Protein phosphatase 2A subunit assembly: the catalytic ... - Nature

5 downloads 0 Views 351KB Size Report
with polyomavirus oncogene, middle tumor antigen. (MT), a viral B-type regulatory subunit. Replacing catalytic subunit threonine 304 or tyrosine 307 with a.
Oncogene (1997) 15, 911 ± 917  1997 Stockton Press All rights reserved 0950 ± 9232/97 $12.00

Protein phosphatase 2A subunit assembly: the catalytic subunit carboxy terminus is important for binding cellular B subunit but not polyomavirus middle tumor antigen Egon Ogris1, Daryl M Gibson and David C Pallas2,3 Division of Cellular and Molecular Biology, Dana-Farber Cancer Insitute and Department of Pathology, Harvard Medical School, Boston, Massachusetts 02115, USA

The carboxy terminus of protein phosphatase 2A (PP2A) catalytic subunit is highly conserved. Seven out of the last nine residues, including two potential in vivo phosphorylation sites, threonine 304 and tyrosine 307, are completely invariant in all known PP2As. Mutational analysis of the carboxy terminus in vivo was facilitated by ecient immunoprecipitation of trimeric PP2A holoenzyme via an epitope-tagged catalytic subunit. The results indicate that the catalytic subunit carboxy terminus is important for complex formation with the PP2A 55 kDa regulatory B subunit, but not with polyomavirus oncogene, middle tumor antigen (MT), a viral B-type regulatory subunit. Replacing catalytic subunit threonine 304 or tyrosine 307 with a negatively charged amino acid abolished binding of the B subunit to the dimeric enzyme core and altered substrate speci®city. Certain other amino acid substitutions of di€erent size and/or charge also abolished or greatly reduced B subunit binding. Substitution of alanine at position 304 or phenylalanine at position 307 did not dramatically reduce B subunit binding or phosphatase activity in vitro, yet the latter substitutions are not found in naturally occurring PP2As. Thus, the wild-type residues are important for a yet unknown function in vivo. Additionally, deleting the carboxy terminal nine amino acids inhibited binding of the B subunit to the dimeric enzyme core, indicating a requirement for one or more of these amino acids for complex formation. MT interaction with the dimeric PP2A enzyme core was not inhibited by any of these mutations. Finally, unlike B subunit, MT does not activate the phosphatase activity of the PP2A heterodimer towards cdc2-phosphorylated histone H1. Keywords: phosphatase; PP2A; polyomavirus middle tumor antigen

Introduction Protein phosphatase 2A (PP2A) is a highly conserved serine/threonine phosphatase involved in the regulation of metabolism, signaling pathways and cell cycle (Cohen, 1989; Lee et al., 1991; Mayer-Jaekel et al.,

Correspondence: DC Pallas Present addresses: 1Institute of Molecular Biology, University of Vienna, Vienna, Austria; 2Department of Biochemistry and Winship Cancer Center, Emory University School of Medicine, Atlanta, GA 30322 Received 12 March 1997; revised 7 May 1997; accepted 7 May 1997

1993; Uemura et al., 1993; Sontag et al., 1993). Its native form consists of a heterodimer between a catalytic 36 kDa subunit termed C and a constant regulatory 63 kDa subunit termed A (Usui et al., 1988), which is usually further complexed with one of several additional regulatory subunits termed B, B', and B'' (Cohen, 1989). The B subunit is known to function in cell cycle progression through mitosis and in cytokinesis (Healy et al., 1991; Mayer-Jaekel et al., 1993; Uemura et al., 1993). In cells stably transformed by the middle tumor antigen (MT) of polyomavirus, MT is found in place of the B subunit in a small portion (*10%) (Ulug et al., 1992; L Haehnel and DC Pallas, unupublished data) of PP2A complexes (Pallas et al., 1990). MT/PP2A complex formation is known to be important for MT-mediated transformation (Grussenmeyer et al., 1987; Pallas et al., 1988; Glenn and Eckhart, 1995; Campbell et al., 1995), but the precise functional consequences of MT association with PP2A are still being elucidated. Detailed mutational analysis has established the A subunit as an integral part of the PP2A holoenzyme, providing structural features for complex formation with the catalytic C and regulatory B-type subunits (Ruediger et al., 1992, 1994). Although the A and C subunits are known to complex stably in the absence of the B subunit, some evidence suggests that the A and B subunits do not stably complex in the absence of C subunit (Ruediger et al., 1994). This result together with data from crosslinking experiments (Kamibayashi et al., 1992) suggests that there may be a requirement for direct B/C subunit interaction to form stable heterotrimer. Catalytic activity and substrate speci®city depend on the subunit composition of the PP2A complex (Cegielska et al., 1994; Kamibayashi et al., 1994; Mayer-Jaekel et al., 1994). For example, the B subunit greatly increases PP2A activity towards cdc2 phosphorylated histone H1 (Sola et al., 1991; Agostinis et al., 1992; Ferrigno et al., 1993; Mayer-Jaekel et al., 1994). PP2A regulation during the development in di€erent tissues may therefore occur in part through di€erential expression of the various subunits (for review, see Mumby and Walter, 1993). Little is known, however, about regulation of PP2A during cell cycle progression (Sontag et al., 1995). It is not likely to occur by di€erential expression of A, B, or C subunits because levels of these subunits appear to remain constant during the cell cycle (Ruediger et al., 1991). Other mechanisms probably exist for more dynamic regulation of PP2A subunit composition and activity. The Xenopus homolog of the mammalian PP2A inhibitor protein, SET, was recently shown to

