Fill and Sign the Standard Survey Work Order Form Outer Banks Engineering

The EMBO Journal vol.15 no.24 pp.7156-7167, 1996 The catalytic subunit of protein phosphatase 2A associates with the translation termination factor eRF1 I I I 1 Natasa Andjelkovic, Stanislaw Zolnierowicz1, Christine Van Hoof 2, Jozef Goris2 and Brian A.Hemmings3 Friedrich Miescher-Institut, PO Box 2543, CH-4002 Basel, Switzerland and 2Afdeling Biochemie, Faculteit der Geneeskunde, Katholieke Universiteit te Leuven, Herestraat 49, 3000 Leuven, Belgium 'Present address: Department of Biochemistry, Faculty of Biotechnology, Medical University of Gdansk, Debinki 1, 80-211 Gdansk, Poland 3Corresponding author By a number of criteria, we have demonstrated that the translation termination factor eRFi (eukaryotic release factor 1) associates with protein phosphatase 2A (PP2A). Trimeric PP2A1 was purified from rabbit skeletal muscle using an affinity purification step. In addition to the 36 kDa catalytic subunit (PP2Ac) and established regulatory subunits of 65 kDa (PR65) and 55 kDa (PR55), purified preparations contained two proteins with apparent Mrs of 54 and 55 kDa. Protein microsequencing revealed that the 55 kDa component is a novel protein, whereas the 54 kDa protein was identified as eRF1, a protein that functions in translational termination as a polypeptide chain release factor. Using the yeast two-hybrid system, human eRF1 was shown to interact specifically with PP2Ac, but not with the PR65 or PR55 subunits. By deletion analysis, the binding domains were found to be located within the 50 N-terminal amino acids of PP2Ac, and between amino acid residues 338 and 381 in the C-terminal part of human eRF1. This association also occurs in vivo, since PP2A can be co-immunoprecipitated with eRF1 from mammalian cells. We observed a significant increase in the amount of PP2A associated with the polysomes when eRF1 was transiently expressed in COS1 cells, and eRF1 immunoprecipitated from those fractions contained associated PP2A. Since we did not observe any dramatic effects of PP2A on the polypeptide chain release activity of eRF1 (or vice versa), we postulate that eRF1 also functions to recruit PP2A into polysomes, thus bringing the phosphatase into contact with putative targets among the components of the translational apparatus. Keywords: eRFI/protein phosphatase 2A/signal transduction/translational termination Introduction Protein phosphatase 2A (PP2A) is implicated in the regulation of many cellular processes including metabolism, signal transduction, growth, development, cell cycle progression and transformation (reviewed in Mumby and Walter, 1993; Mayer-Jaekel and Hemmings, 1994). PP2A encompasses a family of trimeric holoenzymes which consist of a 36 kDa catalytic subunit (PP2Ac) bound to the constant regulatory subunit of 65 kDa (PR65/A) which then associate further with the third, variable regulatory subunit. Several trimeric PP2A holoenzymes have been purified which contain different variable subunits of either 54, 55, 72 or 74 kDa (reviewed in Kamibayashi and Mumby, 1995; Wera and Hemmings, 1995). As documented by in vitro reconstitution assays and by analyzing yeast and Drosophila mutants deficient in regulatory proteins, both the constant and variable subunits are important for controlling PP2A activity and substrate specificity (reviewed in Mayer-Jaekel and Hemmings, 1994; Wera and Hemmings, 1995). For instance, PP2A activity from brain extracts of Drosophila aarl mutants, in which the gene encoding PR55 is disrupted by Pelement insertion, is several fold lower towards histone HI and caldesmon phosphorylated by p34cdc2 as compared with wild-type flies (Mayer-Jaekel et al., 1994). In contrast, phosphorylase