Logo of pnasPNASInfo for AuthorsSubscriptionsAboutThis Article
Proc Natl Acad Sci U S A. Mar 27, 2007; 104(13): 5569–5574.
Published online Mar 21, 2007. doi:  10.1073/pnas.0701005104
PMCID: PMC1838453
Medical Sciences

Rare cancer-specific mutations in PIK3CA show gain of function

Abstract

Fifteen rare cancer-derived mutants of PIK3CA, the gene coding for the catalytic subunit p110α of phosphatidylinositol 3-kinase (PI3K), were examined for their biological and biochemical properties. Fourteen of these mutants show a gain of function: they induce rapamycin-sensitive oncogenic transformation of chicken embryo fibroblasts, constitutively activate Akt and TOR-mediated signaling, and show enhanced lipid kinase activity. Mapping of these mutants on a partial structural model of p110α suggests three groups of mutants, defined by their location in distinct functional domains of the protein. We hypothesize that each of these three groups induces a gain of PI3K function by a different molecular mechanism. Mutants in the C2 domain increase the positive surface charge of this domain and therefore may enhance the recruitment of p110α to cellular membranes. Mutants in the helical domain map to a contiguous surface of the protein and may affect the interaction with other protein(s). Mutants in the kinase domain are located near the hinge of the activation loop. They may alter the position and mobility of the activation loop. Arbitrarily introduced mutations that have no detectable phenotype map either to the interior of the protein or are positioned on a surface region that lies opposite to the exposed surfaces containing gain-of-function mutants. Engineered mutants that exchange acidic or neutral residues for basic residues on the critical surfaces show a gain of function.

Keywords: molecular model, phosphatidylinositol 3-kinase, lipid membrane, protein–protein interaction, activation loop

The catalytic subunit p110α of class I phosphatidylinositol 3-kinase (PI3K) is frequently mutated in cancer (15). The incidence of these cancer-specific point mutations varies with the tissue of origin and is particularly high in mammary and colorectal cancers (1, 2, 612). A conspicuous feature of the p110α mutations is their specific location in the gene of p110α: almost 80% of these mutations map to one of three hot spots in the p110α coding sequence (1). Two of the hot spot mutations are located in the helical domain of p110α at residue positions E542K and E545K, and the third hot spot (residue H1047R) is located in the C-terminal portion of the kinase domain. Besides these frequently occurring hot spot mutations, numerous rare, cancer-specific mutations have also been identified in p110α (1, 3, 7, 8, 10, 11). These rare mutations are widely distributed over the entire coding sequence of p110α. It has been assumed that the prevalence of hot spot mutations reflects the strong selective growth advantage that these mutants provide to the cell. This suggestion is supported by the fact that all three hot spot mutations confer in vitro and in vivo oncogenicity on p110α (1318). A corollary of this interpretation is that the rare mutants would not offer a comparable growth advantage and may therefore show either no phenotype or a loss of function. Consequently, they may be of no relevance to the oncogenic process. We studied 15 of these rare cancer-specific mutants of p110α and found that 14 show gain of function, inducing oncogenic transformation in primary cultures of chicken embryo fibroblasts (CEF), constitutive signaling to Akt, cellular homolog of the oncogene carried by the Akt8 murine retrovirus, and TOR and elevated lipid kinase levels. These observations suggest that even the rare mutants reflect positive selection, most likely by providing a significant growth advantage to the cell. Such growth-promoting activity suggests participation of rare mutations in the cancer process. A map of p110α point mutations on a homology model of the protein suggests the existence of three categories of mutants, defined by the structural and functional domains in which they are located. The data are in accord with the proposal that the three categories of mutants induce a gain of PI3K function through three distinct molecular mechanisms.

Results

Rare Cancer-Derived Mutants of p110α Induce Oncogenic Transformation in CEF.

