The mammalian signalling pathway involving class I PI3K (phosphoinositide 3-kinase), PTEN (phosphatidylinositol 3-phosphatase) and PKB (protein kinase B)/c-Akt has roles in multiple processes, including cell proliferation and apoptosis. To facilitate novel approaches for genetic, molecular and pharmacological analyses of these proteins, we have reconstituted this signalling pathway by heterologous expression in the unicellular eukaryote, Saccharomyces cerevisiae (yeast). High-level expression of the p110 catalytic subunit of mammalian PI3K dramatically inhibits yeast cell growth. This effect depends on PI3K kinase activity and is reversed partially by a PI3K inhibitor (LY294002) and reversed fully by co-expression of catalytically active PTEN (but not its purported yeast orthologue, Tep1). Growth arrest by PI3K correlates with loss of PIP2 (phosphatidylinositol 4,5-bisphosphate) and its conversion into PIP3 (phosphatidylinositol 3,4,5-trisphosphate). PIP2 depletion causes severe rearrangements of actin and septin architecture, defects in secretion and endocytosis, and activation of the mitogen-activated protein kinase, Slt2. In yeast producing PIP3, PKB/c-Akt localizes to the plasma membrane and its phosphorylation is enhanced. Phospho-specific antibodies show that both active and kinase-dead PKB/c-Akt are phosphorylated at Thr308 and Ser473. Thr308 phosphorylation, but not Ser473 phosphorylation, requires the yeast orthologues of mammalian PDK1 (3-phosphoinositide-dependent protein kinase-1): Pkh1 and Pkh2. Elimination of yeast Tor1 and Tor2 function, or of the related kinases (Tel1, Mec1 and Tra1), did not block Ser473 phosphorylation, implicating another kinase(s). Reconstruction of the PI3K/PTEN/Akt pathway in yeast permits incisive study of these enzymes and analysis of their functional interactions in a simplified context, establishes a new tool to screen for novel agonists and antagonists and provides a method to deplete PIP2 uniquely in the yeast cell.
- phosphoinositide 3-kinase (PI3K)
- protein kinase B (PKB)
- Saccharomyces cerevisiae
Generation of PIP3 (phosphatidylinositol 3,4,5-trisphosphate) at the plasma membrane is key to the regulation of mammalian cell proliferation and survival by growth-promoting factors . Many human tumours  have alterations in the PI3K (phosphoinositide 3-kinase) that generates this lipid , in the PTEN (phosphatidylinositol 3-phosphatase) that degrades this lipid  and in a protein kinase [PKB (protein kinase B)/c-Akt] that requires this lipid for its activation .
Class IA PI3K, the best studied and most related to oncogenesis, comprises a regulatory/adaptor subunit (p85) and a catalytic subunit (p110). In humans, seven regulatory subunits, generated by alternative splicing from three genes (p85α, p85β and p55γ) and three p110 catalytic subunit isoforms (α, β and γ), have been described. p85 contains an SH2 domain (Src homology 2 domain) that binds phosphotyrosine residues on receptor-tyrosine kinases, simultaneously recruiting p110, stimulating its activity and generating locally PIP3 from PIP2 (phosphatidylinositol 4,5-bisphosphate) in the plasma membrane.
The reverse reaction – dephosphorylation of PIP3 to PIP2 – is catalysed by PTEN, which is considered a tumour suppressor because loss-of-function mutations are found in overt tumours  and in the germline in diseases marked by hereditary predisposition to cancer . PIP3 is also metabolized to PI(3,4)P2 (phosphatidylinositol 3,4-bisphosphate) by 5-phosphoinositide phosphatases, like SHIP1 and SHIP2 (SH2-containing inositol phosphatase-1 and -2); but PIP3 and PI(3,4)P2 engage the same effectors, for the most part.
The three isoforms (α, β and γ) of PKB/c-Akt are major targets of PIP3, which binds to the N-terminal PH domain (pleckstrin homology domain), thereby anchoring the enzyme to the plasma membrane and inducing a conformational change that exposes its catalytic domain to its upstream activators. Full activation of Akt requires phosphorylation of Thr308 within the activation loop (or T-loop) and also at a more distal site, Ser473. Phosphorylation of Akt1 at Thr308 is mediated by PDK1 (3-phosphoinositide-dependent protein kinase-1), which has a C-terminal PIP3-specific PH domain that facilitates its recruitment to membranes. Hence, Thr308 is called the PDK1 site. The enzyme responsible for phosphorylation of Akt1 at Ser473 (termed the PDK2 site) is controversial. Over time, various investigators have implicated several different classes of protein kinase in PDK2 site phosphorylation , whereas others suggest that autophosphorylation is the cause . Recently, it has been reported that members of the PIKK (phosphatidylinositol kinase-related kinase) family of atypical protein kinases are responsible for PDK2 site phosphorylation of Akt1, including DNA-PKcs (the catalytic subunit of DNA-dependent protein kinase) , the product of the gene mutated in ATM (ataxia telangiectasia) , and the mTOR (mammalian target of rapamycin) orthologue . Once activated, Akt phosphorylates substrates that promote cell survival and stimulate cell growth. Thus hyperactivation of Akt is considered a prerequisite for oncogenesis. Consequently, the PI3K/Akt pathway has become an important target in attempts to develop more efficacious anti-tumour agents .
