Rapamycin Blocks Sexual Development in Fission Yeast through Inhibition of the Cellular Function of an FKBP12 Homolog*

FKBP12 is a ubiquitous and a highly conserved prolyl isomerase that binds the immunosuppressive drugs FK506 and rapamycin. Members of the FKBP12 family have been implicated in many processes that include intracellular protein folding, transport, and assembly. In the budding yeast Saccharomyces cerevisiae and in human T cells, rapamycin forms a complex with FKBP12 that inhibits cell cycle progression by inhibition of the TOR kinases. We reported previously that rapamycin does not inhibit the vegetative growth of the fission yeast Schizosaccharomyces pombe; however, it specifically inhibits its sexual development. Here we show that disruption of the S. pombe FKBP12 homolog,fkh1 +, at its chromosomal locus results in a mating-deficient phenotype that is highly similar to that obtained by treatment of wild type cells with rapamycin. A screen forfkh1 mutants that can confer rapamycin resistance identified five amino acids in Fkh1 that are critical for the effect of rapamycin in S. pombe. All five amino acids are located in the putative rapamycin binding pocket. Together, our findings indicate that Fkh1 has an important role in sexual development and serves as the target for rapamycin action in S. pombe.

FKBP12 is a ubiquitous and a highly conserved prolyl isomerase that binds the immunosuppressive drugs FK506 and rapamycin. Members of the FKBP12 family have been implicated in many processes that include intracellular protein folding, transport, and assembly. In the budding yeast Saccharomyces cerevisiae and in human T cells, rapamycin forms a complex with FKBP12 that inhibits cell cycle progression by inhibition of the TOR kinases. We reported previously that rapamycin does not inhibit the vegetative growth of the fission yeast Schizosaccharomyces pombe; however, it specifically inhibits its sexual development. Here we show that disruption of the S. pombe FKBP12 homolog, fkh1 ؉ , at its chromosomal locus results in a mating-deficient phenotype that is highly similar to that obtained by treatment of wild type cells with rapamycin. A screen for fkh1 mutants that can confer rapamycin resistance identified five amino acids in Fkh1 that are critical for the effect of rapamycin in S. pombe. All five amino acids are located in the putative rapamycin binding pocket. Together, our findings indicate that Fkh1 has an important role in sexual development and serves as the target for rapamycin action in S. pombe.
In addition to their immunosuppressive activity, CsA, FK506, and rapamycin have side effects that may stem, at least in part, from inhibition of the physiological function of the immunophilins. For example, in mammals, FKBP12 functions as a subunit of ryanodine calcium release channels and is thought to modulate intracellular Ca 2ϩ levels in the heart (11)(12)(13). Mice deficient in FKBP12 show severe heart defects associated with loss of function of cardiac ryanodine receptors (14). Similarly, treatment with high doses of FK506 can lead to severe heart failure (15).
Although FKBP12 and cyclophilin-18 are unrelated in primary sequence, both classes of immunophilins exhibit a peptidyl prolyl-cis/trans-isomerization (PPIase) activity that accelerates a rate-limiting step in the folding of peptide and protein substrates in vitro (3, 16 -18). The PPIase activity of the immunophilins is inhibited upon binding to their specific immunosuppressive drugs, suggesting an overlap between the PPIase-active site and the drug-binding site. According to atomic structure analyses of human cyclophilin-18 and FKBP12, both proteins contain a deep hydrophobic binding pocket (19 -21). These pocket structures accommodate the specific immunosuppressive-acting ligands and model tetrapeptides used as pseudosubstrates.
The cellular functions of the immunophilins, as well as the relevance of the PPIase activity within the cellular environment, is not well understood. However, some of the important natural substrates of the immunophilins are now known (reviewed in Ref. 22). For example, the human cyclophilin-18, CyPA, binds the Gag polyprotein of the human immunodeficiency virus, type 1, virion (23)(24)(25). The human FKBP12 protein is physically associated with calcium release channels (11)(12)(13)26), the type I tumor growth factor, transforming growth factor-␤, receptor (27)(28)(29)(30), and the transcription factor YY1 (31).
Genetic studies in the budding yeast Saccharomyces cerevisiae have played a critical role in elucidating the mode of action of the immunosuppressive drugs in higher eukaryotes (reviewed in Refs. 32 and 33). Similar to the effect of rapamycin in T cells and certain non-lymphoid cells, rapamycin treatment of S. cerevisiae cells results in a G 1 cell cycle arrest (34). S. cerevisiae cells contain one FKBP12 homolog, named FPR1 (34), also known as RBP1 (35). Disruption of FPR1 results in slightly slowly growing but viable cells that are completely resistant to rapamycin. This phenotype indicated that FPR1 is a nonessential gene and is the main mediator of the effect of rapamycin (34,35). Later it was shown that Fpr1p forms a complex with rapamycin that binds and inhibits the functions of the TOR1 and TOR2 gene products in cell cycle progression (34, 36 -39). Several proteins that interact physically with Fpr1p in the absence of rapamycin have been identified, and it has been suggested that their activity may be regulated by the interaction with Fpr1p. These include calcineurin (40), the biosynthetic enzyme aspartokinase (41), the high mobility group HMG 1/2 proteins (42), and the transcription factor homolog FAP1 (43).