We chose 15 rare mutants of p110α that had been identified in diverse human cancers and analyzed their biological and biochemical properties (R38H, K111N, N345K, C420R, P539R, E545A, E545G, Q546K, Q546P, H710P, T1025S, M1043I, M1043V, H1047L, and H1047Y). Fig. 1 shows a map of the mutations, illustrating their wide dispersion over the primary sequence of the protein. The mutants were expressed in CEF with the RCAS retroviral vector. Fourteen of the 15 mutants induced oncogenic transformation of the avian cells, as evidenced by focus formation (Fig. 2) and by the ability to propagate and form colonies in agar suspension (not shown). The efficiencies of transformation (EOT), expressed as number of foci per nanogram of DNA, were characteristic of individual mutants, extending from 1.7 to 4.6. Based on specific EOT, these mutants can be further classified into three groups: a class of weak transformers (EOT <2.0), a class of intermediate transformers (EOT 2.0–3.0), and a class of strong transformers (EOT 3.0–4.0) (Table 1 and Fig. 2). For comparison, the EOT of the hot spot mutations is >4.0 foci/ng. One specific cancer-derived mutant, H701P, failed to induce oncogenic transformation when expressed in CEF (Fig. 2). EOTs represent the efficiencies with which individual transforming events are initiated. In addition, the transformed cell foci induced by different mutant groups differ in size (Fig. 2). Focus size reflects the rate of cell replication; EOTs are correlated with the replication rates of the transformed cells, and strong transformers not only initiate transformation more efficiently, they also induce increased rates of cell growth.

Fig. 1.
A map and frequency distributions of the three hot spot mutations of the 15 rare mutations and of eight arbitrary mutations investigated in this study (catalog of Somatic Mutations in Cancer, www.sanger.ac.uk/genetics/CGP/cosmic). Hot spot mutations are ...
Fig. 2.
The EOT of primary CEF by mutants of p110α. The EOT (foci per nanogram of DNA) sets the hot spot mutations apart from the rare mutations and suggests the existence of three categories of rare mutations. Shown are transformation induced by a hot ...
Table 1.
Mutants of p110α grouped by oncogenic potency

Rare Cancer-Derived Mutants of p110α Signal Constitutively Through Akt.

The canonical signaling sequence initiated by PI3K proceeds through Akt, the tuberous sclerosis complex (TSC) proteins TSC1 and TSC2, the Ras-related GTPase Rheb to the TOR kinase (19). Two important downstream targets of TOR are S6K and 4E-BP. Determining the relative levels of phosphorylation of Akt, S6K, and 4EBP in the absence of growth factors assesses the constitutive activity of cellular PI3K signaling. We have probed the status of PI3K signaling in mutant-transfected cells by using Akt and S6K as indicators. All rare mutants that induce oncogenic transformation in cell culture also constitutively activate the downstream signaling cascade as suggested by the phosphorylation of Akt and S6K in the absence of growth factors (Fig. 3). Consistent with this signaling cascade, transformation by the rare mutants is also rapamycin-sensitive [supporting information (SI) Fig. 8]. The single nontransforming mutant (H701P), as well as the wild-type p110α, do not activate signaling under these conditions. The H701P mutation is located at the N terminus of an α-helix in the N-terminal region of the kinase domain, far from the active site and from the gain-of-function mutations.

Fig. 3.
Western blots comparing the phosphorylation levels of Akt and S6K that are induced in CEF by mutant and wild-type p110α. Hot spot mutants are marked with [down-pointing small open triangle], and strongly transforming rare mutants are marked with [diamond]. * marks ...

Rare Mutants with a Gain of Function in Oncogenicity and Signaling Also Show Elevated Lipid Kinase Activity in Vitro.

p110α proteins with specific amino acid substitutions that map to one of the cancer-specific mutational hot spots (E542K, E545K, and H1047R) show increased lipid kinase activity in vitro (1, 13, 14, 20). We have analyzed the lipid kinase activities of the 15 rare cancer-derived mutations and compared them to those of several nononcogenic mutants of p110α (SI Fig. 9). These results show that all rare gain-of-function mutants have elevated lipid kinase activity and that oncogenicity and constitutive signaling in mutants of p110α are roughly correlated with increased enzymatic activity in vitro.