Protein phosphatase 2A subunit assembly E Ogris et al

912

interact with B-type cyclins in vitro (Kellogg et al., 1995). It remains to be determined, however, what the molecular consequences of this interaction are for the regulation of PP2A. Inhibition of PP2A activity in vitro by phosphorylation of the catalytic subunit on tyrosine 307 or on an unidenti®ed threonine residue(s) was reported (Guo and Damuni, 1993; Chen et al., 1992). Some data suggest that these or similar modi®cations may occur in vivo in response to transformation or growth stimulation (Chen et al., 1994). Another posttranslational modi®cation of PP2A, methylation of PP2A on leucine 309 (Rundell, 1987; Lee and Stock, 1993; Favre et al., 1994; Xie and Clarke, 1994) has been reported to undergo cell cycle dependent changes (Turowski et al., 1995). However, neither in the case of PP2A phosphorylation nor PP2A methylation was the potential of these modi®cations to a€ect subunit composition examined. The nine carboxy-terminal amino acids of the PP2A C subunit, residues 301 to 309, contain tyrosine 307, the known site of phosphorylation in vitro by v-Src, and two potential sites of threonine phosphorylation, residues 301 and 304 (Table 1). Seven of these nine residues, including threonine 304 and tyrosine 307, are found in every PP2A C subunit cloned to date (Table 1). Threonine 301 is somewhat less conserved. The fact that tyrosine 307 is accessible to v-Src phosphorylation suggests that one or both of the threonines might also be accessible for phosphorylation. Furthermore, the inhibition of catalytic activity by both tyrosine and threonine phosphorylation is consistent with a potential mechanism wherein a phosphorylated carboxy-terminus (on either tyrosine or threonine) occupies the enzyme active site (Guo and Damuni, 1993; Chen et al., 1992). We therefore undertook a mutational analysis of the C subunit carboxy terminus, focusing on potential in vivo phosphorylation sites. The e€ects of these mutations on both subunit composition and PP2A activity were examined. In this report, we show that the PP2A C subunit carboxy terminus is important for Table 1 Carboxy terminal amino acids in PP2A catalytic subunits from di€erent organisms and from mutants generated for this studya Organism

Sequence of PP2A catalytic subunit carboxy-terminus

Homo sapiens Xenopus Drosophila melanogaster Arabidopsis thaliana S. cerevisiae S. pombe Acetabularia cliftonii

TRRTPDYFL TRRTPDYFL TRRTPDYFL TRKTPDYFL TRKTPDYFL ARRTPDYFL NRRTPDYFL

Single amino acid substitutions present in C subunit mutants:

D *

a

D E A F N Q K K

Selected organisms shown represent all known variations in the individual carboxy-terminal nine amino acids. Residues at positions 301 to 309 are shown. Sequences were obtained from GenBank. Residues identical in all naturally occuring PP2A catalytic subunits are shown in bold type. Single amino acid substitutions introduced into the homo sapiens C subunit using oligonucleotide-directed in vitro mutagenesis are shown in bold below the corresponding amino acid position of the wt sequence. Abbreviations for the amino acid residues are: A, Ala; D, Asp; E, Glu; F, Phe; K, Lys; L, Leu; N, Asn; P, Pro; Q, Gln; R, Arg; T, Thr; and Y, Tyr. The asterisk (*) indicates the replacement of residue 301 with a stop codon

complex formation with the B regulatory subunit but not with polyomavirus MT. Additionally, PP2A complexes containing B subunit or MT have striking di€erences in their ability to dephosphorylate cdc2 phosphorylated histone H1. The results described herein are consistent with the possibility that PP2A subunit composition may be regulated by or may regulate the covalent modi®cation of C subunit carboxy terminal residues.