Although it has been reported that fission yeast (Schizosaccharomyces pombe) can generate PIP3 , PIP3 seems to have no role in the physiology of budding yeast (Saccharomyces cerevisiae) . Nonetheless, the Sacch. cerevisiae genome encodes: (i) two functional PDK1 orthologues (Pkh1 and Pkh2) involved in cell integrity and endocytosis [16,17]; (ii) an apparent PTEN orthologue (Tep1) of uncharacterized biological function [18,19]; (iii) an Akt-like protein kinase (Sch9), which lacks an apparent PH domain, involved in nutrient sensing, ribosome biogenesis, lifespan and cell-size control ; and (iv) clear-cut homologues of the PIKK family, specifically Tor1 and Tor2 (mTOR) , Tel1 (ATM) , Mec1 (ATR)  and Tra1 (most resembles DNA-PKcs) .
To address central questions in the biology of PIP3-dependent signalling and to establish a readily accessible and versatile tool to screen for pharmacological agents that influence this critically important pathway, we devised methods to successfully reconstitute the mammalian PI3K/PTEN/Akt pathway in yeast cells, which is described here. In vivo conversion of the essential PIP2 pool into PIP3 by expression of PI3K impaired yeast growth by altering morphogenesis and vesicular trafficking. The function of PTEN could be readily assessed by its ability to reverse the growth inhibition caused by PI3K. PIP3 generation led to membrane translocation and activation of Akt, enhancing its phosphorylation at both Thr308 and Ser473. The yeast PDK1 orthologues are required for PDK1 site phosphorylation, whereas none of the yeast PIKK family members seems necessary for PDK2 site phosphorylation, implicating some other endogenous enzyme.
Strains, media and growth conditions
The Sacch. cerevisiae strains used in the present study are listed in Table 1. Escherichia coli DH5α F′[K12Δ(lacZYA-argF)U169 deoR supE44 thi-1 recA1 endA1 hsdR17 gyrA96 relA1 (φ80lacZΔM15)F′] was used for routine molecular biology techniques. YPD [1% (w/v) yeast extract, 2% (w/v) peptone and 2% (w/v) dextrose/glucose] broth or agar was the general non-selective medium used for yeast cell growth. Synthetic minimal medium (SGlc) contained 0.17% yeast nitrogen base without amino acids, 0.5% ammonium sulphate and 2% glucose and lacked appropriate amino acids and nucleic acid bases to maintain selection for plasmids. In SGal (synthetic galactose) and SRaf (synthetic raffinose) media, glucose was replaced with 2% (w/v) galactose or 2% (w/v) raffinose respectively. Galactose induction in liquid medium was performed by growing cells in SRaf to mid-exponential phase and then adding galactose to 2% for 6–8 h. Growth assays on plates were performed, typically, by growing transformants overnight in SGlc lacking uracil, leucine or both, adjusting the culture to an absorbance A600 0.5 and spotting samples (5 μl) of the cell suspension and three serial 10-fold dilutions on to the surface of SGlc or SGal plates lacking the appropriate nutrients to maintain selection for plasmids, followed by incubation at 28 °C for 2–3 days. When yeast cells were treated with PI3K inhibitors, transformants were grown overnight in selective medium, adjusted to an A600 of 0.1, and then samples (5 μl) of the cell suspension were used to seed the wells of 96-well microtitre plates containing serial dilutions of either wortmannin (Sigma) or LY294002 (Biomol, Hamburg, Germany) suspended in 100 μl of liquid SGal medium lacking the appropriate nutrients to maintain selection for plasmids.
Transformation of E. coli and yeast and other basic molecular biology methods were carried out using standard procedures. To generate plasmid YCpLG-PI3K, the cDNA encoding PI3K-CAAX was excised from plasmid Psg5/5MycTp110XCAAX  (a gift from M. Collado, Spanish National Cancer Centre, Madrid, Spain) with BamHI and cloned into the same site in yeast vector YCpLG . To produce plasmid YCpLG-PI3KK802R (where K802R stands for Lys802→Arg), bearing a catalytically inactive (‘kinase-dead’) allele of PI3K-CAAX, site-directed mutagenesis was carried out using a DpnI-based strategy  with Turbo PfuI DNA polymerase (Stratagene) and the primers 5′-CAGAACAATGAGATCATCTTTCGAAATGGGGATGATTTACGGC-3′ and 5′-GCCGTAAATCATCCCCATTTCGAAAGATGATCTCATTGTTCTG-3′. Cassettes in which cDNAs encoding either c-Akt, c-AktK179M or an N-myristoylated c-Akt were fused in frame to an HA (haemagglutinin) epitope-tagged version of eGFP (enhanced green fluorescence protein) [HA–eGFP-Akt, HA–eGFP-AktK179M and myr-HA–eGFP-Akt respectively] were excised with HindIII and BamHI from the original Pcefl(X)-derived plasmids that were constructed for expression in mammalian cells  (a gift from M. Lorenzo, Universidad Complutense, Madrid, Spain) and cloned into the corresponding sites in yeast vector pYES2 (Invitrogen), yielding plasmids pYES-GFP-c-Akt, pYES-GFP-c-AktK179M and pYES-myr-GFP-c-Akt respectively. The cDNA encoding PTEN was excised with EcoRI from plasmid Pcmvpten  [a gift from J.M. Paramio, CIEMAT (Centro de Investigaciones Energéticas, Medioambientales y Tecnológicas, Madrid, Spain)] and inserted into the same site in pYES2, generating pYES-PTEN. Plasmid pYES-PTENG129D, expressing a catalytically inactive (‘phosphatase-dead’) allele, was generated by site-directed mutagenesis, as above, but using primers 5′-CACTGTAAAGCTGGAAAGGAACGAACTGGTGTAATG-3′ (upper) and 5′-CATTACACCAGTTCGTTCCTTTCCAGCTTTACAGTG-3′ (lower). To construct plasmid pYES-Tep1-Myc, first the TEP1 coding sequence was amplified by PCR from yeast genomic DNA using the primers 5′-CGGATCCATGAGAGAGGAGGGGAGTG-3′ (upper) and 5′-GGGATCCTATAATTTCCCATTCCAAT-3′ (lower) and cloned into the BamHI site in a yeast vector, pRS306-Myc6, which had been previously generated by inserting a Myc6 epitope into the polylinker in the integrative URA3 vector, pRS306 . Secondly, the resulting TEP1–myc6 fusion was excised using EcoRI and cloned into the same site in pYES2. To construct a PTEN-TEP1 chimaera, we used a now-standard PCR-based method  in which PTEN was amplified using the ‘upper’ primer indicated above, TEP1 was amplified with the ‘lower’ primer indicated above, with the following primers to generate the desired junction: 5′-GCCTCATTAGAAATTCCTGGTCTATAATCCAGAT-3′ and 5′-ATCTGGATTATAGACCAGGAATTTCTAATGAGGC-3′. To construct a plasmid expressing a Tep1–GFP fusion, the TEP1 gene was amplified by PCR using the same oligonucleotides described above, which provide BamHI sites at both ends of the amplicon, and cloned in frame into the BamHI site of the polylinker in pGFP-C-FUS , yielding pTep1-GFP. All constructs were verified by direct DNA sequencing. Other plasmids used in the present study were pLA10H  to express GFP-tagged septin and pRS426-GFP-2xPH(PLCδ) to visualize PIP2 in vivo .