phosphatase activity of PP2A is similar in aarl and control flies. The variable regulatory subunits also represent targets for potential second messengers and viral proteins. Dobrowsky et al. (1993) demonstrated that ceramide activates only trimeric PP2A containing the PR55 subunit whereas the PP2Ac-PR65 dimer is unaffected. Recent data, however, show that neither the constant nor variable regulatory subunits are required for ceramide stimulation of PP2A activity, since both PP2Ac and PP2Ac-PR65 dimer can be stimulated by ceramide in a manner similar to that of the trimeric holoenzyme, suggesting that PP2Ac itself is a target of ceramide action (Law and Rossie, 1995). Furthermore, PP2A associates with transforming antigens of certain DNA tumor viruses, such as polyomavirus small t and middle T, and SV40 small t (Pallas et al., 1990). It is believed that these oncoproteins act to alter PP2A activity by displacing the normal cellular variable regulatory subunits from the trimeric holoenzyme. Some viral proteins interact only with specific forms of PP2A holoenzymes, e.g. SV40 small t antigen is able to replace only the B subunit (PR55), but not the B' subunit (PR61) from trimeric PP2A (Sontag et al., 1994). It was also shown that adenovirus E4orf4 binds to the trimeric PP2A holoenzyme that contains PR55 (Kleinberger and Shenk, 1993). Taken together, these examples illustrate that the activity of PP2Ac is tightly controlled in vivo by regulatory proteins. We developed a strategy for simultaneous purification of different PP2A holoenzymes from rabbit skeletal muscle in order to analyze their subunit structure further. This approach resulted in the purification of two heterotrimeric forms of PP2Ao containing different isoforms of a novel type of variable regulatory subunit (termed PR61) that 7 6 Oxford University Press 7156 i. I PP2A associates with eRFl Homogenate Acidification to pH 5.3 DE52 batch elution with 0.3M NaCI 30-50% (NHd SO4 precipitation DEAE Sepharose (0.05-0.6 M NaCI) pool I pool 3 pool 2 ~~~~~~~~~~~~~I, X minohexyl Sepharose I I Poly-L-lysine Agarose | mnohexyl |Spharose Thiophosphorylase-a Sepharose MonoQ FPLC PP2A0 SDS-PAGE analysis of pool 2 (fractions eluted from DEAE-Sepharose between 0.34 and 0.38 M NaCl) after the thiophosphorylase a-Sepharose purification step revealed that this preparation contained the 36 kDa catalytic and 65 kDa regulatory subunits and several proteins in the range of 54-55 kDa (Figure 2B). Further fractionation by MonoQ FPLC resulted in the separation of PP2A, (trimer containing the 55 kDa regulatory subunit, PR55) and PP2A2 (the dimeric form of the enzyme) from two proteins of 54 and 55 kDa (Figure 2A and C). These two proteins appear to be unrelated to PR55 since they did not crossreact with anti-PR55 antibodies (data not shown). In addition, we identified free PR55cx in this preparation by SDS-PAGE and immunoblot analysis (fractions 37 and 38). This suggests that the PP2A2 identified probably results from the dissociation of PR55, and possibly the 54 and 55 kDa proteins, from the complex (see below). PP2A1+ associated proteins 1 PP2A2 Fig. 1. Schematic outline of the purification protocol used to isolate different PP2A holoenzymes from rabbit skeletal muscle. probably function to target PP2A to nuclear substrates (Tehrani et al., 1996; Zolnierowicz et al., 1996). We also found two novel proteins of 54 and 55 kDa that apparently co-purify with the trimeric PP2A1 holoenzyme following affinity purification. Here we report the identification of the 54 kDa protein as a member of the eRFI family of proteins involved in termination of protein synthesis as well as further examine the functional consequences of its interaction with constituent components