Arbitrary Mutations Introduced into p110α Do Not Result in Oncogenicity, Constitutive Signaling, or Elevated Enzyme Activity.

Because almost all of the cancer-derived mutants of p110α show a gain of function, it was conceivable that wild-type p110α occurs predominantly in an inactive conformation, and that even nontargeted mutations would frequently trigger the transition to the active conformation. We therefore introduced random single amino acid substitutions at five different sites in the p110α coding sequence (E52K, E116K, P217K, G912R, and D1016G). The mutated proteins were expressed with the RCAS vector and tested in CEF for oncogenicity and constitutive signaling through Akt. They were also assayed for lipid kinase activity in vitro. None of these random mutants caused transformation in cell culture or generated a signal in the absence of growth factors (Fig. 4). Moreover, their lipid kinase activities are significantly lower than those of the gain-of-function mutants (SI Fig. 9).

Fig. 4.
Random mutations without phenotype. (A) Five constructs with mutations in arbitrary sites of p110α (P217K, G912R, D1016G, E52K, and E116K) fail to induce oncogenic transformation. The mutant proteins were expressed with the RCAS vector in primary ...

Gain-of-function Mutations of p110α Map to the Surface of the Protein.

Although the 3D structure of p110α has not yet been determined, it is possible to use the known structure of the related p110γ isoform to build a partial model of p110α based on sequence homology between the two isoforms (14, 2123). We have mapped the gain-of-function mutations and the nontransforming, arbitrary mutations shown in Fig. 1 on this homology model (Fig. 5). This model and the nature of the amino acid substitutions provide the following picture: (i) the gain-of-function mutations are grouped in three separate functional domains, the C2 domain, the helical domain and the kinase domain; (ii) two mutations in the C2 domain (N345K and C420R) are located in surface loops that were predicted to be part of the membrane binding region of p110α (21); (iii) three mutation sites in the helical domain (E542, E545, and Q546) are clustered on an exposed face of this domain that could interact with another protein. Similar to the C2 domain mutations, most helical domain mutations insert a basic residue, frequently a lysine (E542K, E545K, and Q546K), making it tempting to speculate that this region may also interact with the cellular membrane. The other mutation site in the helical domain, P539, is located in a loop that packs against the kinase domain in the model. This mutation may perturb the packing of the helical domain with the kinase domain or it may be more exposed in the actual structure. (iv) Five of the kinase domain mutations (M1043I, M1043V, H1047L, H1047Y, and H1047R) and one engineered mutation (D1045K; see below) pack against the hinge region of the activation loop. These mutations may lock the activation loop in the “on” position. The other mutation site (T1025) packs against the base of the catalytic loop and may affect the activity of the enzyme. The two nontransforming kinase domain mutations (Q1033K and D1017G) are completely solvent-exposed and are located far away from the active site (SI Fig. 10). Two of the other nontransforming mutations (H701P and P217K) are located on the opposite site of the molecule, away from the surface regions containing the gain-of-function mutations. A sixth kinase domain mutation (G912R), although located at base of the catalytic loop and near the T1025S mutations, has no phenotype, but may affect the stability of the protein. Together, these observations suggest three main mechanisms for the gain of function in p110α: (i) enhanced binding of the C2 domain to membrane phospholipids, (ii) modified binding of the helical domain to an unknown component or to a membrane, and (iii) increased catalytic activity through interaction with the activation or catalytic loops. The most potent of the kinase domain mutations, H1047R, is located near the hinge region of the activation loop and could affect both the position and the mobility of the activation loop (Fig. 6).

Fig. 5.
Ribbon and molecular surface representation of p110α, showing the locations of gain-of-function mutations on the homology model. The domains of p110α are colored as follows: C2 domain, blue; helical domain, green; N-terminal region of ...
Fig. 6.
Ribbon diagram focused on the catalytic domain of p110α mutant residues. Labeling and color code are as in Fig. 5. The model illustrates the position of H1047R hotspot mutation in proximity to the hinge region of the activation loop and suggests ...

Some Engineered Mutations That Increase the Positive Surface Charge of p110α Show a Gain of Function.