Protein detection by immunoblotting
Standard procedures, as described previously , were used for yeast growth, cell harvesting and cell breakage, as well as for the preparation of protein-containing cell-free extracts, fractionation by SDS/PAGE and transfer on to nitrocellulose membranes. Polyclonal anti-PTEN antibodies (Upstate Biotechnology, Lake Placid, NY, U.S.A.) were used to detect PTEN or the PTEN-Tep1 chimaera in yeast extracts. Yeast MAPK (mitogen-activated protein kinase) Slt2 [orthologue of mammalian ERK5 (extracellular-signal-regulated kinase 5)] was detected with the polyclonal anti-Slt2 antibodies described previously . Rabbit anti-phospho-p44/p42 MAPK (anti-P-Thr202-P-Tyr204) antibodies (New England Biolabs, Hitchin, Herts., U.K.) were used to detect dually phosphorylated forms of Slt2 and of two other yeast MAPKs, Kss1 and Fus3 (orthologues of mammalian ERK1 and ERK2). Anti-phospho-p38 (Cell Signaling Technology, Beverly, MA, U.S.A.) was used to detect dual phosphorylation of yeast MAPK Hog1 [orthologue of mammalian SAPKα (stress-activated protein kinase α)]. Rabbit anti-phospho-Akt(P-Thr308) and rabbit anti-phospho-Akt(P-Ser474) (Cell Signaling Technology) were used to detect the phosphorylation of the corresponding sites in PKB/c-Akt. Yeast CPY (carboxypeptidase Y) was detected with anti-CPY antiserum (Molecular Probes). Yeast actin was detected by cross-reaction with a mouse anti-actin monoclonal antibody (clone C4; MP Biomedicals, Irvine, CA, U.S.A.). GFP fusion proteins were detected using a mouse anti-GFP antibody (JL-8; BD Biosciences, San Jose, CA, U.S.A.). After washing, bound primary antibodies were revealed using horseradish peroxidase-conjugated anti-rabbit or anti-mouse secondary antibodies, as appropriate, and a chemiluminescence detection system (ECL®; Amersham Biosciences). For quantitative measurements on immunoblots, densitometry [in a GS-800 densitometer, model Power Look 2100xL (Bio-Rad), using the software Quantity One] was applied to films and the extent of Thr308 and Ser473 phosphorylation was normalized to the amount of protein recovered in the extract as follows. Ratios of c-AktK179M/c-Akt were calculated by dividing the percentages of the phosphorylation signal of Thr308 or Ser473 of GFP–c-AktK179M relative to GFP–c-Akt by the corresponding percentages of the GFP signal of GFP–c-AktK179M signal relative to GFP–c-Akt. To obtain these ratios, duplicate measurements on at least three different experiments were performed.
Microscopy techniques and immunofluorescence
For fluorescence microscopy of live yeast cells (to visualize GFP), exponentially growing cultures were harvested by brief centrifugation, washed once with sterile water and viewed directly. To observe actin, yeast cells were treated with FITC-conjugated phalloidin (Sigma) as described previously . To monitor vacuolar morphology and bulk fluid-phase endocytosis into the vacuole, staining with FM4-64  and Lucifer Yellow  respectively was performed. To obtain reliable and statistically significant data, ≥200 cells were examined for each condition or experiment. Indirect immunofluorescence with monoclonal anti-PIP3 antibodies (Echelon, Salt Lake City, UT, U.S.A.) at a 1:200 dilution was performed, essentially as described for animal cells , using Cy3-conjugated monoclonal anti-mouse IgG antibodies (Chemicon, Temecula, CA, U.S.A.) at a 1:500 dilution as the secondary antibodies. Cells were examined under an Eclipse TE2000U microscope (Nikon, Tokyo, Japan) and digital images were acquired with an Orca C4742-95-12ER charge-coupled-device camera (Hamamatsu Photonics, Hamamatsu City, Japan) and Aquacosmos Imaging Systems software.