of the PP2A holoenzyme. Results Co-purification of PP2A1 holoenzyme with two proteins of 54 and 55 kDa A modified protocol for PP2A purification from rabbit skeletal muscle (see Figure 1 and Materials and methods) was used to identify novel regulatory and/or associated proteins. The partially purified material obtained from DEAE-Sepharose pools 1, 2 and 3 was analyzed using several antisera developed against the constituent subunits of PP2A reported in earlier publications (Hendrix et al., 1993a,b; Turowski et al., 1995). MonoQ FPLC of fractions corresponding to pool 1 (eluted from DEAE-Sepharose between 0.27 and 0.32 M NaCl) revealed the existence of a trimeric PP2A holoenzyme containing 36 kDa catalytic (PP2Ac) and 65 kDa regulatory (PR65) subunits and a novel type of variable regulatory subunits with apparent Mrs ranging from 56 to 61 kDa. This trimeric holoenzyme apparently corresponds to PP2Ao (Zolnierowicz et al., 1996) according to the classification established previously by Tung et al. (1985). 54 kDa protein that co-purifies with PP2A1 is a member of the eRF1 family of proteins with polypeptide chain release factor activity The 54 and 55 kDa proteins purified by MonoQ FPLC chromatography were used to obtain protein sequence data as described in Materials and methods. Sequences of three tryptic peptides comprising 30 amino acids derived from the 54 kDa protein were obtained: peptide 3, YFDEISQDTGK, peptide 9/10, ILYLTPEQEK and peptide 25/ 26, S/GFGGIGGIL. Comparison of these sequences using the FASTA program (Pearson and Lipman, 1988) revealed 76% homology to the predicted protein sequence encoded by the Saccharomyces cerevisiae SUP45 gene (Breining and Piepersberg, 1986). Human cDNAs corresponding to this protein were isolated using a reverse transcriptionPCR approach (see Materials and methods). Among several clones isolated from human fetal brain library and analyzed by sequencing, only one, termed BBZ.eRFI-4b, contained a full-length open reading frame (1311 bp) as well as 231 bp of the 5'-non-coding region and -2.22 kb of the 3'-non-coding region. This cDNA is identical to the TB3-1 cDNA, originally identified by Grenett et al. (1992) and resequenced by Frolova et al. (1994). cDNAs encoding homologous proteins have been identified recently from Arabidopsis thaliana (Quigley et al., 1994) and Xenopus laevis (Tassan et al., 1993). The report by Frolova et al. (1994) demonstrated that human and Xenopus proteins possess polypeptide chain release factor activity and termed this factor eRFI. Sequence comparison revealed that eRFI protein is highly conserved between species, with yeast and human protein being 67.5% identical (Figure 3). Partial amino acid sequence analysis of seven peptides comprising 71 amino acids derived from the 55 kDa protein did not display any homologies to the sequences present in currently available databases (GenBankTM, release number 95). Currently we are attempting to isolate cDNAs corresponding to this protein to establish its relationship to PP2A and eRFI. mRNA encoding human eRF1 is ubiquitously expressed The levels of transcripts encoding human eRFI were analyzed in poly(A)+ RNA isolated from different human tissues. With the probe corresponding to the complete 7157 N.Andjelkovic et al. A 0 0. CZ- Gi O - Q E =) 0~ (ja 00 : Oins Q - - 0.14 -0.12 -0.10 250 200 8 15010050 - 0.04 - 0.02 r 0.00 0- U) 30 40 60 50 Thiophosphorylase 80 90 MonoQ FPLC a 97 70 Fraction C B 66 X -0.08 < -0.06 97- : .... .. ... 66 - F.:,:: '-'--'-;E _ s . -'x 8 . 