We reported previously (44) that rapamycin does not affect vegetative growth in the fission yeast, Schizosaccharomyces pombe, but severely inhibits its sexual development pathway. S. pombe cells are induced to enter the sexual development pathway under starvation conditions (45). If the sexual development pathway is chosen, cells of opposite mating type conjugate to form diploid zygotes that immediately undergo meiosis and sporulation (45). Rapamycin strongly inhibited sexual development at an early stage, before mating had occurred, but did not affect entrance into stationary phase (44). More recently, we reported that S. pombe contains two TOR homologs, tor1 ϩ and tor2 ϩ (46). tor2 ϩ is an essential gene of as yet unknown function. tor1 ϩ is required under starvation and a variety of other stress conditions that include osmotic and oxidative stresses. Interestingly, none of the studied functions of the S. pombe TOR homologs appear to be inhibited by rapamycin (46). To understand further the response of S. pombe to rapamycin, we isolated and characterized the S. pombe FKBP12 homolog. We found one FKBP12 homolog and named it fkh1 ϩ . Disruption of fkh1 ϩ results in a mating deficient phenotype that is highly similar to that of rapamycin-treated cells. We identified, using a genetic screen, amino acid substitutions in Fkh1 that confer rapamycin resistance. These substitutions occur in conserved residues of FKBP12 and are potentially involved in rapamycin binding. Our analyses of the fkh1 null and rapamycin-resistant mutants suggest that rapamycin exerts its effect on sexual development in S. pombe by inhibiting the function of Fkh1.

EXPERIMENTAL PROCEDURES
Yeast Strains, Media, and Yeast Techniques-Yeast strains used in this paper are described in Table I. Media used are based on those described (47). EMM-N contains no nitrogen; EMM lowG contains 0.1% glucose. Transformation of S. pombe cells was performed by electroporation (48). Rapamycin was added to a final concentration of 0.2 g/ml in liquid or agar-containing media, unless otherwise indicated. An equal volume of the drug vehicle solution (1:1 Me 2 SO:methanol) was used as a control in all experiments. Assays for mating or sporulation efficiency were carried out as follows. Cells were grown at 30°C in EMM medium to the density of ϳ5 ϫ 10 6 -1 ϫ 10 7 cell/ml. The cultures were then washed three times with double distilled water and 5 l containing 5 ϫ 10 6 cells were spotted on EMM, EMM-N, EMM-lowG, or ME medium (see Ref. 44 for detailed description of medium composition). After 3 days of incubation at 30°C, a toothpick was used to pick some of the cells from the center of each patch, and the cells were briefly sonicated and examined microscopically. The percentage of mating was calculated by dividing the number of zygotes, asci, and free spores by the number of total cells. The percentage of sporulation was calculated by dividing the number of asci and free spores by the number of total cells. One zygote or one ascus was counted as two cells and one spore was counted as half-cell. In each experiment 500 -1000 cells were counted. Cell viability after entry into stationary phase was determined as follows. Cells were grown in minimal medium (EMM) or rich medium (YE) at 30°C to confluence, and aliquots were sampled every 24 h. Cell viability was determined by the capacity of cells to form colonies.
FACS Analysis-Cells were stained with the DNA fluorochrome pro-pidium iodide and analyzed by a Becton Dickinson FACSort as described (49). Data were analyzed by Cell Quest software for Macintosh. Disruption of S. pombe fkh1 ϩ -A 1.7-kilobase pair fragment containing the entire fkh1 ϩ gene was amplified by PCR using a genomic S. pombe DNA preparation as a template and primers 101 (5Ј-GCTCA-GAATGATCGACATATACAAC) and 102 (5Ј-CAAACCAGCTACATAG-CACAG). The resulting PCR fragment was cloned into a pGEM-T vector (Promega) to give pGEMT-fkh1. This plasmid was cut with HindIII, within the fourth exon of fkh1 ϩ . The HindIII restriction site lies within the predicted active site and rapamycin binding pocket. The HindIII cut plasmid was ligated with a HindIII fragment containing ura4 ϩ , resulting in the plasmid pfkh1::ura4 ϩ . NotI and SacI were used to release the 3.5-kilobase pair fkh1::ura4 ϩ disruption fragments that were gel-purified and transformed into the homothallic strain TA16. Stable Ura4 ϩ haploids were selected and subjected to PCR analysis with primer 135 (5Ј-GTTATAAACATTGGTGTTGGAACAG) that is complementary to sequences within the ura4 ϩ gene and primer 136 (5Ј-GTTCGAATATAT TCGGTGCGCC) that is complementary to sequences of the fkh1 ϩ locus that are 100 bp downstream of the 3Ј end of the disruption construct. The resultant PCR fragment of 1200 bp confirmed that the disruption cassette integrated into the fkh1 ϩ locus. In addition, we used primer 136 in combination with primer 103 that is complementary to sequences that are 100 bp upstream of the 5Ј end of the disruption construct. The amplification of a single PCR product of the size of 3700 bp is consistent with a single site insertion of the disruption cassette. We analyzed the phenotype of two independently isolated ⌬fkh1 clones and demonstrated that re-introduction of the wild type fkh1 ϩ gene rescued the defects observed in these clones.