The map of the mutants on the partial protein model suggests that increasing the positive surface charge of p110α could lead to a gain of function. We tested this possibility by generating three point mutations that aim to introduce a positively charged amino acid to the surface of p110α. Two of these mutants (D1045K and E579K) showed a gain of function as judged by their ability to transform cells (Fig. 7A). The efficiencies of transformation achieved by these mutants were low (E576K EOT = 0.65, D1046K EOT = 1.6), and they did not significantly increase the phosphorylation level of Akt (Fig. 7B). E576K is located at the edge of a helical domain surface patch that includes E542, E545, and Q546 and could also affect the interaction with another protein or a membrane. As discussed above, D1045K may interact with the activation loop. These observations support the hypothesis that mutations that increase the positive charge on a surface patch of the helical domain or interact with the activation loop have the potential of increasing the activity of p110α.

Fig. 7.
Mutations targeted to the p110 surface. (A) Mutations targeted to the surface of p110α and involving an increase in positive charge are weakly oncogenic (E579K and D1046K). RCAS and wild-type p110α are shown as negative controls, H1047R ...

Discussion

In frequency of occurrence, the hot spot mutants in p110α differ from the rare mutants by at least one order of magnitude (Fig. 1). Their biological and biochemical activities are also significantly elevated above those of the rare mutants: their EOTs place the hot spot mutants in a distinct class above rare mutants. The differences between oncogenic activities of mutants in cell culture are even more pronounced than the EOTs suggest; not only do hot spot mutants induce higher numbers of transformed cell foci, but also individual foci are larger in size and density, suggesting more rapid cell proliferation. The same kind of differences can be observed between the three activity-defined categories of rare mutants: foci induced by the strong mutants are significantly larger and denser than those induced by weak mutants that have a distinctly lower EOT (Fig. 2). The hot spot mutants also consistently show the highest enzymatic activities and strongest effects on Akt. The data suggest a rough correlation between frequency of mutant occurrence in cancer and the extent of mutation-induced gain of function. The selective advantage for cellular replication mediated by a gain of function in PI3K could translate into this correlation between frequency of occurrence and PI3K activity, and the superior activity of the hot spot mutants may account for their higher frequency of occurrence in cancer. However, the fact that most of the rare mutations also induce a significant gain of function suggests these too contribute to the oncogenic process by providing a selective growth advantage to incipient tumor cells.

A study of rare mutants reveals several features commonly correlated with PI3K oncogenicity: enhanced lipid kinase activity, constitutive activation of Akt and TOR-mediated signaling, and high sensitivity of the oncogenic effects to rapamycin. The latter documents the importance of TOR signaling in the transformation process. In clinical settings, rapamycin has been far less effective (24, 25). However, recent studies with the human cancer cell lines that show enhanced PI3K signaling demonstrate that combined inhibition of PI3K and of TOR has a synergistic effect on tumor cell growth, whereas single inhibition of these targets is ineffective (26).

Mapping of the p110α mutations on the structural model of the protein suggests three groups of oncogenic mutants that are defined by their location on distinct functional domains and their differential effect on the activities of p110α. The two mutants in the C2 domain increase basic positive surface charge of that domain and may therefore mediate improved recruitment of p110α to the cell membrane, making lipid kinase activity independent of signals transmitted through the regulatory subunit, p85. Five of the six gain-of-function mutations mapping to the helical domain are clustered on an exposed surface patch of the protein. We speculate that they modify the interaction with another protein or with a lipid-containing membrane. The third group of mutations, located in the kinase domain, cluster close to the hinge region of the activation loop. These mutations may induce a gain of function by changing the position and mobility of the activation loop. This region could also serve as a target for the development of structure-based mutant-specific small molecule inhibitors. Our structure-guided interpretation of mutant effects is in substantial agreement with previous studies that considered the effects of mutant positions and the nature of the substituting amino acids on the function of p110α (14, 23). The possibility that some of the gain-of-function mutations enhance the thermal stability of the protein was also suggested previously and is supported by our observation that some of the transforming p110α mutants are expressed at significantly higher levels that the wild-type protein (23) (M.G. and P.K.V., unpublished work). The existence of more than one molecular mechanism for the mutation-induced gain of function in p110α is further supported by unpublished experiments with p110α constructs carrying double mutations, either in the same or different functional domains of the protein. The gain of function for mutations located in the same domain is not additive, whereas the gains of function resulting from mutations in different domains have an additive effect (L. Zhao and P.K.V. unpublished data). The arbitrarily introduced mutations that show no detectable phenotype map either to the interior of the protein (P217K and G912R) or on an exposed surface facing a different direction from the surfaces containing the gain-of-function mutations (E116K and D1017G). The map positions of these nonfunctional mutations are in accord with our proposed interpretation of the mechanisms governing the mutation-induced gain of function in p110α.

