Fluctuations in AKT and PTEN Activity Are Linked by the E3 Ubiquitin Ligase cCBL

3-Poly-phosphoinositides (PIP3) regulate cell survival, division, and migration. Both PI3-kinase (phosphoinositide-3-kinase) and PTEN (phosphatase and tensin-homolog in chromosome 10) control PIP3 levels, but the mechanisms connecting PI3-kinase and PTEN are unknown. Using non-transformed cells, the activation kinetics of PTEN and of the PIP3-effector AKT were examined after the addition of growth factors. Both epidermal growth factor and serum induced the early activation of AKT and the simultaneous inactivation of PTEN (at ~5 min). This PIP3/AKT peak was followed by a general reduction in AKT activity coincident with the recovery of PTEN phosphatase activity (at ~10–15 min). Subsequent AKT peaks and troughs followed. The fluctuation in AKT activity was linked to that of PTEN; PTEN reconstitution in PTEN-null cells restored AKT fluctuations, while PTEN depletion in control cells abrogated them. The analysis of PTEN activity fluctuations after the addition of growth factors showed its inactivation at ~5 min to be simultaneous with its transient ubiquitination, which was regulated by the ubiquitin E3 ligase cCBL (casitas B-lineage lymphoma proto-oncogene). Protein-protein interaction analysis revealed cCBL to be brought into the proximity of PTEN in a PI3-kinase-dependent manner. These results reveal a mechanism for PI3-kinase/PTEN crosstalk and suggest that cCBL could be new target in strategies designed to modulate PTEN activity in cancer.

The importance of PTEN function in cancer has prompted the study of the mechanisms altering its levels and activity, with most work performed in cancer cells. Certainly, PTEN function can be compromised by heterozygous gene loss, gene mutation,
For IP, WCE were pre-cleared by incubation with protein A (Prot A) or Prot G-sepharose (4 • C, 1 h) (Thermo Fisher). Pre-cleared extracts were incubated with the appropriate antibody (see Ab list) at 4 • C (1-3 h) and incubated for 1 h more with Prot A or G (4 • C). For controls, Prot A/G beads were incubated with WCE without Ab, or with Ab but without WCE. Immunoprecipitates were washed three times with lysis buffer, three times with TBST buffer (TBS buffer [50 mM Tris-HCl, pH 7.5, 150 mM NaCl] containing 0.1% Tween 20), and three times with TBS. In the TBST washes, the IP were incubated for 10 min in an end-over-end rotor (4 • C). The IP were boiled in Laemli buffer, resolved by SDS-PAGE and examined by Western blotting (WB). For the latter, the resolved IP or extracts (50 µg) were transferred onto PVDF or nitrocellulose membranes and probed with the primary Ab and horseradish peroxidase-conjugated secondary Ab (Dako, Agilent). PVDF membranes were used for Ubiquitin and pAKT blots. Blots were developed with ECL reagent (GE Healthcare, Chicago, IL, USA). Immunoblotting band signals were quantitated using FIJI software (NIH, Bethesda, MD, USA).
For PTEN sumoylation and ubiquitination, RIPA cell extracts or dialyzed ubiquitinenriched extracts, respectively, were incubated (90 min, 4 • C) with PTEN XP D4.3-Ab-beads or DA1E Isotype control beads (CST). They were then washed as above (see IP). PTEN ubiquitination or sumoylation in vivo was determined from the signal in ubiquitin or sumoylation blots, as measured using Fiji software (NIH).

PTEN Phosphatase Assay
For the PTEN activity assays, cells were lysed in a detergent buffer (10 mM Tris-HCl [pH 7.4], 150 mM NaCl, 10 mM KCl, 0.5% Nonidet P-40) with protease inhibitors (Roche Applied Science, Penzberg, Upper Bavaria, Germany), okadaic acid, and orthovanadate, but without NaF phosphatase inhibitor. Cells were incubated with lysis buffer at 4 • C for 1 h, then centrifuged (9300 g, 10 min). Cell extracts were pre-cleared by 2 h of incubation with Prot A beads (4 • C) and the resulting supernatant subjected to IP using 138G6 anti-PTEN Ab (Cell Signaling, Danvers, MA, USA) (4 • C, 3 h) and then Prot A (4 • C, 1 h). Duplicated IP were extensively washed. One set was examined by WB using the PTEN Ab, and the other set used for the phosphatase assay performed with the K-4700 Kit (Echelon Biosciences, Salt Lake City, UT, USA) following the manufacturer's instructions. Briefly, lyophilized PI(3,4,5)P 3 (diC16) substrate stock was dissolved (vortexed and sonicated) at low concentration (100 µM) in water. Purified PTEN was washed and suspended in 75 µL reaction buffer (50 mM Tris [pH8], 50 mM NaCl, 10 mM DTT, 10 mM MgCl 2 , 3 µM PI(3,4,5)P 3 ). A PTEN-only control and a no-enzyme control were run in parallel. Reactions proceeded with shaking for 30 min at 37 • C. Since the PTEN was bound to beads, a short 100 g spin allowed for the separation of the enzyme and the product (the latter in the supernatant). The amount of PI(4,5)P 2 produced in the reaction was measured in triplicate (25 µL/each) by ELISA (Echelon Biosciences), using a stranded curve with different amounts of PI(4,5)P 2 . The prepared ELISA assays were incubated (1 h at 37 • C) and the resulting color measured at 450 nm in a plate reader (Thermo Fisher). The values of the known PI(4,5)P 2 standards were used to interpolate the values recorded in the PTEN test reactions. Phosphatase activity values were normalized to the amount of PTEN.

