|Home | About | Journals | Submit | Contact Us | Français|
Many tumours have chronically elevated activity of PI 3-kinase dependent signalling pathways, caused largely by oncogenic mutation of PI 3-kinase itself or loss of the opposing tumour suppressor lipid phosphatase, PTEN. Several PI 3-kinase dependent feedback mechanisms have been identified that may affect the sensitivity of upstream receptor signalling, but the events required to initiate an inhibited state have not been addressed. We show that in a variety of cell types, loss of PTEN via experimental knockdown or in tumour cell lines correlates with a block in IGF1/insulin signalling, without affecting the sensitivity of PDGF or EGF signalling. These effects on IGF/insulin signalling include a reduction of up to five to ten fold in IGF stimulated PI 3-kinase activation, a failure to activate the ERK kinases, and in some cells, reduced expression of IRS1, and both IGF1 and insulin receptors. These data indicate that chronically elevated PI 3-kinase dependent signalling to the degree seen in many tumours causes a selective loss of sensitivity in IGF1/insulin signalling that could significantly reduce the selective advantage of deregulated activation of IGF1/IGF1-R signalling in tumour development.
Cellular behaviour is controlled by external signals through the activation of signal transduction pathways. The sensitivity to stimulation of many cellular signal transduction pathways is dynamic and regulated by feedback and cross-talk with other pathways (Natarajan et al., 2006; Vivekanand & Rebay, 2006). Since most tumours have elevated activity of signalling pathways that control cell proliferation, survival and growth, often through mutation or deregulated expression of growth factors or their receptors, it appears that the effects on tumour development of mutations in a specific pathway will depend upon feedback and cross-talk from other pathways activated in the same tumour cell (Javelaud & Mauviel, 2005; Macrae et al., 2005). A further consequence of signalling cross-talk is that the outcome of using a drug inhibiting a particular signalling pathway will depend not only upon the perceived independent significance of the signal inhibited, but also any compensatory changes in other dependent interacting pathways (Cheung et al., 2003).
The PI 3-kinase signal transduction pathway is activated by numerous diverse stimuli, particularly many peptide growth factors acting through receptor tyrosine kinases. The pathway is characterised by the stimulated activation of class I PI 3-kinases that phosphorylate the abundant membrane phospholipid PtdIns(4,5)P2 to generate the second messenger PtdIns(3,4,5)P3. PtdIns(3,4,5)P3 in turn interacts with downstream targets that are able to recognise selectively and bind the lipid, including the protein kinase Akt, which is known to mediate many of the effects of the pathway on cell survival and growth (Stiles et al., 2002; Stocker et al., 2002). Most human cancers display elevated activity of the PI 3-kinase/Akt signalling pathway, caused most frequently by mutation of the PI 3-kinase subunit p110α, or loss of the opposing tumour suppressor phosphatase, PTEN (Cully et al., 2006; Shaw & Cantley, 2006), leading to increased growth, survival and proliferation of tumour cells. Significantly, however, deregulation of signalling mechanisms upstream of PI 3-kinase is also common in many tumours, particularly the ras small GTPases, and growth factors and their receptors, many of which are found to be activated through over-expression or mutation. The most commonly deregulated receptor systems in terms of tumour numbers include those for the Epidermal Growth Factor family (Shelton et al., 2005), Platelet Derived Growth Factors (Board & Jayson, 2005), Insulin-like Growth Factors (LeRoith & Roberts, 2003) and Hepatocyte Growth Factor (Danilkovitch-Miagkova & Zbar, 2002).
Insulin and insulin-like growth factor 1 (IGF1) regulate cell growth, survival and metabolism, via activation of the insulin and IGF1 receptor tyrosine kinases and phosphorylation of their principal substrates, the Insulin Receptor Substrate (IRS) proteins. The PI 3-kinase pathway then appears to be the principal downstream pathway mediating the cellular effects of insulin and IGF1, most of which are suppressed by pharmacological PI 3-kinase inhibitors. Signalling is via direct recruitment of PI 3-kinase to the tyrosine phosphorylated IRS proteins, with recent data suggesting a central role specifically for the p110α catalytic subunit of PI 3-kinase (Foukas et al., 2006; Knight et al., 2006). Although a large body of evidence has identified mechanisms of signalling cross-talk that may be responsible for clinical insulin resistance, often resulting in inhibitory phosphorylation of the IRS proteins (Pirola et al., 2004; White, 2002) the relevance of these pathways in cancer has received little attention.
