Regulation of the p85/p110α Phosphatidylinositol 3′-Kinase

Our previous studies on the p85/p110α phosphatidylinositol 3-kinase showed that the p85 regulatory subunit inhibits the p110α catalytic subunit, and that phosphopeptide activation of p85/p110α dimers reflects a disinhibition of p110α (Yu, J., Zhang, Y., McIlroy, J., Rordorf-Nikolic, T., Orr, G. A., and Backer, J. M. (1998) Mol. Cell. Biol. 18, 1379–1387). We now define the domains of p85 required for inhibition of p110α. The iSH2 domain of p85 is sufficient to bind p110α but does not inhibit it. Inhibition of p110α requires the presence of the nSH2 domain linked to the iSH2 domain. Phosphopeptides increase the activity of nSH2/iSH2-p110α dimers, demonstrating that the nSH2 domain mediates both inhibition of p110α and disinhibition by phosphopeptides. In contrast, phosphopeptides did not increase the activity of iSH2/cSH2-p110α dimers, or dimers composed of p110α and an nSH2/iSH2/cSH2 construct containing a mutant nSH2 domain. Phosphopeptide binding to the cSH2 domain increased p110α activity only in the context of an intact p85 containing both the nSH2 domain and residues 1–322 (the SH3, proline-rich and breakpoint cluster region-homolgy domains). These data suggest that the nSH2 domain of p85 is a direct regulator of p110α activity. Regulation of p110α by phosphopeptide binding to the cSH2 domain occurs by a mechanism that requires the additional presence of the nSH2 domain and residues 1–322 of p85.

PI 1 3Ј-kinases form a diverse family of lipid kinases that phosphorylate phosphatidylinositol at the D3-position (1). The regulation of the p85/p110 PI 3Ј-kinase is particularly complex. The p85 regulatory subunit contains an N-terminal SH3 domain followed by a proline rich domain, a breakpoint cluster region-homology domain, a second proline-rich domain, and two SH2 domains linked by a putative coiled coil domain (the inter-SH2 or iSH2 domain) that binds to the N terminus of the p110␣ catalytic subunit (2)(3)(4). The binding of proteins such as CDC42 (to the BCR homology domain), Fyn and Lyn (to the proline-rich domains), and p21-ras (to p110␣) increases the activity of p85/p110 dimers in vitro (5)(6)(7). p85/p110 activity is also increased when the two SH2 domains bind to phosphoproteins containing appropriate phosphotyrosyl motifs (8 -10).
We previously examined the effect of p85 on p110␣ in mammalian cells and in vitro (11), and experimentally distinguished two effects of p85 on p110␣: inhibition of its lipid kinase activity and stabilization against thermal inactivation. The inhibition of p110␣ by the p85 subunit is clearly seen during in vitro reconstitution experiments, where dimerization with p85 decreases p110 activity by 80%. The activity of p85/p110␣ dimers is increased when phosphotyrosyl peptides bind to the p85 SH2 domains, but to a level no greater than that of the corresponding amount of monomeric p110␣. These data suggest that phosphopeptide activation of p85/p110 dimers reflects a transition between inhibited and disinhibited states.
In addition to its inhibition of p110␣, p85 stabilizes p110␣ against thermal denaturation. Recombinant p110␣ monomers lose activity rapidly when incubated at 37°C, whereas p85/ p110␣ dimers are stable, and coexpression of p85 with p110␣ in mammalian cells significantly increases the half-life of p110␣ (11). The rapid inactivation of p110␣ at 37°C explains a longstanding discrepancy between mammalian and insect cells (12): monomeric p110␣ is active in insect cells but not mammalian cells because of differences in their culture temperatures (27 versus 37°C). However, the stabilization of p110␣ by p85 in mammalian cells is mimicked by addition of bulky epitope tags to the N terminus of p110␣ (11). Thus, the stabilization by p85 appears to involve the overall conformation of p110␣ rather than the induction of a specific activated state.