Immunofluorescence, Videomicroscopy
IF studies were performed as previously described [19]. Briefly, cells were seeded on slides pretreated with 50 µg/mL Type I-collagen (Sigma-Aldrich). After activation with PDGF or FCS, the cells were fixed with 4% paraformaldehyde (RT • , 10 min) and then permeabilized in PBS with 0.3% Triton X-100 (RT • , 10 min), blocked with PBS plus 10% FCS and 0.01% TX100 (RT • , 90 min), and incubated with PTEN Ab (Millipore, A2B1), (1 h, RT • ). Cy3 anti-mouse secondary Ab was used for detection (Jackson Laboratory, Bar Harbor, ME, USA). DNA was stained with Hoechst 33258 (Molecular Probes, Eugene, OR, USA). Images were acquired at 63× using an SP5 confocal microscope and employing Leica TCS SP5 software. PM-positive cells were selected by measuring the fluorescence signal (in arbitrary units) in a region of interest (ROI) at the PM. When this signal was at least 1.5 times that obtained in the same-size ROI for the adjacent cytosol, the cell was considered PM-positive (i.e., for PTEN). Cells with a similar signal in the PM and cytosol ROI were considered PM-negative. Cells were considered PM-intense when the integrated density of the PM ROI was ≥4 times that of the adjacent cytosol ROI.
For video microscopy, 15 × 10 3 NIH3T3 cells were seeded onto µ-Slide 8 Well dishes (IBIDI). After 24 h, these cells were transfected with pEGFP-N1 or pEGFP-Btk-PH vectors (0.2 µg plus 0.4 µg JetPei/dish). On the next day, the cells were incubated in serum-free RPMI (without phenol red) for 2 h and examined using a Leica DMi8 S epifluorescence microscope equipped with an sCMOS Orca-Flash 4.0 camera and a live cell chamber maintained at 37 • C (5% CO 2 atmosphere). An image was taken prior to activation (t = 0) and recording began after the addition of PDGF (50 ng/mL) or FCS (15%). The recording conditions were: Led 490 nm at 50%, exposure time 150 ms, objective 40×/0.8, magnificationchanger 1.6×. An image was taken every 2 min for 90 min. Signal quantization was performed using FIJI software (NIH).

RT-PCR Analysis, Cell Cycle Progression, and Lentiviral Infection
RNA was extracted using the RNeasy Kit (Qiage, Germantown, MD). cDNA was synthesized using the high capacity cDNA reverse transcription kit (Applied Biosytems, Thermo Fisher). Quantitative real-time PCR was performed using Evagreen qPCR mix (Solis Biodyne, Tartu, Estonia). RQ values were normalized with GAPDH. Cell cycle progression was examined as described [20]. Briefly, the cells were incubated with 20 µM bromodeoxyuridine (BrdU, 1 h), washed and incubated in medium without BrdU, and at different times were stained with anti-BrbU-FITC Ab (BD Biosciences, Franklin Lakes, NJ, USA) and propidium iodide and examined by flow cytometry. For lentiviral production, HEK-293T cells were transfected using JetPei reagent with the pLKO-puromycin lentiviral plasmid in the presence of pMD2.G and psPax2 (Addgene) (48 h). Supernatants containing the viruses were filtered (0.45 µm pore size), supplemented with polybrene (8 µg/mL, Sigma-Aldrich), and added to the cells. Infection was repeated at 24 h. At 72 h, puromycinresistant cells were selected with puromycin (2 µg/mL, 96 h) (Sigma-Aldrich).

Statistical Analysis
Statistical analyses were performed using GraphPad Prism (San Diego, CA, USA). Fluorescence and bioluminescence resonance energy transfer, and the primer and antibody lists are included in Supplementary Information.