Loss of PTEN function in many tumour cells has been shown to lead to elevated levels of PtdIns(3,4,5)P3 and activity of PI 3-kinase dependent signalling, including the PtdIns(3,4,5)P3-dependent protein kinase, Akt. The feedback effects of elevated PtdIns(3,4,5)P3 levels were revealed in PTEN null U87MG glioblastoma cells, known to display greatly elevated levels of the lipid, by investigating the effects on signalling activation caused by chronic reduction in PtdIns(3,4,5)P3 levels. In these cells we were able to address the specificity of any feedback effects on different receptor signalling systems, as they express receptors for IGF1, insulin, PDGF and EGF. Initial experiments looked at the effects on upstream PI 3-kinase activation, of reducing cellular PtdIns(3,4,5)P3 levels, either by re-expression of PTEN or by cellular inhibition of PI 3-kinase using reversible small molecule inhibitors that can be washed away before assaying PI 3-kinase in vitro. These experiments showed that reducing PtdIns(3,4,5)P3 levels caused a dramatic increase in the sensitivity with which IGF1 and insulin activate PI 3-kinase. Either re-expression of PTEN or treatment with the reversible PI 3-kinase inhibitor LY294002 for 24 hours increased the IRS-associated PI 3-kinase activity stimulated by either IGF-1 or insulin up to ten fold (Figures 1A, B, C and S1). In contrast, reduced cellular PtdIns(3,4,5)P3 levels had no significant effect on the activation of PI 3-kinase by PDGF (Figure 1D, E). As a control in these experiments, we also expressed a PTEN mutant, PTEN G129E (Myers et al., 1998), that lacks lipid phosphatase activity, but retains the protein phosphatase activity of the enzyme, and had no effect on the sensitivity of IGF and insulin signalling. In these experiments, LY294002 was generally used at a dose of 10μM, in order to achieve similar effects to physiological levels of PTEN expression. At this concentration in U87MG cells, cellular Akt phosphorylation is only partially reduced, and a dose curve of the effects of LY294002 on insulin stimulated PI 3-kinase activation demonstrates that at this dose, the PI 3-kinase inhibitor resulted in a sub-maximal sensitisation (Figure S2 and data not shown).
Re-expression of PTEN in these PTEN null glioblastoma cells itself inhibits PI3-kinase dependent signalling, and PTEN expression using baculovirus is usually to levels approximately two to five fold higher than those normally found in cells expressing endogenous PTEN, causing an unphysiological block in the activation of Akt and presumably other PI 3-kinase dependent processes. We therefore chose to address the effects of PTEN re-expression on the sensitivity to activation of pathways not principally regulated by PI 3-kinase, using the ERK MAPKs as an example in these cells. Using phosphospecific antibodies to look at the activation of the ERK pathway by IGF1 and insulin, which also requires many of the same upstream components, we found that IGF1 and insulin failed to stimulate the activation of ERK in U87MG cells at a number of time points and concentrations, in contrast to many published results from other cell types, yet the ERK kinases could be activated efficiently by stimulation of U87MG cells with EGF or PDGF (Figure 2 and data not shown). Re-expression of wild-type PTEN or inhibition of PI 3-kinase caused a recovery of the sensitivity of these cells to IGF1 and insulin, without altering the activation of ERK by EGF and PDGF (Figure 2B). In some experiments, this recovery in sensitivity was achieved with only a small reduction in the phosphorylation of ERK in unstimulated cells (Figure 2B), whereas in other experiments, the unstimulated ERK phosphorylation appeared to be more significantly reduced by PTEN activity (See Figure S3). A similar IGF1/insulin specific sensitisation of ERK activation in U87MG cells was achieved by chronic inhibition of PI 3-kinase (Figure 2C).