Expression of PI3K inhibits yeast cell growth and generates PIP3
For expression in yeast, the cDNA encoding p110α was cloned under the control of the galactose-inducible GAL1 promoter. To target the enzyme to the plasma membrane in the absence of the p85 subunit, the -CAAX box of H-ras was fused in frame to the C-terminus of p110α. Wild-type Sacch. cerevisiae cells (YPH499) transformed with this construct grew normally on glucose medium, but failed to grow on galactose medium (Figure 1A). The growth inhibition caused by PI3K expression was alleviated by a point mutation (K802R, Lys802→Arg) in a residue essential for enzyme activity. Hence, the growth inhibition was due to the catalytic function of PI3K and not the result of any non-specific toxic effect of producing this heterologous membrane-targeted polypeptide in yeast.
To determine whether the presence of active PI3K generates PIP3 at the expense of PIP2, both phosphoinositides were visualized in the cells. To detect PIP2, cells were co-transformed with a plasmid expressing a fluorescent PIP2-specific reporter [GFP fused to two tandem copies of the PH domain of mammalian PLCδ1 (phospholipase δ1)]. In yeast, this reporter specifically decorates the plasma membrane . Indeed, in our control cells, the reporter clearly stained the plasma membrane prominently. In contrast, in the cells where catalytically active PI3K was expressed, only cytoplasmic fluorescence was seen, indicating a dramatic depletion of PIP2 from the plasma membrane (Figure 1B). To confirm that PIP3 was produced at the expense of PIP2, the same cells were fixed and stained with an anti-PIP3 antibody. Similar to other methods that failed to detect PIP3 in Sacch. cerevisiae [14,15], no staining was seen in the control cells. In contrast, large patches of fluorescent staining with the anti-PIP3 antibodies were found around the perimeter of the cells expressing functional PI3K (Figure 1C). Thus the PI3K produced was catalytically active, and the most likely cause of its observed inhibitory effect on yeast cell growth was efficient conversion of the PIP2 pool into PIP3.
Expression of PTEN antagonizes the effect of PI3K in yeast
If the growth inhibitory action of PI3K is due to conversion of PIP2 into PIP3, then reversal of this reaction by PTEN should ameliorate growth arrest. Therefore PTEN cDNA was cloned under the control of the GAL1 promoter in a vector compatible with that used to express PI3K. Expression of PTEN alone at a high level, confirmed by immunoblotting with specific anti-PTEN antibodies (results not shown), had no discernible effect on growth or morphology (Figure 2A). However, in cells expressing PI3K, co-expression of PTEN prevented growth arrest. A point mutation (G219E) reported to abrogate the lipid phosphatase activity of PTEN  blocked its ability to reverse PI3K-induced growth inhibition (Figure 2A). Thus the counteracting effect of PTEN on PI3K-initiated signalling was reproduced in yeast cells.
Other authors have suggested that the Sacch. cerevisiae TEP1 gene encodes a PTEN orthologue [18,19]. Tep1 is similar to PTEN through its N-terminal catalytic region (27% identity from residues 1–260), but contains a large insert (residues 104–160) and also differs completely from PTEN over its C-terminal segment. Although Ptn1, the apparent PTEN homologue in fission yeast, hydrolyses 3-phosphoinositides in vitro , no such activity for budding yeast Tep1 has been detected . Normally, TEP1 is expressed only in MATa/MATα diploids undergoing meiosis and sporulation . Therefore TEP1 (fused in frame at its C-terminus to six tandem c-Myc epitopes for immunodetection) was cloned in a vector under GAL1 promoter control. This Tep1 derivative rescues the phenotypes of tep1Δ/tep1Δ cells and has no detectably deleterious effect on haploid cells (results not shown). However, Tep1 expression was unable to alleviate PI3K-induced growth inhibition (Figure 2A), even though immunoblotting with specific anti-c-Myc antibodies confirmed that Tep1 was expressed at a high level (results not shown). Thus, as observed in vitro, Tep1 may be incapable of dephosphorylating PIP3 in vivo. However, a Tep1–GFP fusion was localized exclusively as discrete cytoplasmic spots (Figure 2B), reminiscent of the cytosolic puncta reported for Schizo. pombe Ptn1 , and was never detected at the cell periphery (or even as a diffuse cytoplasmic signal) in any of the ≥200 cells examined. Therefore, even if Tep1 is a functional 3-phosphoinositide phosphatase, it may be incapable of gaining access to the plasma membrane-localized pool of PIP3 generated by PI3K. In contrast, PTEN displayed a diffuse cytoplasmic pattern in yeast, revealed by indirect imunofluorescence using anti-PTEN antibodies (results not shown) and, operationally, must gain access to the cytosolic face of the plasma membrane. Unlike PTEN itself, a PTEN1–190-Tep1261–434 chimaera, in which the unique C-terminal extension of Tep1 was fused in frame to the N-terminal catalytic domain of PTEN, was unable to rescue the growth inhibition caused by PI3K (Figure 2A), even though expression of the chimaera in yeast lysates (detected with anti-PTEN antibodies) was as efficient as that of PTEN itself (results not shown). This result suggests that a localization signal in the C-terminal extension of Tep1 prevents the protein from reaching the plasma membrane.
