Rapamycin, wortmannin, and the methylxanthine SQ20006 inactivate p70s6k by inducing dephosphorylation of the same subset of sites.

Activation of p70s6k in cells stimulated with serum correlates with the phosphorylation of seven sites. Pretreatment of Swiss 3T3 cells with the immunosuppressant rapamycin blocks phosphorylation of four of these sites (Thr229, Thr389, Ser404, and Ser411), whereas phosphorylation proceeds in the remaining three sites (Ser418, Thr421, and Ser424). If rapamycin is added post-serum stimulation, the pattern of phosphorylation is qualitatively similar except that Ser411 is still highly phosphorylated. The inhibitory effect of rapamycin on serum-induced p70s6k activation and the phosphorylation of Thr229, Thr389, Ser404, and Ser411 is rescued by FK506, providing further evidence that the inhibitory effect is exerted through a complex of rapamycin-FKBP12. Wortmannin treatment pre- or post-serum stimulation inhibits phosphorylation of the same set of sites as rapamycin, supporting the argument that both agents act on the same pathway. Likewise, methylxanthine phosphodiesterase inhibitors block p70s6k activation and phosphorylation of the same set of sites as wortmannin and rapamycin. However, other agents that raise intracellular cAMP levels have no inhibitory effect, leading to the hypothesis that the inhibitory actions of methylxanthines on p70s6k activity are not through activating protein kinase A but through inhibition of an upstream kinase. Together the results indicate that there are two kinase signaling pathways that must converge to activate p70s6k and that only one of these pathways is sensitive to rapamycin, wortmannin, and methylxanthine inhibition.

Activation of p70 s6k in cells stimulated with serum correlates with the phosphorylation of seven sites. Pretreatment of Swiss 3T3 cells with the immunosuppressant rapamycin blocks phosphorylation of four of these sites (Thr 229 , Thr 389 , Ser 404 , and Ser 411 ), whereas phosphorylation proceeds in the remaining three sites (Ser 418 , Thr 421 , and Ser 424 ). If rapamycin is added postserum stimulation, the pattern of phosphorylation is qualitatively similar except that Ser 411 is still highly phosphorylated. The inhibitory effect of rapamycin on serum-induced p70 s6k activation and the phosphorylation of Thr 229 , Thr 389 , Ser 404 , and Ser 411 is rescued by FK506, providing further evidence that the inhibitory effect is exerted through a complex of rapamycin-FKBP12. Wortmannin treatment pre-or post-serum stimulation inhibits phosphorylation of the same set of sites as rapamycin, supporting the argument that both agents act on the same pathway. Likewise, methylxanthine phosphodiesterase inhibitors block p70 s6k activation and phosphorylation of the same set of sites as wortmannin and rapamycin. However, other agents that raise intracellular cAMP levels have no inhibitory effect, leading to the hypothesis that the inhibitory actions of methylxanthines on p70 s6k activity are not through activating protein kinase A but through inhibition of an upstream kinase. Together the results indicate that there are two kinase signaling pathways that must converge to activate p70 s6k and that only one of these pathways is sensitive to rapamycin, wortmannin, and methylxanthine inhibition. p70 s6k and p85 s6k represent two isoforms of the same kinase that are encoded by a common gene and are identical except for a 23-amino acid extension at the amino terminus of p85 s6k (see Refs 1 and 2). Furthermore, both isoforms lie on a p21 ras -p42 mapk /p44 mapk independent signal transduction pathway (3,4), which bifurcates at the level of the receptor (4). Whereas p70 s6k seems to be restricted to the cytoplasm (5), the aminoterminal extension of p85 s6k harbors a nuclear localization signal that constitutively targets it to the nucleus (5). The major substrate of the kinase in both compartments of the cell appears to be the 40 S ribosomal protein S6 (see Ref. 6), whose multiple phosphorylation in the cytoplasm has been implicated in the selective translational up-regulation of a family of essential gene products (7,8). Consistent with this finding, either microinjection of neutralizing antibodies into cells (9) or treatment of cells with the immunosuppressant rapamycin, which selectively blocks p70 s6k /p85 s6k phosphorylation and activation (10 -12), severely impedes cell cycle progression.