Cloning of fkh1 ϩ cDNA-The cDNA of fkh1 ϩ was isolated by PCR amplification from an S. pombe cDNA library (50) with the primers 92 (5Ј-GGAATTCCATATGGGTGTCGAAAAGCAAGTTATTTC; underlined is the NdeI restriction site) and 81 (5Ј-TGACCAATGGCGAA-GAAGTCC). The PCR product of 486 bp was cloned under the control of the thiamine-repressible nmt1 promoter in pREP1 (50). In all experiments cells were grown under derepression conditions (in the absence of thiamine) for full activity of the nmt1 promoter. The cDNA of fkh1 ϩ was also cloned into the S. cerevisiae expression vector pCM189 (51) using primers 92 and 81. In addition, the S. cerevisiae FKBP12 homolog, FPR1, was cloned into the pCM189 vector and pCM189Ј (a vector that differs from pCM189 only in that it contains LEU2 as a selective marker and not URA3) using primers 106 (5Ј-ATAAGAATG-CGGCCGCCGGATCCCGCTCGAGGTCG) and 110 (5Ј-ATAAGAATGC-GGCCGCCAATTAAGGCTCAGATACTTACC). The NotI sites in both primers are underlined. pCM189-fkh1 ϩ and pCM189-FPR1 were transformed into the S. cerevisiae strains JK9-3d (MAT␣ leu2-3,2-112 trp1-1 ura3-52 his4 rme1 HMLa) and JK9-3d␣ ade2 fpr1::ADE2 (52), the kind gift of J. Heitman, Duke University Medical Center. fkh1 ϩ cDNA was also isolated during a screen for S. pombe genes that could suppress the rapamycin-sensitive phenotype in S. cerevisiae. The S. cerevisiae strain, RS188N (MATa leu2-3,2-112 trp1-1 ura3-1 ade2-1 his3-11, 15 can1-100), was transformed with a S. pombe cDNA library constructed using a S. cerevisiae high copy number expression vector in which expression of inserts is regulated by the strong ADH1 promoter (53). Transformants were plated onto minimal medium plates containing 0.1 g/ml rapamycin at 30°C. 25 rapamycin-resistant colonies were isolated from over 10 5 transformants. Of these, 8 exhibited rapamycin resistance upon re-streak on rapamycin-containing plates, and in 4 the rapamycin resistance phenotype was dependent on the presence of the plasmid. 2 Sequence analysis revealed that one of these, pR22, encodes for fkh1 ϩ . This clone contained the entire open reading frame of fkh1 ϩ flanked by 22 and 150 bp at 5Ј and 3Ј ends of the open reading frame, respectively.
Western Blot Analysis of Fkh1-Total protein extracts were prepared from mid-log wild type (TA16), ⌬fkh1 (TA59), and wild type (TA16) cells transformed with pREP1-fkh1 ϩ , following the method described (47). Aliquots of whole cell extracts containing 40 g of protein were fractionated by SDS-polyacrylamide gels and transferred to membrane filters. The immobilized proteins were detected using the PerkinElmer Life Sciences ECL system. The Fkh1 proteins were detected with polyclonal antibodies raised against S. cerevisiae FKBP12, the kind gift of J. Heitman, Duke University Medical Center.
Isolation of Rapamycin-resistant fkh1 Mutants-fkh1 mutants were obtained by PCR-based mutagenesis. Conditions for PCR-based random mutagenesis of fkh1 ϩ were essentially as described (54). Briefly, 5 ng of The ligation product was used for PCR amplification with primers 189 (5Ј-GAATAAGT-CATCAGCGGTTGTTTC) and 190 (5Ј-TCATCCATGCGGCCAATCTT-GTCG). These DNA fragments containing the mutated fkh1 ϩ cDNA flanked by pREP1 sequences were co-transformed with pREP1 into the S. pombe strain TA77 (leu1-32 ura4-D18 fkh1::ura4 ϩ h 90 ). Transformants were plated on minimal medium and after 4 days of incubation at 30°C replica-plated to minimal medium with or without 0.2 g/ml rapamycin. After an additional 5 days of incubation at 30°C, the plates were exposed to iodine vapor. Iodine vapor is routinely used to detect sporulating colonies. Spores are darkly stained by iodine vapor, whereas vegetative cells remain unstained. In the presence of rapamycin (44) or in ⌬fkh1 colonies (this study), no dark staining is observed since the sexual development pathway is blocked prior to conjugation. Plasmid DNA was isolated from ⌬fkh1 transformants that stained dark in the presence of rapamycin and used for re-transformation of TA77 and transformation of bacterial cells for plasmid amplification. Plasmids that conferred rapamycin resistance phenotype upon re-transformation were further subjected to DNA sequence analysis.