Sublethal concentrations of LY294002 partially alleviate PI3K-induced growth arrest
Wortmannin and LY294002 are two relatively specific inhibitors of mammalian PI3K, effective over the nanomolar range . The sole PI3K in Sacch. cerevisiae, Vps34, converts phosphatidylinositol into phosphatidylinositol 3-phosphate, but cannot use PIP2; at the nanomolar level, neither wortmannin nor LY294002 has any effect on Vps34 or on protein trafficking or cell growth, largely due to their poor penetration through the yeast cell wall . At much higher doses, wortmannin shows some toxicity against yeast . Likewise, over the millimolar range, LY294002 was inhibitory to the growth of our strain (IC50∼1.5 mM). However, at this concentration, LY294002 actually stimulated the growth of cells expressing mammalian PI3K by 3–5-fold (Figure 3). Therefore compounds that inhibit human PI3K, but are generally non-toxic, could be selected from chemical libraries by screening for those that fully restore growth to yeast cells expressing p110α. Moreover, any compound that acts on yeast will surely enter an animal cell.
PI3K expression disrupts the cytoskeleton and alters secretory and endocytic trafficking
Because p110α expression efficiently converted plasma-membrane PIP2 into PIP3 and prevented growth, it provided a novel means to examine the physiological consequences of this perturbation and thereby shed light on the essential functions of PIP2 in yeast. Upon shift to galactose, asynchronous cultures of PI3K-expressing cells did not arrest with a uniform cellular or nuclear morphology, assessed by microscopy and flow cytometry (results not shown), indicating that loss of plasma membrane PIP2 does not block cell-division cycle progression at a specific stage.
PIP2 has been implicated as a regulator of both the actin  and septin  cytoskeletons. Our system provided a unique means to confirm and extend those conclusions. In control cells, the actin cytoskeleton showed a highly polarized distribution, with actin filaments in the mother cell and actin patches restricted to the bud (Figure 4A, left panel). In the same strain expressing PI3K, this asymmetry was lost because actin patches and diffuse actin staining were found in more than 80% of the mother cells (Figure 4A, right). Similarly, in control cells, a fluorescently tagged septin was incorporated exclusively into the collar-like structure found at the mother-bud neck (Figure 4B, left panel), whereas in the same strain expressing PI3K, abnormal septin-containing structures were seen in >50% of cells (Figure 4B, right panel). Thus PIP2 in the plasma membrane is important for maintaining proper architecture of both the actin and septin cytoskeletons. Since the functions of both structures are essential for yeast cell viability, their disruption is sufficient to explain the growth inhibitory effect of PI3K expression.
Specific phosphoinositides have been implicated in both secretory and endocytic transport in yeast . To examine secretion, we monitored the precursor forms of vacuolar CPY because their productive processing and maturation depends on their efficient delivery by Golgi-derived vesicles. In control cells, both the p1 and p2 precursor forms, and also the mature (m) form, were observed by immunoblotting (Figure 4C, left panel), whereas in cells expressing PI3K, the ER (endoplasmic reticulum)-specific precursor (p1) accumulated at the expense of the p2 form (Figure 4C, right), although the effect was rather mild and the amount of mature CPY generated was little affected. Using similar means, maturation of other vacuolar (alkaline phosphatase Pho8) and exocytic (cell-wall protein Gas1) marker proteins was unaffected (results not shown).
Endocytic transfer of plasma-membrane material to the yeast vacuole membrane can be monitored by staining with a lipophilic fluorescent dye, FM4-64, and fluid-phase endocytosis into the vacuole can be evaluated by uptake of a water-soluble fluorescent dye, Lucifer Yellow. In control cells, both fluorochromes selectively stained the membrane and lumen of the vacuoles respectively (Figure 4D, upper panels). In the same cells expressing PI3K, no distinctive vacuolar membranes were seen and no dye import was observed (Figure 4D, lower panels), indicating that endocytosis was severely impaired.
PI3K expression activates the cell-integrity MAPK, Slt2
Cell-membrane biogenesis and cell-wall biogenesis require the transport of secretory vesicles along actin cables to the bud tip . Damage to the cell envelope promotes the expression of genes necessary to co-ordinate appropriate repair of the cell wall via the cell-integrity pathway, which activates the MAPK, Slt2/Mpk1 (yeast orthologue of mammalian Erk5) . Hence, as an independent and sensitive means to assess whether PI3K expression affects actin-mediated secretory vesicle transport, activation (dual phosphorylation) of Slt2 was examined using phospho-specific antibodies. In control cells (carrying two empty expression vectors), Slt2 was present largely in its unactivated (unphosphorylated) state, whereas in cells expressing plasmid-borne PI3K, a marked increase in activated (dually phosphorylated) Slt2 was readily detected (Figure 5). Moreover, just as co-expression of PTEN ameliorated PI3K-induced growth inhibition, Slt2 activation in cells expressing PI3K was greatly reduced by the co-expression of active PTEN, but not catalytically inactive PTEN (Figure 5). In contrast, PI3K expression did not activate the MAPKs of the pheromone response pathway, Fus3 (Erk1 orthologue), or the filamentous growth response pathway, Kss1 (Erk2 orthologue), as judged by immunoblotting with anti-phospho-Erk antibodies, nor the hyperosmotic stress response pathway, Hog1 (p38 SAPK orthologue), monitored with anti-phospho-p38 antibodies (results not shown).