These experiments indicate that in U87MG cells, PtdIns(3,4,5)P3 dependent feedback mechanisms regulate the upstream sensitivity of the IGF-1/insulin signalling system, but not that of PDGF or EGF. To investigate potential mechanisms mediating these feedback effects, we looked at the effects of PtdIns(3,4,5)P3 levels on the expression of upstream components of the IGF/insulin signalling pathway in U87MG cells by immuno-blotting, including the IGF1 and insulin receptors and the IRS proteins. This work showed that expression of PTEN or pharmacological PI 3-kinase inhibition over a 24 hour period increased the expression levels of both IGF1 and insulin receptors and IRS1 (Figures 2B and C, 3A and B). We did not detect expression of IRS2 in U87MG cells. We also found that in some experiments, expression levels of total Akt kinase were increased by PTEN or PI 3-kinase inhibition, but this was not always reproducible (eg Figure S3). Since the level of the IRS proteins and the insulin receptor is known to be regulated through several mechanisms, we looked at the abundance mRNA encoding these proteins in U87MG cells by quantitative RT-PCR. The expression of both messages was increased about 3-fold by 24h incubation with the PI 3-kinase inhibitor, LY294002 (Figure 3C).
The use of reversible inhibitors of PI 3-kinase in these experiments allowed us to address the time-scale over which sensitisation was achieved, and compare the observed sensitisation of the pathway with effects on different possible mechanistic components. In these experiments, untreated insensitive cells were compared with those treated with PI 3-kinase inhibitor for either 24 hours or 30 minutes, and finally with cells that had been treated with inhibitor for 24 hours, but from which the inhibitor had been removed for the last hour before stimulation (termed ‘wash out’ protocol in figure labels). These experiments showed that PI 3-kinase activation by insulin was sensitised around three-fold by the short 30 minute treatment with the inhibitor and a similar result was achieved by a 24 hour ‘wash-out’ treatment, with the inhibitor removed 30 minutes before stimulation. Only cells chronically exposed to the inhibitor for 24 hours and remaining in the drug during stimulation were fully sensitised by around ten-fold, with respect to insulin induced PI 3-kinase activation (Figure 4A).
Similar experiments looking at the activation of ERK by IGF1/insulin showed, in contrast, that the sensitisation of ERK activation by PI-103 required the continued presence of the PI 3-kinase inhibitor (Figure 2C). This was further analysed in PI 3-kinase inhibitor timecourses, although the interpretation of these experiments was affected by a rapid inhibitory effect of PI 3-kinase inhibitors (especially the more potent PI-103) on insulin induced ERK activation (Figure 4B and data not shown), consistent with previous reports defining a role for PI3-kinase in this process (Sajan et al., 1999). This initial drop in ERK activity caused by PI 3-kinase inhibitors was followed by a gradual recovery in the basal ERK activity and a rapid development of sensitivity to IGF1 stimulation (Figure 4B). Analysis of Akt/PKB activation in these experiments showed a remarkable sensitisation after chronic PI-103 treatment (6h or 17h), displaying strong activation of the kinase in response to IGF1 despite the continued inhibition of PI 3-kinase (Figure 4B). Similar experiments performed with the mTOR/raptor inhibitor rapamycin showed that this did not lead to the same sensitisation seen with PI-103 (Figure S4).
In parallel experiments, the contribution of different candidates to these chronic and acute effects was investigated. The elevated expression of IRS1 and the insulin receptor required chronic PI 3-kinase inhibition and was not reversed rapidly by removal of the inhibitor (Figure 4B and C). We also investigated two inhibitory serine phosphorylation events upon IRS1 that have been reported to regulate insulin/IGF sensitivity, S307 and S636/9, and found that these sites were relatively highly phosphorylated in untreated U87MG cells but reduced greatly by short or long-term treatment with PI 3-kinase inhibitor, and recovered rapidly after removal of the inhibitor (Figure 4C). Thus for the sensitisation of PI 3-kinase activation by IGF and insulin, chronic effects including changes in the expression level of signalling components and acute effects, including changes in regulatory phosphorylation events, appear to contribute similar degrees of sensitisation during PI 3-kinase inhibition in U87MG cells. For the sensitisation of ERK activation by IGF and insulin, acute effects including changes in regulatory phosphorylation events appear to predominate.