43 43 31 . ..... k55-kDa 54-kDa 31 C 4 5 6 7 8 9 10 C 36 37C 48 49 50 51 52 53 54 55 56575838 70 71 72 73 74 Fig. 2. Co-purification of trimeric PP2AI holoenzyme with two proteins of 54 and 55 kDa. (A) Elution profile and protamine-stimulated phosphorylase phosphatase activity of MonoQ FPLC fractions. Absorbance at 280 nm (A280) and phosphorylase phosphatase activity of the MonoQ fractions are presented as closed and open circles, respectively. The first peak of activity represents PP2A1 while the second, smaller peak corresponds to the dimeric form of the holoenzyme, PP2A2. (B) SDS-PAGE analysis of pool 2 after thiophosphorylase a chromatography. (C) SDS-PAGE analysis of MonoQ fractions of PP2A1. C = previously purified PP2AI holoenzyme loaded as standard. human eRFI cDNA (BBZ.eRF1-4b) multiple transcripts were detected of ~2, 2.5 and 4 kb (Figure 4). mRNA encoding human eRFI appears to be ubiquitously expressed, with the highest transcript levels found in lung, skeletal muscle and placental tissues. Of the three classes of transcripts detected, the 4 kb species showed the highest level of expression in all tissues. Quantitation by ImageQuant software showed that 2.5 and 2 kb classes of transcripts were expressed ~2- to 4-fold less than the high molecular weight transcript. This mRNA distribution is different from that reported for ClI, the X.laevis eRF1 homolog, where a much more restricted pattern of expression was found, with both mRNA and the protein being completely absent in liver (Tassan et al., 1993). Human eRF1 interacts with the catalytic subunit of PP2A in the yeast two-hybrid system In order to assess which subunit of PP2A associates with human eRFI, we used the yeast two-hybrid system (Fields and Song, 1989). Another protein, termed eRF3, was previously shown to bind to eRFI and stimulate its activity in polypeptide chain termination in S.cerevisiae and X.laevis (Stansfield et al., 1995; Zhouravleva et al., 1995). Therefore, we extended the two-hybrid analysis to see whether eRF3 could be an interaction partner as well. Human PP2Aca, PR65a, PR55x, eRFI and eRF3 cDNAs were fused with the yeast transcriptional activator GAL4 DNA binding or transactivation domains (as described in Materials and methods). Expression of fusion proteins was checked in double transformants, and the interaction of PP2A subunits with human eRFI and eRF3 evaluated by monitoring the expression of two different reporter genes, lacZ and HIS3. Transcriptional activation of reporter genes driven by wild-type GAL4 protein, or brought about by the interaction between SV40 large T and p53, were used as positive controls. To exclude intrinsic transcrip7158 tional activation capacity or non-specific binding of either molecule to unrelated proteins, co-transformations with the empty vector or vector encoding human lamin C fused to the opposite domain of GAL4 were used as negative controls. These experiments showed that human eRFI binds to eRF3 (Figure 6), as one would expect based on the previous reports from S.cerevisiae and X.laevis studies (Stansfield et al., 1995; Zhouravleva et al., 1995). They also showed that the human eRFI specifically interacts with PP2Ac, but not with PR65 or PR55, in both reporter systems, since only PP2Ac-eRF1 double transformants were positive as confirmed by f-galactosidase assays (Figure 6) and did not require histidine for growth (data not shown). From this analysis, we conclude that the PP2A subunit that binds eRFI is the catalytic subunit itself. On the other hand, eRF3 failed to bind to either of the three PP2A subunits tested in this system (Figure 6). Identification of domains required for the interaction of PP2Ac and eRF1 In addition to the analysis of interactions of full-length proteins in the two-hybrid system, we attempted to map the regions on both eRFI and PP2Ac required for this association. For this purpose, a series of N- and C-terminally truncated versions of both proteins was