PI3K expression causes relocalization and activation of c-Akt in yeast
In animal cells, receptor-induced activation of PI3K produces PIP3, thereby stimulating membrane recruitment and PDK1-mediated phosphorylation of PKB/c-Akt . To determine whether PI3K-induced PIP3 production in yeast has similar effects, cDNAs expressing GFP–c-Akt1, a catalytically inactive derivative, GFP–c-Akt1K179M, and a constitutively membrane-targeted (N-myristoylated) variant, myr-GFP–c-Akt1, were cloned under the control of the GAL1 promoter in a vector compatible with that used to express PI3K. None of these fusions alone caused any morphological abnormality or growth impairment on galactose medium, nor was any MAPK activated (results not shown). Both active and kinase-dead GFP–c-Akt1 displayed diffuse nuclear and cytoplasmic fluorescence (Figure 6A, left panels). In cells co-expressing PI3K, however, both fusions showed a distinct rim of fluorescence around the periphery of every cell (Figure 6A, right panels). This pattern was indistinguishable from that displayed by the membrane-targeted myr-c-Akt–GFP, in the absence or presence of PI3K (Figure 6A, bottom panels). Thus, as in animal cells, PIP3 production recruits c-Akt to the plasma membrane (and, to a lesser extent, to some, as yet unidentified, membrane-bound cytoplasmic organelles).
To correlate membrane translocation of c-Akt1 with its phosphorylation state, extracts of cells carrying the three different c-Akt1 fusions, in the absence and presence of PI3K, were resolved by SDS/PAGE and analysed by immunoblotting with phospho-specific antibodies.
First, readily detectable basal phosphorylation at the PDK1 site (Thr308) and at the PDK2 site (Ser473) was observed in cells expressing c-Akt1 alone. For kinase-dead c-Akt1K179M expressed under the same conditions, phosphorylation at Thr308 was reduced only moderately (ratio of c-AktK179M/c-Akt was 0.75±0.22), whereas phosphorylation at Ser473 was not significantly different (ratio of c-AktK179M/c-Akt was 1.11±0.25), indicating that autophosphorylation is not required for either of these modifications (Figure 6B). Secondly, in the presence of active (but not catalytically inactive) PI3K, we reproducibly observed that phosphorylation at both Thr308 and Ser473 was substantially elevated in both active and kinase-dead c-Akt1. The myr-c-Akt1 derivative was always expressed at a level significantly lower than that of either active or kinase-dead c-Akt1, perhaps explaining the observed reduction in Thr308 phosphorylation, and phosphorylation at Thr308 was only moderately increased upon PI3K expression. No Ser473 phosphorylation of myr-c-Akt1 was detectable in the absence or presence of PI3K, suggesting perhaps that the kinase responsible for Ser473 phosphorylation cannot attack c-Akt restricted to the plasma membrane.
Yeast PDK1 orthologues are responsible for phosphorylation of c-Akt at Thr308
To determine whether modification of c-Akt1 at Thr308 requires Pkh1 and Pkh2, the phosphorylation state of GFP–c-Akt1 and GFP–c-Akt1K179M was examined in wild-type cells and in otherwise isogenic pkh1ts pkh2Δ cells after shift to a restrictive temperature. Inactivation of Pkh1 in cells lacking Pkh2 abolished phosphorylation at Thr308 in both proteins, and the effect was specific because phosphorylation at Ser473 was unaffected (Figure 6C). Either Pkh1 or Pkh2 are competent to phosphorylate c-Akt1 at Thr308 because deletion of either PKH1 or PKH2 alone did not eliminate Thr308 phosphorylation (Figure 6D). Inactivation of two endogenous targets of Pkh1 and Pkh2, the Ypk1 and Ykr2/Ypk2 protein kinases (functional orthologues of mammalian SGK) [16,17] had no effect on the state of c-Akt modification because phosphorylation at Thr308 (Figure 6D) and Ser473 (results not shown) was unaffected when ypk1ts ykr2Δ cells were shifted to restrictive temperature.
The PDK2 site-directed kinase is not Tor2 nor any other PIKK family member
We found that kinase-dead c-Akt1K179M is still phosphorylated at Ser473 (Figures 6B and 6C), ruling out autophosphorylation as the sole basis for PDK2 site phosphorylation. The possibility that the most closely related, endogenous yeast AGC kinase, Sch9 , cross-phosphorylates c-Akt1 at the PDK2 site was ruled out by the fact that there was no discernible difference in the level of c-Akt1 phosphorylation at Ser473 in sch9Δ cells compared with otherwise isogenic SCH9+ cells (results not shown).
It has been suggested that various members of the PIKK family of atypical protein kinases (DNA-PKcs, ATM and mTOR) are responsible for PDK2 site phosphorylation in c-Akt1. To address the role of Tor1 and Tor2 in phosphorylation of Ser473 in c-Akt1 in yeast, cells were treated with the TOR-specific inhibitor rapamycin at concentrations that inhibit growth, but no reduction in the level of Ser473 (or Thr308) phosphorylation in either c-Akt1 or c-Akt1K179M was seen (results not shown). It has been appreciated recently, however, that Tor1 and Tor2 can each be found in discrete complexes, so-called TORC1 and TORC2, and that only TORC1 is rapamycin-sensitive . Hence, as a more definitive approach, we examined the state of Ser473 phosphorylation in tor1Δ tor2Δ cells carrying a plasmid expressing a Tor2ts mutant, which were shifted to restrictive temperature for periods of time more than sufficient to inactive the thermolabile Tor2 and prevent cell growth. By 9 h after shift, 100% of the mutant cells had lost cortical actin polarization (a sensitive measure of Tor2 function ), whereas all of the control TOR1+ TOR2+ cells retained normal actin polarization (results not shown). In the mutant cells, there was no detectable reduction in Ser473 phosphorylation in either c-Akt1 or c-Akt1K179M, even 9 h after shift to restrictive temperature, compared with TOR1+ TOR2+ cells under the same conditions (Figure 7A). A null mutation (tel1Δ) in the yeast ATM orthologue, a non-essential gene, did not reduce Ser473 phosphorylation of c-Akt1 (Figure 7 B) or c-Akt1K179M (results not shown). Similarly, a shift of temperature-sensitive mutations in the yeast ATR orthologue (mec1ts) or the apparent yeast DNA-PKcs orthologue (tra1ts), both essential genes, did not significantly reduce Ser473 phosphorylation in c-Akt1 (Figure 7B) or c-Akt1K179M (results not shown). Thus, although a kinase capable of phosphorylating efficiently the PDK2 site in c-Akt1 clearly exists in yeast, it does not appear to correspond to any member of the PIKK family encoded in the Sacch. cerevisiae genome.
