Rac1-mediated membrane raft localization of PI3K/p110β is required for its activation by GPCRs or PTEN loss
We aimed to understand how spatial compartmentalization in the plasma membrane might contribute to the functions of the ubiquitous class IA phosphoinositide 3-kinase (PI3K) isoforms, p110α and p110β. We found that p110β localizes to membrane rafts in a Rac1-dependent manner. This localization potentiates Akt activation by G-protein-coupled receptors (GPCRs). Thus genetic targeting of a Rac1 binding-deficient allele of p110β to rafts alleviated the requirement for p110β-Rac1 association for GPCR signaling, cell growth and migration. In contrast, p110α, which does not play a physiological role in GPCR signaling, is found to reside in nonraft regions of the plasma membrane. Raft targeting of p110α allowed its EGFR-mediated activation by GPCRs. Notably, p110β dependent, PTEN null tumor cells critically rely upon raft-associated PI3K activity. Collectively, our findings provide a mechanistic account of how membrane raft localization regulates differential activation of distinct PI3K isoforms and offer insight into why PTEN-deficient cancers depend on p110β.https://doi.org/10.7554/eLife.17635.001
The phosphoinositide 3-kinase (PI3K) pathway has been widely studied as a major regulator of cellular proliferation, metabolism and migration (Fruman et al., 1998; Engelman et al., 2006; Cain and Ridley, 2009). PI3Ks catalyze phosphorylation of the 3’OH group on phosphoinositides at the plasma membrane, which induces translocation of the Ser/Thr kinase Akt. Once recruited to the plasma membrane, Akt becomes activated by PDK1 and mTOR2 dependent phosphorylation events (Alessi et al., 1997; Sarbassov et al., 2005). Activation of PI3K signaling frequently occurs in human cancers (Vivanco and Sawyers, 2002).
The PI3Ks are organized into three classes based on their structural similarities and mechanism of activation. The most extensively studied class I PI3Ks are activated by growth factor receptors. Of these, class IA enzymes are activated via receptor tyrosine kinases (RTKs), small G-proteins and G-protein-coupled receptors (GPCRs) and PI3K signaling is functionally antagonized by the PIP3 degrading PI3 phosphatase PTEN (phosphatase and tensin homolog) (Sang et al., 2012). p110α and p110β are the catalytic class IA isoforms with a ubiquitous expression pattern (Jia et al., 2009). Heterodimerization with the regulatory p85 subunit stabilizes p110 isoforms but at the same time keeps their kinase activity in check through inhibitory interactions (Yu et al., 1998; Zhang et al., 2011). Although both p110α and p110β form heterodimers with p85 and catalyze the same biochemical reaction upon activation, studies have indicated that they have distinct functions (Foukas et al., 2006; Zhao et al., 2006; Jackson et al., 2005; Jia et al., 2008; Guillermet-Guibert et al., 2008; Ni et al., 2012). p110α appears to mediate bulk of PI3K signaling downstream of RTKs whereas p110β preferentially signals downstream of GPCRs. Notably p110α and β isoforms are involved in mutually exclusive signaling complexes with distinct members of small GTPase protein superfamily, Ras and Rac1 respectively (Gupta et al., 2007; Fritsch et al., 2013). However, we are far from a complete understanding of the molecular mechanisms that account for the full spectrum of their functional differences.
In cancer, p110α is the main isoform required for transformation by oncogenes (Zhao et al., 2006) and the encoding PIK3CA is frequently found to have suffered from one of two common activating hotspot mutations in many tumor types (Samuels et al., 2004). This reliance of tumors on p110α may be partially explained by the fact that it has much higher specific activity compared to p110β (Zhao et al., 2006; Beeton et al., 2000; Knight et al., 2006). Indeed in a GEM model of breast cancer driven by Her2, knockout of p110α blocked tumor formation while knockout p110β actually caused tumors to develop more quickly- a surprising result explained by a competition model where the less active p110β binds to the same limiting number of p85/p110 binding sites on Her2 and thus actually lowers signaling output compared to the case where only p110α is expressed and hence can bind to all of the PI3K binding sites on the receptor (Utermark et al., 2012). Notably tumors that have lost expression of PTEN are largely dependent on p110β (Jia et al., 2008; Ni et al., 2012; Wee et al., 2008), an unexpected outcome given its weaker kinase activity. While it is possible that some loss of PTEN tumors feature activation of GPCRs thus explaining their p110β dependence (Rodriguez et al., 2016), that does not appear to be a unifying theme.
The plasma membrane is the site of activation for numerous signaling cascades, and its organizing principles contribute to the specificity and potency of these sequences of biochemical reactions (Kusumi et al., 2012). The lipid raft hypothesis proposes that the preferential association among sphingolipids, sterols, and specific proteins confers upon the lipid bilayer the potential for lateral segregation of key signaling molecules, dramatically transforming the cell membrane from a passive bystander to a dynamic entity facilitating signaling through subcompartmentalization (Lingwood and Simons, 2010). Sphingolipid and cholesterol enriched membrane rafts play a key signaling role in T cell activation via the T cell synapse (Gaus et al., 2005), B cell activation (Gupta and DeFranco, 2007), focal adhesions, cell migration (Gaus et al., 2006), membrane trafficking in polarized epithelial cells (van Meer et al., 1987) and hormone signaling (Márquez et al., 2006).
Fluorescence correlation spectroscopy data implicate involvement of raft nanodomains in recruitment of Akt to the cell membrane upon PIP3 production (Lasserre et al., 2008), and FRET-based Akt activity reporters uncovered a preferential activation of Akt in membrane rafts (Gao and Zhang, 2008). On the other hand PTEN was shown to localize selectively to nonraft membrane microdomains in human embryonic kidney cells, possibly further restricting Akt activation in these regions (Gao et al., 2011). However, although membrane microdomain compartmentalization is believed to be a crucial mechanism for achieving signaling specificity, the role of spatial partitioning in class IA PI3K signaling has remained elusive.
To address these questions, we combined simultaneous knock-out of p110α and p110β with genetic targeting that enabled us to express unique class IA PI3K isoforms directed to specific plasma membrane microdomains. Using this strategy, we investigated Akt activation, cellular proliferation and migration in response to growth factors, when expression of particular class IA PI3K isoforms was directed to membrane rafts or nonraft regions of the plasma membrane. We found that raft targeting of either p110α or p110β potentiates GPCR mediated activation of Akt. In addition, we determined that p110β required Rac1 binding for raft localization and Gβγ association for activation downstream of GPCRs, whereas raft-targeted p110α was dependent on EGFR activity. Notably we found that PI3K signaling was also dependent on raft integrity in PTEN null cancer cells. Taken together, these results indicate that any class IA PI3K catalytic isoform, when targeted to GPCR signaling permissive membrane microdomains, could activate Akt and maintain downstream PI3K signaling through distinct mechanisms. Finally, our data has novel implications concerning the optimal methods for inhibiting Akt activation in PTEN null tumors.