In this paper, we examine the regulation of p110␣ by p85 in detail. We find that the iSH2 domain of p85 is sufficient to bind p110␣ but does not affect its activity. Inhibition of p110␣ requires the presence of the nSH2 domain linked to the iSH2 domain. Phosphopeptide binding to the nSH2 domain can directly modulate p110␣ activity. In contrast, phosphopeptide binding to the cSH2 domain modulates p110␣ activity by a mechanism that requires residues 1-322 of p85 (the SH3, BCR homology and proline-rich domains) and the nSH2 domain. These data suggest that the nSH2 domain is the principle regulator of p85/p110␣ activity.

Construction of Mutant p85 and p85
Fragments-The GST-nSH2/ iSH2/cSH2 construct (in pGEX-3X) has been previously described (8). The GST-iSH2 construct was created by PCR amplification of residues 431-600 of human p85, followed by subcloning into pGEX-2T. The nSH2/iSH2 construct was created by introducing two STOP codons after the codon corresponding to p85 residue 600 in the GST-nSH2/ iSH2/cSH2 construct. The GST-iSH2/cSH2 construct was produced by PCR amplification of residues 431-724 of p85, followed by subcloning into pGEX-2T. Full-length N-terminal HA-tagged p85 (10) was directly subcloned into pGEX-2T, and brought into frame by digestion with BamHI (present in the pGEX multiple cloning site) and EagI (present in the HA tag), filling in with Klenow polymerase, and the blunt ligation. Point mutations in the nSH2 domain (R358A) and cSH2 domain (R659A) of the pGEX-nSH2/iSH2/cSH2 construct were introduced, and XhoI/EcoRI fragments from these mutated constructs were * This work was supported by grants from the American Diabetes Association, and National Institutes of Health Grant GM55692 (to J.M.B.). 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 U.S.C. Section 1734 solely to indicate this fact.
‡ Supported by a fellowship from the Howard Hughes Medical Foundation.
§ Established Scientist of the American Heart Association, New York Affiliate and a recipient of a Scholar Awards from the Irma T. Hirschl Trust. To whom correspondence should be addressed: Dept. of Molecular Pharmacology, Albert Einstein College of Medicine, 1300 Morris Park Ave., Bronx, NY 10461. Tel.: 718-430-2153. 1 The abbreviations used are: PI, phosphatidylinositol; GST, glutathione S-transferase; BCR, breakpoint cluster region; HA, hemagglutinin.
subcloned into pGEX-p85 to produce pGEX-p85-R358A and pGEX-p85-R659A. All mutations were produced with the Quick-Change mutagenesis kit and confirmed by sequencing.
p110␣ Binding Assays-Recombinant N-myc-p110␣ was absorbed onto anti-Myc (Oncogene Science)-Protein G beads. N-myc-p110 beads or control beads were washed 3 times in 10 mM Tris, pH 7.5, 100 mM NaCl, 1 mM EDTA, and then incubated with nSH2/iSH2/cSH2 or iSH2 fragments (10 g) for 1 h at 4°C. After 3 more washes, absorbed proteins were analyzed by Western blotting with anti-iSH2 antibodies (13) followed by 125 I-Protein A. Alternatively, GST-fusion proteins were immobilized on glutathione-Sepharose, washed, and incubated with Nonidet P-40 lysates from HEK 293T cells transfected with myc-p110␣ (20 g DNA/10 cm dish, lysed 48 h after transfection). The samples were washed and absorbed proteins were analyzed by blotting with monoclonal anti-myc antibodies (Oncogene Science)/rabbit anti-mouse secondary antibodies followed by 125 I-Protein A, and reblotting with rabbit anti-GST antibodies.
p110␣ Inhibition and Activation Assays-35 l of lysate from control Sf-9 cells or Sf-9 cells expressing N-myc-p110␣ were incubated with 5 g of recombinant p85 or p85 fragment for 1 h at 4°C. The mixtures were then assayed directly for PI 3-kinase activity using sonicated phosphatidylinositol as a substrate as described previously (11). Alternatively, mixtures of p110␣ and p85 or fragments of p85 were incubated in the absence or presence of a bisphosphotyrosyl peptide derived from the Tyr-608/Tyr-628 region of IRS-1 (1 M) and assayed for PI 3Ј-kinase activity as described previously (10,11).