Results PP2A C-subunit carboxy-terminal mutations Initially, negatively charged amino acids were introduced at the highly conserved threonine 304 (T304D) and tyrosine 307 (Y307E) in an attempt to mimic constitutive phosphorylation while alanine (T304A) or phenylalanine (Y307F) was introduced to create unphosphorylatable versions (Table 1). For comparison, a less conserved threonine at position 301 was also changed to a negatively charged amino acid (T301D). To study more thoroughly the functional constraints on substitutions at these positions, other mutants introducing neutral or positively charged amino acids were also constructed (Table 1). Finally, a mutant lacking the nine carboxy-terminal amino acids was created (T301stop). As described in Materials and methods, wt control and all mutant C-subunits were epitope-tagged on their amino-terminus with a sequence recognized by the antibody, 12CA5. All proteins were stable when translated in vitro (data not shown). Characterization of mutants in vitro Prior to analyses in vivo, several C subunit mutants were characterized in vitro with respect to two properties: (1) competence to form the core A/C heterodimer and (2) catalytic activity. To assay ability to associate with A subunit, 35S-labeled C subunits translated in vitro were tested for their ability to bind bacterially produced GST-A subunit fusion protein (GSTsubA; Figure 1) and GST alone. Phosphorimager quantitation of the results shown in Figure 1 indicates that the mutants bind to the A-subunit at a level 550% of wt. Thus, the carboxy-terminal nine amino acids are not essential for proper folding sucient to allow formation of the core PP2A A/C dimer. To test for catalytic activity, phosphatase assays (MayerJaekel et al., 1994) were performed on immunoprecipitates of the translated mutants using the peptide substrate, Kemptide. All mutants assayed demonstrated substantial phosphatase activity (data not shown). Immunoprecipitation of trimeric PP2A complexes from cells To examine PP2A complex formation in vivo, epitopetagged wt and mutant C subunits were introduced into MT-expressing NIH3T3 cells. Because MT is produced at a low level relative to PP2A in these cells, most PP2A complexes still contain B subunit. These cells allow the simultaneous study of the ability of the A/C

Protein phosphatase 2A subunit assembly E Ogris et al

Y307E

Y307F

T304D

T304A

wt-36

T301stop

heterodimer to complex with cellular B subunit or with MT. A major hurdle in the study of PP2A has been the diculty of obtaining antisera capable of immunoprecipitating the entire trimer complex. No previous study has reported ecient immunoprecipitation of trimeric PP2A containing the 55 kDa B subunit. Instead, many

studies have relied on biochemical puri®cation and/or size fractionation of PP2A complexes. The epitope-tag used in the present study was designed to circumvent this problem, and to provide a method of distinguishing endogenous and recombinant C subunits. To test the ability of anti-tag antibody (12CA5) to immunoprecipitate trimeric PP2A complexes, anti-tag immunoprecipitates from MT-transformed NIH3T3 cells expressing epitope-tagged wt C subunit were probed by immunoblotting for the presence of additional subunits (wt-36 lanes; Figures 2a and b). The A and B regulatory subunits as well as MT were readily detected, indicating that both A/C/B and A/C/MT trimeric complexes had been immunoprecipitated. C-subunit mutants di€er in their ability to form MT- or B subunit-containing complexes in vivo

C subunits GSTonly GSTsubA

+

+ +

+ +

+ +

+ +

Similar analysis of mutant PP2A C-subunit complexes in vivo revealed that C subunit carboxy-terminal residues are important for subunit assembly. Figure 2a shows results of probing immunoprecipitates of wt and mutant C-subunits for the presence of the C, A, and B subunits and for MT. Consistent with results obtained in vitro (Figure 1), all mutants bound substantial amounts of the A subunit. In contrast to the ®nding that all mutants form A/C heterodimers, several mutants were completely defective in binding B subunit. No B subunit was detectable in immunoprecipitates of T304D and Y307E, even on long exposures. Thus, the introduction of an acidic residue at either of these positions prevents B subunit association. On the other hand, substitutions of smaller, uncharged amino acids at these positions did

+ +

+

Figure 1 Complex formation of mutant PP2A C subunits with A subunit in vitro. Wt and mutant C subunits translated in vitro in the presence of 35S-methionine were incubated with Glutathione S-Transferase beads (GST) or GST-A subunit fusion protein beads (GSTsubA) following a protocol described previously by Sontag et al. (1993). After washing beads once with NP40 lysis bu€er and three times with Tris-bu€ered saline (pH 8.0), bound C subunits were detected by SDS ± PAGE and autoradiography. A representative assay is shown. The C subunits, indicated by the bracket, migrate as a doublets in these gels. This pattern of migration on SDS ± PAGE has been noted previously for endogenous PP2A C subunit (Campbell et al., 1995; Turowski et al., 1995) and does not appear to be due to degradation. The C subunit mutants are referred to by the position mutated preceded by the wt amino acid and followed by the introduced residue. wt36: wt C subunit. The lanes shown are from the same experiment but analysed in parallel on two identical gels