Generation of Isogenic MEFs expressing p110α or p110β
To facilitate the study of the individual p110 isoforms, we first generated immortalized p110αflox/flox; p110βflox/flox mouse embryonic fibroblasts (MEFs). Expression of endogenous p110α and p110β is almost entirely lost upon transduction of the cells with an adenovirus (AdCre) expressing the Cre-recombinase (Figure 1A). These double knock-out (DKO) MEFs show a reduction of phosphorylated Akt (p-Akt) (Figure 1A, compare lanes 1–5 with lanes 6–10), have a blunted response in promotion of p-Akt and p-S6 to a variety of growth signals (Figure 1B, compare lanes 2–5 with lanes 7–10) and essentially cease to proliferate (Figure 1C), which is consistent with the key role of the PI3K pathway in proliferation and signaling.
To generate isogenic MEFs expressing either p110α or p110β, we ectopically expressed comparable levels of wild-type (wt) HA-p110α or HA-p110β in DKO MEFs (henceforth DKO 'add-back' is used to denote these MEFs) (Figure 1D). Notably, ectopic expression of either p110 isoform could rescue PI3K signaling defects and growth inhibition in the DKO MEFs (Figure 1—figure supplement 1A,B). The resulting isogenic DKO add-back MEFs were highly sensitive to small molecule inhibitors targeting the unique isoform expressed, in growth factor signaling and proliferation assays (Figure 1E and Figure 1—figure supplement 1A,B). These cells have allowed us to compare the two commonly expressed p110 isoforms in much more detail than has previously been possible.
Rac1 binding constitutes a raft localization signal for p110β
Using this robust molecular genetic system, we attempted to address how spatial compartmentalization might contribute to the specificity of the activation of p110β by GPCRs. To determine whether wt p110β becomes differentially localized to membrane microdomains upon GPCR activation, we first analyzed triton sensitive and resistant membrane fractions in p110α-wt and p110β-wt DKO add-back MEFs. Our fractionation experiments demonstrated that p110β resides in detergent resistant membranes in addition to soluble and detergent sensitive fractions where p110α is primarily detected (Figure 2—figure supplement 1A). A detergent-free fractionation based on density gradient centrifugation in HMECs confirmed the notion that there are distinct pools of p110β localized to raft and nonraft membrane microdomains (Figure 2A). Of note, Rac1, which is known to interact with p110β displayed a similar fractionation pattern. p110α in comparison, can only be found in nonraft membranes. In addition to MEFs and HMECs; we analyzed HEK293, DU145, PC3, MCF7 and BT549 cancer cell lines by examination of triton sensitive and resistant membrane fractions and detected p110β in raft as well as nonraft fractions (Figure 2—figure supplement 1B).
As a canonical class IA PI3K isoform, p110β functions as a GPCR effector and interacts with the Gβγ component of the tripartite G-protein (Guillermet-Guibert et al., 2008; Dbouk et al., 2012). Physical association of p110β with Rac1 is also required to potentiate signaling (Fritsch et al., 2013). To test whether Rac1 or Gβγ may be responsible for localization of p110β to membrane rafts, we introduced point mutations to p110β that abolish its binding to Rac1 or Gβγ (Fritsch et al., 2013; Dbouk et al., 2012) (Figure 2B). Next, we generated DKO add-back MEF lines that express p110β-wt, Gβγ (GBM) or Rac1 (RBM) binding mutants as their only class IA PI3K isoform (Figure 2C). p110β-GBM or RBM DKO add-backs displayed retarded proliferation in comparison to either wt or p110β-wt DKO add-back MEFs (Figure 2D) and were largely deficient in migration through Transwell inserts towards chemoattractants (Figure 2E). Starvation and serum or GPCR agonist stimulation experiments demonstrated that although p110β deficient in either Gβγ or Rac1 binding, instigated Akt phosphorylation upon serum stimulation, both were significantly impaired in elevating p-Akt, in response to two GPCR agonists; lysophosphatidic acid (LPA) and sphingosine 1-phosphate (S1P) (Figure 2F). Interestingly, fractionation of raft versus nonraft membrane microdomains revealed that only p110β-GBM but not p110β-RBM localizes to rafts suggesting that Rac1 binding to p110β might constitute a raft localization signal (Figure 2G). We next attempted to verify this in HMECs, where p110β could readily be detected in triton resistant membrane fractions. We knocked-down Rac1 expression using Rac1 specific siRNAs and analyzed triton sensitive and resistant membrane microdomains. Downregulation of Rac1 led to a displacement of p110β from membrane rafts whereas raft and nonraft membrane marker distributions were unchanged (Figure 2H).
To determine whether raft targeting could compensate for Gβγ or Rac1 binding deficiencies in p110β, we constructed raft and nonraft-targeted p110β vectors using protein lipidation sequences derived from the Lyn tyrosine kinase and Kras respectively (Gao and Zhang, 2008) (Figure 3A) and generated p110β-Lyn and p110β-Ras DKO add-back MEFs. Analysis of homogenized samples fractionated via density gradient centrifugation confirmed enriched localization of p110β-Lyn to rafts and p110β-Ras to nonraft membrane microdomains (Figure 3B). Elution of triton sensitive and triton resistant membrane fractions further demonstrated enrichment of the targeting plasmids at the desired microdomains (Figure 3C). p110β-Ras DKO add-back MEFs displayed a blunted response in Akt phosphorylation upon serum starvation and LPA stimulation whereas both p110β-wt and p110β-Lyn DKO add-backs had significant levels of p-Akt upon LPA stimulation (Figure 3D). Similarly, an increase in membrane-associated p-Akt was observed, when p110β-Lyn but not p110β-Ras DKO add-back MEFs were stimulated with LPA (Figure 3—figure supplement 1).
After confirming the involvement of raft-localized p110β in GPCR mediated activation of Akt, we added Lyn-domain to all versions of p110β (Figure 4A) and generated p110β-Lyn, p110β-Lyn GBM or p110β-Lyn RBM DKO add-back MEF lines (Figure 4B, left). As expected, add-back of p110β-wt enriched at rafts (Figure 4B, right) induced p-Akt in response to serum or LPA stimulation (Figure 4C, left). Interestingly, raft targeting restored p-Akt in p110β-Lyn RBM, but not in p110β-Lyn GBM DKO add-back MEFs upon LPA stimulation (Figure 4C, right). Taken together, these results strongly suggest that p110β-Rac1 association regulates raft recruitment of p110β, and imply that the Gβγ interaction with p110β might be independently required for its activation.
While p110β-Lyn is enriched in rafts, it is still moderately localized in nonraft regions, possibly through p85 SH2/SH3 domain-mediated molecular interactions. Therefore we tested whether raft localization is critical for compensating Rac1-binding deficiency in p110β-Lyn RBM. To this end, we selectively disrupted the formation of membrane rafts with the cholesterol-depleting agent, methyl-β-cyclodextrin (MβCD) in starvation-LPA stimulation experiments. Both p110β-Lyn and p110β-Lyn RBM failed to induce p-Akt when rafts were disrupted by MβCD, a defect which was rescued upon addition of excess cholesterol to the cells (Figure 4D). Moreover, the raft-excluded p110β-Ras allele was relatively insensitive to cholesterol depletion upon stimulation with either serum or PDGF, and promoted Akt phosphorylation in the presence of MβCD (Figure 4E). Notably LPA stimulation failed to induce p-Akt in either condition, presumably because p110β-Ras is excluded from rafts.