Phosphopeptide Binding Studies-p85 fragments containing point mutations in the nSH2 or cSH2 domains (3 g) were incubated in the presence of varying concentrations of unlabeled bisphosphotyrosyl peptide plus 0.5 Ci of a 125 I-Bolton-Hunter-labeled bisphosphotyrosyl peptide containing the photoactivatable amino acid benzoylphenylalanine (10). After 30 min at 4°C, the samples were irradiated on ice at 350 nm for 1 h and analyzed as described previously (10).

RESULTS
The C-terminal Half of p85 Is Sufficient for Inhibition of p110␣-p110␣ is maximally active as a monomer, and is inhibited by dimerization with p85 (11). To determine the domains of p85 required for this inhibition, we compared the activity of p110␣ when reconstituted with a fragment of p85 containing the C-terminal half of p85 (nSH2-iSH2-cSH2), or the iSH2 domain of p85 alone (Fig. 1A). Work from several laboratories have identified the iSH2 domain of p85 as the region that binds to p110␣ (14 -17). Consistent with these observations, both the nSH2-iSH2-cSH2 and iSH2 fragments of p85 bound to immobilized p110␣-beads, but not to control beads (Fig. 1B). Incubation of p110␣ with the nSH2-iSH2-cSH2 fragment inhibited its activity by approximately 80% (Fig. 1C, lane b), and subsequent incubation of bisphosphotyrosyl peptide with either the p85/p110␣ or nSH2-iSH2-cSH2/p110␣ dimers increased their activity by 100% (Fig. 1D, lanes a and b). Surprisingly, the iSH2 domain could bind p110␣ ( Fig. 1B) but had no effect on its activity (Fig. 1C, lane c). These data show that binding by the iSH2 domain to p110␣ is not sufficient to inhibit its catalytic activity, and suggest that the SH2 domains exert an additional constraint on p110␣ activity.
Inhibition of p110␣ by Single SH2-iSH2 Fragments-Dhand et al. (14) predicted that the iSH2 domain of p85 should exist as two antiparallel helices, which would place the SH2 domains of p85 is close apposition. It is possible that the two SH2 domains are close enough contact each other, thereby imposing a strained conformation of the iSH2 domain. This strain would be relieved by conformational changes in either SH2 domain upon phosphopeptide binding. Alternatively, each SH2 domain could independently affect the conformation of the iSH2 domain or p110␣. Both models are consistent with previous data showing partial activation from either SH2 domain (10). However, the first model predicts that both SH2 domains would be required for inhibition of p110␣, whereas the second model predicts that the iSH2 domain linked to either SH2 domain would partially inhibit p110␣.
To distinguish between these models, we measured p110␣ activity in the presence of p85 fragments containing the iSH2 domain linked to single SH2 domains ( Fig. 2A). We first established that the fragments bound to p110␣. We incubated recombinant nSH2/iSH2 or iSH2/cSH2 with control beads or beads containing immobilized p110␣ (Fig. 2B, left panel). Alternatively, we incubated recombinant p110␣ (produced in HEK 293T cells) with immobilized GST-nSH2/iSH2 and GST-iSH2/cSH2 (Fig. 2B, right panel). Both constructs bound to p110␣, although binding by the nSH2/iSH2 construct was slighter higher than binding by the iSH2/cSH2 construct. We next measured p110␣ activity in the presence of the two constructs. The nSH2-iSH2 fragment inhibited p110␣ to the same extent as the nSH2-iSH2-cSH2 fragment, suggesting that contact between the SH2 domains is not required for inhibition of p110␣ (Fig. 2C, lanes b and c). Moreover, the activity of both nSH2-iSH2/p110␣ dimers and nSH2-iSH2-cSH2/p110␣ dimers were increased 100% by tyrosine phosphopeptides (Fig. 2D,  lanes a and b). These data show that the nSH2-iSH2 domain fragment is sufficient to mediate inhibition and phosphopeptide disinhibition of p110␣. Surprisingly, dimerization of p110␣ with the iSH2-cSH2 fragment was only slight inhibitory (Fig. 2C, lane d), and no increase in activity was seen in the presence of tyrosine phosphopeptide (Fig. 2D, lane c). These data suggest that the Cterminal SH2 domain contributes little to the inhibition of p110␣, which is primarily because of the N-terminal SH2 domain.