constructed (Figure 5) and tested in the same experimental setup as described above. All of the C-terminal deletion mutants of PP2Ac (PP2Ac 259, PP2Ac'29 and PP2Ac1-159), but none of the N-terminal deletion mutants (PP2AC50309, PP2Ac'00°309 and PP2Ac150309) were able to interact with eRFI (Figures SA and 6), suggesting that the region essential for binding was within the N-terminal 50 amino acid residues of the protein. Differences in binding of the C-terminal truncations of PP2Ac to the full-length eRFI suggest that the C-terminal portion of the protein also contains sequences that influence this PP2A associates with eRF1 50 Hum eRF1 MADDPSAADRNVEIWKIKRLXKSLEAARGNGTSHISLIZPPRDQISRVAK Xen eRFI ...................... Sc Sup45 Ara eRFl MDNNVESKI. ....V. .. VQ. E .--.K.I ... .GL.PLYQ. .......... ....... G..T ... M.R. .VA. . . . A 309 PP2AcI 309 IGA..IZI,aL 1 50 309 100 Hum eRll MLADEFGTASNIKSRVNRLSVLGAITSVQQRLKLYNKVPPNGLVVYCGTI Xen eRFI ........... ... Sc Sup45 .T. ..S.... T..K .....TL. Ara eRfl ... . Y .Q.Q... S....A . . 150 Hum eRFI Xen eRFi Sc Sup45 I. .D. TF.I ..Y Ara eRfl .oDD ..... T * . .- .. e. . . *.. .. ... . . ..... . @@v@@ . ..... . A ..... @^.. e . . Xan Atll sbR U'Lr-1 . .. .. .. .. .. .. .. .. . .. . . . .. Sc Sup45 M.. .Q.T...SVS.... T .. Ara eRfl .M..N.T.S .. . .. . . . . . , 309 . = 259 1 PP2Ac1-259 IGAL4DBD . 1 , 250 Hum eRFi NYVRKVAETAVQLFI--SGDKVNVAGLVLAGSADFKTELSQSDaTDQRLQ Xen eRFi ..................... Sc Sup45 ........ V... .. --TN . K. .. D.AK.EL..P..A Ara eRfl .....T..L.T.FY.NPATSQP..S. ..EL..P... 300 , 309 E PP2Ac'50309 VIDGSGALFGTLQGNTRWVLHFrTVDLPKXKGRGGQSALRFARLR&EKRH . ......... L4 ....... 200 Hum eRF$ 100 PP2Ac'00309 150 e . V-.SE..QA.D. P-.NE ..ES.D . . 3O9 PP2Ac5O9 L.D. .... L.T... Binding to eRFI + PP2Ac . 209 PP2Acl-209 GAL408t 1 159 PP2Acl-159 GAUDB' .. Hum eRFl SKVLKLVDISYGGENGFNQAIELSTEVLSNVKFIQEKKLIGRYFDEISQD Xen eRFl ............. . eRF1 B 437 n . . Sc Sup45 C..II. ...........I. A.A.A. ... .....LEA ...... .. A.I ..K..!..... eRF1 1-437 IGAL4UT Binding to PP2Ac + Ara eRfl A.I.NV . . 437 51 350 Hum eRFI TGKYCFGVEDTLKALEMGAVEILIVYENLDIMRYVLHCQGTENE-E1 YF Xen eRPl ........... T. R.N.S.S.. .-T ..L Sc Sup45 .. ..Y.ID... DLL.... K.. .7.. ETI..TFK---DA.DBEVIK. Ara eRfl . ... ... ............. T W..... N.E.KNNT.G.IV-.RXL + eRFI51 437 . ... 400 Hum eRF1 TPEQEKDKSHYTDKZTGQOELIUSHPLLEWrANNYKKFGATLEIVTDKS S. Xen eRFi ........ X Sc Sup45 AZPEA ....A. .A. .A. MDVVSEE..1.X.L.A ...N. ..FI .. Ara eRfl GKD. NNQ.N.H.A ..NA.L.VQ.K . E..R. ....F..N.. Hum eRFl QEGSQFVKGFGGIGGILRYRVDFQGMEYQGGDDEFFDLDDY Xen eRFI ........ V....., Sc Sup45 S ..A...T.....AM .. K.N.EQL-VDESZ..YY.E.EGSDYDFI Ara eRfl . CR. L . QL.MTRFDELSDGEVYE.S. 437 437 437 435 437 93 GAL493I eRF197 150 eRF1I501Q7 1 404d 0111 NIL-1 S 9.5 7.544.4 ~ _ -- 2.4 4 kb 2.5 kb 2 kb 1.4 Fig. 4. Analysis of eRFI mRNA levels in human tissues. Human tissue blot (Clontech) loaded with 2 ,tg of poly(A)+ RNA isolated from the indicated human tissues was hybridized with a human eRFl cDNA probe corresponding to clone BBZ.eRFI-4 labeled to a specific activity of -1 X 109 c.p.m./j.g DNA. Three hybridizing fragments of 4, 2.5 and 2 kb were detected. Hybridization and washing of the blot were performed as described in the manufacturer's instructions. The blot was exposed to Kodak XAR-5 film for 9 h at -80'C with intensifying screens. 411 J eRF1-411 IGAL4T"DI 1 eRFI 1-381 + 381 1 GAL4TADI 1 Fig. 3. Alignment of amino acid sequences of eRFI from different species. Amino acid sequence of human eRFI (Hum eRFI) is aligned with corresponding sequences of X.laevis (Xen eRF1), S.cerevisiae (Sc Sup45) and A.thaliana (Ara eRFI). Amino acid sequences determined for the rabbit homolog are underlined. Dots represent identical residues, while dashes represent spaces introduced to optimize the alignment. The region in the C-terminal half of human eRFI which was identified as essential for binding to PP2Ac in the two-hybrid system analysis is presented in bold type. 437 = 338 eRFI 1-338 IGAL4TAOIFig. 