Gβγ interaction is critical for p110β activation downstream of GPCRs
To identify the phenotypic consequences of raft-targeted, Gβγ or Rac1 binding deficient p110β expression, we first determined the growth rate of our p110β DKO add-back MEF lines under limiting amounts of mitogenic stimuli. Deficiency in Gβγ binding, significantly reduced rates of proliferation while p110β-Lyn RBM supported growth comparable to p110β-Lyn (Figure 5A).
Next, we investigated the function of raft-targeted p110β in cellular migration, which is known to be regulated by GPCR and Rac1 dependent signaling (Dorsam and Gutkind, 2007; Bid et al., 2013). Specifically we analyzed the efficiency of our DKO add-back MEFs in wound healing assays. Wound closure was monitored up to 12 hr upon scratching. p110β-Lyn RBM DKO add-back MEFs migrate more efficiently into the wound than p110β-Lyn GBM DKO add-back cells (Figure 5B,C). This observation is consistent with a distinct activating role of Gβγ on p110β, which cannot be compensated by relocalization of the molecule to rafts. Transwell migration assays using the same add-back lines reinforced the notion that selective raft targeting is compensatory for p110β-RBM whereas p110β-Lyn GBM DKO add-backs are defective in migration (Figure 5D,E). Taken together, this data is again consistent with the idea that the primary function of p110β-Rac1 binding lies in the placement of p110β in membrane rafts and thus creating a GPCR signaling competent PI3K module.
Raft Targeted p110α facilitates GPCR Signaling and has redundant functions with p110β
To test our hypothesis that Rac1 binding localized p110β to GPCR rich membrane microdomains where it could be further activated by a second signal, we attempted to see if we could render p110α, which is not normally GPCR responsive, sensitive to GPCR signaling by changing its membrane localization. To this end, we generated vectors that selectively target wt p110α to subdomains of the plasma membrane. Following our previous work, p110α was targeted to membrane rafts by using the Lyn-derived targeting motif and to nonraft membranes by the motif derived from Kras (Figure 6A). Analysis of homogenized samples fractionated via density gradient centrifugation confirmed enriched localization of p110α-Lyn to rafts, whereas p110α-Ras was more prominently localized to nonraft regions of the plasma membrane (Figure 6B). Elution of triton sensitive and triton resistant membrane fractions further demonstrated enrichment of the targeting plasmids at the desired microdomains (Figure 6—figure supplement 1A). Surprisingly, use of LPA led to a significant phosphorylation of Akt in cells expressing raft-targeted p110α-Lyn in a manner comparable to that seen in p110β-wt DKO add-back MEFs (Figure 6C). Activation of Akt was observed to a much lesser extent in cells expressing the nonraft membrane targeting p110α-Ras allele. Similarly, an increase in membrane-associated p-Akt was observed, when p110α-Lyn but not p110α-Ras DKO add-back MEFs were stimulated with LPA (Figure 6—figure supplement 1B).
We also examined p110 isoform localization in the PTEN null PC3 prostate cancer cell line that is dependent on p110β function for Akt activation and growth (Ni et al., 2012; Hill et al., 2010). In these cells, we knocked down p110β expression using p110β-specific shRNAs. Downregulation of p110β reduced p-Akt and p-S6 levels (Figure 6—figure supplement 2A). We then tested if expression of membrane targeting p110α alleles could restore Akt activation and downstream S6 phosphorylation. p110α-Lyn but not p110α-Ras expression compensated for p110β loss and restored p-Akt and p-S6 in p110β knockdown cells (Figure 6—figure supplement 2B, compare lanes 7 with 8), suggesting that raft-targeted p110α might have redundant functions with wt p110β.
EGFR activity is critical for signaling mediated by raft targeted-p110α
Because raft localization of p110α is sufficient for activation of Akt in LPA signaling, we next focused on the mechanism by which p110α is activated downstream of GPCRs. Previous reports suggest that GPCR activation can also trigger widespread downstream signaling via modulation of various growth factor receptor pathways (Hsieh and Conti, 2005). For instance, LPA stimulation promotes metalloprotease-mediated transactivation of EGFR, and EGFR activity itself is responsible for major growth-promoting effects of GPCR signaling (Prenzel et al., 1999). Since p110α has no Gβγ or Rac1 binding site, we hypothesized that GPCR mediated EGFR activity might be critical for raft-localized p110α to modulate Akt function in LPA signaling. To test this idea, we selectively inhibited Gβγ, EGFR or PDGFR with gallein, lapatinib and crenolanib respectively, in serum starved and LPA stimulated p110α-Lyn or p110β-wt DKO add-back MEFs. EGFR activation as well as Akt and the downstream S6 phosphorylations were observed for both MEF lines upon stimulation with LPA (Figure 6D). In support of our hypothesis, EGFR but not Gβγ or PDGFR inhibition reduced p-Akt/S6 in p110α-Lyn DKO add-back MEFs (Figure 6D, compare lanes 3, 4 and 5). In contrast, Gβγ activity was critical for p110β-wt mediated GPCR signaling while EGFR inhibition had only a modest effect (Figure 6D, compare lanes 8 and 9). Consistent with these results, p110α-Lyn DKO add-back MEFs were more resistant to gallein in cellular proliferation assays (Figure 6E, left), whereas p110β-wt DKO add-backs were refractory to EGFR inhibition by lapatinib (Figure 6E, right). These results suggest that the EGFR activation by LPA stimulated GPCRs is required for raft-localized p110α to promote Akt phosphorylation and activation; whereas the canonical Gβγ pathway was essential for p110β-mediated GPCR signaling.
Raft dependent PI3K function is essential for PTEN null cancer cells
Our results thus far suggest that raft integrity might play a more significant role for p110β dependent signaling. We and others have demonstrated that tumors deficient of phosphatase and tensin homolog (PTEN) are often dependent on the p110β isoform. Therefore we employed PTEN intact and PTEN null cancer cell line pairs and attempted to disrupt rafts by inhibiting cholesterol function. First, we treated PTEN positive DU145 and PTEN null PC3 prostate cancer cells with increasing doses of the cholesterol-depleting agent MβCD. PC3 cells were significantly more sensitive to MβCD treatment than DU145 cells (Figure 7—figure supplement 1A,B). Next we used PTEN wt MCF7 and PTEN null BT549 breast cancer cells to further test our hypothesis. Along the same line, BT549 breast cancer cells were more susceptible to MβCD than MCF7 cells (Figure 7—figure supplement 1C,D). As an independent approach, we utilized simvastatin, an FDA-approved inhibitor of the enzyme HMG-CoA reductase, which catalyzes the rate-limiting step (HMG-CoA to mavelonic acid) in cholesterol biogenesis in order to lower cholesterol levels. Once again, PC3 cells were considerably more sensitive to inhibition of cholesterol synthesis than DU145 cells. A simvastatin concentration of 1 μM was sufficient to potently inhibit proliferation of PC3 cells, which did not impede growth of DU145 cells (Figure 7A,B). Next we treated PTEN wt MCF7 and PTEN null BT549 cells with simvastatin. Treatment of BT549 cells with simvastatin led to a significant impediment of proliferation at 1–2 μM concentrations, which was not growth inhibitory for MCF7 cells (Figure 7C,D). Another PTEN wt breast cancer cell line, T47D was similarly refractory to growth inhibitory effects of simvastatin (Figure 7—figure supplement 1E,F). These results suggest that PTEN null
cancers might be more sensitive to inhibition of raft function which in turn negatively influences p110β dependent PI3K signaling. In an attempt to restore PI3K pathway activation, we expressed constitutively active, oncogenic mutants of p110α, H1047R and E545K in PC3 cells. Notably this lead to an activation of Akt as anticipated (Figure 7E). Next, we treated these cells with the cholesterol-lowering agent, simvastatin. Although control PC3 cells were quite susceptible to simvastatin treatment, PC3-p110α H1047R or E545K expressing cells proliferated more robustly in the presence of previously deleterious doses of the inhibitor (Figure 7E and F). A biochemical analysis of these cells showed that Akt activation was repressed in PC3 cells treated with simvastatin and this repression was relieved by p110α-H1047R or E545K expression (Figure 7G). These results suggest that raft specific inhibition of PI3K function in PC3 cells can be partially alleviated by expression of constitutively active oncogenic alleles of p110α, which presumably utilize raft independent modes of activation. As a complementary approach, we re-expressed either a wt or a catalytically dead version of human Pten (Pten C124S) in PC3 and BT549 cells. Restoration of wt Pten lead to a reduction in Akt activation as judged by levels of pAkt (Figure 7—figure supplement 2A and D). Interestingly, Pten expressing PC3 and BT549 cells became more resistant to increasing concentrations of simvastatin, while cells expressing Pten C124S were almost as sensitive to simvastatin as the parental cell lines (Figure 7—figure supplement 2B,C,E and F, E and F). These results are in support of the notion that loss of Pten in certain cancers might lead to a dependence on raft mediated PI3K signaling.