These results show that several feedback mechanisms exist in U87MG cells that together can potently and selectively desensitise upstream IGF1/insulin signalling in response to high levels of PtdIns(3,4,5)P3. However, in order to address the widespread relevance of such effects, we first investigated the sensitivity of a number of cells lines, including several lacking PTEN, in some cases partially replicating published results (Batty et al., 2004; Jackson & Yee, 1999). These experiments showed that NIH3T3 fibroblasts, MCF7 breast carcinoma cells and MDCK kidney epithelial cells, all expressing PTEN, gave a robust activation of ERK in response to IGF or insulin (Figure (Figure55 and and6).6). However, similar to the U87MG glioblastoma cells (Figure 2A), the PTEN null prostate carcinoma cell line PC3 and the breast carcinoma cell line MDA-MB-468 failed to activate ERK detectably in response to IGF1 or insulin, although these cells could activate ERK in response to EGF and activate Akt or PI 3-kinase in response to IGF and/or insulin (Figure 6 and data not shown). Similarly, the PTEN null astrocytoma cell line 1321N1 and the germ cell tumour cell line NCCIT activated ERK detectably in these experiments, but this activation was only to a degree much weaker than that induced by EGF in these cells or by IGF and insulin in NIH3T3 or MCF7 cells (Figure 5).
These results together suggest that elevated levels of PtdIns(3,4,5)P3 and activated downstream signalling induced by PTEN loss may cause a desensitisation of multiple signalling pathways activated by IGF1/insulin signalling, through changes in receptor proximal signalling that may occur in many cell types. We determined to test this hypothesis in NIH3T3 fibroblasts and MDCK kidney epithelial cells in which PI 3-kinase dependent signalling was elevated by RNA-mediated knockdown of PTEN expression. These cells were chosen for these experiments due to the robust activation of ERK induced in these cells by IGF1 and insulin, their non-transformed nature and ease of transfection. These experiments, comparing the sensitivity of ERK activation by IGF1, insulin, PDGF and EGF, showed that when PTEN expression was reduced by RNA interference, both cell types displayed normal activation of ERK by PDGF and EGF, but were significantly suppressed in their ability to activate ERK in response to IGF1 or insulin (Figure 6A and B). In contrast, cells transfected in parallel with scrambled RNA sequences activated ERK normally in response to all four stimuli. In NIH3T3 cells this loss in sensitivity was achieved with no significant change in the phosphorylation of ERK in unstimulated cells, whereas in MDCK cell experiments, the unstimulated ERK phosphorylation appeared to be substantially increased by loss of PTEN, but was not increased further by IGF1 stimulation. The analysis of both cell types lacking PTEN also showed a modest reduction in the expression of IRS1 and IRS2 (greater reduction in MDCK cells), but no significant reduction in the expression of the insulin receptor (Figure 6A and B). When the activation of PI 3-kinase was analysed in these cells, it was found that PTEN siRNA and reduced expression caused an almost five-fold reduction in the activation of PI3-kinase by IGF1, but did not affect its activation by PDGF (Figure 6C). These results indicate that the elevation of PtdIns(3,4,5)3 dependent signalling caused by loss of PTEN is sufficient to desensitise significantly IGF1 induced ERK and PI 3-kinase activation.
To extend this conclusion, we generated further NIH3T3 cells with activated PI 3-kinase dependent signalling by expression of constitutively active mutants of the catalytic subunit of PI 3-kinase p110α and the downstream kinase Akt1. These experiments gave results very similar to those obtained with the PTEN knockdown cells, displaying a desensitisation in ERK activation by IGF and insulin, but not EGF or PDGF. In contrast, cells transfected in parallel with empty expression vector activated ERK normally in response to all stlimuli (Figures (Figures6D6D and S5). Hence the desensitising effects are likely to reflect predominantly changes induced by Akt and its downstream targets.