5. Schematic representation of deletion mutants of human PP2Aca (A) and human eRFI (B) tested in the two-hybrid system and their ability to retain the interaction. Regions of the molecules required for binding in the two-hybrid system, located between amino acid residues 1 and 50 in PP2Acca and 338 and 381 in eRFI, respectively, are highlighted. interaction. N-terminal deletion mutants of eRFI (eRF151437, eRF193437 and eRF150-437) still retained this interaction and, of the three C-terminal deletions of eRFI (eRF 1 1,411 eRF1-381 and eRF 11-338), only the largest one, truncating the protein at Thr338, failed to bind to PP2Ac (Figures 5B and 6). This indicates that the putative binding domain on eRFI lies between Thr338 and Asn381. The minimally sufficient polypeptides defined from these experiments (PP2Ac'-159 and eRFI 1-381) are still able to interact with each other (Figure 6). Surprisingly, the interaction involves a region in eRF 1 that is poorly conserved between species. Currently we are investigating the conservation of this interaction by evaluating eRFI from a number of species in the two-hybrid system. Immunoprecipitates of eRF1, but not eRF3, contain PP2A activity To test whether complex formation between PP2A and eRFI occurs in mammalian cells, we examined whether these proteins co-immunoprecipitate. Association between PP2Ac and PR65 under the same experimental conditions was used as a positive control. Extracts from COS-1 cells transiently transfected with human eRFI, eRF3 or PR65cx tagged with the hemagglutinin (HA) epitope at the 7159 N.Andjelkovic et aL 1. SV40 large T/p53 2. PP2Ac/PR65 3. PP2Ac/eRF1 4. PR65/eRF1 5. PR55/eRF1 6. pGBT9/eRF1 7. PP2Ac/pGAD424 D E 21 8. PP2Ac/eRF151-437 PP2Ac/eRF193-437 PP2Ac/eRF1 150.437 _ C._ < 1. 0) U, U) 0 0-9 fl m C.) 0 CD) 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 5 nI II PP2Ac/eRF1 41 PP2Ac/eRF11381 PP2Ac/eRF1I-338 PP2Ac5°309/eRF1 PP2Acl°°-309/eRfl PP2Ac150-309/eRF1 PP2Ac1259/eRFI PP2Ac1225/eRF1 PP2Ac'159/eRF1 20. PP2Ac1 159/eRF1381 ~~~21. eRF3/eRF1 23. eRF3/PP2Ac 24. 25. eRF3/PR65 eRF3/PR55 1 2 3 4 5 6 7 8 9 10111213141516171819202122232425 SFY526 (lacZ reporter) cotransformants Fig. 6. Quantitative analysis of the interactions between PP2A subunits (PP2Ac, PR65ax and PR55a) and eRFl and eRF3 termination factors in the yeast two-hybrid system. Interactions were scored for activation of the lacZ reporter gene in double transformants of S.cerevisiae SFY562 strain. ,B-Galactosidase activity was determined in liquid assays using ONPG as a substrate (see Materials and methods). Numbers represent the mean values (± SEM) from three independent experiments carried out in duplicate assays. N-termini were subjected to immunoprecipitation with the anti-HA tag monoclonal antibody 12CA5. We measured PP2A activity in transfected COS-1 cell extracts and the immunoprecipitates using a 32P-labeled peptide (Kemptide Val6, Ala7) as a substrate (Figure 7). Assays were performed in the presence and absence of 10 nM okadaic acid, which is used typically to distinguish between PP1 and PP2A activities. The specific activity of PP2A in the extracts (expressed as mU/mg protein) was in the same range for all transfected cells (Figure 7A). As shown in Figure 7B, 12CA5 immunoprecipitates from cells transfected with HA-tagged eRFI or PR65 contained significant okadaic acid-sensitive, PP2A-like phosphatase activity as compared with immunoprecipitates from mocktransfected cells. PP2A activity associated