Collectively, the nature of the amino acid substitutions that induce a gain of function in the helical or C2 domain suggests that exposure at the surface of p110α and positive charge may be the two most important and perhaps more generally applicable requirements for activating the oncogenic potential of p110α. For mutations in the kinase domain, interaction with the activation loops appears to be a mechanism that could lock the enzyme in the “on” position. The cellular transforming ability of two mutants that were engineered to meet the two criteria of either surface location and positive charge or proximity to the activation loop supports this suggestion. The failure of these mutants to induce constitutive signaling through Akt remains unexplained.

Our observations on the rare cancer-derived mutants of p110α focus attention on the molecular mechanisms for the mutant-induced gain of function.

Materials and Methods

Cell Culture and Transformation Assays.

Primary cultures of CEF were prepared from White Leghorn embryos obtained from Charles River Breeding Laboratories (Preston, CT). For focus assays, DNA was transfected into CEF by using Lipofectamine Plus reagent (Invitrogen, Carlsbad, CA). Focus assays with infectious retroviral vectors were performed as described (27, 28). Rapamycin (Calbiochem, La Jolla, CA) was added at a concentration of 5 ng/ml in the nutrient agar overlay of the infected cells. Controls received DMSO vehicle instead of rapamycin. Transfected or infected cells were fed every other day with nutrient agar for 10 or 14 days and stained with crystal violet.

Plasmid Construction.

Wild-type human p110α was cloned into SfiI sites of the pBSFI cloning vector as described (29). The mutant constructs containing nucleotide substitutions described in this article were generated by using the QuikChange XL II site directed mutagenesis kit (Stratagene, La Jolla, CA) and the appropriate oligonucleotides (for the primer sequences, see SI Table 2). The wild-type and mutated PIK3CA genes were subsequently cloned as SfiI DNA fragments into the avian retrovirus vector RCAS(A)-SfiI.

Serum Starvation.

For serum starvation, CEF were cultured in Ham's F-10 medium (Sigma, St. Louis, MO) containing 0.5% FBS and 0.1% chicken serum for 40 h. The medium was then replaced with serum-free F-10 medium, and the cells were further incubated for 2 h. Cells were then harvested for protein analysis.

Western Blotting.

Western blotting was performed as described (29), with minor modifications. Cells were lysed in 1× Passive Lysis buffer (Promega, Madison, WI) containing a protease inhibitor mixture (Roche, Pleasanton, CA) and 1 mM PMSF/50 mM NaF/1 mM Na3VO4. Lysates containing 20–30 μg of protein were separated by SDS/PAGE and transferred to Immobilon P membranes (Millipore Billerica, MA). The membranes were blocked with 5% BSA/1× TBS/0.05% Tween-20 for 1 h at room temperature and incubated overnight at 4°C with primary antibodies. Anti-p110α antibody (sc-7174) was purchased from Santa Cruz Biotechnology (Santa Cruz, CA), and anti-Akt, anti-phospho-Akt (Ser-473), anti-S6K, anti-phospho-S6K (Thr-389) antibodies were obtained from Cell Signaling Technology (Beverly, MA). After rinsing with Tris-buffered saline-Tween 20 (TBS-T), the blots were incubated with secondary antibodies in TBS-T for 1 h at room temperature. The reactive bands were visualized by chemiluminescence (Pierce, Rockford, IL).