Activation of p70 s6k /p85 s6k is associated with multiple phosphorylation of the enzyme, which can be monitored as the slower migration of a family of bands on SDS-PAGE, 1 which collapse into a single band following treatment with phosphatase or rapamycin (10,13). Initially, phosphorylation at four major sites, Ser 411 , Ser 418 , Thr 421 and Ser 424 , which are clustered in a putative autoinhibitory domain, was found to parallel p70 s6k activation (14). Substitution of these four residues with acidic amino acids mimicked phosphorylation at these sites (15). Unexpectedly, the acidic form of the kinase was still inactivated by rapamycin, and more surprisingly, inactivation was shown to be associated with the dephosphorylation of a distinct set of sites, with no effect observed on the phosphorylation of the four identified sites (15). Recently, these novel rapamycin-sensitive sites have been identified as Thr 229 in the catalytic domain and Thr 389 and Ser 404 located in a linker domain that couples the catalytic and autoinhibitory domains. 2 Each of the sites was found to reside in an atypical trypsin cleavage product, which largely explained the difficulty in their identification. Earlier rapamycin studies (15) led to the hypothesis that at least two sets of phosphorylation events regulate kinase activity, one of which was controlled by mitogens and a second that appeared to be constitutively activated. However, in re-examining this data, it was noted that in the presence of serum, the parent and the acidic mutant exhibited a similar mobility shift when analyzed by Western blots of SDS-PAGE (15). Furthermore, in both cases, rapamycin treatment caused a similar increase in electrophoretic mobility of the protein.
These data suggested that one or more of the rapamycin-sensitive phosphorylation sites was involved in the p70 s6k mobility shift (15). Since a similar mobility shift is observed during the rapid activation of p70 s6k by mitogens (4,10), which is blocked by rapamycin (10), this finding raises the question of whether the rapamycin-sensitive phosphorylation sites are also regulated by mitogens.
Since p70 s6k activation has not yielded to in vitro reconstitution (3,16), a number of indirect approaches have been exploited in an attempt to identify key regulatory points in this * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  signaling pathway (4,17,18). The establishment of regulatory points has been extremely valuable in elucidating other signal transduction pathways (19). In the case of p70 s6k , this line of study has led to the finding that the antibiotic wortmannin, which blocks phosphatidylinositol 3-OH kinase activation, also blocks p70 s6k activation (17,18). Though the role of phosphatidylinositol 3-OH kinase in regulating this pathway has been questioned (4,20), wortmannin clearly inhibits a key step in the p70 s6k pathway. In contrast to rapamycin, wortmannin does not inhibit TPA activation of p70 s6k , leading to a model in which the wortmannin block has been placed upstream of the site of rapamycin action (18). Exploiting a similar line, Monfar et al. (21) have recently demonstrated that raising intracellular levels of cAMP in T cells either blocks interleukin 2-induced p70 s6k activation or causes the immediate inactivation of the kinase in post-interleukin 2-stimulated cells. These findings have led to an expansion of the model above in which activation of protein kinase A acts as a key negative regulator of p70 s6k as well as phosphatidylinositol 3-OH kinase, whose activation is also blocked by raising intracellular cAMP levels (21). To raise cAMP levels, Monfar et al. (21) co-treated cells with forskolin, an activator of adenylate cyclase, and IBMX, a phosphodiesterase inhibitor. However, methylxanthines, such as IBMX, have been shown to also act as protein kinase inhibitors (22). Indeed, previous studies have demonstrated that treatment of cells with methylxanthines alone is sufficient to block S6 phosphorylation, whereas raising intracellular cAMP with prostglandin E 1 , a potent adenylate cyclase agonist, had no inhibitory effect on this response (23). These latter findings raise the possibility that methylxanthines themselves may block p70 s6k activation and, if selective, could be potential tools for identifying upstream kinases in this pathway.
The mapping of individual phosphorylation sites involved in regulating p70 s6k enables the assessment of these models and allows for the determination of whether the newly identified sites of phosphorylation are also implicated in p70 s6k activation. Here we have employed p70 s6k mutants as well as twodimensional phosphopeptide analysis of endogenous p70 s6k to establish whether any of the newly identified rapamycin-sensitive sites of phosphorylation are also involved in mitogenic activation of p70 s6k . Next we have assessed whether rapamycin and wortmannin inhibition of p70 s6k activation is paralleled by the dephosphorylation of the same or a distinct set of sites. Finally, we have examined whether methylxanthines are capable of blocking p70 s6k activation, as opposed to other agents that raise intracellular cAMP levels, and have examined the sites of p70 s6k phosphorylation affected by such treatment.

EXPERIMENTAL PROCEDURES
Plasmid Constructs and Mutagenesis-Site-directed mutagenesis resulting in the mutations of Ser 3 Asp or Ala (residues 411, 418, and 424) and Thr 3 Glu or Ala (residue 421) was performed as described previously (15). To epitope-tag the p70 s6k constructs, we introduced by polymerase chain reaction the sequence GAG CAG AAG CTT ATC TCC GAG GAG GAC CTG, corresponding to the Myc 9E10 epitope (24) after the initiation codon of the wild type p70 s6k (p2B4) (25). Exchanging Xba-BglII fragments (nucleotides 1-911) between wild type p70 s6k and p70 s6k D 3 E or p70 s6k A 4 allowed the introduction of the Myc tag in the mutant constructs.