EGF and Serum Trigger Reverse the Fluctuations in AKT and PTEN Activity
To study the change in endo-PTEN activity after the addition of GF, immortalized human HEK-293T cells, which express the WT forms of PTEN and PI3-kinase [21]), were examined. The cells were first preincubated without serum (2 h) to reduce the basal activa-Cells 2021, 10, 2803 6 of 21 tion of intracellular signaling pathways, and then activated for short periods (0-120 min) using FCS. Endo-PTEN was immunoprecipitated from the activated cell extracts and its phosphatase activity determined by measuring the conversion of PIP 3 into PIP 2 . PIP 3 levels were monitored in parallel with Thr308pAKT and Ser473pAKT levels [1,22].
FCS triggered an early increase in Thr308pAKT (at~5 min) followed by a reduction (at 10-15 min), then a second peak (at~30-60 min) and subsequent reduction (Figure 1a). In well-resolved gels, the Thr308pAKT signal was detected as two bands, with one band perhaps resulting from an additional Akt modification. Endo-PTEN activity was tested in parallel and a similar fluctuation seen, with a reduction at~5 min and a second at~30-60 min following the addition of FCS (i.e., coinciding with pAKT peaks) ( Figure 1b). Indeed, the simultaneous representation of Thr308pAKT and PTEN activity revealed maximum Thr308pAKT levels to coincide with reductions in PTEN activity (Figure 1c). Optimal AKT activation is known to require the phosphorylation of AKT at Ser473 [23]. Ser473pAKT variations were similar but were detected slightly later than those of Thr308pAKT (Figure 1a, c). Over the 2 h examination period, two (sometimes three) complete fluctuation cycles (with inverse AKT and PTEN peaks and troughs) were detected.
Cells activated with EGF were examined similarly. As with FCS, EGF induced fluctuating AKT and PTEN activities (Supplementary Figure S1a To determine whether these fluctuations also occurred in primary cells, a similar assay was performed using fresh MEF. As with the HEK-293T cells, FCS stimulation triggered Thr308pAKT signal fluctuations with two peaks at~5 min and~30-60 min (Figure 1d,e; Supplementary Figure S1e). Ser241PDK1 phosphorylation is known to be dependent on PIP 3 [24]; the cells showed high Ser241PDK1 levels after serum starvation, but they also fluctuated after the addition of GF (Figure 1d,e). The Ser241PDK1 and PDK1 blots showed two bands perhaps reflecting more than one modification of PDK1. Ser473pAKT levels also fluctuated in serum-activated MEF (Figure 1d,e), while PTEN, AKT, and PI3-kinase protein levels remained unchanged ( Figure 1d). These results show that Thr308pAKT levels and PTEN activity fluctuate in GF-activated primary cells, and that PIP 3 /pAKT peaks coincide with PTEN activity troughs.

PDGF and Serum Induce Transient PIP 3 Recruitment to the Plasma Membrane
In an attempt to confirm the PIP 3 fluctuations at the PM following growth factor receptor (GFR) activation, NIH3T3 cells expressing green fluorescent protein (GFP) fused to the Btk-pleckstrin homology (PH)-domain were subjected to live imaging. Since the Btk-PH domain binds selectively to PIP 3 [25], it is possible to examine PIP 3 levels at the cell membrane [26]. Nuclear PIP 3 levels were not assessable since GFP-Btk-PH (like GFP) constitutively localizes to the nucleus [26].
min following the addition of FCS (i.e., coinciding with pAKT peaks) ( Figure 1b). Indeed, the simultaneous representation of Thr308pAKT and PTEN activity revealed maximum Thr308pAKT levels to coincide with reductions in PTEN activity (Figure 1c). Optimal AKT activation is known to require the phosphorylation of AKT at Ser473 [23]. Ser473pAKT variations were similar but were detected slightly later than those of Thr308pAKT ( Figure  1a, c). Over the 2 h examination period, two (sometimes three) complete fluctuation cycles (with inverse AKT and PTEN peaks and troughs) were detected. To compare phosphatase activity at different times, the PIP 2 levels at each time were normalized to the mean PIP 2 level (equivalent to 1). For controls (Ctr IP), extracts were incubated with protein A. (c) The pAKT signal from assays as in (a) was measured, corrected for the AKT and β−actin content, and normalized to the mean pAKT signal for each assay (equivalent to 1). The graphs represent the relative PTEN activity and pAKT levels (Thr308 or Ser473) at each time point (mean ± SD, n = 7). (d) MEF were grown to confluence, maintained confluent for 24 h, and then incubated in FCS-deprived medium (2 h) before treatment with FCS (15%) for different times. Extracts were examined as in (a). (e) The graph shows the Ser473pAKT, Thr308pAKT and Ser241pPDK1 signals normalized to AKT and PDK1 and to β−actin (mean ± SD, n = 7). Dashed lines indicate time points left out from the original gels (originals blots are included as Supplementary Information). * p < 0.05, ** p < 0.01, *** p < 0.001, Student's paired t test.
NIH3T3 cells (immortal embryonic fibroblasts) are highly adherent and flat, and therefore optimal for video microscopy. They showed pAKT fluctuations upon addition of FCS or PDGF (see below). In control cells transfected with GFP, no apparent changes in GFP localization were detectable after adding FCS or PDGF (Supplementary Videos Cells 2021, 10, 2803 8 of 21 S1 and S2). In contrast, in cells expressing GFP-Btk-PH, both FCS and PDGF induced a rapid recruitment of the PH domain (that binds to PIP 3 ) at the PM (Supplementary Videos S3 and S4). This first Btk-PH burst (within minutes) at the membrane was sharper after the addition of PDGF than after FCS. PDGF binds to a single receptor type resulting in synchronous PI3K activation, while FCS activates different GF receptors (those of LPA, insulin, sphingosine 1-P, etc.) with different activation kinetics. In the videos, the second PIP 3 peak was subtler, possibly due to the basal PIP 3 levels detected on cell extensions throughout the study period. Nonetheless, a higher, sustained Btk-PH recruitment at the cell border was detectable in several sequential frames during the 30-60 min observation period, corresponding to the second pAKT peak (Figure 2a tion of PDGF than after FCS. PDGF binds to a single receptor type resulting in synchronous PI3K activation, while FCS activates different GF receptors (those of LPA, insulin, sphingosine 1-P, etc.) with different activation kinetics. In the videos, the second PIP3 peak was subtler, possibly due to the basal PIP3 levels detected on cell extensions throughout the study period. Nonetheless, a higher, sustained Btk-PH recruitment at the cell border was detectable in several sequential frames during the 30-60 min observation period, corresponding to the second pAKT peak (Figure 2a

pAKT Fluctuations Require PTEN Expression
Given the complementary patterns of AKT and PTEN activity, it was postulated that PTEN inactivation might help maximum PIP3 levels be reached (and maximal AKT activity) upon the addition of GF. This hypothesis was tested in normal MEF by the depletion of PTEN with siRNA, which abrogated the fluctuation in pAKT in response to FCS ( Figure  3a).