Using appropriate expression vectors, we reconstructed the mammalian PI3K/PTEN/PKB signalling pathway in Sacch. cerevisiae. The ability of PTEN to reverse the toxic effect of PIP3 production by PI3K could be convincingly recapitulated in yeast. Thus the strains constructed provide a simplified model for genetic analysis of these enzymes and to screen for compounds that are PI3K antagonists and PTEN agonists. Our approach also provided a novel means to deplete specifically the plasma-membrane PIP2 pool without blocking the production of other phosphoinositides. Finally, this system offered unequivocal confirmation that membrane recruitment of PKB/c-Akt is PIP3-dependent.
High-level expression of membrane-targeted p110α caused severe growth arrest, whereas a catalytically inactive mutant did not. Before the present study, two other groups expressed mammalian PI3K in yeast. In one study, expression of class I PI3K from the GAL1 promoter did not significantly interfere with growth . In the other , heterologously expressed PI3K was somewhat deleterious. In our work, we used a prenylatable C-terminal CAAX box to direct p110α to its substrate in the plasma membrane, most likely explaining its more potent growth-inhibitory action. Restoration of growth to these cells provides a straightforward positive selection to identify new inhibitors of PI3K that are efficacious in vivo. Co-expression of PTEN, but not a catalytically inactive mutant, relieved PI3K-induced growth arrest. Such cells provide a tool for testing the activity of various PTEN alleles in vivo and another positive growth-based assay to screen for compounds that enhance the function of crippled PTEN variants. Therefore this inexpensive yeast-based system may be useful in developing anti-cancer chemotherapeutics.
Using a PIP2-specific fluorescent reporter and PIP3-specific antibodies, we showed that PI3K expression caused a dramatic reduction in the PIP2 pool and a concomitant marked increase in the PIP3 content of the plasma membrane. Despite the presence of other elements in the p110α polypeptide that might perturb cell function (its Ras-binding domain, its lipid-binding C2 domain, the engineered CAAX box etc.), production of catalytically inactive p110α at the same level was innocuous. Therefore depletion of PIP2 appears to be the sole cause of the growth-inhibitory effects of PI3K expression, indicating that the plasma-membrane pool of PIP2 is essential for cell viability. Our results confirm, and significantly extend, conclusions drawn from globally eliminating all cellular PIP2 by shifting temperature-sensitive phosphatidylinositol 4-phosphate 5-kinase (mss4) mutants to restrictive temperature .
PI3K-mediated reduction in plasma-membrane PIP2 content altered both actin and septin architecture. The actin perturbation, by itself, could explain the growth arrest, based on prior reports showing that Mss4 is critical for proper actin organization  and that correct functioning of the actin cytoskeleton is essential for yeast cell viability . However, proper septin assembly is also essential . Our data provide the first demonstration that PIP2 itself is critical for the maintenance of septin structure in vivo, despite evidence to the contrary about the phosphoinositide binding specificity of yeast septins in vitro . The block to endocytosis caused by PI3K-induced PIP2 depletion was much more dramatic than the moderate impairment we found in secretory transport. This result is again consistent with, but confirms in a unique way, the cumulative evidence that efficient endocytosis requires PIP2 at the plasma membrane , whereas secretory transport requires phosphatidylinositol 4-phosphate on the Golgi membrane .
PI3K expression uniquely activated Slt2/Mpk1, the MAPK of the cell-integrity signalling pathway , but none of the three other MAPKs present. This pathway is initiated by the protein kinase, Pkc1 (yeast orthologue of mammalian PRK2), which is activated by GTP-bound Rho1, a small GTPase. Rom2, the guanine nucleotide-exchange factor for Rho1, is recruited to the plasma membrane, in part, via its PH domain, which has been shown to bind PIP2 and PIP3 equally well .
Overall, our results indicate that, for at least some essential cellular functions (actin assembly, septin assembly, endocytosis etc.), PIP3 cannot functionally substitute for PIP2. Indeed, PIP3 is normally never made in budding yeast (Sacch. cerevisiae) [15,47]. In contrast, PIP3 is present in fission yeast (Schizo. pombe), but kept at a low level, by action of a PTEN-like phosphatase, Ptn1 . The purported Sacch. cerevisiae PTEN orthologue, Tep1, could not alleviate the toxicity of PIP3 production, even when overexpressed. Moreover, we failed to detect PIP3 in tep1Δ mutants using the same anti-PIP3 antibodies that readily detected this lipid in Tep1-containing cells expressing PI3K (I. Rodríguez-Escudero, M. Molina and V.J. Cid, unpublished results). Thus Tep1 is probably not a functional 3-phosphoinositide phosphatase in vivo, in agreement with its lack of detectable enzyme activity in vitro . However, given that Tep1–GFP is restricted to cytoplasmic puncta, the possibility that Tep1 cannot gain access to the PIP3 generated upon PI3K expression has not been ruled out.