Cell Culture, DNA Transfection, and Radioactive Labeling-Swiss mouse 3T3 cells were grown, maintained, and arrested in G O as described previously (26). For in vivo 32 P i labeling, the medium was exchanged on day 7 for 15 ml of phosphate-free DMEM containing 0.1% bovine serum albumin (for experiments employing wortmannin, bovine serum albumin was omitted from the medium), after 6 h 1 mCi of 32 P i was added, and cells were incubated overnight in the presence of the radioactive label before the various treatments were initiated. For preparation of extracts, the medium was removed from the plate, and the cells were washed twice with ice-cold buffer A (120 mM NaCl, 20 mM NaF, 10 mM pyrophosphate, 5 mM EGTA, 1 mM EDTA, 1 mM benzami-dine, 0.1 mM phenylmethylsulfonyl fluoride, 30 mM 4-nitrophenyl phosphate, 50 mM Tris-HCl, pH 8.0) and then extracted in the same buffer containing 1% Nonidet P-40. Cell extracts were centrifuged at 12,000 ϫ g for 15 min at 4°C, and the supernatant was immediately frozen in liquid N 2 and stored at Ϫ70°C. Human embryonic kidney 293 cells were seeded at 10 6 cells/10-cm dish in DMEM supplemented with 10% (v/v) FCS. After 24 h, cells were transiently transfected with 10 g of parent Myc-p70 s6k DNA or mutant Myc-p70 s6k DNA for 20 h employing a modified calcium phosphate precipitation method (27). After transfection overnight, cells were washed twice with DMEM and arrested in G O for 24 h in DMEM. Cell extracts were prepared as above.
Quantitation of 32 P i Incorporation and Analysis of Phosphopeptides-Cell extracts were quickly thawed, the kinase was immunoprecipitated with M5 antiserum (28), the precipitate was subjected to SDS-PAGE, and p70 s6k was localized by autoradiography. After the 32 P i -labeled p70 s6k was excised from the gel, the amount of 32 P i incorporated into the kinase was determined by Cerenkov counting, followed by its electroelution from the gel. Two-dimensional tryptic phosphopeptide analysis was performed as described previously (14) with the following modifications. The electroeluted protein was precipitated with 15% (w/v) trichloroacetic acid and performic acid oxidized prior to digestion (29), and no endo-Lys C digestion was performed before the addition of trypsin (3 ϫ 5 g over 24 h). The phosphopeptides were separated on the same thin layer plates, but the cellulose used by the manufacturer (Merck) was changed, resulting in slower mobility of phosphopeptides in the chromotography dimension. Also, the resultant chromatograms were visualized using a PhosphorImager and Image-Quant Software (Molecular Dynamics) rather than long term autoradiography on film.
Immunoblotting-Equal amounts of total protein from extracts were subjected to SDS-PAGE and electrophoretically transferred to an Imobilon P membrane (Millipore), and the resultant membrane was incubated overnight with M1 antibody at 4°C, with gentle agitation after blocking with 5% skim milk (Fluka) (9). The blot was decorated with an anti-rabbit horseradish peroxidase-conjugated secondary antibody and developed using the ECL system (Amersham Corp.).
Protein Determination and in Vitro Kinase Assay-The protein concentration of cell extracts was measured using a Bio-Rad D/C protein assay kit with bovine serum albumin as the standard. For immunoprecipitation of p70 s6k , equal amounts (20 g) of total protein from cell extracts were diluted in buffer A to a final volume of 200 l, and then 5 l of p70 s6k M5 antibody was added. After incubation of the mixture on ice for 2 h, the immunocomplexes were incubated with protein A-Sepharose beads for 30 min at 4°C with gentle shaking and then pelleted by centrifugation. The protein A immunocomplexes were washed four times in buffer A, washed one time with dilution buffer (30), and resuspended in 5 l of dilution buffer containing 1 mM dithiothreitol. The suspended protein A complexes were assayed for p70 s6k activity as described previously (28). Immunocomplex assays of the Myc epitope-tagged p70 s6k were performed as above except that 9E10 anti-Myc antibody and protein G-Sepharose beads were used. For total S6 kinase activity, equal amounts of total protein from extracts were assayed directly without immunoprecipitation (28). For MAP kinase assays, 150 g of cell lysate was immunoprecipitated with 10 l of a specific p42 mapk polyclonal antibody. Immunoprecipitates were then assayed for MBP activity as described previously (31). Total MAP kinase assays were carried out as above except without immunoprecipitation.