RESULTS
Identification of the FKBP12 Homolog in the S. pombe Genome-Most of the S. pombe genome has been sequenced through the coordination of the Sanger Center, UK. Based on sequence comparisons, we identified one S. pombe FKBP12 homolog on chromosome II and named it fkh1 ϩ (for FKBP12 homolog). The open reading frame of fkh1 ϩ is interrupted by 4 introns of 182, 128, 105, and 48 base pairs. fkh1 ϩ encodes a putative 112-amino acid protein with a predicted mass of 12 kDa. We cloned the fkh1 ϩ cDNA by PCR amplification using a fission yeast cDNA library as a template (see "Experimental Procedures"). Sequence analysis confirmed that the 4 introns predicted in the genomic sequence are spliced out in the cDNA clone.
Analysis of the predicted amino acid sequence encoded by fkh1 ϩ reveals that this gene is very similar to its S. cerevisiae homolog, FPR1 (72% overall identity). The similarity between fkh1 ϩ and the human FKBP12 homolog is comparable to the similarity between FPR1 and the human FKBP12 (55% overall identity). fkh1 ϩ encodes all the amino acids required for rapamycin binding as predicted by the high resolution structure of the human FKBP12-rapamycin complex (Ref. 21 and see Fig. 5).
When Expressed in S. cerevisiae, fkh1 ϩ Functions Similarly, but Not Identically, to the S. cerevisiae FKBP12 Homolog-The S. cerevisiae FKBP12 protein, Fpr1p, binds to rapamycin. FKBP12-rapamycin complexes bind the TOR proteins and thus inhibit some of their functions (see Introduction). Since the S. pombe fkh1 ϩ gene shows a significant level of homology with FPR1, we examined whether fkh1 ϩ can replace FPR1 in mediating the effect of rapamycin in S. cerevisiae. To this goal, we expressed fkh1 ϩ cDNA in S. cerevisiae using ADH1 promoterdriven vector, pCM189 (51). Wild type and ⌬fpr1 S. cerevisiae cells were transformed with pCM189-fkh1 ϩ , and the transformants were streaked onto plates containing 0.08 g/ml rapamycin (Fig. 1). As described previously, the wild type S. cerevisiae cells did not form colonies in the presence of rapamycin, whereas ⌬fpr1 cells were completely resistant to the lethal effect of the drug (34) (Fig. 1). Expression of fkh1 ϩ in ⌬fpr1 cells restored rapamycin sensitivity (Fig. 1, plate 2), indicating that fkh1 ϩ , like FPR1, is capable of mediating the effect of rapamycin in S. cerevisiae cells. It is therefore most likely that the gene product of fkh1 ϩ forms a toxic complex with rapamycin that binds and inhibits the S. cerevisiae TOR proteins.
Unexpectedly, following a prolonged incubation, cells expressing pCM189-fkh1 ϩ exhibited slow growth in the presence of rapamycin, either in the genetic background of wild type or ⌬fpr1 cells (Fig. 1, plates 3 and 4). Thus, the expression of fkh1 ϩ under the strong ADH1 promoter can slightly increase rapamycin resistance in S. cerevisiae cells. Overexpression of FPR1 from the same expression vector did not exhibit such an effect (see Fig. 2A), consistent with previous findings (55).
We also screened an S. pombe cDNA library for genes that can confer rapamycin resistance in S. cerevisiae cells (see "Experimental Procedures"). Wild type S. cerevisiae was transformed with the S. pombe cDNA library, and the transformants were plated on rapamycin-containing plates. Sequence analysis of one of the isolated cDNA clones revealed that it encoded fkh1 ϩ . The weak rapamycin resistance phenotype conferred by overexpression of fkh1 ϩ was observed in several S. cerevisiae strains, including RS188N and JK9 -3d (see "Experimental Procedures" for full genotypes), indicating that this suppression activity is not strain-specific.
We examined the ability of fkh1 ϩ to suppress rapamycin sensitivity at different drug concentrations ranging from 10 to 150 ng/ml. S. cerevisiae wild type cells transformed with either pCM189-fkh1 ϩ or pCM189-FPR1 were streaked on rapamycincontaining plates, and their growth was monitored. Cells transformed with pCM189-fkh1 ϩ grew faster on 10 ng/ml rapamycin than on 100 ng/ml rapamycin ( Fig. 2A) and did not form colonies on 150 ng/ml rapamycin (data not shown). pCM189-FPR1 had no significant suppression activity at any drug concentration.