PI3K Assay.

In vitro PI3K activity was analyzed as described (29), with minor modifications. The immune complexes were prepared by incubating 400 μg of protein with 5 μl of anti-p110α antibody for 2 h at 4°C, followed by 2 h of incubation with Protein A-agarose. The beads were washed three times with lysis buffer and twice with TNE (10 mM Tris, pH 7.5/100 mM NaCl/1 mM EDTA). Subsequently, the immune complexes were incubated with 50 μl of kinase reaction buffer containing 20 mM Hepes (pH 7.5), 10 mM MgCl2, 200 μg/ml phosphatidylinositol (sonicated), 60 μM ATP, and 200 μCi/ml (1 Ci = 37 GBq) [γ-32P]ATP for 5 min at room temperature. The reaction was terminated by adding 80 μl of 1 M HCl, and the phosphorylated lipids were extracted with 160 μl of chloro-form/methanol (1:1). Samples were dried down, dissolved in chloroform, and spotted onto Silica Gel 60 TLC plates (EMD, San Diego, CA). The plates were developed in a borate buffer system (30) and visualized by autoradiography. Quantification of signals was analyzed by using phosphoimager (Molecular Dynamics, Sunnyvale, CA) and processed with ImageQuant software (Molecular Dynamics).

Modeling and display of the p110α structure. The p110α homology model was generated by using the SwissModel server and is based on the published structures of p110γ (PDBid: 1e8y, sequence ID 38%) and working with the known sequences of the α and γ isoforms. The model is analyzed and displayed with the aid of the Pymol program (DeLano Scientific, Palo Alto, CA).

Supplementary Material

Supporting Information:

Acknowledgments

The work of M.G. and P.K.V. is supported by grants from the National Cancer Institute. The work of M.-A.E. is supported by the National Institutes of Health, Protein Structure Initiative, Grant U54 GM074898. We thank Lynn Ueno for expert technical assistance and Kathleen M. Alexander for help with the manuscript. We also thank Leyna Zhao for permission to mention her unpublished observations on double mutants. Sohye Kang reviewed a draft of the manuscript and made valuable suggestions. This is manuscript no. 18601 of The Scripps Research Institute.

Abbreviations

PI3K
phosphatidylinositol 3-kinase
EOT
efficiency of transformation
CEF
chicken embryo fibroblasts.

Footnotes

The authors declare no conflict of interest.

This article contains supporting information online at www.pnas.org/cgi/content/full/0701005104/DC1.