a

Y307Q

Y307K

T304N

T304K

T301D

wt-36

GREonly

T301stop

Y307E

Y307F

T304D

T304A

wt-36

GREonly

b

C subunit

A subunit

B subunit

PYMT

Figure 2 Ecient immunoprecipitation of trimeric PP2A complexes allows analysis of the ability of mutant C subunits to complex with A and B subunits or with MT in vivo. (a and b) Lysates from cells containing only control vector (GREonly) or epitope-tagged wt (wt-36) or mutant C subunits (see Results) were immunoprecipitated with 12CA5 antibody and analysed by SDS ± PAGE and immunoblotting. (a and b) represent two di€erent experiments that di€er only in the mutants tested. The blots were sequentially probed with antibodies recognizing the C (12CA5), A, and B PP2A subunits, and with anti-MT antibody. (a) The C subunits, indicated by the bracket, migrated as doublets in this gel. Whether doublets or a single band are seen varies from gel to gel (compare with panel b). This behavior on SDS ± PAGE has been observed previously (Campbell et al., 1995; Turowski et al., 1995), and does not appear to be due to degradation. The relative ratio of the two bands and the distance between the bands varies with individual mutants and between gels, for unknown reasons. A somewhat lower level of associated A subunit was consistently seen with the mutants, Y307E and T301stop. The data shown are representative of results obtained from four independent experiments. (b) The data shown are representative of results obtained from three independent experiments

913

Protein phosphatase 2A subunit assembly E Ogris et al

914

not hamper B subunit binding; T304A and Y307F bound B subunit at levels equal to or greater than that seen for wt. Analysis of the carboxy-terminal deletion mutant, T301 stop, demonstrated that one or more C subunit carboxy-terminal residues are essential for B subunit association; no B subunit could be detected in immunoprecipitates of this mutant, even on long exposures. Similar results with all these mutants were obtained with cell lines expressing mutant C subunits in the absence of MT (E Ogris, K Conroy and DC Pallas, unpublished data), indicating that MT does not contribute to the observed e€ect. Because MT acts like a B-type subunit, binding to the A/C heterodimer mutually exclusive with the B subunit, we investigated the ability of the C-subunit mutants to bind MT (Figure 2a). All of the mutants bound MT as eciently as wt. Variation in MT binding seen in Figure 2a paralleled the variable expression of MT in the individual cell lines (data not shown). Thus, unlike for the B subunit, the carboxy-terminal nine amino acids appear to be dispensable for MT binding. To further study the importance of phosphorylatable carboxy-terminal residues for B subunit binding, additional mutants were analysed (Figure 2b). All the mutants associated with the A subunit at wt levels and with MT in proportion to the level of MT in the individual cell lines. B subunit association, however, was again a€ected. Introduction of a basic residue at position 304 (T304K) greatly decreased B subunit association, but not to zero. Substitution of asparagine for threonine 304 (T304N) or lysine (Y307K) or glutamine (Y307Q) for tyrosine 307 had as severe an e€ect as introduction of acidic amino acids, indicating that a negative charge is not required for disruption of B subunit association. Finally, replacement of the less conserved threonine at position 301 with an aspartic acid (T301D) decreased but did not abolish B subunit binding. Di€erence in substrate speci®city between MT and normal cellular PP2A complexes When immunoprecipitates of PP2A complexes from cell lines expressing epitope tagged T304A,T304D, Y307F, Y307E or wt C subunits were tested for phosphatase activity towards phosphorylase a as a substrate, catalytic activities of 55% wt level or greater were obtained for the mutant complexes (Table 2). Thus, the almost complete inhibition (90%) of C subunit activity found by Chen and coworkers upon phosphorylation of tyrosine 307 in vitro (Chen et al., 1992) was not reproduced by substitution of a negative charge at that position. When histone H1 phosphorylated on cdc2 sites was used as a substrate, T304D and Y307E had less than 5% of wt activity, while T304A and Y307F showed activity at least that of wt (Table 2). Thus, the loss of B subunit binding (Figure 2a) correlates with a dramatic decrease in phosphatase activity against cdc2 phosphorylated histone H1. This result con®rms previous ®ndings of others regarding B subunit modulation of PP2A speci®city (Sola et al., 1991; Agostinis et al., 1992; Ferrigno et al., 1993; Mayer-Jaekel et al., 1994), but, in addition, indicates that mutation of the C subunit can indirectly generate the same phenotype. Because MT is found in place of the B subunit in some heterotrimeric PP2A complexes in MT-trans-

Table 2 Activity of PP2A complexes containing mutant C subunits towards phosphorylase a and cdc2-phosphorylated histone H1 substratesa

Mutant T304A T304D Y307F Y307E

Phosphatase activity (%wt)b phosphorylase a cdc2-phosphorylated (Mean+s.d.) histone H1 (Mean+s.d.) 55+13 103+42 114+36 87+36