One possibility to explain the dosage-dependent suppression of fkh1 ϩ is that the gene product, Fkh1, forms a complex with rapamycin that does not inhibit the S. cerevisiae TOR proteins as efficiently as Fpr1p-rapamycin complexes. According to such a model, Fpr1-rapamycin complexes would compete with Fkh1rapamycin complexes. Thus overproduction of Fpr1p is expected to counteract the weak rapamycin resistance associated with fkh1 ϩ . To investigate this, wild type cells were co-transformed with pCM189Ј-FPR1 and pCM189-fkh1 ϩ , and the resulting transformants were streaked on rapamycin-containing plates. The results demonstrate that increased levels of FPR1 abolished the fkh1 ϩ -dependent rapamycin resistance (Fig. 2B). These findings support our suggestion that the slight decrease in rapamycin sensitivity in cells overexpressing Fkh1 results from reduced ability in inhibiting the TOR proteins.
Cells Disrupted for fkh1 ϩ Exhibit a Mating Deficient Pheno- type-To study the cellular function(s) of fkh1 ϩ and to examine its possible role in the response to rapamycin in S. pombe, we used the ura4 ϩ gene to construct a homothallic strain that carried a disruption of the fkh1 ϩ gene. Ura4 ϩ stable haploids were isolated, and PCR analysis confirmed that a single integration event had occurred at the fkh1 locus (see "Experimental Procedures" and data not shown). Consistently, no product of the fkh1 ϩ gene is observed by Western blot analysis (see Fig. 3B).
Analysis of two independently isolated ⌬fkh1 clones (fkh1::ura4 ϩ leu1-32 ura4-D18 ade6-M216 h 90 ) showed that the growth rate and cell morphology of ⌬fkh1 strains were indistinguishable from the parental wild type strain (data not shown). However, the ability of ⌬fkh1 strains to enter the sexual development pathway was greatly diminished. S. pombe cells are induced to undergo sexual development under starvation conditions, the first stage being the mating of two haploid cells to form the diploid zygote. Conventionally, either low nutrient medium such as ME medium, low nitrogen medium, or low glucose medium are used. We found that under any of these conditions, the mating efficiency of ⌬fkh1 strains was reduced by 10 -40-fold compared with wild type isogenic strains ( Fig. 3A and data not shown). The sterile phenotype of ⌬fkh1 was suppressed when we re-introduced fkh1 ϩ (Fig. 3A). Interestingly, introduction of FPR1 on an S. pombe expression vector also suppressed the mating defect of ⌬fkh1 (Fig. 3A). Thus, although FPR1 does not seem to have a role in the sexual development pathway in S. cerevisiae, it can fulfill the function of fkh1 ϩ when expressed in S. pombe.
We overexpressed fkh1 ϩ in S. pombe by placing it downstream of nmt1 promoter (see "Experimental Procedures"). Overexpression of fkh1 ϩ was verified by Western analysis (Fig.  3B). Overexpression of fkh1 ϩ in wild type cells (TA16) did not affect either growth or the efficiency of sexual development (Fig. 3A). Thus, whereas fkh1 ϩ is required for sexual development, its natural level is not a limiting factor for this process.
Nitrogen starvation is also a signal for diploid cells to enter meiosis, a later stage in the sexual development pathway (reviewed in Ref. 45). To examine whether fkh1 ϩ plays a role in meiosis/sporulation, we constructed a diploid strain homozygous to the disruption in fkh1 ϩ (TA94). These diploid cells were induced to undergo meiosis/sporulation under nitrogen starvation. Microscopic examination revealed that normal asci containing four spores were formed with the same efficiency (50%) as in wild type cells (TA07, 54%). Thus, although fkh1 ϩ is required for mating it is not required for meiosis/sporulation. When S. pombe cells enter stationary phase they become smaller and round and can maintain their viability over long periods (56). Some of the S. pombe mutants that are impaired in sexual development are also impaired in their ability to acquire normal stationary phase morphology and physiology. These include mutants of the cAMP-dependent pathway (57,58) and mutants of the Spc1-Wis1 stress-activated mitogenactivated protein kinase pathway (59 -61). In such mutants, the sterile phenotype may stem from an inability to sense or respond properly to starvation conditions. Recently, we reported (46) that null mutants of the S. pombe TOR homolog, tor1, are defective both in sexual development and entrance into stationary phase. In contrast, rapamycin-treated cells can enter stationary phase properly (44). Here we found that ⌬fkh1 cells arrested growth as relatively small cells and remained viable over a long period (see "Experimental Procedures" and data not shown). We therefore conclude that fkh1 ϩ is not required for entrance into stationary phase.
Finally, in some sterile mutants, the defect in mating is associated with an inability to arrest in G 1 in response to nitrogen starvation (46,59). We analyzed the DNA content of starved ⌬fkh1 cells and rapamycin-treated cells. The results shown in Fig. 3C demonstrate that neither ⌬fkh1 nor rapamycin-treated cells are defective in their ability to arrest in G 1 under nitrogen starvation conditions.