RESULTS
Autoinhibitory Domain-Recently, two mutant p70 s6k constructs were described in which the four phosphorylation sites in the autoinhibitory domain, Ser 411 , Ser 418 , Thr 421 , and Ser 424 (Fig. 1A) were mutated to either acidic residues, (p70 s6k D 3 E), or neutral residues, (p70 s6k A 4 ) (15). Transient expression of these mutants or the parent enzyme in human 293 cells demonstrated that p70 s6k D 3 E was as active as the parent construct, whereas the activity of p70 s6k A 4 was severely reduced (15), consistent with these sites playing a major role in regulating p70 s6k activity. However, if these are the only sites of phosphorylation associated with mitogen-induced kinase activation, then the activity of these two constructs should be mitogenindependent. To assess this possibility, p70 s6k , p70 s6k A 4 , and p70 s6k D 3 E were each transiently expressed in human 293 cells, the cultures were arrested in G O , and the S6 kinase activity of each construct was measured before and after serum stimulation. Even though the basal activity of p70 s6k D 3 E is higher than p70 s6k and that of p70 s6k A 4 is lower, both constructs are activated by serum (Fig. 1B). Activation was associated in all three cases with a decrease in mobility of the kinase on SDS-PAGE (data not shown). These results indicate that kinase activation may be dependent on phosphorylation of other sites.
Serum-induced Phosphorylation Sites-Consistent with the above data, in initial phosphopeptide analysis studies other phosphopeptides were detected in some preparations (14), and more recent studies have demonstrated that other phosphorylation sites play a critical role in regulating kinase activity (15). To determine whether additional phosphorylation sites were associated with p70 s6k activation, quiescent Swiss 3T3 cells were stimulated with serum for 15 min in the presence of 32 P i . Serum treatment leads to a rapid activation of p70 s6k as measured by its ability to phosphorylate 40 S ribosomal protein S6 in vitro ( Fig. 2A, lanes 1 and 2) or its slower migration on Western blots of one-dimensional SDS-PAGE ( Fig. 2A, lanes 3  and 4). The shift in electrophoretic mobility corresponds to an increase in the incorporation of phosphate into the protein (13). In quiescent cells, two forms of p70 s6k can be clearly distinguished on Western blots, designated i and ii, which, upon serum stimulation, exhibit decreased mobility, designated iii and iv ( Fig. 2A, compare lanes 3 and 4). Analysis of twodimensional tryptic phosphopeptide maps of p70 s6k from quiescent cells revealed the presence of phosphate in Ser 411 , Ser 418, Thr 421 , and Ser 424 (Fig. 2B), the four sites residing in the autoinhibitory domain (14). Serum stimulation leads to a rapid increase in the amount of 32 P i incorporated into these four sites, and the appearance of three additional phosphopeptides (Fig. 2C), which have been shown to be singly phosphorylated at Thr 229 , Thr 389 , and Ser 404 . 2 Thr 229 is situated in the catalytic domain, whereas Thr 389 and Ser 404 are located in the linker domain, which couples the catalytic and autoinhibitory domains (Fig. 1A). The appearance of Thr 229 , Thr 389 , and Ser 404 was unexpected, in that they were not detected in 3T3 cells in earlier studies (14), but it is consistent with the results described in Fig. 1B and the observation that one or more of these sites appear to be involved in the mobility shifts detected in 293 cells transiently overexpressing p70 s6k (15). In the initial study, the failure to detect these phosphopeptides in 3T3 cells was probably due to the fact that none of these sites reside in a predicted tryptic peptide, 2 suggesting the presence of a contaminating protease. Taken together the data demonstrate that all seven phosphorylation sites are rapidly induced upon mitogenic stimulation.
Effect of Rapamycin-The three peptides phosphorylated at  (51). B, the activity of parent Myc-p70 s6k (1), Myc-p70 s6k A 4 (2), or p70 s6k D 3 E (3) expressed in human 293 cells was determined in an immunocomplex assay before (stippled bars) and after (solid bars) stimulation with 10% FCS for 1 h. All activities were normalized for p70 s6k expression levels by quantitation of Western blots probed with the 9E10 anti-Myc antibody, and decorated with rabbit antimouse IgG followed by 125 I-conjugated protein A (Amersham Corp.). Quantitation of bound 125 I-protein A was done using a PhosphorImager and ImageQuant software (Molecular Dynamics).

FIG. 2. Serum-induced activation of p70 s6k in Swiss 3T3 cells.