pAKT Fluctuations Require PTEN Expression
Given the complementary patterns of AKT and PTEN activity, it was postulated that PTEN inactivation might help maximum PIP 3 levels be reached (and maximal AKT activity) upon the addition of GF. This hypothesis was tested in normal MEF by the depletion of PTEN with siRNA, which abrogated the fluctuation in pAKT in response to FCS (Figure 3a). PTEN-Luc levels were found to be within the range of endo-PTEN (Supplementary Fig  S2). PTEN-Luc expression in PTEN −/− PC3 cells restored the pAKT fluctuation and creased the size of the pAKT peaks compared to those recorded for control PC3 cells (F ure 3b, top). The overexpression of PTEN to 10x normal using PRK5-WT-PTEN also stored pAKT fluctuations, and reduced the mean pAKT levels (Figure 3b, bottom). Th results show that PTEN activity induces fluctuations in pAKT. In a reverse approach, after confirming that cancer cells lacking PTEN expression (PC3 prostate cancer cells) showed a single AKT activation wave after the addition of serum (Figure 3b), the consequences of reconstituting PTEN expression were investigated. PTEN expression was examined using different recombinant (r)-PTEN cDNAs, and PTEN-Luc levels were found to be within the range of endo-PTEN (Supplementary Figure S2). PTEN-Luc expression in PTEN−/−PC3 cells restored the pAKT fluctuation and increased the size of the pAKT peaks compared to those recorded for control PC3 cells (Figure 3b, top). The overexpression of PTEN to 10× normal using PRK5-WT-PTEN also restored pAKT fluctuations, and reduced the mean pAKT levels (Figure 3b, bottom). These results show that PTEN activity induces fluctuations in pAKT.
The requirement of PTEN phosphatase activity for pAKT fluctuations was confirmed by comparing PC3 cell reconstitution with WT or inactive-C124S-PTEN. Reconstitution with C124S-PTEN yielded a single-wave pAKT pattern, as seen in PTEN null PC3 cells (Supplementary Figure S3a). Additionally, the treatment of HEK-293T with a PTEN inhibitor (BpV) prior to FCS stimulation flattened pAKT levels (Supplementary Figure S3b).  Figure S3d). PTEN might affect the progression of the cell cycle in a phosphatase-independent manner [27], but since constitutive PI3-kinase expression induces a similar phenotype as PTEN loss [20], the present results support the idea that PTENmediated PI3-kinase/AKT fluctuations are required for optimal cell cycle progression.

