SC–Pi was expected because the tRNA nucleus–cytosol distribution of wild-type cells returned to an even distribution between 2 and 3.5 hr of growth in SC–Pi. Since tRNAGlu-D accumulates in all of the nuclei of the majority of Pi-deprived “wild-type” heterokaryons, similar to amino-acid-deprived “wild-type” heterokaryons (Shaheen and Hopper 2005), the tRNA that accumulated in response to Pi deprivation entered the nucleus via retrograde tRNA transport of cytosolic tRNA.
The anticipated cause of tRNA nuclear accumulation within pho88Δ cells was low intracellular levels of Pi caused by defective Pi uptake. Therefore, the tRNA accumulated within the nuclei of pho88Δ cells was also expected to have entered the nucleus through retrograde transport. Heterokaryons homozygous for pho88Δ (“88Δ”) were subjected to FISH analysis. Consistent with our prediction, 95.2% of “88Δ” heterokaryons accumulated tRNAGlu-D in all nuclei (Figure 5A, column 5; Table 4). The data indicate that the tRNA accumulated within the nucleus of pho88Δ cells entered via retrograde tRNA transport.
Amino-acid-starvation-induced tRNA nuclear accumulation requires the β-importin, Mtr10, as mtr10Δ cells exhibit an even distribution of tRNA when grown in SC lacking amino acids (Shaheen and Hopper 2005). If Mtr10 is necessary for Pi-deprivation-induced tRNA nuclear accumulation, then Pi-deprived mtr10Δ cells will exhibit an even distribution of tRNA. When grown in SC–Pi at 23° for 1.5 hr, mtr10Δ cells showed the same tRNA nucleus–cytosol distribution as mtr10Δ cells grown in SC (Figure 5B, columns 2 and 1, respectively). The data indicated that Pi-deprivation-induced tRNA nuclear accumulation was Mtr10 dependent and therefore occurred by a mechanism shared with amino-acid-starvation-induced tRNA nuclear accumulation.
We undertook these studies to systematically query nonessential yeast genes for synthetic interactions with los1Δ to learn more about tRNA subcellular dynamics and we discovered a role for the availability of Pi in this process. SGA technology uncovered 131 candidate interactions with los1Δ. While 6 candidates were identified in both the 22° and the 37° assays, only 3 (lrp1Δ, rpl13bΔ, arc1Δ) of these had bona fide interactions as determined by tetrad analysis and growth assays. Of the 97 candidates evaluated by tetrad dissection, eight novel interactions were uncovered (Table 2). The SGA false-positive rate for los1Δ is 90.7% compared to the 25–50% success rate reported for other baits (Tong et al. 2004). The false-positive rate of the candidates that appeared in both screens (3 of 6) is 50%. This indicates that many false positives could have been eliminated by repeating the assays at both 22° and 37°. However, we note that PHO88 would not have been analyzed if SGA had been performed only at 37°. Since we previously noted that amino acid starvation causes tRNA nuclear accumulation (Shaheen and Hopper 2005), one may have expected to uncover SHR3, a likely equivalent to Pho88 for amino acid permeases, as a candidate. However, deletion of SHR3 was inviable by large-scale deletion (Winzeler et al. 1999) and was not included in the assay.
The eight novel synthetic interactions uncovered in this study affect amino acid synthesis, translation, mRNA and tRNA processing, and Pi uptake. The variety of cellular processes affected by the identified deletions and the data indicating that some of the identified deletions do not affect tRNA subcellular dynamics, raises the question as to why these synthetic interactions with los1Δ occur. RPL12A and RPL13B encode ribosomal proteins. The synthetic interactions between los1Δ and mutations of protein synthesis machinery, including tef2Δ and gcd11 (Hellmuth et al. 1998; Grosshans et al. 2000), are likely the result of the cumulative impairment of translation by reduced availability of cytoplasmic tRNA (los1Δ) and inefficient ribosomes. Aro7 is required for synthesis of tyrosine and phenylalanine. Deletion of ARO7 affects amino acid synthesis, and reduced amino acid availability may also affect the rate of protein synthesis. The genetic interaction between lrp1Δ and los1Δ may be caused by multiple defects in cellular RNA levels because previous studies uncovered synthetic lethal interactions between lrp1Δ and a variety of genes encoding defective tRNA, mRNA, and rRNA processing activities (Hieronymus et al. 2004; Davierwala et al. 2005). Our studies uncovered a single synthetic-enhanced growth interaction between snt309Δ and los1Δ. Since Snt309 is a component of spliceosomes (Chen et al. 1999), it is unclear why snt309Δ and los1Δ would have enhanced genetic interactions.