150+81 2+2 107+14 3+2

a Anti-epitope tag immunoprecipitates prepared from extracts of cells expressing epitope-tagged C subunits were assayed as described in Materials and methods using phosphorylase a and cdc2-phosphorylated histone H1 as substrates. bPhosphatase activities of immune complexes containing mutant C subunits are presented as a percentage of the activity obtained in parallel for immune complexes containing wt C subunit

formed cells, we were interested in determining whether it might mimic the e€ects of the mutations that abolish B subunit binding. Cayla and coworkers had previously analysed immunoprecipitates of an undetermined mixture of polyomavirus middle and small tumor antigens (Cayla et al., 1993). They found that the ratio of PP2A activity in these immunoprecipitates towards phosphorylase a and a cdc2-phosphorylated retinoblastoma protein peptide was similar to that obtained for biochemically puri®ed A/C heterodimer. However, because the amount of PP2A in their tumor antigen immunoprecipitates was not measured, they only obtained a ratio of activities towards these two substrates. Thus the direct e€ect of the tumor antigens on PP2A activity towards each of these substrates was not known nor the relative contribution of the two tumor antigens. In the present study, the ability to directly immunoprecipitate equivalent amounts of catalytic subunit complexed primarily with B subunit or with MT allowed the comparison of these heterotrimers from the same cells. MT and anti-tag immunoprecipitates were prepared from cells expressing MT and epitope tagged wt Csubunit. The anti-tag immunoprecipitates contain PP2A complexes predominantly complexed with B subunit (E Ogris, K Conroy, and DC Pallas, unpublished data), while the MT immunoprecipitates contain C and A subunits, but no B subunit (Pallas et al., 1990). The two immunoprecipitates were normalized for amount of C-subunit by immunoblotting and activity towards phosphorylase a and cdc2phosphorylated histone H1 was measured as described in Materials and methods. MT decreased activity towards both substrates (Table 3). Approximately threefold less speci®c activity was observed with MT-associated PP2A relative to B subunitcontaining PP2A when phosphorylase a was used as substrate. However, 40-fold less activity towards the histone H1 substrate was observed with MTassociated PP2A relative to B subunit-containing PP2A, indicating that MT does not activate the A/ C heterodimer towards cdc2-phosphorylated histone H1. If papovavirus small tumor antigens have a similar e€ect on these substrates, it may explain in part how ST stimulates large tumor antigen (LT)directed viral replication in vivo by replacing some B subunits, given that LT is positively activated in replication by phosphorylation of a cdc2-speci®c site (McVey et al., 1989).

Protein phosphatase 2A subunit assembly E Ogris et al

Table 3 MT-associated PP2A activity towards phosphorylase a and cdc2-phosphorylated histone H1 substratesa

Experiment 1 2 3 4 5 6 7 Mean+s.d.

MT-associated phosphatase activity (%)b cdc2-phosphorylated phosphorylase a histone H1 25 35 7 40 15 62 33

2 2 1 5 2 4 2

31+18

2.5+1.4

a

PP2A activity present in MT immunoprecipitates and anti-C subunit (epitope tag) immunoprecipitates was measured using phosphorylase a and cdc2-phosphorylated histone H1 as substrates as described in Materials and methods. bMT-associated activity is presented as a percentage of the activity obtained for an equivalent amount of catalytic subunit immunoprecipitated directly with anti-epitope tag antibody (contains predominantly B subunit)

Discussion Analysis of PP2A subunit composition in vivo has been hampered by the inability to eciently immunoprecipitate trimeric PP2A complexes containing the B regulatory subunit. We report here the successful immunoprecipitation of catalytically active trimeric complexes utilizing an epitope-tagged C subunit. This approach has been used in the present study to analyse the e€ects of C subunit mutations on trimeric complex formation and activity. It should also facilitate future studies such as analysis of PP2A subunit composition across the cell cycle or in response to speci®c stimuli. Previous evidence suggests that the A and B subunits do not stably complex in the absence of C subunit (Ruediger et al., 1994). Our results strongly suggest that the molecular basis of this dependence is, at least in part, a requirement for an interaction between the C subunit carboxy terminus and the B subunit. This interaction would explain the observation by ourselves (E Ogris, Q Feng, L Chou and DC Pallas, unpublished data) and others (Ulug et al., 1992) that antisera directed to the C subunit carboxy terminus immunoprecipitate PP2A containing MT but not B subunit. The formal possibility also exists that mutation of, or antibody binding to, the C terminus could be having an indirect e€ect. Binding of B subunit to the C subunit carboxy terminus correlates with activation of PP2A towards cdc2-phosphorylated histone H1, and might be part of the molecular mechanism of this regulation. However, the nine carboxy-terminal residues are not simply inhibiting cdc2 site dephosphorylation in the absence of B subunit, because the mutant deleted in these residues (T301 stop) is not highly activated towards cdc2 phosphorylated histone (E Ogris, I Mudrak, K Conroy and DC Pallas, unpublished data). The fact that substitution of alanine for threonine at position 304 does not decrease B subunit binding indicates that this threonine is not essential for complex formation with this subunit. However, because changing the threonine to certain other amino acids such as aspartate abolishes B subunit association, this residue is clearly in a position to interfere directly or indirectly with binding if modi®ed appropriately. Tyrosine 307 may be involved