A, p70 s6k activity in extracts from quiescent cells (lane 1) or cells stimulated with 10% FCS for 15 min (lane 2) was assayed after immunoprecipitation with p70 s6k M5 antibody as described under "Experimental Procedures." Extracts (20 g of total protein) from quiescent cells (lane 3) or cells serum-stimulated for 15 min (lane 4) were subjected to Western blot analysis. Derivatives i-iv represent increasingly phosphorylated forms of p70 s6k . Cells labeled with 32 P i were incubated in the absence (B) or presence (C) of 10% FCS for 15 min. Immunoprecipitated p70 s6k was subjected to two-dimensional tryptic phosphopeptide mapping following performic acid oxidation as described under "Experimental Procedures." Identified phosphorylation sites of p70 s6k are numbered according to the p70 s6k sequence (51), and the origin is indicated with an arrow.
Thr 229 , Thr 389 , and Ser 404 are rapidly dephosphorylated in response to rapamycin when p70 s6k is transiently expressed in 293 cells (15). To establish whether this is the case with endogenous p70 s6k in 3T3 cells, quiescent cells were stimulated with serum in the presence of the inhibitory macrolide. As previously shown (10), such treatment abolished p70 s6k activation as measured by its ability to phosphorylate S6 in an immune complex assay (Fig. 3A, compare lanes 1-3). However, in contrast to earlier findings (10), phosphate continued to be incorporated into the protein, albeit at a much lower rate (Fig.  3B, compare lanes 1-3). Analysis of the tryptic phosphopeptides derived from this material revealed that the incorporation of phosphate into Thr 229 , Thr 389 , and Ser 404 was severely inhibited by rapamycin pretreatment, but surprisingly, the same was true for Ser 411 (compare Fig. 3C with Fig. 2C), which was not sensitive to rapamycin when p70 s6k was transiently expressed in 293 cells (15). In contrast to the four rapamycinsensitive sites, the amount of phosphate in Ser 418 , Thr 421 , and Ser 424 continued to increase (compare Fig. 3C with Fig. 2C), consistent with the finding that rapamycin does not completely block the phosphorylation of the kinase (Fig. 3B). If, instead, rapamycin is added post-serum stimulation, the kinase is inactivated within 30 min (Fig. 3A) but is still highly phosphorylated (Fig. 3B). Tryptic phosphopeptide analysis reveals that the phosphorylation states of Ser 418 , Thr 421 , and Ser 424 remain constant, whereas Thr 389 and Ser 404 are no longer visible, and Thr 229 as well as Ser 411 are reduced (compare Fig. 3D and Fig.  2C). These data indicate the possible involvement of two separate pathways in the activation of p70 s6k by serum, only one of which is sensitive to rapamycin. Furthermore, the difference in phosphorylation of Ser 411 following addition of rapamycin preand post-serum treatment, suggests that the phosphorylation on Ser 411 may be dependent on Thr 229 , Thr 389 , and/or Ser 404 .
Rescue by FK506 -Inhibition of p70 s6k activity by rapamycin has been shown to be blocked by an excess of FK506 (10), arguing that this effect is exerted through a complex with the immunophillin FKBP12. If the inhibitory effect of the rapamycin-FKBP12 complex is exerted through the dephosphorylation of Thr 229 , Thr 389 , and Ser 404 as well as Ser 411 , these sites should become rephosphorylated in the presence of FK506. As the S6 kinase activity of p70 s6k in either immune complex assays or assays of total cell extracts is quantitatively equivalent (32), all subsequent assays have been carried out in cell extracts. The results show that p70 s6k activity in extracts from serum-stimulated cells that have been treated with rapamycin for 30 min can be rescued by the addition of an excess of FK506, consistent with the presence of mitogen-activated p70 s6k kinase. FK506 alone has no effect on serum-induced p70 s6k activity (Fig. 4A). In parallel, FK506 rescues the phosphorylation of Thr 389 and Ser 404 as well as Thr 229 and Ser 411 (compare Fig.  4, B and C) and, as expected, FK506 alone has little effect on the serum-induced tryptic-phosphopeptide pattern (Fig. 4D). Thus, reversal of the rapamycin inhibitory effect by FK506 is regulated through the same sites of phosphorylation.