Our data show that loss of PTEN can lead to a selective loss of sensitivity in IGF1/insulin signalling through multiple mechanisms. We find that loss of PTEN causes a remarkable change in the IGF induced activation of PI 3-kinase; a five fold change in NIH3T3 cells when PTEN is knocked down, and up to ten fold when PTEN is re-expressed in PTEN null U87MG glioblastoma cells. We also find that loss of PTEN caused an almost complete block in the activation of the ERK MAP kinases by IGF/insulin when knocked down, and correlated with an insensitivity to activation by these stimuli in a variety of tumour cells. This is likely to be significant given the requirement for ERK activity in some cellular responses to IGF1 and insulin, including the protection from apoptosis and transcriptional responses (Keeton et al., 2005; Kiepe et al., 2005; Parrizas et al., 1997).
Many studies have investigated signalling pathways regulating the sensitivity of insulin and IGF signalling, particularly in the context of diabetes-related insulin resistance. These studies have identified several mechanisms of feedback and cross-talk that may be relevant to our work, with particular attention focusing on the mTOR/S6K mediated inhibitory phosphorylation and degradation of IRS1 (Harrington et al., 2004; Um et al., 2004) and additional pathways downstream of mTOR (Fisher & White, 2004; Harrington et al., 2005; Pirola et al., 2004; Um et al., 2006). Other pathways identified include the regulation of IRS-2 transcription by PTEN expression (Simpson et al., 2001), inhibitory phosphorylation of IRS1 by the PI 3-kinase dependent protein kinases, PKCζ (Liu et al., 2001; Ravichandran et al., 2001), and even direct phosphorylation of IRS1 via the protein kinase activity of PI 3-kinase itself (Lam et al., 1994). However, these mechanisms have generally not been considered in the setting of tumour development, and whether they affect the sensitivity of IGF/insulin signalling in cells lacking PTEN has not been addressed. One significant recent study showed that in cells including MCF7 cells, which express PTEN but have an activating mutation in PIK3CA and modest basal activation of Akt, rapamycin could activate Akt by blocking mTOR mediated feedback inhibition of IGF1 and IRS1 signalling (O’Reilly et al., 2006). This indicates that even without the dramatic activation of PI3K-Akt-mTOR often observed, signalling through IGF1 is in a partially feedback inhibited state. However, we show here that MCF7 cells can, for example, still robustly activate ERK and Akt in response to IGF and insulin. These data strongly support our conclusions, but address only a subset of the feedback pathways relevant in cells lacking PTEN and did not address whether further feedback inhibition was induced by PTEN loss and further elevated PtdIns(3,4,5)P3. Together with our own data however, this suggests that inhibitory feedback of IGF/insulin signalling mediated by PI3K acts over a very large range of pathway activities. A clearer understanding of the sensitivities of feedback signalling would seem to require the development of more quantitative assays to compare pathway activation in different cells (Cutillas et al., 2006).
Our data indicate that several mechanisms are likely to contribute to the desensitisation of IGF/insulin signalling caused by PTEN loss in different cell types. In U87MG cells, expression levels of IRS1, and both IGF1 and insulin receptors were increased by PTEN expression or PI 3-kinase inhibition, with transcription of IRS1 and the insulin receptor being increased by PI 3-kinase inhibition. Inhibitory serine/threonine phosphorylation of IRS1 was also reduced. In some experiments in these cells and also in NIH3T3 and MDCK cells, PTEN expression or PI 3-kinase inhibition correlated with a small increase in the expression levels of total Akt kinase, although this was not always evident (Figure S3 and data not shown). The similarity of the effects of PTEN expression to the effects of pharmacological PI 3-kinase inhibition, and the lack of effect of the selectively lipid phosphatase dead PTEN G129E mutant, strongly suggests that the principal mechanism of action of PTEN in these experiments is through its lipid phosphatase activity and the regulation of cellular PtdIns(3,4,5)P3 levels. Since both LY294002 and PI-103 may also inhibit the downstream PI kinase-related kinase, mTOR (Knight et al., 2006), it appears that many of the effects of PTEN may be mediated by this conserved PI3K-Akt-mTOR signalling pathway. In support of this, some of the effects of PTEN, LY294002 and PI-103 were also caused in these experiments by rapamycin, an inhibitor of mTOR-raptor activity, particularly those on IRS1 protein and mRNA levels in U87MG cells (Figure S4 and data not shown). The effect of rapamycin on IGF induced ERK activation in U87MG cells was difficult to interpret, as rapamycin treatment caused a small but rapid activation of ERK (Figure S4). As this ERK activity was not further stimulated by IGF1, it seems probable that the sensitisation of ERK caused by PTEN and PI 3-kinase inhibition is independent of mTOR-raptor, but this remains equivocal. Significantly in this context, either knockdown of PTEN, or expression of constitutively active mutant of either PI 3-kinase p110α or Akt1 was sufficient to desensitise ERK activation by IGF and insulin in NIH3T3 cells.