in B subunit binding. Although the hydroxyl group of tyrosine 307 is dispensable for B subunit binding, the ®nding that three non-aromatic amino acid substitutions no longer bind detectable B subunit suggests that the aromatic ring may be important. The dependence of B subunit association on C subunit carboxy-terminal residues could theoretically provide a basis for mechanisms regulating B subunit binding to the A/C heterodimer. This would introduce a dynamic dimension in the regulation of PP2A complex formation. Proteolytic clipping of the C subunit carboxy-terminus, a reaction that is known to occur in cell extracts, or modi®cation of carboxyterminal amino acids by phosphorylation or methylation (Lee and Stock, 1993; Favre et al., 1994; Xie and Clarke, 1994), could possibly regulate B subunit association. According to this hypothesis, the fact that MT association with the A/C heterodimer is fundamentally di€erent than for the B subunit might provide an explanation for how MT could escape normal cell cycle control of PP2A. It will be very interesting to assay the e€ects of these C subunit mutations on the formation of PP2A trimeric complexes containing other B-type subunits in order to determine whether the association of these subunits is a€ected similarly or di€erently compared to the B subunit and to MT. MT is known to mimic normal cellular proteins such as growth factor receptors in its actions, except with altered regulation. It would not be unlikely to ®nd one or more cellular B-type subunits that, like MT, display binding independent of the C subunit carboxy-terminus. This possibility is strengthened by recent data of McCright and Virshup obtained using the yeast two-hybrid system, which suggest that the B'' regulatory subunit, unlike the B subunit, may bind to the A subunit independently of C subunit (McCright and Virshup, 1995). It is also possible that B subunit interaction with the C subunit carboxy-terminus in vivo interferes with addition or removal of regulatory covalent modifications on the C subunit. For example, when bound, the B subunit may prevent methylation or demethylation of the carboxy-terminal leucine by sterically hindering the association of the relevant enzyme with the necessary residue(s). MT or another B-type subunit that does not interact with the C subunit carboxyterminus might not disallow the same reactions, leading to di€erential regulation. Such a possibility points to the importance of analysing the modi®cation state of PP2A associated with di€erent B-type subunits including MT. Why are threonine 304 and tyrosine 307 absolutely conserved? Substitution of alanine for the threonine or phenylalanine for the tyrosine does not greatly a€ect B subunit binding or enzyme activity, yet these substitutions have not been found in nature despite the large number of PP2A cDNAs analysed. It is possible that these residues serve a constitutive function that we are not aware of. Greater e€ects on activity may be found using a substrate other than the ones we used, or perhaps these precise residues are important for binding of other subunits, or for recognition of the catalytic subunit by the PP2A methyl transferase or demethylase. An alternative, but not mutually exclusive, possibility is that there is an essential role for reversible modi®cation of these residues during the cell

915

Protein phosphatase 2A subunit assembly E Ogris et al

916

cycle. With the ability to immunoprecipitate PP2A complexes now in hand, studies to test several of these possibilities are currently underway.

Materials and methods Plasmids and mutagenesis Wild-type (wt) PP2A C subunit cDNA (Stone et al., 1988) with an NcoI site containing the start ATG (KS Campbell, TM Roberts and DC Pallas, unpublished) was cloned into pcDNA I Amp (Invitrogen), together with a doublestranded oligonucleotide which introduced at the 5' end the coding sequence for a nine amino acid peptide (TyrPro TyrAspValProAspTyrAla) from in¯uenza hemagglutinin followed by the thrombin recognition site (LeuValProArgGlySer). Site-directed mutagenesis was performed on C subunit cDNAs cloned in pcDNA I Amp vector according to the manufacturer's instructions using the Muta-Gene Phagemid In Vitro Mutagenesis Kit (BIO-RAD). Deletion of the Cterminus was performed by standard cloning techniques. The entire cDNA of every mutant was sequenced to con®rm successful mutagenesis and to ensure that no additional mutation occurred. Wt and mutant C subunit cDNAs including the epitope tag sequence were cloned into the dexamethasone-inducible vector, pGRE 5-2 (Mader and White, 1993). An inducible vector was chosen to try to minimize the potential deleterious e€ects of wild-type and mutant C subunits (if any) while lines were being carried in culture, and to provide for an uninduced control in analyses of their e€ects. Cells and cell culture NIH3T3 lines expressing wt polyomavirus MT and a geneticin resistance gene (Cherington et al., 1986) were transfected by the calcium phosphate precipitation method (Sambrook et al., 1989). Twenty mg of vector only or vector containing wt or mutant C subunits were cotransfected together with 2 mg of a plasmid conferring resistance to hygromycin B. Selection medium contained 300 mg/ml hygromycin B and 400 mg/ml geneticin. Clones were isolated after 14 days and expanded to mass cultures. After 24 h induction with 25 mM dexamethasone, individual clones were tested for expression of epitope-tagged protein by immunoblotting with 12CA5 antibody. Cell lines producing the recombinant protein were maintained at 378C in Dulbecco's modi®ed Eagle's medium (DMEM)/ 10% calf serum containing hygromycin B and geneticin. All C subunits expressed at 10 ± 50% of the level of endogenous wt C subunit. Of note, basal levels were substantial, with very little induction in the presence of dexamethasone.