Effect of Wortmannin-The antibiotic wortmannin has been shown to specifically block activation of both phosphatidylinositol 3-OH kinase (18) and p70 s6k (17,18), though its specificity for the p70 s6k pathway versus the p42 mapk /p44 mapk pathway has been recently challenged (33,34). Furthermore, it has been hypothesized that the inhibitory effect is exerted upstream of the rapamycin block as TPA-induced activation of p70 s6k , presumably through protein kinase C, is insensitive to wortmannin but is still blocked by rapamycin (18). However, wortmannin may also act on a rapamycin-independent signaling pathway, exerting its inhibitory effect on p70 s6k through a rapamycin-insensitive phosphorylation site. To test the role of wortmannin, we first assessed its ability to induce inactivation of p70 s6k in cells pre-treated with serum. Under these conditions, the IC 50 for p70 s6k inactivation is between 50 and 100 nM wortmannin (Fig. 5A, inset). In cells either pretreated with 200 nM wortmannin or treated with the same concentration of the antibiotic 15 min post-serum stimulation, p70 s6k activation is either blocked or returns to basal levels within 30 min (Fig.  5A). The inhibitory effect on kinase activity is paralleled by an increase in the mobility of the kinase analyzed on Western blots of SDS-PAGE (Fig. 5B). To examine the effect of wortmannin on the phosphorylation pattern of p70 s6k , cells were either pretreated with wortmannin and then stimulated with serum or stimulated with serum followed by subsequent addition of the antibiotic. The results are very similar to those obtained with rapamycin (Fig. 3, C and D); wortmannin pretreatment severely suppressed the phosphorylation of Thr 229 , Thr 389 , Ser 404 , and Ser 411 , with phosphorylation proceeding in the rapamycin-insensitive sites, whereas treatment following serum stimulation leads to dephosphorylation of Thr 389 and Ser 404 with less of an effect on Thr 229 and Ser 411 (Fig. 5, C and  D, respectively). These data support a model in which both agents are acting on the same signaling pathway.
Effects of SQ20006 -Recent studies have shown that raising cAMP levels in T cells by applying forskolin, an adenylate cyclase agonist, together with IBMX, an inhibitor of cAMP-dependent phosphodiesterase, blocks p70 s6k activation (21). However, previous studies in Swiss 3T3 cells had shown that phosphodiesterase inhibitors alone, but not other agents that raise cAMP levels, are responsible for blocking mitogen-induced S6  (A and B, lane 4, and D). A, total p70 s6k activity was measured in an in vitro immunocomplex kinase assay as in Fig. 2. One unit of kinase incorporates 1 pmol of P i /min into S6. B, the amount of 32 P i incorporated into p70 s6k was quantitated by Cerenkov counting following immunoprecipitation of the kinase as described under "Experimental Procedures." C and D, two-dimensional tryptic phosphopeptide maps of 32 P i -labeled p70 s6k from cells pretreated (C) or posttreated (D) with rapamycin was carried out as in Fig. 2. phosphorylation, the target of p70 s6k (23). The results in Fig. 6 demonstrate that neither 8-bromo-cyclic AMP nor forskolin has an effect on serum-induced p70 s6k activation, whereas IBMX had a small but significant inhibitory effect. In contrast, the methylxanthine SQ20006, a more potent inhibitor of phosphodiesterase and S6 phosphorylation (23,35), severely suppresses p70 s6k activation. If SQ20006 is added post-serum stimulation, it also induces p70 s6k inactivation as measured by its ability to phosphorylate S6 in vitro (Fig. 7A) or by its increased electrophoretic mobility on SDS-PAGE (Fig. 7B). Under these conditions SQ20006 has no effect on p42 mapk /p44 mapk activation (Fig. 7A, inset). The effect on p70 s6k activity and mobility shift suggests that SQ20006 is operating on the same phosphorylation sites as wortmannin and rapamycin. To examine this possibility the effect of SQ20006 on the pattern of p70 s6k phosphorylation was analyzed. The rapamycin-sensitive sites of phosphorylation are also sensitive to SQ20006, exhibiting approximately the same qualitative pattern if added before or after serum stimulation (Fig. 7, C and D). These results support earlier conclusions that the inhibitory effect of phosphodiesterase inhibitors is not through raising cAMP levels but instead through blocking an upstream kinase (23).
As wortmannin does not block TPA activation of p70 s6k , whereas rapamycin does, each agent has been argued to attack a unique target in the p70 s6k signaling pathway. If both agents operate on the same pathway, the wortmannin target would be situated more proximal to the cell surface receptor, with the rapamycin target situated downstream. To determine where SQ20006 acts in this pathway, cells were stimulated with TPA in the presence or absence of all three agents. The results show that wortmannin has no effect on TPA activation of p70 s6k , as shown by others (17), whereas rapamycin and SQ20006 block kinase activity (Fig. 8). In contrast to p70 s6k activation, all three agents have no effect on p42 mapk activation (Fig. 8). These results suggest that SQ20006 operates very similarly to rapamycin, possibly inhibiting a common target. DISCUSSION From the data presented here, it is evident that the phosphorylation of Thr 229 , Thr 389 , and Ser 404 are largely responsible for the mobility shifts observed on Western blots following mitogenic stimulation. The most likely reason these sites were not detected in the initial analysis (14) was that all three reside in atypically cleaved tryptic peptides. 2 This problem is further compounded by the fact that the cleavage efficiency of these peptides is poor and varies between batches of trypsin. The ability of the marcrolide to induce an equivalent increase in mobility of the wild type p70 s6k and the p70 s6k D 3 E mutant on SDS-PAGE (15) is consistent with the conclusion that the mobility shifts in p70 s6k are due to the rapamycin-sensitive sites. Furthermore, the phosphorylation of the rapamycin-insensitive sites, like the rapamycin-sensitive sites, appears to have a large impact on kinase activity (Fig. 1). Thus, as has been described for other kinases (36), only a subset of sites cause changes in electrophoretic mobility, and therefore, interpretations concerning the extent of p70 s6k phosphorylation and activation by mobility shift should be treated cautiously.