Together, our data indicate that multiple pathways downstream of PtdIns(3,4,5)P3 play roles regulating the upstream sensitivity of IGF/insulin signalling. Our direct analyses of the effects of PTEN knockdown come from two untransformed cell lines, NIH3T3 fibroblasts and MDCK kidney epithelial cells, in which loss of PTEN caused the activation of ERK by IGF and insulin to be almost abolished and a dramatic drop in the activation of PI 3-kinase by IGF1. These data are supported by evidence that four out of five PTEN null cell lines of different lineages examined failed to activate ERK in response to IGF and insulin, and that in U87MG cells re-expression of PTEN led to a sensitisation of IGF/insulin stimulated ERK activation. These data strongly suggest that the selective suppression of IGF/insulin signalling by PTEN loss may be a general characteristic shared by many cell lineages and therefore of importance in many different tumours. Significantly, however, the relative significance of specific mechanisms in this feedback appears to vary greatly between cell types. For example, PTEN knockdown in MDCK cells caused a drop of approximately 90% in the expression levels of IRS2, whereas in NIH3T3 cells, knockdown of PTEN caused a more modest reduction in expression of the IRS proteins, of approximately 50%.
The efficient use of anticancer drugs targeting specific signalling molecules requires a reliable diagnostic understanding of which tumours are likely to be sensitive to which drugs. Or data suggest that in many circumstances, the selective advantage to a tumour caused by deregulation of IGF-R signalling would be much less in a tumour lacking PTEN or with an activating mutation in PI 3-kinase. Indeed, what limited data are available regarding the loss of PTEN and over-expression of the IGF1-R indicate a negative correlation between these events in metastatic prostate cancer (Hellawell et al., 2002). Since anticancer therapies are being developed to target the IGF1-R (Mitsiades et al., 2004), our work suggests that the efficacy of these inhibitors could be greatly affected by the status of the PI 3-kinase/PTEN/Akt pathway in some target tumours.
Stock cultures of U87-MG cells were maintained in Minimum Essential Medium (MEM) supplemented with 10% foetal-calf serum, 2mM glutamine and 1X non-essential amino acids at 37°C in a humidified atmosphere with CO2 at 5%. Stocks of NIH3T3 fibroblasts, MDCK canine kidney epithelial cells, 1321N1 astrocytoma, PC3 prostate carcinoma, MDA-MB-469 breast carcinoma and NCCIT germ cell tumour cells were maintained in DMEM + 10% FBS. Cell clones expressing activated Akt or short hairpin RNAs were derived by transfection of linearised vectors using Fugene-6 (Roche) followed by selection of single cell derived clones in 96 well plates with G418 (0.5mg/ml) for 3 weeks. Unfortunately, it was not found possible to derive stable clones expressing active mutants of PI 3-kinase p110α. For LY294002 wash-out experiment, the 24hr LY294002 treatment was removed, cells washed twice with fresh medium and then incubated in this medium for an hour prior to the subsequent 10min stimulation with 150nM insulin. Expression vectors encoding constitutively active mutants of Akt/PKB and p110α PI 3-kinase were kindly provided by Dario Alessi and Bart Vanhaesebroeck. Adapted baculoviruses containing the PTEN cDNA downstream of a CMV promoter were prepared in SF9 cells, using standard protocols developed for recombinant protein expression in insect cells, and added to low confluence U87MG cell cultures for 24 hours at 5% (v/v) culture volume. The use of fluorescently marked proteins and functional studies show that this routinely led to relatively even expression of target proteins in well over 95% of the cultured U87MG cells as previously described (Leslie et al., 2001). A mixture of four short interfering (si) RNAs targeting murine Pten, individual siRNAs and scrambled controls were purchased from Dharmacon. Vectors for the stable expression of PTEN shRNA were made using PTEN previously published PTEN siRNA sequences (Mise-Omata et al., 2005) (see Supplementary Table S1) incorporated into the pRNA-H1/NEO shRNA vector (GenScript).