c-Akt expressed in yeast displayed a strictly PI3K-dependent translocation from a diffuse nucleocytoplasmic distribution to a distinct pattern of rim-staining around the plasma membrane. Given the vast body of evidence that the PH domain of c-Akt binds preferentially to PIP3 [and PI(3,4)P2] , this result provides independent proof that PIP3 was generated at the plasma membrane in yeast upon PI3K expression, corroborating the analysis we performed with the PIP3-specific antibodies. Immunoblotting with phospho-specific antibodies against P-Thr308 and P-Ser473 showed that phosphorylation at both sites was substantially stimulated upon PI3K expression. Even in the absence of PI3K expression, there was readily detectable basal phosphorylation at both sites; nonetheless, c-Akt1 overexpression was innocuous to the yeast cell. Using commercial antibodies (Santa Cruz Biotechnology, Santa Cruz, CA, U.S.A.) directed against the consensus phosphoacceptor motif preferred by c-Akt1, we found no difference in the pattern of bands when immunoblots of extracts of control yeast cells were compared with yeast cells expressing c-Akt1 (I. Rodríguez-Escudero, F. Roelants, J. Thorner, M. Molina, V. J. Cid, unpublished results). These findings suggest that high-level expression of c-Akt1 alone does not perturb yeast cell physiology because efficient substrates are absent. In this regard, among the important downstream targets of c-Akt in animal cells are members of the FOXO (Forkhead box O) family of transcription factors . However, the residues directly phosphorylated by c-Akt1 in their metazoan counterparts are not conserved in any of the four forkhead transcription factors encoded by the Sacch. cerevisiae genome (Fkh1, Fkh2, Fhl1 and Hcm1), at least as we align these proteins.
Sch9 is the Sacch. cerevisiae protein kinase that most closely resembles c-Akt1. Like AGC kinases that are targets of PDK1 in animal cells , Sch9 contains recognizable matches to the consensus PDK1 site (corresponding to Thr308 in c-Akt) and to the consensus PDK2 site (corresponding to Ser473 in c-Akt) [16,57]. Despite the similarities, the phospho-specific antibodies directed against c-Akt that we used did not detect any background band in lysates of control yeast cells. Apparently, limited sequence divergence within the motifs and in the immediately flanking residues is sufficient to prevent cross-reaction with Sch9. Moreover, all of the behaviours we report here for c-Akt were identical in sch9Δ cells (I. Rodríguez-Escudero, F. Roelants, J. Thorner, M. Molina, V. J. Cid, unpublished results) and, hence, none of our findings can be attributed to endogenous Sch9.
In addition to its PIP3-dependent translocation to the plasma membrane, other aspects of c-Akt regulation were also recapitulated in yeast cells. Phosphorylation of the PDK1 site (Thr308) in c-Akt in vivo was strictly dependent on the Sacch. cerevisiae PDK1 orthologues Phk1 and Phk2, as we showed previously in vitro . Moreover, mammalian PDK1 rescues the lethality of pkh1Δ pkh2Δ cells . Thus our current findings provide a further demonstration that PDK1 function has been highly conserved. We also found robust phosphorylation of the PDK2 site (Ser473), indicating that a PDK2-like activity exists in Sacch. cerevisiae. This modification was not due to obligatory autophosphorylation because Ser473 phosphorylation was still prominent in catalytically inactive c-Akt. Despite recent claims in animal cells that the major PDK2-site kinases are members of the atypical PIKK family, we found that phosphorylation of Ser473 in c-Akt was not reduced in yeast cells in which the mTOR-like (Tor1 and Tor2) or the other PIKK-like proteins (Tel1, Mec1 or Tra1) were inactivated using appropriate temperature-sensitive or null mutations. If, like PDK1, the function of the PDK2-site kinase has been highly conserved, then our results suggest that PDK2-site phosphorylation may not be a physiologically relevant function of mTOR, ATR or DNA-PKcs. Given that the number of protein kinases in Sacch. cerevisiae is substantially lower than that in animal cells, our findings may help narrow the search for the still elusive PDK2-site kinase.
We are grateful to C. Rivas, M. Collado, M. Lorenzo, J. Paramio, N. Hashash, R. Cha, C. Kiesecker, J. Workman, K. Matsumoto, M. Hall, D. Lew and D. Alessi for providing plasmids and strains. This work was supported by grant BIO2004-02019 from the Comisión Interministerial de Ciencia y Tecnología (CICYT, Spain) to M. M., by a fellowship from Comunidad Autónoma de Madrid (CAM, Spain) to I. R.-E., by a grant from the University of California–University of Madrid collaborative programme titled Complutense New Del Amo Program (NDAP) to V. J. C. and J. T., and by the NIH (National Institutes of Health) research grant GM21841 to J. T.
Abbreviations: ATM, ataxia telangiectasia; CPY, carboxypeptidase Y; DNA-PKcs, the catalytic subunit of DNA-dependent protein kinase; ERK, extracellular-signal-regulated kinase; HA, haemagglutinin; MAPK, mitogen-activated protein kinase; mTOR, mammalian target of rapamycin; PDK1, 3-phosphoinositide-dependent protein kinase-1; PH, domain, pleckstrin homology domain; PI3K, phosphoinositide 3-kinase; PIKK, phosphatidylinositol kinase-related kinase; PI(3,4)P2, phosphatidylinositol 3,4-bisphosphate; PIP2, phosphatidylinositol 4,5-bisphosphate; PIP3, phosphatidylinositol 3,4,5-trisphosphate; PKB, protein kinase B; PLCδ1, phospholipase δ1; PTEN, phosphatidylinositol 3-phosphatase; SH2, domain, Src homology 2 domain
- The Biochemical Society, London