Signal transduction via the PI3K pathway requires a sequence of highly orchestrated molecular events transpiring at the plasma membrane. Nevertheless, it is still unclear how local signaling is modulated and how membrane microdomain architecture might influence the signal output. We have used a molecular genetic system that allowed us to express specific membrane-targeted isoforms of either p110α or p110β in an isogenic p110α/p110β DKO background. This genetic system enabled functional comparisons between wild type and membrane-targeted alleles of class IA PI3Ks. Interestingly for p110β, raft targeting alleviates the requirement for Rac1 binding. However, Gβγ association is unable to be rescued by raft targeting. These results indicate that Rac1-p110β interaction may be necessary for localization of p110β to GPCR signaling permissive membrane microdomains, whereas its association with Gβγ remains crucial for the activation of its lipid kinase activity upon ligand binding to GPCRs. Interestingly, even p110α, which is usually unresponsive to GPCR activation, becomes responsive if properly localized. Finally, our data has implications concerning the optimal methods for inhibiting Akt activation in PTEN null cancers (Figure 7H).
Our DKO add-back MEF lines were extremely well behaved when tested by the guidance of existing literature: e.g. they were responsive to the relevant PI3K class IA isoform-specific small molecule inhibitors for both signaling and growth. Our efforts in enriching p110α and p110β in membrane rafts and nonraft regions using targeting sequences derived from the Lyn kinase and Kras were also clearly successful as judged by several means of membrane fractionation. Minimal off-target localization with these p110 constructs might be attributed to the indirect interactions of p85 mediated via its SH2 and/or SH3 domains.
The resulting targeting constructs were utilized to study the role(s) of microdomain compartmentalization in growth factor signaling. Serum starvation and stimulation experiments revealed that GPCR activation could be communicated to Akt when either p110α or p110β was adequately localized in membrane rafts. One fundamental difference between these PI3Ks however, is in their intrinsic membrane binding properties. p110β, by virtue of its association with Rac1, is naturally localized to membrane rafts. On the other hand, p110α does not reside in rafts under physiological conditions. However, when it is artificially targeted to the rafts, p110α is capable of functioning downstream of GPCRs. Moreover, raft-targeted p110α acts in a redundant manner with p110β in PTEN null PC3 cells.
While membrane raft localization is essential for p110 activation by GPCRs, whether it occurs via Rac binding for p110β in the physiological case or via a Lyn-tag for p110α, the second step in p110 activation by GPCRs can be more varied. In the physiological case p110β is activated via its interaction with Gβγ. However, we found that EGFR activity is essential for raft-localized p110α mediated Akt phosphorylation. EGFR is enriched in raft fractions, and its activity has been intimately linked to GPCR signaling. LPA stimulation, for instance, was reported to induce EGFR phosphorylation and transactivation (Daub et al., 1997). Membrane bound metalloproteases and non-receptor tyrosine kinases like Src and Pyk2 are implicated in this regulation (Prenzel et al., 1999; Andreev et al., 2001). Our results suggest that LPA stimulated-EGFR could contribute to activation of raft-localized p110α, whereas the canonical Gβγ association is the prominent form of activation for p110β. However, we cannot rule out a modest contribution of EGFR in raft-localized p110β activation, as we detected residual activation of Akt in cells expressing raft-targeted, Gβγ binding deficient p110β upon LPA treatment. Notably p110β may also be dependent on Rac mediated raft localization for its activation in the non-physiological conditions of PTEN loss. However, in the case of p110β activation in response to PTEN loss yet another mechanism may be involved as the second step in p110β activation- further experimentation will be required to test this point.
Our findings confirm the existing literature that both Rac1 and Gβγ binding are instrumental for p110 function downstream of GPCRs (Fritsch et al., 2013; Dbouk et al., 2012) and allow us to place them in a two-step model for p110 activation. Raft targeting of mutant versions of p110 enabled us to infer that Rac1 binding, at least in a PTEN wild-type setting, is crucial for localization of p110β into signaling permissive membrane microdomains. Gβγ interaction on the other hand, provides the basis for p110β activation upon ligand-induced stimulation of GPCRs. We further verified the involvement of raft microdomains in LPA mediated GPCR signaling via cholesterol deprivation experiments. Cholesterol depletion specifically inhibited activating Akt phosphorylations whereas Erk phosphorylation was largely unaffected. These results indicate that activation of the MAPK signaling cascade by GPCRs is not crucially dependent on raft integrity, reinforcing the notion that critical role of membrane compartmentalization may be unique to PI3K/Akt signaling.
The higher abundance of PTEN in nonraft membrane regions seen here and in earlier work (Gao et al., 2011) might be an important determinant in localizing signaling as it may prevent unnecessary propagation of signaling cascades initiated in rafts to nonraft regions. Thus the reported selective localization of PTEN in nonraft domains suggests that it might exert relatively little control on raft recruitment of p110β and activation of Akt, while preventing the resulting signals from propagating outside raft domains. The elevated signaling potency observed when either p110α or β is targeted to rafts could be explained by a regional lack of PTEN dependent negative regulation. In tumor cells lacking PTEN, such p110β dependent signaling could spread to larger portions of the cells greatly strengthening PI3K signaling. However the fact that p110β signaling in PTEN null cells is still dependent on raft integrity suggests that the second, Rac-independent 'Gβγ like' signal activating p110β is likely to initiate in rafts in the absence of PTEN. We do not feel that this second signal is Gβγ itself in most PTEN null cells as Akt activation is often not blocked by GPCR antagonists.