Role of individual SH2 domains in phosphopeptide regula-tion of p110␣/nSH2-iSH2-cSH2 dimers-We have previously shown that in dimers containing p110␣ and intact p85, both SH2 domains contribute to the increase in activity caused by tyrosine phosphopeptides (10). We therefore examined the activity of p110␣ when bound to nSH2-iSH2-cSH2 constructs containing disabling point mutations in the conserved FLVRES motifs of the N-terminal or C-terminal SH2 domains (R358A FIG. 2. Inhibition of p110␣ by the iSH2 domain linked to single SH2 domains. A, nSH2/iSH2/cSH2, nSH2/iSH2 and iSH2/cSH2 structures. B, left panel: control beads or myc-p110 beads were incubated with recombinant nSH2/iSH2 or iSH2/cSH2 (10 g) as in Fig. 1, and bound proteins were analyzed by blotting with anti-iSH2 antibodies; right panel: control glutathione-Sepharose beads or beads containing immobilized GST-nSH2/iSH2 or GST-iSH2/cSH2 (5 g) were incubated with lysates from HEK 293T cells expressing myc-p110␣. Bound proteins were analyzed by sequential blotting with antimyc and anti-GST antibodies. C, p110␣ was incubated for 30 min at 4°C with recombinant nSH2/iSH2/cSH2, nSH2/ iSH2, or iSH2/cSH2. The mixtures were then assayed for PI 3Ј-kinase activity, and expressed as % of p110␣ alone. The data are the mean Ϯ S.E. from four experiments. D, the mixtures were incubated for an additional 1 h in the absence or presence of 1 M bisphosphotyrosyl peptide, and PI 3Ј-kinase kinase activity was measured. The data are expressed as stimulation over the activity of each mixture in the absence of phosphopeptide. The data are the mean Ϯ S.E. from four experiments.
The inability of the cSH2 domain to mediate phosphopeptide activation of the nSH2(R358A)-iSH2-cSH2/p110␣ dimers was inconsistent with our earlier experiments showing that both the nSH2 and cSH2 domains contributed to phosphotyrosyl peptide activation of p85/p110␣ dimers (10). The discrepancy could be because of differential post-translational processing of the constructs in bacteria as opposed to insect cells. Alternatively, the cSH2 domain might function differently in the nSH2-iSH2-cSH2 fragment as opposed to intact p85. We therefore produced bacterial GST-fusion proteins containing full-length wild-type p85, as well as p85 containing disabling mutations in the nSH2 and cSH2 domains (R358A and R659A, respectively; Fig. 4A). All of the bacterial p85 constructs inhibited p110␣ to the same extent as p85 produced in baculovirusinfected Sf-9 cells (p85 Sf-9 ) (data not shown). Incubation of phosphopeptides with either p110␣/GST-p85 dimers or p110␣/ p85 Sf-9 dimers increased activity by 120 -130% (Fig. 4B, lanes a  and b). Phosphopeptides also increased the activity of GST-p85 constructs containing mutant SH2 domains; the activity of nSH2-iSH2-cSH2(R659A)/p110␣ dimers increased by 90%, and the activity of nSH2(R358A)-iSH2-cSH2/p110␣ dimers increased by 50% (Fig. 4B, lanes c and d).
These data confirm our earlier report that both N-terminal and C-terminal SH2 can partially mediate phosphopeptide activation of p85/p110␣ dimers. They also show that the different results obtained with wild-type p85 and the nSH2-iSH2-cSH2 fragment were not because of aberrant processing in bacterial cells. Instead, they suggest that the residues 1-322 of p85 play a role in the regulation of p85/p110␣ by phosphopeptides. Activation by the cSH2 domain occurred only in the context of intact p85, whereas activation by the nSH2 domain occurred in the absence of the cSH2 domain or residues 1-322 of p85 (the SH3, BCR-homology and proline-rich domains). DISCUSSION We have previously shown that p85 is an inhibitor of p110␣ activity, and that binding of tyrosyl phosphopeptides to the p85 SH2 domains relieves this inhibition (11). This study demonstrates that the SH2 domains of p85 are critical for the inhibitory effects of p85 on p110␣. Consistent with previous reports, we find that the iSH2 domain of p85 is sufficient to bind to p110␣ (14 -17). However, iSH2 domain binding alone does not affect p110␣ activity. Instead, the presence of an SH2 domain linked to the N terminus of the iSH2 domain is required for inhibition of p110␣.
