RAPT 1 , a mammalian homolog of yeast Tor , interacts with the FKBP 12 / rapamycin complex ( phosphatidylinositol 3 - kinase )

Rapamycin is a potent immunosuppressant that blocks the G1/S transition in antigen-activated T cells and in yeast. The similar effects of rapamycin in animal cells and yeast suggest that the biochemical steps affected by rapamycin are conserved. Using a two-hybrid system we isolated mammalian clones that interact with the human FK506/rapamycinbinding protein (FKBP12) in the presence of rapamycin. Specific interactors, designated RAPTI, encode overlapping sequences homologous to yeast Tor, a putative novel phosphatidylinositol 3-kinase. A region of 133 amino acids of RAPT1 is sufficient for binding to the FKBP12/rapamycin complex. The corresponding region in yeast Tor contains the serine residue that when mutated to arginine confers resistance to rapamycin. Introduction of this mutation into RAPT1 abolishes its interaction with the FKBP12/rapamycin complex. Rapamycin and FK506 are structurally related immunosuppressants that block distinct steps in T-cell activation. FK506 interferes with the early induction of lymphokine gene expression stimulated by the binding of antigen to the T-cell receptor, whereas rapamycin blocks subsequent lymphokine-induced cell division (1-4). Rapamycin specifically blocks the cell cycle in G1 in lymphocytes (1, 5-7), certain nonlymphoid cells (8-10), and Saccharomyces cerevisiae (11). The molecular details of how these drugs act are only partially understood. Both drugs bind to and inhibit the peptidyl-prolyl cis-trans isomerase activity of FK506/ rapamycin-binding protein (FKBP12). However, inhibition ofpeptidyl-prolyl cis-trans isomerase activity is not sufficient for immunosuppressive activity (1, 5, 12). Rather, these immunosuppressants confer gain-of-function to their cognate binding proteins. FKBP12 complexed with FK506 binds to and inhibits the Ca2+-dependent phosphatase calcineurin (13-16). Calcineurin, in turn, regulates the nuclear localization of transcription factors required for lymphokine gene expression (17, 18). Genetic analyses in yeast identified a putative novel phosphatidylinositol (PI) 3-kinase as the possible target of the FKBP12/rapamycin complex. Mutations in TORI (also called DRRI) or TOR2, genes specifying different forms of the putative PI 3-kinase, render yeast resistant to rapamycin (11, 19-21). These studies raise the question of whether Tor directly associates with the FKBP12/rapamycin complex and whether inhibition of Tor activity is sufficient for the antiproliferative activity of rapamycin in yeast and mammalian cells. The results presented here demonstrate that the mammalian homolog of yeast Tor, RAPT1, interacts with FKBP12/rapamycin and defines a region of 133 amino acids, the RAPT1-binding domain, sufficient for this interaction. The serine-to-arginine mutation in Tor that confers resistance of yeast to rapamycin when introduced into the RAPTI1binding domain prevents formation of the FKBP12/ rapamycin/RAPT1 complex. Based on these results we conclude that RAPT1 is a target of the FKBP12/rapamycin complex. MATERIALS AND METHODS Components of the Two-Hybrid System. The yeast strain L40 and the two-hybrid plasmids containing the lexA DNAbinding domain (pBTM116) and the VP16 activation domain (pVP16) were a gift from Stan Hollenberg (22). A PCR product containing the gene encoding human FKBP12 was cloned into the EcoRI andBamHI sites ofpBTM116, creating either an in-frame fusion between lexA and FKBP12 (plexAhFKBP12) or the same fusion separated by six glycine residues (plexA-G6-hFKBP12). The lexA-FKBP12 fusions used to identify the FKBP12/rapamycin interactors are referred to as "baits." Construction of the flcbl-2 Allele. VBY567 (MATa his3A200 trpl -901 leu2-3, 112 ade2 L YS2::(lexAop)4-HIS3 URA3::(lexAop)8-lacZ gal4 gal80 fkbl-2::ADE2) was constructed as follows: a linear fragment containing the disruption marker ADE2 flanked by FKBI noncoding sequences was used to transform yeast strain L40, selecting for adenine prototrophy. Ade+ yeast transformants were tested for