Notably we have found that knock down of PTEN in our MEFs does not render their growth dependent on p110β or sensitive to raft disruption suggesting that while PTEN loss may be necessary for p110β activation (in rafts as demonstrated by the data in Figure 7) it is not sufficient. Similarly the Vanhaesebroeck group has found that not all tumors arising in mice with global knockout of a single PTEN allele are dependent on p110β (Berenjeno et al., 2012). Thus additional parameters seem to be involved in optimal activation of p110β in a PTEN null setting.
Use of simvastatin, an FDA approved cholesterol lowering drug, hampered proliferation of PTEN null cancer cells which might pinpoint a plausible 'addiction' of these cancer cells to raft mediated PI3K function. Based on these findings, a combinatorial treatment regimen employing PI3K inhibitors and cholesterol lowering statins might have the potential to treat a broad range of cancers with minimal undesirable side effects.
In summary, our findings establish a novel distinction between p110α and p110β, and offer new insight into the mechanistic basis of p110 activation by GPCR signaling. Here, we propose that recruitment and activation constitute two distinct steps in PI3K/Akt signaling downstream of GPCRs (Figure 7H). Once properly placed, either p110α or β molecules can be activated via alternative GPCR signaling cascades owing to the versatility of signaling components activated by a liganded GPCR. Our data warrant further work on the role of membrane partitioning in regulation of the PI3K/Akt pathway and offer novel therapeutic aspects concerning treatment of PTEN null cancers.
Materials and methods
Vector construction and shRNA sequencesRequest a detailed protocol
Lyn and Kras tagged constructs (Gao and Zhang, 2008) were generated by in frame fusion of the N-terminal part of the Lyn kinase (GCIKKSKRKDKD, for myristoylation and palmitoylation) at the 5’ end or the C-terminal part of Kras (KKKKKSKTKCVIM, CAAX motif for prenylation) at the 3’end of a pBABE p110α or p110β vector with an N-terminal HA-tag respectively. p110β Gβγ (K532D/K533D) and Rac1 (S211D/K230A) binding mutants were generated using QuikChange site-directed mutagenesis kit (Agilent, Santa Clara, California) and sequence verified. The sequence of the shRNA targeting human p110β transcripts is as follows; CATTCAGCTGAACAGTAGCAA. shGFP sequence is GCAAGCTGACCCTGAAGTTCAT. pBabeL-Pten wt and pBabeL-Pten C124S plasmids were kind gifts from William Sellers (Addgene plasmids #10785 and #10931).
Generation of immortalized mouse embryonic fibroblasts (MEFs), cell culture and generation of stable cell linesRequest a detailed protocol
HEK293, p110αflox/flox; p110βflox/flox MEFs and their derivatives were grown at 37°C in 5% CO2 in Dulbecco’s modified Eagle’s medium (DMEM, including 4.5 g/L D-glucose, L-glutamine and 110 mg/L sodium pyruvate) supplemented with 8% fetal bovine serum (FBS, Gemini-Bio, West Sacramento, California) and penicillin, streptomycin (100 IU/ml and 100 µg/ml respectively, Gibco). Human mammary epithelial cells (HMECs) were generated and cultured as described (Zhao et al., 2005). DU145, PC3, MCF7 and BT549 cells were acquired from ATCC and were not further authenticated. They were cultured in RPMI medium (Gibco, Waltham, Massachusetts) supplemented with 8% FBS under standard conditions. All cell lines used were negative for mycoplasma contamination.
p110αflox/flox; p110βflox/floxMEFs were prepared from embryos at embryonic day 13.5 post-fertilization. These primary MEFs were immortalized using the standard 3T3 protocol (Meek et al., 1977). Floxed MEFs were treated with AdCre (Iowa Viral Vector Core, Iowa City, Iowa) to generate knock-out cells or with AdLacZ for control. Add-back MEF lines were generated by introducing the construct of interest into the cells first and then treating them with two rounds AdCre infection. For RNAi experiments, PC3 cells were treated with 1 µg/ml of doxycycline for 48 hr.
Control siRNAs (AM4629) and siRNAs targeting human Rac1 (ID 164723) were ordered from Thermo Fisher (Waltham, Massachusetts, transfected into HMECs at 50 nM concentration using lipofectamine 2000 (Invitrogen, Waltham, Massachusetts) according to the manufacturer’s instructions. 48–72 hr post transfection; cells were harvested and subjected to immunoblot or fractionation assays.
Amphotropic retroviruses were produced by transfection of HEK293 cells with packaging plasmids encoding Vsv-g, gag-pol and a retroviral vector encoding the gene of interest using lipofectamine 2000 (Invitrogen) according to the manufacturer’s instructions. For production of amphotropic lentiviruses, same cells were transfected with Vsv-g, Delta 8.9 and pLKO.tet on vector encoding an shRNA sequence of interest under standard instructions. Culture supernatants containing retro or lentivirus were collected twice at 48 hr and 72 hr posttransfection, pooled together, sterile filtered (Thermo Scientific) and then used immediately.
Growth factor stimulation and drug treatmentRequest a detailed protocol
Cells were starved for 4 hr and stimulated with insulin (2 µg/ml), EGF (20 ng/ml), LPA (20 µM, all Sigma), PDGF (20 ng/ml), FGF (10 ng/ml, both Miltenyi Biotech, Germany), or 5% FBS for 5 min. Cells were washed with and scraped into ice-cold PBS and then lysed in RIPA buffer (Westnet, Canton, Massachusetts) containing protease and phosphatase inhibitor cocktails (both Roche Diagnostics, Indianapolis, Indiana), 1 mM sodium orthovanadate (Cell Signaling, Danvers, Massachusetts) and 1 mM dithiothrietol (DTT, Bio-Rad, Hercules, California). p110α specific-inhibitor BYL-719 (1 µM), p110β specific-inhibitor TGX-221 (2 µM), pan-PI3K inhibitor GDC-0941 (1 µM), pan-Akt inhibitor MK-2206 (1 µM, all Selleck Chem, Houston, Texas) were used in proliferation experiments. In growth factor stimulation experiments BYL-719, TGX-221, Gβγ inhibitor Gallein (5 µM, Santa Cruz, Dallas, Texas), EGFR inhibitor Lapatinib (5 µM), PDGFR inhibitor Crenolanib (0.2 µM, both Selleck), MβCD (5 mM) and water soluble cholesterol (2 mM, both Sigma, St. Louis, Missouri) were added to serum starved cells 30 min before stimulation.
Antibodies and western blottingRequest a detailed protocol
10–20 µg of total protein was separated with 7.5–12% SDS-PAGE and transferred to nitrocellulose membranes (Bio-Rad). The membranes were blocked with 5% non-fat milk (LabScientific, Highlands, New Jersey) in TBS for 30 min and then incubated with primary antibodies in TBS containing 5% milk and 0.1% Tween-20 overnight at 4°C. After rinsing with TBS-0.1% Tween, membranes were incubated with the secondary antibodies diluted in 5% milk-TBS for 1.5 hr at RT. Bands were detected by using Odyssey CLx Scanner. Band intensity quantifications were performed using Image Studio 3.1. Normalizations were performed either relative to the actin or Akt band intensities. Western blots were performed with antibodies against p110α (#4249), p110β (#3011), Akt (#4691), phospho-Akt T308 (#13038), phospho-Akt S473 (#9271), phospho-Erk 1/2 T202/Y204 (#9101), phospho-S6 S235/236 (#2211), phospho-S6 S240/244 (#5364), phospho-EGFR Y1068 (#2236), S6 (#2217), HA (#2367), TfnR (transferrin receptor) (#13208), Nup (nucleoporin) (#2598), EGFR (#4267), PTEN (#9188) (all Cell Signaling), Rac1 (05–389) (Millipore, Billerica, Massachusetts), β-actin (A5441), α-tubulin (T5168) (Sigma), Caveolin1 (MA3-600) (Thermo Fisher) and Gq (sc-392) (Santa Cruz). IRDye 800CW Goat anti-mouse or anti-rabbit secondary antibodies (Li-Cor, Lincoln, Nebraska) were used.