with eRFl accounts for ~1% (0.4-1.6% range) of the total cytoplasmic PP2A activity. Immunoprecipitation of eRF3 did not bring down PP2A activity significantly higher than the background. This result suggests that the interactions of PP2A and eRF3 with eRF1 are mutually exclusive. PP2Ac and eRF1 are associated in vivo in mammalian cells To determine which PP2A subunits were present in the complex with eRFl, HA-eRFI immunoprecipitates were subjected to Western blot analysis with rabbit polyclonal anti-peptide antisera specific for different subunits of PP2A (see Materials and methods). These experiments showed that the catalytic subunit of PP2A (PP2Ac) can be detected in immunoprecipitates from HA-eRFl- or HA-PR65-transfected, but not from mock-transfected cells (Figure 8, lower panel). We looked for the presence of other established regulatory subunits of PP2A in eRF1 immunoprecipitates, and were able to detect PR65 (Figure 8, upper panel), but not PR55 or PR72 (data not shown). From this analysis, we conclude that eRFI apparently complexes with the core dimer of PP2A (consisting of 7160 PP2Ac and PR65) to form a novel trimeric complex, which is much less abundant than other previously reported complexes of PP2A. These experiments provide the first evidence for an in vivo association between the catalytic subunit of PP2A and eRFI protein in mammalian cells, which is to our knowledge the first report of PP2A interacting with a protein involved in the regulation of protein synthesis. Expression of HA-tagged eRF1 in COS-1 cells increases the amount of PP2A associated with the polysomes We have looked at the distribution of PP2A in fractionated exponentially growing COS-1 cells, as well as in COS-1 cells transiently transfected with HA-tagged eRFl. Ribosomes (80S) from these cells were obtained by high speed centrifugation of cell-free extracts through a 38% sucrose cushion (described in Materials and methods). We performed controls using antibodies specific to ribosomal (S6) and cytosolic (regulatory RII subunit of protein kinase A) proteins to evaluate successful separation (data not shown). PP2A distribution was analyzed in total cell-free extracts, and in the sucrose and ribosomal fractions. The analysis was carried out by Western blotting and activity measurements using 32P-labeled peptide (Kemptide Val6, Ala7) as a substrate (Figure 9A and B). These experiments showed that in untransfected COS-1 cells, PP2A present in the polysomes was a very small portion of total cytoplasmic PP2A activity (1-2%), which is in agreement with estimates from studies performed using rabbit reticulocyte lysates (Foulkes et al., 1983). However, overexpression of eRFi significantly increases the amount of PP2A present in the polysomes, suggesting that PP2A can be recruited to the polysomes by increasing the amount of free eRF1 available to bind to PP2Ac (Figure 9A). The data from activity measurements were confirmed subsequently by Western blot analysis (Figure 9B) of PP2A associates with eRFl A HA-eRF1 E I11 E 3500- Cl ::. 3000- eRF1- PP2Ac ow 5000. _Os= 1 I 10002: ' I 2500- wj 2000< 1500- a. W.'wI WB a-PR65 PR65 C,) Ul) 0 HA-PR65 C MOCK 4000- I~ ~ ~ ~ ~ ~ *a I1. Mock HA-eRF1 HA-eRF3 HA-PR65 -OA M + 10 nM OA B 3 E 1- u a. U) 0 a. Mock HA-eRF1 HA-eRF3 HA-PR65 -OA + 10 nM OA Fig. 7. (A) Specific PP2A activity in extracts of COS-1 cells transiently transfected with HA-tagged eRFl, eRF3 or PR65. (B) PP2A activity in corresponding anti-HA tag immunoprecipitates from 100 tg extracts. Activity measurements were performed in the absence (gray bars) and presence (black bars) of 10 nM okadaic acid, a potent PP2A inhibitor, using 32P-labeled Kemptide Val6. Ala7 as substrate as described in Materials and methods. Numbers