rapamycin resistance to confirm that the wild-type FKBI allele was replaced by ADE2. Library Screen for Rapamycin Target Genes. VBY567 strains harboring each binding-domain fusion were transformed with a mouse embryonic day 10.5 cDNA PCR library (22) generated by random-primed synthesis of poly(A)+ RNA. The library transformants were plated onto medium lacking tryptophan and leucine, allowed to grow for 3 days at 30°C, pooled, and stored as frozen stocks in 50% (wt/vol) glycerol. An aliquot was thawed and outgrown in liquid glucose medium for one doubling and then plated onto phosphate buffer-containing (pH 7.0) plates under the conditions described, at 106 per 15-cm plate. Blue colonies (His+ LacZ+) were picked after 4-11 days of incubation at 30°C. Each candidate was retested for growth on medium lacking histidine (minus His) with or without rapamycin. Candidate clones that grew in the presence of rapamycin and failed to grow on medium without rapamycin were further tested for the plasmid dependence of their growth phenotypes (23). DNA samples prepared from such candidates were used to transform Escherichia coli strain MC1066 [Rec+ AlacI-Ax74 galUgalK strepAr, trpC9830 leuB6pyrF: :TnS(kanr)] carrying a leuB mutation that is complemented by yeast LEU2. PlasmidDNA samples were prepared from Leu+ bacteria and then used to transform yeast strains to test interaction of the candidate library clones against different baits. Nonspecific Abbreviation: PI 3-kinase, phosphatidylinositol 3-kinase. *Present address: Vertex Pharmaceuticals, Inc., 40 Allston Street, Cambridge, MA 02139. tTo whom reprint requests should be addressed. 12574 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact. Proc. Natl. Acad. Sci. USA 91(1994) 12575 interactors (those conferring His+ LacZ+ when paired with an unrelated bait fusion) were eliminated, while clones specific for FKBP12 bait fusions were retained. Candidate clones that have fulfilled all the above criteria were analyzed by DNA sequencing on both strands. Sequence Analyses. The homology between mouse Raptl and yeast Tor was found by searching the National Center for Biotechnology Information sequence data bases using the BLAST network service. Subsequent sequence alignments were performed by the Clustal comparison method using the analysis software MEGALIGN (DNAstar, Madison, WI). Construction of the Serine-to-Arginine Mouse Raptl Mutation. The mouse Raptl serine-to-arginine mutation was constructed by oligonucleotide mutagenesis. Coding and noncoding strand oligonucleotides containing the mutation were GAAGAGGCAAGACGCTTGTAC and GTACAAGCGTCTTGCCTCTTC. PCRs were performed using these oligonucleotides in combination with oligonucleotides GAGTTTGAGCAGATGTTTA and the M13 universal primer, which are sequences in the pVP16 vector, 5' and 3' of the mouse Raptl insert, respectively. pVP16 containing mouse Raptl was used as the template for PCR. The PCR product, digested with BamHI and EcoRI, was cloned into the BamHI and EcoRI sites in pVP16. The resulting clone was sequenced to verify that the clone contained the serine-to-arginine mutation and no others. Isolation and Characterization of Human RAPTI Clones. The human RAPTI homolog was isolated by screening 3 x 106 A phage plaques of an Epstein-Barr virus-transformed B-lymphocyte cDNA library (24) with a mouse Raptl probe at 30% formamide/5x standard saline citrate (SSC)/5x Denhardt's solution/1% SDS/200 ,ug of salmon sperm DNA per ml. Ten overlapping clones of human RAPTI were obtained. Northern Blot Analysis. The multiple tissue Northern blots (containing 2 ,Mg of human RNA per lane) were from Clontech. Hybridizations were at 42°C in 5x SSPE/5x Denhardt's solution/30% formamide/1% SDS/200 yg of denatured salmon sperm DNA per ml. Washes were at O.1x SSC/0.1% SDS at 550C. The blot was exposed for 5 days prior to autoradiography. The levels of RNA loaded in each lane were independently monitored by hybridizing the same blots with a human G3PDH probe and were found to be similar in all lanes, with the exception of skeletal muscle, which had 2-3 times the signal (data not shown).