Membrane preparation and Opti-Prep density gradient fractionationRequest a detailed protocol
2–4 confluent 150 mm dishes (BD, San Jose, California) of cells grown in 2% FBS to limit excessive amounts of growth factors, were rinsed with ice-cold PBS and scraped into homogenization buffer (20 mM Tris-HCl, pH 7.8, 250 mM sucrose (Sigma), 1 mM CaCl2 and 1 mM MgCl2). Cells were pelleted by centrifugation at 4°C for 5 min at 250 g and resuspended in 1 ml of cold homogenization buffer supplemented with protease and phosphatase inhibitors. Cells were then subjected to mechanical disruption with 15 strokes of a tight pestle in a tissue grinder (Kimble/Kontes, Rockwood, Tennessee). Homogenates were then passed through a 22 g needle 20 times. Lysates were then centrifuged at 500 g for 10 min at 4°C to get rid of intact cells, nuclei and cell debris. The resulting supernatant was centrifuged at 52,000 g for 1 hr at 4°C. Supernatants were stored (soluble fraction) and the pellets were subjected to 1% Triton X100, 50 mM Tris-HCl pH 7.4, 150 mM NaCl, 5 mM EDTA supplemented with protease and phosphatase inhibitors. After 20 min incubation on ice, lysates were centrifuged at 52,000 g for 1 hr at 4°C. Supernatants were collected as the triton-sensitive fraction. Remaining membrane pellets were further extracted with 100 mM N-octyl glucoside (Santa Cruz), 50 mM Tris HCl pH 7.4, 150 mM NaCl supplemented with protease and phosphatase inhibitors. Following 20 min of incubation on ice, samples were centrifuged at 16,000 g for 30 min at 4°C. Supernatants were collected as the detergent resistant membrane (DRM) fraction.
For the detergent-free Opti-Prep (Sigma) density gradient fractionation, 4 confluent 150 mm dishes of cells grown in 2% FBS were rinsed and scraped into the homogenization buffer. Cells were pelleted by centrifugation at 4°C for 5 min at 250 g and resuspended in 600 μl of cold homogenization buffer supplemented with protease and phosphatase inhibitors. Cells were then subjected to mechanical disruption with 15 strokes of a tight pestle in a tissue grinder (Kimble/Kontes). Homogenates were then passed through a 22 g needle 20 times. At 4°C, lysates were centrifuged at 1000 g for 10 min. The resulting postnuclear supernatant was transferred into a separate tube. The pellet was again lysed by the addition of 400 µl cold homogenization buffer supplemented with protease and phosphatase inhibitors followed by sheering 20 times through a 22 g needle. After centrifugation at 1000 g for 10 min at 4°C, the second postnuclear supernatant was combined with the first. 1 ml of 50% Opti-Prep (Sigma) was added to the combined postnuclear supernatants and placed in the bottom of a 5 ml centrifuge tube (Beckman Coulter, Brea, California). 400 μl each of 20%, 17.5%, 15%, 12.5%, 10%, 7.5% and 5% Opti-Prep solutions were poured on top of the lysates. Discontinuous gradients were then centrifuged for 90 min at 100000 g at 4°C using an SW-55 rotor in a Beckman ultracentrifuge. After centrifugation, 400 μl fractions were collected and distribution of proteins was analyzed by immunobloting.
ImmunofluorescenceRequest a detailed protocol
Cells grown on coverslips were fixed with 4% formaldehyde (Polysciences, Inc, Warrington, Pennsylvania) in PBS at RT for 15 min. Coverslips were washed in PBS 3 times for 5 min and blocked in blocking buffer (1xPBS, 5% normal goat serum (Cell Signaling), 0.3% Triton X-100) for 1 hr. Cells were incubated with primary antibodies diluted in 1xPBS, 1% BSA, 0.3% Triton X-100 overnight at 4°C. Followed by rinsing 3x 5 min with PBS, coverslips were incubated with secondary antibodies diluted in 1xPBS, 1% BSA, 0.3% Triton X-100 for 1–2 hr at RT in dark. Cells were then mounted on glass slides using ProLong Gold with DAPI (Cell Signaling) mounting medium and imaged with a Nikon TE2000-U epifluorescence microscope with appropriate filters. Corrected total cell membrane fluorescence was determined for p-Akt T308 and p-Akt S473 signals using Image J (National Institute of Health, Bethesda, Maryland. Rabbit anti–p-Akt T308, anti-p-Akt S473 and mouse anti-HA were all from Cell Signaling. Alexa 488 anti–rabbit, Alexa 594 anti–mouse from Invitrogen. Quantifications are average of 3 independent experiments of at least 10 cells per experimental condition.
Cellular proliferation assaysRequest a detailed protocol
MEFs were seeded in 6-well tissue culture plates (Corning, Corning, New York) at a density of 1 × 104 cells per well. Cancer cells were seeded in 12-well plates (Corning) at a density of 1 × 104 cells per well. At the indicated time points, cells were washed with PBS, fixed in 10% acetic acid (Fisher), 10% ethanol for 15 min and then stained with 0.2% crystal violet (Sigma), 10% ethanol for 20 min, washed with distilled water and air-dried. Cell-associated dye was extracted with 2 ml of 10% acetic acid for 15 min and the optical density (OD) was measured at 595 nm. Values were normalized to the OD at day 0 or to vehicle treatment at day seven for each sample. Each data point was determined in either duplicates or triplicates. For experiments performed with limited supplementation, DMEM with 2% FBS was used. For experiments with cancer cells, RPMI with 4% FBS was used. Data shown are the mean of at least three independent experiments with standard deviation.
Activation of simvastatinRequest a detailed protocol
Simvastatin (Sigma) needs to be activated by opening of the lactone ring before use in cell culture. Briefly, four milligrams of simvastatin were dissolved in 0.1 ml of 100% ethanol, with subsequent addition of 0.15 ml of 0.1 N NaOH. The solution was heated at 50°C for 2 hr in a water bath and then neutralized with HCl to pH 7.2. The resulting solution was brought to a final volume with sterile PBS (10 mM), and aliquots were stored at −80°C until use.
Wound healing assaysRequest a detailed protocol
Cells were seeded at 5 × 105 cells/well in 6-well dishes in DMEM, 8% FBS, incubated for 24 hr then serum-starved for 10 hr. A wound was induced into the cell monolayer using a 200 μl pipette tip. Cells were incubated in DMEM, 2% FBS, at 37°C. In triplicate experiments, wounds were imaged at the indicated time points for 12 hr at 3 positions along each wound. Recovery percentages indicate the percentage ratio of distance traversed by the leading edge of migrating cells to the initial wound, determined with ImageJ.