Immunoprecipitations and immunoblotting Logarithmically growing cells were induced 24 h with 25 mM dexamethasone. A 150 mm dish of each cell line was lysed in 1.0 ml NP40 lysis bu€er (1% Nonidet P-40; 10% (vol/vol) glycerol; 135 mM NaCl; 20 mM Tris, pH 8.0; 1 mM phenylmethylsulfonyl ¯uoride; 0.03 units/ml aprotinin) for 15 min with rocking at 48C. Lysates were scraped and cleared at 13 000 g and normalized for the expression of epitope tagged C subunits by immunoblotting. Control, vector-only cell lysate was used in an amount equal to that required for the lowest expressing C subunit cell line. Epitope-tagged C subunits were immunoprecipitated with 12CA5 antibody crosslinked to protein A-Sepharose beads (Harlow and Lane, 1988). Immune complexes were washed once with NP40 lysis bu€er, three times with Tris-bu€ered saline, and then were analysed by 10% SDS ± PAGE (Laemmli, 1970). Immunoblotting (Towbin et al., 1979) was performed with the mouse monoclonal anti-tag antibody, 12CA5 (1 : 1000), rabbit anti-B subunit and anti-A subunit antibodies from our laboratory or the laboratory of Dr Brian Wadzinski (1 : 5000), and mouse monoclonal anti-MT antibody, F4 (1 : 2500) (Pallas et al., 1986). Immunoblots were developed with enhanced chemiluminescence (Amersham). Phosphatase assay Phosphatase activity present in anti-epitope tag immunoprecipitates or anti-MT immunoprecipitates prepared from cells was assayed using phosphorylase a and Histone H1. g-32P-labeled phosphorylase a substrate was prepared from phosphorylase b according to the manufacturer's (GibcoBRL) instructions. Histone H1 was phosphorylated by mitotic p34cdc2 puri®ed from Nocodazole arrested HeLa cells as described (Mayer-Jaekel et al., 1994). Lysates used for immunoprecipitation were equilibrated according to epitope-tagged C subunit expression levels. Assays were performed at a linear range and with subsaturating amounts of each substrate.

Acknowledgements We thank Brian Wadzinski for antibodies to the PP2A A and B subunits, Brian Hemmings for the C subunit cDNA, John White for the pGRE vector, Kasey Nelson and Ingrid Mudrak for technical assistance, and Thomas Roberts, Geo€rey Cooper, Jeremy Green, Robert Jordan and Anh Nguyen-Huynh for critical reading of the manuscript. This work was supported by a National Institutes of Health grant to DCP (CA57327). During much of these studies EO was supported by an Erwin SchroÈdinger Fellowship from Austrian Fonds zur FoÈrderung der Wissenschaftlichen Forschung.

References Agostinis P, Derua R, Sarno S, Goris J and Merlevede W. (1992). Eur. J. Biochem., 205, 241 ± 248. Campbell KS, Auger KR, Hemmings BA, Roberts TM and Pallas DC (1995). J. Virol., 69, 3721 ± 3728. Cayla X, Ballmer-Hofer K, Merlevede W and Goris J. (1993). Eur. J. Biochem., 214, 281 ± 286. Cegielska A, Sha€er S, Derua R, Goris J and Virshup DM. (1994). Mol. Cell. Biol, 14, 4616 ± 4623. Chen J, Martin BL and Brautigan DL. (1992). Science, 257, 1261 ± 1264. Chen J, Parsons S and Brautigan DL. (1994). J. Biol. Chem., 269, 7957 ± 7962.

Cherington V, Morgan B, Spiegelman BM and Roberts TM. (1986). Proc. Natl. Acad. Sci. USA, 83, 4307 ± 4311. Cohen P. (1989). Annu. Rev. Biochem., 58, 453 ± 508. Favre B, Zolnierowicz S, Turowski P and Hemmings BA. (1994). J. Biol. Chem., 269, 16311 ± 16317. Ferrigno P, Langan TA and Cohen P. (1993). Mol. Biol. Cell., 4, 669 ± 677. Glenn GM and Eckhart W. (1995). J. Virol., 69, 3729 ± 3736. Grussenmeyer T, Carbone-Wiley A, Scheidtmann KH and Walter G. (1987). J. Virol., 61, 3902 ± 3909. Guo H and Damuni Z. (1993). Proc. Natl. Acad. Sci. USA, 90, 2500 ± 2504.