Recently Chung et al. (17), employing specific point mutants of the PDGF receptor, provided evidence for two separate signaling pathways leading to p70 s6k activation. One pathway was regulated through tyrosines 740 and 751 and hypothesized to be mediated through activation of phosphatidylinositol 3-OH kinase, whereas the second pathway was regulated by phosphorylation of tyrosines 1009 and 1021, apparently signaling through protein lipase C␥. Interestingly, the phosphatidylinositol 3-OH kinase inhibitor wortmannin only blocked signaling from tyrosines 740 and 751 and not tyrosines 1009 and 1021, while rapamycin blocked p70 s6k activation through both pathways (17). These results led to the hypothesis that, in the pathway mediated by tyrosines 740 and 751, the rapamycin block lies downstream of the wortmannin block (18). The hypothesis that both agents inhibit p70 s6k activation through different components, which are located on the same signaling pathway, is consistent with their ability to block the same set of phosphorylation sites in the kinase (Figs. [3][4][5]. Indeed, this same set of phosphorylation sites is also sensitive to SQ20006 treatment (Fig. 7).
Earlier studies had shown that phosphodiesterase inhibitors can inhibit or ablate serum-induced S6 phosphorylation (23,37). Recent results from Monfar et al. (21) employing the cAMP elevating agents forskolin and IBMX, demonstrated that the two agents together prevent interleukin 2-induced p70 s6k activation in a T cell line, leading them to conclude that this  1  and 2, B). Whole cell lysates were subjected to S6 kinase assays as described previously under "Experimental Procedures." B-D, two-dimensional tryptic phosphopeptide maps of p70 s6k were analyzed from cells treated as in A, lanes 2, 3, and 4, respectively, which had been prelabeled with 32 P i . Two-dimensional tryptic phosphopeptide analysis was carried out as in Fig. 2. inhibitory effect was exerted through raising intracellular levels of cAMP. The results presented here demonstrate that, in Swiss 3T3 cells, raising cAMP levels either by use of the nonhydrolyzable analogue 8-bromo-cAMP or an adenylate cyclase agonist has no effect on serum-induced p70 s6k activation (Fig.  6), consistent with earlier studies on S6 phosphorylation (23). However, IBMX alone had a 25% inhibitory effect on kinase activation, whereas the more potent phosphodiesterase inhibitor SQ20006 (35) had a more pronounced effect. Recent studies have shown that the structurally related 2-aminopurine analogue, olomoucine, selectively inhibits a number of cell cycleregulated kinases both in vitro and in vivo (38). Olomoucine however, had no effect on serum-induced p70 s6k activation while blocking p42 mapk /p44 mapk activation (data not shown), suggesting that SQ20006 or similar structural analogues may be useful tools in specifically analyzing the p70 s6k signal transduction pathway.
SQ20006 has a very similar p70 s6k inhibitory profile to that of rapamycin. However, earlier studies showed that SQ20006, at concentrations that completely inhibit p70 s6k activation, ablate serum-induced up-regulation of protein synthesis (23). In contrast, rapamycin only has a marginal effect on global protein synthesis (7). Although having only a small effect on general protein synthesis, rapamycin selectively suppresses the translational up-regulation of a family of mRNA that are characterized by having a polypyrimidine tract at their 5Ј transcriptional start site (7). Taken together this suggests that SQ20006 is inhibiting the function of at least one other cell component that is involved in the up-regulation of translation. Since the inhibitory effect of SQ20006 appears to be exerted at initiation (23), this component may be one of the specific factors involved in initiation of translation (39).