Our findings indicate that fkh1 ϩ is required specifically for an early stage of the sexual development pathway. Like treatment with rapamycin (44), ⌬fkh1 cells are specifically defective in their ability to undergo mating but can undergo meiosis/ sporulation. As in rapamycin-treated cells, ⌬fkh1 cells do not show other defects associated with abnormal response to starvation, such as entrance into stationary phase or arrest in G 1 in response to nitrogen deprivation. Taken together, the phenotype of ⌬fkh1 cells is extremely similar to that of rapamycintreated cells.
Isolation of Rapamycin-resistant fkh1 Mutants-The close similarity between the phenotypes of ⌬fkh1 cells and rapamycin-treated cells suggests that the direct target of rapamycin in  3. Phenotype of fkh1-disrupted cells. A, cell patches of wild type (WT) strain (TA16) and ⌬fkh1 strain (TA59) carrying no plasmids or transformed with pREP1-fkh1 ϩ , pREP1-FPR1, or vector only, were replica-plated onto ME plates (conditions that induce mating) in the presence or absence of rapamycin. After 3 days of incubation at 25°C, the efficiency of mating was determined microscopically. B, detection of the Fkh1 protein was performed by immunoblotting with polyclonal antibodies raised against the S. cerevisiae Fpr1 protein. Cells transformed with pREP1-fkh1 ϩ (OP) exhibited increased expression of a 12-kDa protein compared with wild type (WT), consistent with the predicted molecular weight of Fkh1. In protein extracts of ⌬fkh1 cells (⌬) no 12-kDa band was observed. C, exponentially growing wild type (TA06) and ⌬fkh1 (TA96) cells in minimal medium were collected, washed, and resuspended in nitrogen-free (EMM-N) medium. After 3 days of incubation at 25°C, aliquots were removed, and the DNA content of individual cells was measured by FACS.
S. pombe cells is Fkh1. We hypothesized that if Fkh1 is the target for rapamycin action, then we might isolate fkh1 mutants that can confer rapamycin resistance. Such mutants are expected to be impaired in rapamycin binding but to retain activity necessary for the sexual development pathway.
We randomly mutated fkh1 ϩ cDNA using error-prone PCR (see "Experimental Procedures"). Strain TA77 (leu1 ura4 ⌬fkh1 h 90 ) was transformed with a mixture of plasmids bearing mutated fkh1 cDNAs, plated on minimal medium in the absence of rapamycin. After colonies had developed they were replicaplated onto 0.2 g/ml rapamycin-containing plates. Colonies that underwent sexual development despite the presence of rapamycin were identified by exposure to iodine vapor (see "Experimental Procedures"). Of 25,000 transformants, 28 clones showed plasmid-dependent sporulation on rapamycincontaining plates. Sequence analysis revealed that 16 of these 28 clones carried single missense mutations at one of the following positions: Phe-47, Cys-49, Leu-56, Ile-92, or Phe-100. The remaining 12 mutants carried 2-4 missense mutations. In each case at least one mutation occurred at one of the critical positions Phe-47, Cys-49, Leu-56, Ile-92, or Phe-100 (see Table  II). Since iodine vapor analysis suggested that all mutations conferred similar rapamycin resistance phenotype, representatives of mutants of each of the five critical mutations were chosen for further analysis as follows: F47S, C49R, L56F, I92F, and F100L. Quantitative assessment of the mating efficiencies of ⌬fkh1 cells expressing these fkh1 mutants demonstrated that all fully complemented the mating deficiency phenotype and conferred a complete rapamycin resistance phenotype (Fig. 4).
Alignment of the amino acid sequence of Fkh1 with that of HuFKBP12 reveals that all the fkh1 rapamycin-resistant mutations fall into or adjacent to the predicted rapamycin binding pocket (21). Four of the five mutations, F47S, L56F, I92F, and F100L, correspond to amino acid residues of HuFKBP12 that most closely interact with rapamycin in the atomic structure of the HuFKBP12-rapamycin complex (see Fig. 5). In particular, Phe-47 and Phe-100 correspond to the HuFKBP12 amino acid residues that surround the portion of rapamycin that penetrates most deeply into the protein (21). Cys-49 in Fkh1, corresponding to Phe-48 in HuFKBP12 and Cys-55 in S. cerevisiae Fpr1p, is the only amino acid out of the five identified that is not mapped to the very core of the rapamycin binding pocket but resides in a close vicinity. The identification of rapamycinresistant fkh1 mutants carrying mutations near or at the predicted rapamycin-binding pocket argues that the rapamycin resistance of these mutants is due to impaired rapamycin binding.