Two mechanisms could explain the inhibition of p110␣ by the nSH2/iSH2 versus iSH2 fragments. The iSH2 domain is predicted to form a coiled-coil domain (14). The presence of an nSH2 domain at the N-terminal end of the iSH2 domain could exert a conformational strain on the coiled-coil and alter its interactions with p110␣. Alternatively, the nSH2 domain may directly contact p110␣, inhibiting its activity. In both mechanisms, conformational changes induced by phosphoprotein binding to the nSH2 domain would relieve the inhibition of p110␣. We have no direct experimental evidence to distinguish these hypotheses at this time. However, we have noticed that unlike the iSH2 domain itself, a GST-iSH2 fusion protein binds p110␣ and inhibits its activity by 50%. Inhibition by attachment of a bulky GST moiety to the N terminus of the iSH2 domain is consistent with the first mechanism. Also consistent with this model is a recent paper by Jimenez et al. (19a) describing an oncogenic truncated p85 molecule. Expression of this mutant with p110␣ caused an increase in activity as compared with wild-type p85, which we would interpret as the loss of p85-induced inhibition of p110␣. Since the nSH2 domain is present in the truncation mutant, the loss of inhibition would seem to be because of a conformational change in the iSH2 domain caused by the removal of its extreme C terminus. On the other hand, Cooper and Kashishian reported a direct interaction between p110␣ and the p85 nSH2 domain in transfected cells (19), which would be consistent with the second mechanism.
Importantly, the iSH2 domain itself neither activates nor inhibits p110␣. In contrast, others have suggested that iSH2 domain binding to p110␣ provides critical activating interactions that are required for p110␣ activity in mammalian cells (20), and attachment of the iSH2 domain to p110 has been used to produce a constitutively active enzyme (21). However, attachment of bulky moieties such as GST or a tris-HA tag to the N terminus of p110␣ increases the activity of monomeric p110␣ activity in mammalian cells by stabilizing the protein (11). Given that the iSH2 domain has no effect on p110␣ activity in vitro, we think it likely that the iSH2-p110␣ chimera is active because of the attachment of a bulky group, rather than the provision of specific activating interactions.
A surprising finding in this study is the marked difference in the roles of the nSH2 and cSH2 domains. The nSH2/iSH2 fragment inhibits p110␣, and nSH2/iSH2-p110␣ complexes are activated by phosphotyrosine peptides. In contrast, iSH2/cSH2 fragments bind p110␣ but have little effect on its activity. Phosphopeptide binding to the cSH2 domain does contribute to p110␣ activation, but only in the context of the entire p85 protein. Thus, phosphopeptide modulation of p110␣ via the cSH2 domain appears to be distinct from modulation via the nSH2 domain.
Previous studies have suggested that intramolecular interactions may occur between the SH3 domain and the prolinerich domain of p85 (22). In this case, the cSH2 domain, SH3-PRD domains, and the nSH2 domain may form a compact structure (Fig. 5). Our data would suggest that the nSH2 domain is the major regulator of p110␣ activity, and that occupancy of the nSH2 domain induces a conformational change (23,24) that is transmitted to the iSH2 domain and/or p110␣ (Fig. 5A). In contrast, phosphopeptide occupancy of the cSH2 domain may induce a conformational change that is transmitted to the regulatory nSH2 domain by way of residues 1-322 of p85 (the SH2, Bcr and proline-rich domains) (Fig. 5B). This model would predict that disruption of intramolecular interactions within the SH3 and proline-rich domains of p85 would minimize phosphopeptide-induced activation via the cSH2 domain, but would not affect activation via the nSH2 domain.
Experiments to test this hypothesis are in progress.
In summary, we have shown that the iSH2 domain of p85 mediates binding to p110␣, whereas the inhibitory effects of p85 on p110␣ are largely mediated by an additional constraint imposed by the nSH2 domain. Phosphopeptide occupancy of the nSH2 domain can directly modulate p110␣ activity. In contrast, modulation of p110␣ activity by the cSH2 domain occurs by a mechanism that requires residues 1-234 of p85 as well as the nSH2 domain. These studies highlight the complexities of p110␣ regulation by p85.