It is clear from the data presented here that the signaling events leading to p70 s6k activation are not all converging through a rapamycin/wortmannin/SQ20006-sensitive pathway. Instead, there appear to be at least two independent pathways required for p70 s6k activation, only one of which is blocked by the inhibitors employed here. The sites targeted by this pathway have been recently identified as Thr 229 , Thr 389 , and Ser 404 2 (Fig. 2C). In this study, the primary target of rapamycin was Thr 389 , despite the fact that phosphorylation of residues equivalent to Thr 229 play a key role in regulating the activity of a number of kinases (40). Thr 389 is situated in an unusual uncharged sequence flanked by aromatic residues: VFLGFT 389 YVAPS. Comparison of the sequence surrounding Thr 389 with sequences found to be substrates for known kinases reveals no obvious candidate for the Thr 389 kinase (41). The motifs surrounding Thr 229 and Ser 404 are similar, suggesting that all three sites may be regulated by a novel kinase. In contrast to the rapamycin-sensitive phosphorylation sites, the sites located in the autoinhibitory sequence all exhibit (S/T)P motifs. The most likely candidate for regulating these sites initially appeared to be the p42 mapk /p44 mapk (42). However, a number of studies have strongly argued against this possibility (3,4,43). It may be instead that these sites are regulated by another member of this family such as Jun kinase (44)  . Cell extracts were prepared and either assayed for (A) S6 kinase activity or (B) the mobility of p70 s6k on Western blots as described under "Experimental Procedures." C, two-dimensional tryptic phosphopeptide maps of p70 s6k from cells pretreated with wortmannin for 30 min prior to stimulation with 10% FCS for 15 min; D, maps of p70 s6k from cells first stimulated with 10% FCS for 15 min followed by the addition of wortmannin for 30 min. Cells were prelabeled with 32 P i , and extracts were prepared as in A and B. Two-dimensional phosphopeptide analysis was carried out as described in Fig. 2. nase (46), each of which lies on a distinct signaling pathway. Cell cycle-dependent kinases, as well, phosphorylate SP motifs and thus may also be implicated.
The complexity of p70 s6k activation by phosphorylation is further emphasized by the observation that phosphorylation events mediated by these pathways may be interdependent. The results presented here raise the possibility that the phosphorylation of Thr 229 and Ser 411 may be dependent on phosphorylation of Thr 389 and/or Ser 404 (Figs. 3-6). Such an interpretation may also hold true for phosphorylation of Thr 421 . This site is only found phosphorylated in the tryptic peptide Thr 421 -Lys 427 when Ser 424 is also phosphorylated (14). In contrast, the singly phosphorylated form of this peptide is only phosphorylated on Ser 424 , implying that phosphorylation at the threonine residue is dependent on prior phosphorylation at the serine. Point mutations of individual sites, combined with phosphopeptide maps will help to resolve the importance of each site in hierarchal phosphorylation. Knowledge of the phosphorylation sites and their mutual interdependence will be an important tool in identifying the upstream kinases that regulate p70 s6k activation.
Recent findings have indicated that p70 s6k /p85 s6k plays a critical role in cell cycle progression (9,10,47). Furthermore, it appears to mediate this effect through the phosphorylation of 40 S ribosomal protein S6 and the subsequent translational up-regulation of a family of mRNA transcripts that encode for components of the protein synthetic machinery (7,8). Up-regulation of specific translational components (48,49) or obstruction of gene products that regulate their function (50) can transform cells or increase their susceptibility to transformation. These observations may explain why p70 s6k is among the most highly conserved mammalian enzymes, having the identical sequence in man, mouse, rat, and rabbit. A highly regulated mechanism of p70 s6k control would be consistent with the loss of this control leading to a constitutive growth state. The use of specific inhibitors should provide invaluable tools in probing regulatory pathways that govern p70 s6k /p85 s6k activation and mechanisms that link the kinase to translational control. FIG. 7. Effect of SQ20006 on p70 s6k activity and phosphorylation state. A, whole cell lysates from Swiss 3T3 cells, first stimulated with 10% FCS for 15 min and then treated with 1.2 mM SQ20006 for the indicated times were subjected to S6 kinase assays as described under "Experimental Procedures." A (inset), Total MAP kinase activity from cell extracts of time point 0 and 30 min was done as described under "Experimental Procedures." A, equal amounts of protein (16 g) from cell extracts at indicated times were analyzed by Western blotting as described under "Experimental Procedures." C and D, two-dimensional tryptic phosphopeptide maps of p70 s6k from cells pretreated with 1.2 mM SQ20006 for 30 min prior to stimulation with 10% FCS for 15 min (C) or from cells first stimulated with 10% FCS for 15 min followed by the addition of 1.2 mM SQ20006 for 30 min (D). 32 P i -labeled p70 s6k was subjected to the two-dimensional phosphopeptide analysis as described in Fig. 2.   FIG. 8. Effect of rapamycin, wortmannin, and SQ20006 on TPA activation of p70 s6k . Swiss 3T3 cells were incubated in the absence or presence of 0.5 M TPA for 30 min following pretreatment for 30 min with either 1.2 mM SQ20006 (SQ), 200 nM wortmannin (WM), or 5 nM rapamycin (RAP). p70 s6k and MAP kinase activity were measured following immunoprecipitation of equal amounts of protein from each cell extracts as described under "Experimental Procedures."