References

1. Samuels Y, Wang Z, Bardelli A, Silliman N, Ptak J, Szabo S, Yan H, Gazdar A, Powell SM, Riggins GJ, et al. Science. 2004;304:554. [PubMed]
2. Samuels Y, Velculescu VE. Cell Cycle. 2004;3:1221–1224. [PubMed]
3. Campbell IG, Russell SE, Choong DY, Montgomery KG, Ciavarella ML, Hooi CS, Cristiano BE, Pearson RB, Phillips WA. Cancer Res. 2004;64:7678–7681. [PubMed]
4. Bader AG, Kang S, Zhao L, Vogt PK. Nat Rev Cancer. 2005;5:921–929. [PubMed]
5. Karakas B, Bachman KE, Park BH. Br J Cancer. 2006;94:455–459. [PMC free article] [PubMed]
6. Velho S, Oliveira C, Ferreira A, Ferreira AC, Suriano G, Schwartz S, Jr, Duval A., Carneiro F, Machado JC, Hamelin R, et al. Eur J Cancer. 2005;41:1649–1654. [PubMed]
7. Lee JW, Soung YH, Kim SY, Lee HW, Park WS, Nam SW, Kim SH, Lee JY, Yoo NJ, Lee SH. Oncogene. 2005;24:1477–1480. [PubMed]
8. Levine DA, Bogomolniy F, Yee CJ, Lash A, Barakat RR, Borgen PI, Boyd J. Clin Cancer Res. 2005;11:2875–2878. [PubMed]
9. Wu G, Xing M, Mambo E, Huang X, Liu J, Guo Z, Chatterjee A, Goldenberg D, Gollin SM, Sukumar S, et al. Breast Cancer Res. 2005;7:R609–R616. [PMC free article] [PubMed]
10. Broderick DK, Di C, Parrett TJ, Samuels YR, Cummins JM, McLendon RE, Fults DW, Velculescu VE, Bigner DD, Yan H. Cancer Res. 2004;64:5048–5050. [PubMed]
11. Saal LH, Holm K, Maurer M, Memeo L, Su T, Wang X, Yu JS, Malmstrom PO, Mansukhani M, Enoksson J, et al. Cancer Res. 2005;65:2554–2559. [PubMed]
12. Bachman KE, Argani P, Samuels Y, Silliman N, Ptak J, Szabo S, Konishi H, Karakas B, Blair BG, Lin C, et al. Cancer Biol Ther. 2004;3:772–775. [PubMed]
13. Kang S, Bader AG, Vogt PK. Proc Natl Acad Sci USA. 2005;102:802–807. [PMC free article] [PubMed]
14. Ikenoue T, Kanai F, Hikiba Y, Obata T, Tanaka Y, Imamura J, Ohta M, Jazag A, Guleng B, Tateishi K, et al. Cancer Res. 2005;65:4562–4567. [PubMed]
15. Bader AG, Kang S, Vogt PK. Proc Natl Acad Sci USA. 2006;103:1475–1479. [PMC free article] [PubMed]
16. Isakoff SJ, Engelman JA, Irie HY, Luo J, Brachmann SM, Pearline RV, Cantley LC, Brugge JS. Cancer Res. 2005;65:10992–11000. [PubMed]
17. Samuels Y, Diaz LA, Jr, Schmidt-Kittler O, Cummins JM, Delong L, Cheong I, Rago C, Huso DL, Lengauer C, Kinzler KW, et al. Cancer Cell. 2005;7:561–573. [PubMed]
18. Zhao JJ, Liu Z, Wang L, Shin E, Loda MF, Roberts TM. Proc Natl Acad Sci USA. 2005;102:18443–18448. [PMC free article] [PubMed]
19. Engelman JA, Luo J, Cantley LC. Nat Rev Genet. 2006;7:606–619. [PubMed]
20. Liu Z, Roberts TM. Cell Cycle. 2006;5:675–677. [PubMed]
21. Walker EH, Perisic O, Ried C, Stephens L, Williams RL. Nature. 1999;402:313–320. [PubMed]
22. Kang S, Bader AG, Zhao L, Vogt PK. Cell Cycle. 2005;4:578–581. [PubMed]
23. Stephens L, Williams R, Hawkins P. Curr Opin Pharmacol. 2005;5:357–365. [PubMed]
24. Faivre S, Kroemer G, Raymond E. Nat Rev Drug Discov. 2006;5:671–688. [PubMed]
25. Easton JB, Houghton PJ. Oncogene. 2006;25:6436–6446. [PubMed]
26. Fan QW, Knight ZA, Goldenberg DD, Yu W, Mostov KE, Stokoe D, Shokat KM, Weiss WA. Cancer Cell. 2006;9:341–349. [PMC free article] [PubMed]
27. Duff RG, Vogt PK. Virology. 1969;39:18–30. [PubMed]
28. Bos TJ, Monteclaro FS, Mitsunobu F, Ball AR, Jr, Chang CH, Nishimura T, Vogt PK. Genes Dev. 1990;4:1677–1687. [PubMed]
29. Aoki M, Schetter C, Himly M, Batista O, Chang HW, Vogt PK. J Biol Chem. 2000;275:6267–6275. [PubMed]
30. Walsh JP, Caldwell KK, Majerus PW. Proc Natl Acad Sci USA. 1991;88:9184–9187. [PMC free article] [PubMed]

Articles from Proceedings of the National Academy of Sciences of the United States of America are provided here courtesy of National Academy of Sciences
PubReader format: click here to try

Formats:

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...

Links

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...