Maximum pAKT Levels Correlate with PTEN Ubiquitination
To study the mechanism of PTEN inactivation, a number of post-translational modifications (PTMs) were taken into account. PTEN SUMO-1 modification increases PTEN localization at the PM [11], whereas C-terminal PTEN phosphorylation (pPTEN) and PTEN ubiquitination can inactivate PTEN [10,28].
No substantial changes in endo-PTEN CT-phosphorylation were detected in HEK-293T cells stimulated with FCS for 0-90 min (Figure 4a). To examine ubiquitination, endo-PTEN was immunoprecipitated from FCS-activated HEK-293T extracts and examined in blots. This revealed a fluctuation in endo-PTEN ubiquitination after FCS stimulation although the ubiquitination signal was very weak (Supplementary Figure S4a). To improve the detection of ubiquitinated-PTEN, the ubiquitinated proteins in whole cell extracts (WCE) were concentrated in Tandem Ubiquitin-Binding-Entities (TUBES) peptide columns [17] (diagram in Supplementary Figure S4b). PTEN was immunoprecipitated from ubiquitinenriched extracts and examined in blots. The addition of FCS increased ubiquitinated-PTEN levels at~5 min, coinciding with the first pAKT peak. They then fell before increasing again at the time of the second pAKT peak and remained high at later time points (60-90 min) (Figure 4b). Subsequent PTEN blots yielded a nearly indistinguishable pattern (Figure 4b).
The ladder pattern obtained in these blots suggests that, in some of the PTEN Lys residues, multiple ubiquitins had been incorporated.
Sumoylation was also transient, and did not coincide with PTEN ubiquitination or the pAKT peaks, but co-occurred with a reduction in pAKT (Figure 4b). Subsequent PTEN blots identified sumoylated and non-modified PTEN (Figure 4b). PTEN secondary blots also identified a fraction of PTEN to be sumoylated prior to stimulation; this was better detected in SUMO2,3 blots (see below). Therefore, after the addition of FCS, PTEN underwent early ubiquitination coinciding with the maximum pAKT levels (and low PTEN activity, Figure 1). Posterior sumoylation coincided with the fall in pAKT levels. Left blots show Thr308pAKT levels in WCE. Extracts (1 mg) were also incubated with anti-PTEN Ab for IP and pPTEN levels determined in WB (right panels). For controls (Ctr), extracts were incubated with Prot A. The graph shows the normalized pPTEN signal (corrected for PTEN levels) with respect to the mean pPTEN signal (considered 1) (mean ± SD, n = 3). (b) HEK-293T cells were activated as in (a); Thr308pAKT levels were examined in WCE (top panels). Extracts were also enriched for ubiquitinated proteins in TUBES columns. The eluted extracts were incubated with PTEN Abbeads for IP and these examined by WB (top left). Secondary PTEN blots confirmed the retention of more ubiquitinated-PTEN at 5 min in the TUBES columns (right). PTEN was also immunoprecipitated from WCE and its sumoylation examined by WB. Subsequent PTEN blots showed the efficiency of PTEN IP. For controls, extracts were incubated with control beads. Maximum ubiquitination (ellipses) and sumoylation (rectangles) are indicated. Left blots show Thr308pAKT levels in WCE. Extracts (1 mg) were also incubated with anti-PTEN Ab for IP and pPTEN levels determined in WB (right panels). For controls (Ctr), extracts were incubated with Prot A. The graph shows the normalized pPTEN signal (corrected for PTEN levels) with respect to the mean pPTEN signal (considered 1) (mean ± SD, n = 3). (b) HEK-293T cells were activated as in (a); Thr308pAKT levels were examined in WCE (top panels). Extracts were also enriched for ubiquitinated proteins in TUBES columns. The eluted extracts were incubated with PTEN Ab-beads for IP and these examined by WB (top left). Secondary PTEN blots confirmed the retention of more ubiquitinated-PTEN at 5 min in the TUBES columns (right). PTEN was also immunoprecipitated from WCE and its sumoylation examined by WB. Subsequent PTEN blots showed the efficiency of PTEN IP. For controls, extracts were incubated with control beads. Maximum ubiquitination (ellipses) and sumoylation (rectangles) are indicated.

FCS, PDGF, and EGF Addition Alters PTEN Localization at pAKT Peaks and Troughs
PTEN sumoylation was examined in blots of PTEN immunopurified from WCE. Sumoylation was also transient, and did not coincide with PTEN ubiquitination or the pAKT peaks, but co-occurred with a reduction in pAKT (Figure 4b). Subsequent PTEN blots identified sumoylated and non-modified PTEN (Figure 4b). PTEN secondary blots also identified a fraction of PTEN to be sumoylated prior to stimulation; this was better detected in SUMO2,3 blots (see below). Therefore, after the addition of FCS, PTEN underwent early ubiquitination coinciding with the maximum pAKT levels (and low PTEN activity, Figure 1). Posterior sumoylation coincided with the fall in pAKT levels.

FCS, PDGF, and EGF Addition Alters PTEN Localization at pAKT Peaks and Troughs
Since PTEN ubiquitination and sumoylation might alter PTEN localization [11,12], tests were made to see whether PTEN modifications induced by FCS concurred with changes in PTEN localization. Cells were activated with FCS or EGF for short time periods (0-90 min) and then subjected to subcellular fractionation and the PTEN in the cell fractions examined by WB. At all times, the majority of the PTEN signal was found to remain in the cytosol (>80%) (Figure 5a). In contrast, the amount of PTEN at the PM was found to be reduced at 5 and 30-to-60 min after the addition of FCS or EGF, when pAKT levels were at their highest. PTEN relocated back to the PM later, when pAKT levels were low (Figure 5a, Supplementary Figure S5a Since PTEN ubiquitination and sumoylation might alter PTEN localization [11,12], tests were made to see whether PTEN modifications induced by FCS concurred with changes in PTEN localization. Cells were activated with FCS or EGF for short time periods (0-90 min) and then subjected to subcellular fractionation and the PTEN in the cell fractions examined by WB. At all times, the majority of the PTEN signal was found to remain in the cytosol (>80%) (Figure 5a). In contrast, the amount of PTEN at the PM was found to be reduced at 5 and 30-to-60 min after the addition of FCS or EGF, when pAKT levels were at their highest. PTEN relocated back to the PM later, when pAKT levels were low ( Figure  5a, Supplementary Figure S5a   The changes in PTEN localization were confirmed by immunofluorescence (IF). PTEN Ab specificity (in IF) was tested by PTEN depletion in human and mouse cells; only in the latter cells did PTEN depletion reduce the PTEN IF signal (Supplementary Figure S5c). The IF analysis was therefore performed in murine NIH3T3 cells. These were cultured for 2 h without serum and then activated with FCS or PDGF, which induced pAKT fluctuation with maximum levels at 5 and 30 min (Figure 5b). The proportion of cells with a PTEN signal at the PM was reduced at 5 and 30 min, i.e., when the pAKT signal was strong (Figure 5b). At these times, a moderate increase in the nuclear PTEN signal was detected. In contrast, at 10 and 60-to-90 min (when pAKT levels were low), PTEN localized to the PM in a large proportion of the cells (Figure 5b). Therefore, cell activation with FCS, PDGF, or EGF reduced PTEN levels at the PM at~5 min, coinciding with the times when pAKT levels were highest (and PTEN-ubiquitinated); PTEN relocation to the PM occurred in tandem with low pAKT levels (and PTEN sumoylation).
The involvement of NEDD4-1, cCBL, and CBL-b in PTEN ubiquitination was examined using siRNA. Whereas NEDD4-1 depletion did not alter PTEN ubiquitination or the changes in pAKT after FCS addition, both cCBL and CBL-b depletion reduced PTEN ubiquitination soon after FCS addition and flattened the Thr308pAKT levels ( Figure 6b). Since PTEN sumoylation takes place after PTEN ubiquitination (see above), the effect of cCBL depletion on sumoylation was also examined. cCBL depletion reduced the small PTEN sumoylation levels seen in quiescence (clearly detected with SUMO2,3 Ab) and abrogated PTEN sumoylation after GF addition (Supplementary Figure S6a,b). These observations show that cCBL is essential for the early ubiquitination of PTEN after cell stimulation with GF, and for later PTEN sumoylation.  Thr308pAKT levels were examined in WCE. Cell extracts were enriched for ubiquitinated proteins by purification in TUBES columns. The eluted ubiquitin-enriched extracts were used for PTEN IP with Ab-fused Sepharose beads. For controls (Ctr IP), extracts were incubated with isotype-matched control Ab beads; PTEN ubiquitination was examined by WB. Dashed lines indicate lanes that were cut out from the original blots (included as Supplementary Information). UBC12 shRNA efficiency was assessed by q-PCR; MLN-4924 activity via the examination of p27kip levels upon cell incubation without FCS (16 h). pAKT signals were measured, corrected for the AKT level, and normalized to the mean pAKT signal in controls (equivalent to 1). The endo-PTEN ubiquitination signal (60-120 KDa) was measured, corrected for cellular PTEN levels (in WCE), and compared to the mean PTEN ubiquitination signal (considered 1). The graph shows the pAKT signal vs. the ubiquitinated PTEN signals (mean ± SD, n = 3).