Protein phosphatase 2A subunit assembly E Ogris et al

Harlow E and Lane D. (1988). Antibodies: a laboratory manual. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY. Healy AM, Zolnierowicz S, Stapleton AE, Goebl M, DePaoli-Roach AA and Pringle JR. (1991). Mol. Cell. Biol., 11, 5767 ± 5780. Kamibayashi C, Estes R, Lickteig RL, Yang SI, Craft C and Mumby MC. (1994). J. Biol. Chem., 269, 20139 ± 20148. Kamibayashi C, Lickteig RL, Estes R, Walter G and Mumby MC. (1992). J. Biol. Chem., 267, 21864 ± 21872. Kellogg DR, Kikuchi A, Fujii-Nakata T, Turck CW and Murray AW. (1995). J. Cell. Biol., 130, 661 ± 673. Laemmli UK. (1970). Nature, 227, 680 ± 685. Lee J and Stock J. (1993). J. Biol. Chem., 268, 19192 ± 19195. Lee TH, Solomon MJ, Mumby MC and Kirschner MW. (1991). Cell, 64, 415 ± 423. Mader S and White JH. (1993). Proc. Natl. Acad. Sci. USA, 90, 5603 ± 5607. Mayer-Jaekel RE, Ohkura H, Ferrigno P, Andjelkovic NKS, Uemura T, Glover DM and Hemmings BA. (1994). J. Cell. Sci., 107, 2609 ± 2618. Mayer-Jaekel RE, Ohkura H, Gomes R, Sunkel CE, Baumgartner S, Hemmings BA and Glover DM. (1993). Cell, 72, 621 ± 633. McCright B and Virshup DM. (1995). J. Biol. Chem., 270, 26123 ± 26128. McVey D, Brizuela L, Mohr I, Marshak DR, Gluzman Y and Beach D. (1989). Nature, 341, 503 ± 507. Mumby MC and Walter G. (1993). Physiol Rev., 73, 673 ± 699. Pallas DC, Cherington V, Morgan W, DeAnda J, Kaplan D, Scha€hausen B and Roberts TM. (1988). J. Virol., 62, 3934 ± 3940.

Pallas DC, Schley C, Mahoney M, Harlow E, Scha€hausen BS and Roberts TM. (1986). J. Virol., 60, 1075 ± 1084. Pallas DC, Shahrik LK, Martin BL, Jaspers S, Miller TB, Brautigan DL and Roberts TM. (1990). Cell, 60, 167 ± 176. Ruediger R, Hentz M, Fait J, Mumby M and Walter G. (1994). J. Virol., 68, 123 ± 129. Ruediger R, Roeckel D, Fait J, Bergqvist A, Magnusson G and Walter G. (1992). Mol. Cell. Biol., 12, 4872 ± 4882. Ruediger R, Van Wart Hood JE, Mumby M and Walter G. (1991). Mol. Cell. Biol., 11, 4282 ± 4285. Rundell K. (1987). J. Virol., 61, 1240 ± 1243. Sambrook J, Fritsch EF and Maniatis T. (1989). Molecular cloning: a laboratory manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Sola MM, Langan T and Cohen P. (1991). Biochim. Biophys. Acta, 1094, 211 ± 216. Sontag E, Fedorov S, Kamibayashi C, Robbins D, Cobb M and Mumby M. (1993). Cell, 75, 887 ± 897. Sontag E, Nunbhakdi-Craig V, Bloom GS and Mumby MC. (1995). J. Cell Biol., 128, 1131 ± 1144. Stone SR, Mayer R, Wernet W, Maurer F, Hofsteenge J and Hemmings BA. (1988). Nucleic Acids Res., 16, 11365. Towbin H, Staehelin T and Gordon J. (1979). Proc. Natl. Acad. Sci. USA, 76, 4350 ± 4354. Turowski P, Fernandez A, Favre B, Lamb NJ and Hemmings BA. (1995). J. Cell. Biol., 129, 397 ± 410. Uemura T, Shiomi K, Togashi S and Takeichi M. (1993). Genes Dev., 7, 429 ± 440. Ulug ET, Cartwright AJ and Courtneidge SA. (1992). J. Virol., 66, 1458 ± 1467. Usui H, Imazu M, Maeta K, Tsukamoto H, Azuma K and Takeda M. (1988). J. Biol. Chem., 263, 3752 ± 3761. Xie H and Clarke S. (1994). J. Biol. Chem., 269, 1981 ± 1984.

917