Transwell migration assaysRequest a detailed protocol
Cells were seeded in duplicates onto 8-µM PET Transwell migration chambers (Corning) at a density of 1 × 105 cells/well in media containing 1% FBS and media containing 4% FBS + 20 μM LPA + 5 µM S1P were placed in the lower chamber as chemoattractant. Cells were allowed to migrate to the lower side of the inserts for 18 hr. Cells which did not migrate, were removed with a cotton swab on the upper side and then the inserts were fixed in 10% acetic acid, 10% ethanol for 15 min and then stained with 0.2% crystal violet, 10% ethanol for 20 min, washed with distilled water, air-dried and then scanned. Cell-associated dye was then extracted with 2 ml of 10% acetic acid for 15 min and the OD was measured at 595 nm. Data shown are the mean of three independent experiments with standard deviation.
Statistical analysisRequest a detailed protocol
Two-tailed Student’s t-test was used for differential comparison between two groups. T-tests and IC50 calculations were performed in GraphPad Prism.
Src and Pyk2 mediate G-protein-coupled receptor activation of epidermal growth factor receptor (EGFR) but are not required for coupling to the mitogen-activated protein (MAP) kinase signaling cascadeJournal of Biological Chemistry 276:20130–20135.https://doi.org/10.1074/jbc.M102307200
RAC1: an emerging therapeutic option for targeting cancer angiogenesis and metastasisMolecular Cancer Therapeutics 12:1925–1934.https://doi.org/10.1158/1535-7163.MCT-13-0164
Phosphoinositide 3-kinases in cell migrationBiology of the Cell 101:13–29.https://doi.org/10.1042/BC20080079
Signal characteristics of G protein-transactivated EGF receptorThe EMBO Journal 16:7032–7044.https://doi.org/10.1093/emboj/16.23.7032
G-protein-coupled receptors and cancerNature Reviews Cancer 7:79–94.https://doi.org/10.1038/nrc2069
The evolution of phosphatidylinositol 3-kinases as regulators of growth and metabolismNature Reviews Genetics 7:606–619.https://doi.org/10.1038/nrg1879
Phosphoinositide kinasesAnnual Review of Biochemistry 67:481–507.https://doi.org/10.1146/annurev.biochem.67.1.481
Spatiotemporal analysis of differential Akt regulation in plasma membrane microdomainsMolecular Biology of the Cell 19:4366–4373.https://doi.org/10.1091/mbc.E08-05-0449
Condensation of the plasma membrane at the site of T lymphocyte activationThe Journal of Cell Biology 171:121–131.https://doi.org/10.1083/jcb.200505047
Integrin-mediated adhesion regulates membrane orderThe Journal of Cell Biology 174:725–734.https://doi.org/10.1083/jcb.200603034
Lipid rafts and B cell signalingSeminars in Cell & Developmental Biology 18:616–626.https://doi.org/10.1016/j.semcdb.2007.07.009
G-protein-coupled receptor signaling and the EGF network in endocrine systemsTrends in Endocrinology & Metabolism 16:320–326.https://doi.org/10.1016/j.tem.2005.07.005
PI 3-kinase p110beta: a new target for antithrombotic therapyNature Medicine 11:507–514.https://doi.org/10.1038/nm1232
Should individual PI3 kinase isoforms be targeted in cancer?Current Opinion in Cell Biology 21:199–208.https://doi.org/10.1016/j.ceb.2008.12.007
Dynamic organizing principles of the plasma membrane that regulate signal transduction: commemorating the fortieth anniversary of Singer and Nicolson's fluid-mosaic modelAnnual Review of Cell and Developmental Biology 28:215–250.https://doi.org/10.1146/annurev-cellbio-100809-151736
Raft nanodomains contribute to Akt/PKB plasma membrane recruitment and activationNature Chemical Biology 4:538–547.https://doi.org/10.1038/nchembio.103
Estrogen receptors in membrane lipid rafts and signal transduction in breast cancerMolecular and Cellular Endocrinology 246:91–100.https://doi.org/10.1016/j.mce.2005.11.020
The functions and regulation of the PTEN tumour suppressorNature Reviews Molecular Cell Biology 13:283–296.https://doi.org/10.1038/nrm3330
Sorting of sphingolipids in epithelial (Madin-Darby canine kidney) cellsThe Journal of Cell Biology 105:1623–1635.https://doi.org/10.1083/jcb.105.4.1623
The phosphatidylinositol 3-Kinase AKT pathway in human cancerNature Reviews Cancer 2:489–501.https://doi.org/10.1038/nrc839
PTEN-deficient cancers depend on PIK3CBPNAS 105:13057–13062.https://doi.org/10.1073/pnas.0802655105
Kari AlitaloReviewing Editor; University of Helsinki, Finland
In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.
Thank you for submitting your article "Rac1-mediated membrane raft localization of PI3K/p110β is required for its activation by GPCRs or PTEN loss" for consideration by eLife. Your article has been favorably evaluated by Charles Sawyers as the Senior Editor and two reviewers, one of whom is a member of our Board of Reviewing Editors. The reviewers have opted to remain anonymous.
The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.
The reviewers found the study interesting and the manuscript excellent. Its results illuminate a novel two-step mechanism that differentiates the recruitment and activation of the PI3K p110β isoform and relative roles of GPCR and RAC proteins.
1) The correlation of PTEN loss with raft-dependent PI3K function employs pairs of PTEN positive and negative tumor cells lines that differ in a number of other parameters. Conclusions regarding the negative regulation of PI3K signaling by PTEN in rafts vs. nonraft membrane regions should be strengthened by employing PTEN silencing in the p110 mutant and various add-back MEF cells.
2) The behavior of p110α when targeted to lipid rafts is interesting, with LPA induced activation being shown to operate via EGFR. It would be helpful to show that LPA is indeed inducing EGFR activation under these conditions (as in Figure 6D), by demonstrating increased tyrosine phosphorylation.https://doi.org/10.7554/eLife.17635.016
1) The correlation of PTEN loss with raft-dependent PI3K function employs pairs of PTEN positive and negative tumor cells lines that differ in a number of other parameters. Conclusions regarding the negative regulation of PI3K signaling by PTEN in rafts vs. nonraft membrane regions should be strengthened by employing PTEN silencing in the p110 mutant and various add-back MEF cells.
As an alternative approach, more in conjunction with the experiments using PC3 and BT549 cells in the original draft; we have generated Pten re-expression lines using the parental Pten-null PC3 and BT549 cells. Re-expression of Pten lead to a reduction in p-Akt in both PC3 and BT549 cells, whereas expression of a phosphatase-dead version of Pten did not alter levels of p-Akt. Interestingly, restoration of Pten expression rendered PC3 and BT549 cells more resistant to simvastatin treatment. These results complemented the data obtained upon use of oncogenic p110α mutants in simvastatin treated PC3 cells. Both PC3 and BT549 cells could be partially saved from the adverse effects of simvastatin via expression of Pten, which attenuates the activity of Akt that functions downstream of PI3K signaling. These results further corroborated the notion that Pten loss might be critical in lipid raft mediated p110beta activation. You can now find these results in Figure 7—figure supplement 2.