Purified cCBL Ubiquitinates Purified PTEN In Vitro
Since cCBL was more abundant than CBL-b in the HEK-293T cells, the mechanism of PTEN regulation by CBL molecules was studied by focusing on cCBL. Tests were made to determine whether purified cCBL ubiquitinated purified PTEN in vitro. A ubiquitination reaction was first optimized using cCBL purified from HEK-293T cells; cCBL was tested prior to and upon its Tyr phosphorylation by an active form of the Src-family kinase Lck (Y505F-Lck) [18] (Supplementary Figure S6c). Phospho (p)-cCBL or non-modified cCBL were incubated with ubiquitin, E1 ligase, and E2 ligases (UBE2H5B or UBE2N) in an ATPcontaining buffer. This reaction revealed the cCBL auto-ubiquitination capacity, which was greater for p-cCBL (Supplementary Figure S6d).
To test for PTEN ubiquitination by cCBL, r-cCBL (from baculovirus), and PTEN purified from bacteria were used (Supplementary Figure S6e). r-cCBL ubiquitinated PTEN in vitro (Figure 6c). The high MW band detected in PTEN blots even in the absence of cCBL (Figure 6c, d) might correspond to bacterially aggregated PTEN. The capacity of cCBL to modify PTEN was compared to that of WWP1; cCBL ubiquitinated PTEN in vitro to an extent similar to WWP1 (Figure 6d). PTEN ubiquitination can reduce PTEN phosphatase activity [28]; indeed, cCBL-mediated ubiquitination in vitro induced an~20% reduction in PTEN phosphatase assays (Supplementary Figure S6f). Together, cCBL expression is essential for in vivo PTEN ubiquitination at~5 and~30-60 min after the addition of GF. It is also required for later PTEN sumoylation. The cCBL ubiquitination of PTEN might be direct, as purified r-cCBL was able to ubiquitinate r-PTEN in vitro.