Different synthetic interactions between los1Δ and pho88Δ were observed when SGA or direct mutagenesis was employed to generate double mutants. The cold-sensitive nature of the synthetic slow-growth interaction observed in los1Δ pho88Δ cells may explain why PHO88 was found as a SGA candidate when the array was performed at 22° but not at 37°. Tetrad dissection was performed at 23° on YEPD medium, which has low levels of Pi (Yoshida et al. 1989b). This may have contributed to the absence of los1Δ pho88Δ progeny. If low temperature and/or medium prevented the appearance of los1Δ pho88Δ colonies, then tetrad dissection of pho88Δ by los1Δ crosses on high-Pi medium at 30° may produce los1Δ pho88Δ colonies. Tetrad dissection also involves a starvation period to induce meiosis, whereas direct mutagenesis occurs on rich media. Thus, it is also possible that los1Δ pho88Δ spores are unable to germinate after this starvation.
Deletion of PHO88, GTR1, or GTR2 caused nuclear accumulation of mature tRNA. Simulating the Pi-uptake defect observed in pho88Δ cells by growing wild-type cells in SC–Pi caused transient, temperature-dependent tRNA nuclear accumulation without causing accumulation of intron-containing tRNAs. The data indicate that tRNA nuclear accumulation was not due to impaired export of newly synthesized tRNAs and that Pi-deprived wild-type and pho88Δ cells accumulated mature tRNA only in nuclei that entered through tRNA retrograde transport. Consistent with amino-acid-starvation-induced tRNA redistribution, the data also demonstrated that the tRNA nuclear accumulation observed in Pi-deprived cells was Mtr10 dependent.
Although we have demonstrated that the mature tRNA accumulating within nuclei of pho88Δ and Pi-deprived wild-type cells entered the nuclei via the retrograde transport pathway, the mechanism for the accumulation remains unclear. Accumulation could have occurred due to increased retrograde transport, blocked re-export, or simultaneous occurrence of both increased retrograde transport and blocked re-export. Mtr10 may function in a signal transduction pathway regulating retrograde transport and/or re-export or as part of a tRNA importing complex.
Nuclear accumulation of tRNA that occurred due to PHO88 deletion largely paralleled the accumulation that occurred in Pi-deprived wild-type cells. However, pho88Δ cells constitutively accumulated tRNA within nuclei, whereas the accumulation observed during Pi deprivation was transient in nature. This difference may be explained by the differences in how the cells were impaired for Pi uptake. In the absence of Pho88, cells were able to constitutively acquire very low levels of Pi, sufficient for slow growth (Yompakdee et al. 1996). In contrast, Pi deprivation caused wild-type cells to undergo a two-phase response. During the first phase, when tRNA nuclear accumulation occurred, polyphosphate is mobilized (Mouillon and Persson 2006), Pho84 is highly expressed, and Pho5 is slightly expressed. The second phase, when tRNA returned to normal distribution, was marked by increased expression of Pho5. Similarly, recent data demonstrated that acute glucose starvation causes tRNA nuclear accumulation, whereas an even distribution of tRNA occurred during prolonged starvation (Whitney et al. 2007). This indicates that cells may have acclimatized to prolonged starvation and redistributed previously accumulated tRNA via re-export. Cells with PHO88 deletions may never reach the extremely low levels of intracellular Pi required for this transition. As Pi-deprivation-induced tRNA nuclear accumulation was independent of Pho4 and Pho81, the signaling pathway(s) controlling tRNA distribution during the Pi deprivation time course have yet to be determined.
Although a goal of this work was to uncover tRNA export pathways that function in parallel to Los1 to export newly synthesized tRNA, such a pathway was not identified. This unidentified exporter(s) may have been missed by SGA or not included in the deletion collection, or it may be encoded by an essential gene that has yet to be recognized as a tRNA transporter. Instead, we successfully uncovered novel connections between los1Δ and mRNA processing (lrp1Δ) and stable RNA processing (snt309Δ) and additional connections to the ribosome (rpl13bΔ, rpl12aΔ) and amino acid synthesis (aro7Δ). The novel synthetic interactions between los1Δ and gtr1Δ, gtr2Δ, or pho88Δ identified in this work provided a clear connection between tRNA subcellular dynamics and phosphate availability. Further studies are required to understand the signal transduction and the mechanism by which Pi levels regulate tRNA nucleus–cytosol distribution.
We thank H. Shaheen, M. Whitney, A. Murthi, and K. Stauffer for valuable scientific interactions. This work was supported by a fellowship to R.L.H. from the American Heart Association, by a grant to A.K.H. from the National Institutes of Health, and by grants to C.B. from the Canadian Institute of Health Research, Genome Canada, and Genome Ontario.
Communicating editor: S. Sandmeyer
- Received January 21, 2007.
- Accepted March 12, 2007.
- Copyright © 2007 by the Genetics Society of America