represent the mean values (+ SEM) from three independent experiments carried out in duplicate assays. PP2A and eRFI, which showed a significant increase of PP2Ac and PR65 in COS- 1 cells following overexpression of eRFI. The slower migrating band cross-reacting with eRFI antibody represents the HA-tagged form. In contrast to previous reports on the S.cerevisiae homolog Sup45 (Stansfield et al., 1992), mammalian eRFl was shown not to be present exclusively in the polysomal fraction, but rather equally distributed between cytoplasmic and polysomal fractions, which may point to its relatively loose attachment to the 40S subunit in mammalian cells. Subsequent immunoprecipitation experiments from fractionated COS-1 cells described above, followed by activity measurements and immunoblotting, showed that PP2A dimer and associated okadaic acid-sensitive phosphatase activity are present in immunoprecipitated fractions of eRFI, confirming that indeed increased PP2A detected in the polysome fraction was associated with eRFI (Figure 9C and D). Effects of different forms of PP2A on polypeptide chain release factor activity of eRF1 Since we initially did not observe any effect of eRF1 on basal or protamine-stimulated activity of PP2A, we attempted to look for possible effects of PP2A on polypep- I2 3 4 5 6 7 WB o.-eRF1 WB a PP2Ac 8 Fig. 8. Western blot analysis of 12CA5 immunoprecipitates from COS-1 cells transiently transfected with HA-tagged human eRFI or PR65a. The upper panel was probed with peptide-specific antisera against PR65 (Ab 65 177/196), the middle panel with antisera against eRFI (Ab eRF1424/437) and the lower panel with antisera against PP2A catalytic subunit (Ab C1/20). Recombinant PR65oa, eRFI and PP2Ac, 25 ng each (lane 1), 12CA5 immunoprecipitate from 100 .tg mocktransfected cell extracts (lane 2), 12CA5 immunoprecipitates from 50, 100 and 250 .g HA-eRFI-transfected cell extracts (lanes 3, 4 and 5, respectively), 12CA5 immunoprecipitates from 50, 100 and 250 p.g HA-PR65-transfected cell extracts (lanes 6, 7 and 8, respectively). tide chain release factor activity of eRFI. Therefore, we measured the stop codon-dependent release of formyl[35S]methionine from the formyl [35S]methiony1-tRNAfmetAUG-80S substrate complex mediated by eRF in an in vitro termination assay (Tate and Caskey, 1990). In these experiments, we used recombinant human histidinetagged eRFI (His-eRFI) and GST-tagged eRF3 (GSTeRF3) purified to apparent homogeneity, as well as purified preparations of PP2Ac, PP2A-, and PP2Aj. His-eRF1 was active as a release factor on its own, but treatment with equimolar concentrations of different forms of PP2A did not have any dramatic effects on the activity of the recombinant protein (Figure 10). As previously reported for the S.cerevisiae and X.laevis homologs (Stansfield et al., 1995; Zhouravleva et al., 1995), GST-eRF3 was able to stimulate eRFI release factor activity (data not shown). We did not observe any significant effects of different PP2A preparations on eRF3-stimulated activity of His-eRF1 (data not shown). We also tested the release activity of a MonoQ-purified preparation of PP2A that contains eRFI (MQ I), as confirmed by Western blotting analysis with Ab eRF142 4437. eRF1 present in this preparation was also active in termination assays, although to a somewhat lower extent than the recombinant protein, but the activity in the presence of 10 nM okadaic acid (Sigma) was not significantly different from that in untreated samples (Figure 10). To understand further the interaction of PP2A with eRF1, we determined release activity in immunoprecipitates of HA-eRFI from transfected COS-l cells. The data so far available indicate that the activity of eRFI is extremely low (