We were curious if the fkh1 mutants were also impaired in rapamycin binding in S. cerevisiae cells. If so, the fkh1 mutants could not restore rapamycin sensitivity in S. cerevisiae ⌬fpr1 mutants. We cloned each of the fkh1 mutants into an S. cerevisiae expression vector and transformed it into ⌬fpr1 strain (Fig.  6). Although the wild type fkh1 ϩ gene could efficiently restore rapamycin sensitivity in ⌬fpr1 mutants, the F47S mutant com-pletely failed to restore rapamycin sensitivity, suggesting that this mutation is strongly impaired in rapamycin binding. ⌬fpr1 cells transformed with F100L and C49R grew well on plates containing 25 ng/ml rapamycin but poorly on plates containing 100 ng/ml rapamycin, suggesting that the F100L and C49R mutants are partially impaired in rapamycin binding in S. cerevisiae cells. Somewhat surprisingly, the mutant L56F efficiently restored rapamycin sensitivity, suggesting that it can efficiently form a toxic complex with rapamycin in S. cerevisiae.
Our findings indicate that whereas all the fkh1 mutants confer complete rapamycin resistance in S. pombe, they are not identical in their ability to restore rapamycin sensitivity in ⌬fpr1 S. cerevisiae cells. Of all the mutations only the F47S mutation completely abolished the ability of fkh1 to restore rapamycin resistance in ⌬fpr1. The differences in the behavior of the fkh1 mutants in the two yeast systems may not be surprising since in S. pombe rapamycin seems to inhibit directly the FKBP12 function, whereas in S. cerevisiae rapamycin exerts its effect by forming a complex with FKBP12 that inhibits the TOR proteins. DISCUSSION The immunophilins, FKBPs and cyclophilins, are highly conserved from bacteria to human and have been found to be both widely distributed and abundantly expressed. In vitro, these proteins exhibit PPIase activity that accelerates the refolding of denatured proteins (16 -18). Given these properties, it has been suggested that immunophilins may play a general role in protein folding (reviewed in Ref. 3). However, more recent studies strongly suggest that immunophilins play specialized roles, dependent on their ability to selectively bind to other proteins. For example, the mammalian FKPB12 specifically interacts with the ryanodine calcium release channel, altering its sensitivity to Ca 2ϩ and stabilizing its closed state (11)(12)(13). Studies in S. cerevisiae are consistent with the suggestion that the immunophilins do not carry out a general, housekeeping, role, since mutants lacking all the immunophilins are viable (62).
Little is known about the functions of the immunophilins in S. pombe, a yeast that is distantly related to S. cerevisiae. Only two members of the family have been subjected to detailed analysis as follows: wis2 ϩ , a heat shock-inducible 40-kDa cyclophilin that is involved in cell cycle regulation (63), and fkb39 ϩ , a 39-kDa FKBP homolog that is localized to the nucleus (64). The isolation and expression of a cyclophilin-18  Phe-100  Leu  2  Phe-100  Ser  1  Phe-100  Ile  3  Phe-47  Ser  8  Phe-47  Tyr  1  Cys-49  Arg  18  Cys-49  Ser  1  Leu-56  Phe  1  Ile-92  Phe  1  Ile-92  Thr  1 FIG. 4. fkh1 rapamycin-resistant mutants. ⌬fkh1 cells (TA77) transformed with vector only, fkh1 ϩ , or fkh1 bearing the indicated mutations, were streaked on EMM medium with (ϩR) or without (ϪR) 0.2 g/ml rapamycin and incubated at 30°C for 4 days. The cells were then exposed to iodine vapor. Since iodine stains only spores, only cells that underwent sexual development are stained with dark color. The mating frequency of each strain was scored under the microscope, and the averages of the scores obtained in two independent experiments are shown.
homolog has been reported (63,65). In the present study we isolated the S. pombe FKBP12 homolog fkh1 ϩ and demonstrated that it is specifically required for an early step of the sexual development pathway.
In S. cerevisiae, the FKBP12 homolog, FPR1, has a critical role in mediating the effect of rapamycin to the TOR proteins (see Refs. 38 and 39 and reviewed in Ref. 66). The amino acid sequence encoded by fkh1 ϩ is highly similar to that of FPR1. Consistently, fkh1 ϩ can replace FPR1 in mediating the effect of rapamycin in S. cerevisiae (Fig. 1). Slight differences do appear between fkh1 ϩ and FPR1, since overexpression of fkh1 ϩ but not of FPR1 can slightly reduce the sensitivity to rapamycin in S. cerevisiae cells ( Fig. 2A). We suggest that fkh1 ϩ reduces rapamycin sensitivity in S. cerevisiae by forming Fkh1-rapamycin complexes that are not as efficient in inhibiting the S. cerevisiae TOR proteins as Fpr1p-rapamycin complexes (Fig. 2B).
One of the key observations that led to the currently accepted model for rapamycin mode of action in S. cerevisiae was that cells disrupted for FPR1 are viable and rapamycin-resistant (34,35). Unlike this finding, disruption of fkh1 ϩ does not result in rapamycin resistance in S. pombe. Unexpectedly, the phenotype of ⌬fkh1 mutants highly resembled that of cells treated with rapamycin; ⌬fkh1 cells are defective in an early step of the sexual development pathway, before mating occurs, but show no defects in later steps such as meiosis or sporulation. In S. pombe, sexual development is a process induced only under starvation conditions (see Introduction). However, neither the sterility of rapamycin-treated cells (44) nor the sterility of ⌬fkh1 mutants is associated with defects in response to starvation. The strong similarity between the phenotype of ⌬fkh1 mutants and that of rapamycin-treated wild type cells suggests that rapamycin inhibits sexual development directly by inhibiting the cellular function of fkh1 ϩ .