PI3-Kinase Controls cCBL Association with EGFR
To test whether the addition of GF brings PTEN and cCBL into proximity, cells were activated with EGF and the recruitment of cCBL or PTEN to EGFR examined by IP and immunoblotting. As PTEN band resolves close to the IgG heavy chain (used in IP), analyses were performed using PTEN-Luc (which has a higher MW). The low PTEN levels increase by PTEN-Luc expression in HEK-293T cells did not affect pAKT fluctuation (Supplementary Figure S7a). EGFR IP followed by PTEN blotting showed that a fraction of PTEN was bound constitutively to EGFR (Figure 7a, left); in contrast, EGF induced a transient translocation of cCBL to EGFR (middle). A small fraction of cCBL was constitutively bound to PTEN (Figure 7a, right); this basal complex looks not be bound to EGFR since no cCBL was found associated with EGFR prior to stimulation (Figure 7a, middle). Thus, EGFR binds to PTEN constitutively and to cCBL only after EGF addition.
EGF is known to induce PI3-kinase translocation to EGFR [36]. Since the PI3-kinase regulatory subunit (p85 PI3K ) binds to cCBL after T cell activation [15], it was postulated that EGF might trigger a p85 PI3K -mediated recruitment of cCBL to EGFR. Biochemical examination of the endogenous p85 PI3K /cCBL complex was difficult due to the non-specific bands of p85 PI3K Ab in blots. The involvement of p85 PI3K in cCBL recruitment to EGFR was tested by modification of the p85α and p85β levels (Supplementary Figure S7b). p85 PI3K overexpression (indicated as ↑p85α p85β) enhanced, and p85 PI3K depletion (↓p85α p85β) abrogated, cCBL association to EGFR (Figure 7b). The EGFR/PTEN and cCBL/PTEN complex levels, however, were not proportional to p85 PI3K levels (Supplementary Figure S7c).
Since a fraction of PTEN is bound constitutively to EGFR, EGF should bring p85 PI3K into proximity with PTEN. To test this, IP/WB was avoided given the non-specific p85α Ab bands in blots. Instead, PTEN/p85 PI3K proximity was examined by fluorescence energy transfer (FRET) [37][38][39]. The efficiency of FRET in detecting p85 PI3K complexes was confirmed using fluorescent forms of the PI3-kinase catalytic and regulatory subunits (using the SH3-p85 PI3K domain as a negative control) (Supplementary Figure S7d,e). FRET also confirmed the formation of a PI3-kinase α and β complex [21] mediated by p85α PI3K but not by p50α PI3K (Supplementary Figure S7f). abrogated, cCBL association to EGFR (Figure 7b). The EGFR/PTEN and cCBL/PTEN complex levels, however, were not proportional to p85 PI3K levels (Supplementary Figure S7c).  Energy transfer between p85α PI3K and PTEN was examined taking advantage of the PTEN-Luc bioluminescence signal. Titration of the bioluminescence energy transfer (BRET) between PTEN-Luc and YFP-p85α PI3K revealed a positive interaction (Figure 7c). Moreover, the addition of EGF induced a transient increase in the BRET signal at 5-15 min showing that EGF increases p85α PI3K and PTEN proximity (Figure 7d).
It was postulated that the depletion of cCBL by reducing PTEN ubiquitination would impair PTEN detachment from the PM. The cCbl-depleted NIH3T3 cells showed a modest but significant increase in the proportion with PM-bound PTEN compared to controls (Figure 7e, Supplementary Figure S7g).
Thus, EGF stimulates a p85 PI3K -dependent translocation of cCBL to EGFR, where a fraction of PTEN is constitutively bound. Once close, cCBL can modulate PTEN ubiquitination/inactivation and its detachment from the PM, thus allowing maximum PIP 3 /pAKT levels to be reached. After PTEN detachment from the PM, PTEN undergoes a sumoylation process that likely induces its later membrane localization and the reduction of pAKT levels simultaneous with PTEN sumoylation. The capacity of PI3-kinase to bring cCBL to the receptor and thus induce transient PTEN inactivation and internalization reveals that crosstalk occurs between the enzymes regulating PIP 3 (PI3-kinase and PTEN) and highlights cCBL as a new target for the control of PTEN activity.