We also have followed the reviewers’ suggestions and have employed PTEN silencing in MEFs. These results were surprising to us. According to our western blot analysis, we have obtained 2 MEF lines with near complete knock-out of PTEN. As expected, a significant increase in p-Akt and p-S6 levels were detected accompanying the loss of PTEN expression (Figure A). We used these PTEN knock-down MEF lines alongside with control cells to study phenotypes associated with PI3K inhibition and simvastatin treatment. Initially we treated these cells with isoform specific inhibitors of class IA PI3Ks; BYL719 (a p110α specific inhibitor), KIN193 (a p110β specific inhibitor), CAL101 (a p110δ specific inhibitor) or GDC001 (a pan-PI3K inhibitor). wt MEF cells are inherently more dependent on p110α mediated signaling. A modest concentration of p110α inhibitor (1μM) significantly inhibited cellular growth of exponentially growing cells. Comparable amounts of p110β or p110δ specific inhibitors do not significantly affect cellular growth (Figure B). Of note, knock-down of PTEN with 2 shRNAs did not alter this prevalent isoform dependence. PTEN silenced MEFs are still p110α dependent for growth and do not display a significant growth reduction upon inhibition of p110β or p110δ isoforms. In addition, all these MEF lines were exquisitely sensitive to pan-PI3K inhibition by GDC. We have also verified this biochemically in a short-term inhibitor treatment assay analyzing p-Akt and p-S6 in the same MEF lines. Treatment with BYL719 for 2 hours almost completely wiped out p-Akt or p-S6 in either control or PTEN silenced MEF lines whereas KIN193 displayed a much reduced potency in downregulating Akt and S6 phosphorylations in control or PTEN knock-down MEFs (Figure D). In line with the assays using isoform-specific inhibitors, treatment of either control or PTEN silenced MEFs with simvastatin lead to a similar level of growth inhibition. In brief, PTEN loss in untransformed MEFs rendered them neither p110β dependent nor more sensitive to interference with raft function. These results were indeed intriguing given the strong correlation between PTEN loss and the dependence on the p110β isoform.
Our data suggest that although PTEN loss may be essential for progression into a more p110β dependent state, it does not seem to be sufficient for this transition. Indeed, in an animal model of prostate tumor formation induced by Pten loss, in only one out of the three mouse prostate lobes ablation of p110β, but not that of p110α, impeded tumorigenesis with a concomitant diminution of Akt phosphorylation (Jia et al. Nature, 2008). The Vanhaesebroeck lab also found tumor formation driven by PTEN loss was not always blocked by ablation of p110β activity in a different GEM model. Clearly, additional parameters seem to be involved in optimal activation of p110β in a PTEN null setting. At the present time we have not incorporated this data into the text – we feel that negative data is not sufficient to make strong inferences. However if the reviewers or editors prefer we are willing to incorporate this data (probably as a point in the Discussion and supplemental figure).
2) The behavior of p110α when targeted to lipid rafts is interesting, with LPA induced activation being shown to operate via EGFR. It would be helpful to show that LPA is indeed inducing EGFR activation under these conditions (as in Figure 6D), by demonstrating increased tyrosine phosphorylation.
We have performed the experiment suggested by the reviewers. Western blot analysis for Y1068 phosphorylation on EGFR yielded results which were in complete agreement with previous reports describing activatory EGFR phosphorylation in response to LPA. We have now included a p-EGFR panel in Figure 6D displaying corresponding activation of EGFR upon LPA stimulation.https://doi.org/10.7554/eLife.17635.017
Article and author information
National Institutes of Health (P50 CA168504-01A1)
- Jean J Zhao
- Thomas M Roberts
National Institutes of Health (P50 CA165962-01A1)
- Jean J Zhao
- Thomas M Roberts
National Institutes of Health (CA172461-01)
- Jean J Zhao
Stand Up To Cancer (Translational Research Grant SU2C-AACR-DT0209)
- Jean J Zhao
- Thomas M Roberts
National Institutes of Health (P01-CA50661)
- Thomas M Roberts
National Institutes of Health (CA30002)
- Thomas M Roberts
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
We thank Dr. Jennifer Spangle for critical reading of the manuscript. We thank the Nikon Imaging Center at Harvard Medical School for technical assistance and the use of instruments. This work was supported by National Institutes of Health Grants P01-CA50661 (to TMR), CA30002 (to TMR), P50 CA168504-01A1 (to JJZ and TMR), P50 CA165962-01A1 (to JJZ and TMR), CA172461-01 (to JJZ), and Stand Up to Cancer Dream Team Translational Research Grant SU2C-AACR-DT0209 (to JJZ and TMR).
- Kari Alitalo, University of Helsinki, Finland
- Received: May 8, 2016
- Accepted: September 22, 2016
- Version of Record published: October 4, 2016 (version 1)
© 2016, Cizmecioglu et al.
This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.
- Page views
Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
- Cancer Biology
CD73 is an ectonucleotidase overexpressed on tumor cells that suppresses anti-tumor immunity. Accordingly, several CD73 inhibitors are currently being evaluated in the clinic, including in large randomized clinical trials. Yet, the tumor cell-intrinsic impact of CD73 remain largely uncharacterized. Using metabolomics, we discovered that CD73 significantly enhances tumor cell mitochondrial respiration and aspartate biosynthesis. Importantly, rescuing aspartate biosynthesis was sufficient to restore proliferation of CD73-deficient tumors in immune deficient mice. Seahorse analysis of a large panel of mouse and human tumor cells demonstrated that CD73 enhanced oxidative phosphorylation (OXPHOS) and glycolytic reserve. Targeting CD73 decreased tumor cell metabolic fitness, increased genomic instability and suppressed poly ADP ribose polymerase (PARP) activity. Our study thus uncovered an important immune-independent function for CD73 in promoting tumor cell metabolism, and provides the rationale for previously unforeseen combination therapies incorporating CD73 inhibition.
- Biochemistry and Chemical Biology
- Cancer Biology
Nutrient stress in the tumor microenvironment requires cancer cells to adopt adaptive metabolic programs for survival and proliferation. Therefore, knowledge of microenvironmental nutrient levels and how cancer cells cope with such nutrition is critical to understand the metabolism underpinning cancer cell biology. Previously, we performed quantitative metabolomics of the interstitial fluid (the local perfusate) of murine pancreatic ductal adenocarcinoma (PDAC) tumors to comprehensively characterize nutrient availability in the microenvironment of these tumors (M. R. Sullivan, Danai, et al., 2019). Here, we develop Tumor Interstitial Fluid Medium (TIFM), a cell culture medium that contains nutrient levels representative of the PDAC microenvironment, enabling us to study PDAC metabolism ex vivo under physiological nutrient conditions. We show that PDAC cells cultured in TIFM adopt a cellular state closer to that of PDAC cells present in tumors compared to standard culture models. Further, using the TIFM model, we found arginine biosynthesis is active in PDAC and allows PDAC cells to maintain levels of this amino acid despite microenvironmental arginine depletion. We also show that myeloid derived arginase activity is largely responsible for the low levels of arginine in PDAC tumors. Altogether, these data indicate that nutrient availability in tumors is an important determinant of cancer cell metabolism and behavior, and cell culture models that incorporate physiological nutrient availability have improved fidelity to in vivo systems and enable the discovery of novel cancer metabolic phenotypes.