Does Fkh1 form a toxic complex with rapamycin that inhibits the functions of the S. pombe TOR proteins? We have recently determined that the S. pombe tor1 ϩ gene is required under various stress conditions, including starvation, whereas tor2 ϩ is an essential gene (46). Since rapamycin does not affect entrance into stationary phase or response to stress conditions, it appears that most, if not all, of the S. pombe TOR functions are not inhibited by the drug. Sterility is the sole phenotype common to rapamycin treatment and loss of function of TOR activ-ity. However, the sterile phenotypes of tor1 mutants and rapamycin-treated cells seem to be unrelated. In tor1 mutants the sterile phenotype is likely to be associated with the inability to respond to starvation conditions, whereas rapamycin-treated cells are specifically defective in the sexual development pathway (46). Are the phenotypes of tor1 and fkh1 mutants related? Thus far we have failed to show any genetic link between tor1 and fkh1 mutants; overexpression of tor1 ϩ does not alleviate the sterility of ⌬fkh1 cells, and overexpression of fkh1 ϩ does not alleviate the sterility of ⌬tor1 cells. 3 The rapamycin-binding site of FKBP12, revealed through structure analyses of HuFKBP12 (21), is composed of aromatic side chains that form a hydrophobic pocket. In this work we have exploited the yeast genetic system to identify residues in Fkh1 involved in the response to rapamycin. We identified 5 amino acid residues in Fkh1 that are critical for the effect of rapamycin in S. pombe. Four of the five amino acids identified, Phe-47, Leu-56, Ile-92, and Phe-100, correspond to amino acid residues of HuFKBP12 that most closely interact with rapamycin. The fifth amino acid, Cys-49, corresponding to Phe-48 in HuFKBP12, is not mapped to the very core of the ligandbinding pocket but in a close vicinity. The C49R mutation, however, conferred complete rapamycin resistance in S. pombe, and our studies in S. cerevisiae suggested that it diminished rapamycin binding (Fig. 6). Since all the rapamycin-resistant fkh1 mutants are mutated at amino acid residues potentially important for rapamycin binding, it is likely that the rapamycin resistance phenotype stems from an inability of the mutant proteins to bind the drug. This suggestion awaits further support from binding experiments of rapamycin to the wild type and mutant Fkh1 proteins.
Notably, all the rapamycin-resistant fkh1 mutants completely restored mating in ⌬fkh1 mutants (Fig. 4). This finding suggests that the Fkh1 rapamycin-binding site and the putative active site do not completely overlap. In particular, mutation at the Phe-100 amino acid residue did not impair the cellular activity of Fkh1 in sexual development, despite its being one of the most conserved amino acid residues in FKBP12 sequences (21). The effects of mutations at positions corresponding to Phe-100 in human FKBP12 (F99Y) and in S. cerevisiae FPR1 (F106Y) have been studied previously (67,68). Like the F100L mutation in Fkh1, neither the F99Y mutation nor the F106Y mutation affected the protein function in vivo. The F99Y mutant supported the ryanodine channel function (67), and the F106Y mutant complemented the slow growth phenotype observed in ⌬fpr1 (68). It is also interesting to note that the HuFKBP12 F99Y mutant and the S. cerevisiae F106Y mutant show reduced PPIase activity in the in vitro peptide cleavage assay (67,68). More recently, however, it was demonstrated that the S. cerevisiae Phe-106 mutant retained PPIase activity in a different assay that uses ribonuclease T1 as a substrate (68). It has thus been suggested by Dolinski 5. A comparison of FKBP12 sequences and positions of rapamycinresistant fkh1 mutants. The predicted amino acid sequence of fkh1 ϩ (SpFKBP) is aligned with the S. cerevisiae Fpr1p (ScFKBP) and the human FKBP12 homolog (HuFKBP). Asterisks denote the positions in SpFKBP12 in which single point mutations confer rapamycin resistance phenotype. Boxes indicate the positions that are predicted to interact most closely with rapamycin according to structural studies of the HuFKBP12-rapamycin complex (21). that the Phe-106 mutant may retain PPIase activity toward large substrates, leaving it an open question whether the PPIase activity observed in vitro is relevant for the cellular activity in vivo.
The finding that fkh1 ϩ is required for sexual development provides a novel, genetically amenable system to study the possible role of this immunophilin in cellular function. In the future we intend to exploit this system to identify the substrate(s) for Fkh1 cellular function in vivo. We have demonstrated in this work the ability to isolate fkh1 mutants that are rapamycin-resistant. In the future, a similar approach will be utilized to isolate fkh1 mutants that are defective in their ability to support sexual development, thus determining the amino acid residues critical for Fkh1 activity in this pathway.