Discussion
The aim of this study was to understand how PTEN activity is regulated after cell activation by growth factors, and to determine the mechanism linking PI3-kinase and PTEN activities for controlling PIP 3 levels. It was found that both the activity of the PI3-kinase effector AKT and the membrane PIP 3 levels fluctuate following the addition of GF, with a first peak at~5 min and a second one at~30-60 min. Between these peaks, membrane PIP 3 levels and pAKT levels decrease. The time frame under analysis was chosen by taking into account previous studies showing that PTEN is active at the membrane at~90 min following the addition of GF [21]. Within this period, PTEN showed activity fluctuations complementary to those of AKT. When AKT was activated, PTEN was inhibited, and vice versa, suggesting that AKT and PTEN activity variations are probably linked. Indeed, PTEN expression was essential for the fluctuation in AKT activity. PTEN underwent a ubiquitination event early after the addition of GF, at the time of maximum pAKT levels. cCBL expression was required for PTEN early ubiquitination in cells activated by GF. So, either PTEN is ubiquitinated directly by cCBL (as cCBL is able to ubiquitinate PTEN in vitro), or cCBL brings a ubiquitin E3 ligase to PTEN. The ubiquitination of PTEN reduced its activity and correlated with its detachment from the membrane. These results show that maximum PIP 3 /pAKT levels require not only PI3-kinase to be active but also the inactivation of PTEN. It is here shown that the addition of EGF triggers the PI3-kinasedependent recruitment of cCBL to the EGFR, where it encounters PTEN. Thus, PI3-kinase induces PTEN inactivation by bringing the cCBL E3 ligase to the EGFR, which in turn results in PTEN ubiquitination.
Although the transient nature of AKT activation by GF was known, the fluctuation and the complementary behavior of PTEN and AKT activity, were not. Most previous studies included a single or two-three activation time points [2,21,22]. Dalle Pezze et al. [40] did, however, include a detailed kinetic study (as we do here) using HeLa cells activated with insulin; their blots also showed Thr308pAKT fluctuation, but the authors passed over this observation as their focus was mTOR.
AKT/PTEN fluctuation was detected in normal murine and immortalized fibroblasts, in human PTEN-reconstituted PC3 cells, and in HEK293T cells, showing that it is a generalized process. It could be argued that the two pAKT peaks (in the 0-90 min timeframe) could be linked to two different inputs of PI3-kinase activation, such as by GFR and GTPase activation [16,41]. However, these inputs do not explain the reductions in pAKT levels. The present assessment of PTEN phosphatase activity showed that PTEN activity also fluctuates, and that pAKT modulation requires PTEN activity. PTEN depletion abrogated AKT fluctuations, which were rescued by expression of WT-PTEN (but not inactive-PTEN) in PTEN−/− cells. These transient fluctuations in pAKT are not irrelevant since the depletion of PTEN (here) and the expression of a constitutive, active PI3-kinase [20] both eliminate them, resulting in delayed exit from G2/M and slower progression through the cell cycle.
One of the best-established PTEN regulatory mechanisms is the phosphorylation of its C-terminal cluster of four Ser and Thr residues, which keeps PTEN inactive and stable [10,13]. In the present work, phospho-PTEN levels remained constant after the addition of FCS (0-90 min), ruling out CT-phosphorylation as the mechanism behind PTEN inactivation after the addition of GF. The present work does not, however, exclude a potential reduction in the phosphorylation of individual CT-Ser and Thr residues since the Ab used for the analysis does not distinguish fully phosphorylated PTEN from single residue phosphorylation defects [42].
Ubiquitination might also reduce PTEN activity [28]. Indeed, PTEN ubiquitination at~5 min after GF addition concurred with PTEN showing low phosphatase activity. PTEN ubiquitination was unaffected by interference with cullin E3 ligases or by NEDD4 depletion, but it was nearly abrogated by cCBL or CBL-b depletion. cCBL and CBL-b also cooperate for optimal GFR downregulation [43]. The involvement of NEDD4 in PTEN ubiquitination in vivo has been controversial [31,32] and the presented observations support that NEDD4 is not involved in GF-induced early PTEN ubiquitination. Since purified cCBL was able to ubiquitinate purified PTEN in vitro, cCBL might ubiquitinate PTEN directly. Future descriptive studies will examine which PTEN Lys residues are modified in a cCBL-dependent manner.
The reduction in pAKT levels correlated with an increase in PTEN sumoylation and PTEN localization to the PM. PTEN binding to PIP 2 in the PM triggers an allosteric activation of the phosphatase [44] which could account for the increase in PTEN activity once localized to the PM. PTEN sumoylation seems to also require the former PTEN ubiquitination process since cCBL depletion reduced PTEN ubiquitination and sumoylation.
To gain insight into how cCBL and PTEN might encounter one another, the association of these molecules, and with EGFR, was examined. Whereas a fraction of PTEN was constitutively bound to EGFR, cCBL recruitment to EGFR only occurred after the addition of EGF. Cell activation induces PI3-kinase translocation to GFR [36] but also cCBL binding to p85 PI3K [15]. Here it was postulated, and confirmed, that cCBL recruitment to EGFR (where PTEN is bound) is controlled by p85 PI3K . This model was challenged by testing two predictions made in view of these results. The first was that the addition of EGF should also bring p85 PI3K and PTEN into proximity. This was confirmed using bioluminescence energy transfer assays. The second was that, since cCBL depletion abrogates early GF-induced PTEN ubiquitination, which correlated with its detachment from PM, cCBL depletion might increase PTEN localization at the PM. This was confirmed by immunofluorescence.
It remains to be determined whether cCBL and p85 PI3K stay bound to PTEN in its transit to the nucleus. Previous studies have shown that p85 PI3K , p110α, p110β, and PTEN all form part of a large (~600-kDa) macromolecular complex (identified by gel filtration), supporting the idea that PTEN and p85 PI3K might transit to the nucleus in complex [21,45]. cCBL might also internalize bound to EGFR and PTEN.

Conclusions
The present results reveal a mechanism for the control of endogenous PTEN activity upon the addition of GF. Shortly after its addition, the activity of the PI3-kinase effector AKT, and that of PTEN, show complementary fluctuations in behavior. PTEN is needed for these fluctuations since its depletion results in the flattening of pAKT levels. Moreover, PTEN reconstitution in PTEN-deficient cells restores pAKT fluctuation, which is needed for optimal progression through the cell cycle. AKT/PTEN co-regulation involves the EGF-induced recruitment of PI3-kinase to EGFR, and in turn that of cCBL. The expression of cCBL E3 ligase is essential for PTEN ubiquitination in cells. cCBL mediated PTEN ubiquitination in vitro reduced its phosphatase activity and correlated with PTEN detachment from the PM. The present results show that cCBL recruitment to the EGFR is mediated by PI3-kinase. Thus, by inducing cCBL translocation to the EGFR, PI3-kinase promotes the inactivation of PTEN allowing maximum PIP3/pAKT levels to be reached. After cCBL-regulated PTEN ubiquitination, PTEN undergoes a process of sumoylation and re-localizes to the membrane, a process that occurs with a simultaneous reduction in pAKT levels. The biological significance of these results is that the crosstalk of PI3-kinase and PTEN, mediated by cCBL, guarantees both the optimal and transient nature of PIP 3 bursts. Based on these results, the reduction of PTEN ubiquitination, and the boosting of PTEN sumoylation, might help to increase PTEN activity in cancer cells.