Quantitative RT-PCR was used to determine the expression of IRS1 and the Insulin receptor, and used EST plasmids for the encoding genes and a β2-microglobulin control as experimental standards, adapting the method of Darragh et al (Darragh et al., 2005). These ESTs were obtained from the IMAGE consortium (EST numbers Ins-R: 4823710 and 5725876; IRS1: 6570696 and 6053288; β2 microglobulin: 5502428. Cellular RNA was purified using a Qiagen RNeasy kit, and reverse transcribed using an iScript cDNA synthesis kit (BioRad) both following the manufacturer’s instructions. Quantitative PCR was carried out using an iCycler and all reagents were also from BioRad. The PCR primers used are described in Supplementary Table S1.
Assays of cellular PI 3-kinase using PI as a substrate in vitro used an adaptation of a previous method (Batty et al., 2004). Cells were washed once in ice cold PBS and lysed in buffer containing 25mM HEPES (pH 7.4), 150mM NaCl, 1mM EDTA, 1mM EGTA, 1mM MgCl2, 5mM beta-glycerophosphate, 50mM NaF, 5mM sodium pyrophosphate, 1mM Na3VO4, 1mM DTT, 0.2mM PMSF, 1mM benzamidine, 1mg/ml leupeptin and 1% v/v NP-40. 0.5mg of cellular protein was mixed for 1 hour with 1μg anti-IRS-1 antibodies pre-bound to Protein G-Sepharose beads. These immunocomplexes were then washed twice in lysis buffer, twice in lysis buffer without detergent (NP-40) and finally twice in reaction buffer (120mM NaCl, 5mM beta-glycerophosphate, 2.5mM MgCl2, 1mM EGTA, 0.2mM EDTA, 25mM HEPES (pH 7.4), 1mM DTT, 0.2mM PMSF, 1mM benzamidine, and 1mg/ml leupeptin). Reactions were then started with the addition of pre-mixed substrate vesicles (100μM phosphatidylethanolamine, 100μM phosphatidylinositol, prepared by sonication) and 5μM ATP, including 5μCi [γ-32P] ATP in reaction buffer to the washed immunocomplexes to a final volume of 50μl. After incubation at 37° for 30 minutes, reactions were stopped by the addition of 600μl 80:40:1 (by volume) CH3OH:CH3Cl:12M HCl. After the addition of a further 200μl CHCl3 and 320μl 0.1M HCl, samples were mixed by vortexing, and separated by centrifugation at 14,000g for 1 minute. After removal of the upper phase, the lower phase was washed with a further 780μl of synthetic upper phase and the upper phase again removed to waste. The final lower phase, containing the product lipids, was then neutralized with 15μl 1.0M NH4OH in methanol and dried under vacuum. Lipids were finally dissolved in 30μl of 2:1 CHCl3:CH3OH (by volume) and separated by thin layer chromatography, before final detection of 32P PI(3)P using a Fuji Phosphorimager. Analysis of these data was performed using Aida image analysis software and was initially validated by comparison with duplicate data generated by scintillation counting of 32P PI(3)P samples in silica scraped from TLC plates as described in the original protocol. An example of the primary data from these assays is shown in Figure S1.
We would like to thank Rodolfo Marquez for the PI 3-kinase inhibitor PI-103 and Doreen Cantrell, Bart Vanhaesebroeck and Dario Alessi for expression constructs. This work was supported by the Medical Research Council, the Association for International Cancer Research and the Dundee DSTT consortium (Astra Zeneca, Boehringer Ingelheim, GlaxoSmithKline, Merck and Co., Merck KGaA and Pfizer).