ABSTRACT
Signal transduction is regulated by the lateral segregation of proteins into nanodomains on the plasma membrane. However, the molecular mechanisms that regulate the lateral segregation of cell surface receptors, such as receptor tyrosine kinases, upon ligand binding are unresolved. Here we used high-resolution spatial mapping to investigate the plasma membrane nanoscale organization of the epidermal growth factor (EGF) receptor (EGFR). Our data demonstrate that in serum-starved cells, the EGFR exists in preformed, cholesterol-dependent, actin-independent nanoclusters. Following stimulation with EGF, the number and size of EGFR nanoclusters increase in a time-dependent manner. Our data show that the formation of EGFR nanoclusters requires receptor tyrosine kinase activity. Critically, we show for the first time that production of phosphatidic acid by phospholipase D2 (PLD2) is essential for ligand-induced EGFR nanocluster formation. In accordance with its crucial role in regulating EGFR nanocluster formation, we demonstrate that modulating PLD2 activity tunes the degree of EGFR nanocluster formation and mitogen-activated protein kinase signal output. Together, these data show that EGFR activation drives the formation of signaling domains by regulating the production of critical second-messenger lipids and modifying the local membrane lipid environment.
The epidermal growth factor (EGF) receptor (EGFR) is a single transmembrane domain protein that possesses intrinsic tyrosine kinase (TK) activity. Ligand binding to the extracellular domain induces conformational changes that promote activation of the intracellular TK domain. The kinase domain then autophosphorylates a number of tyrosine residues in the C-terminal region of the protein, creating docking sites for adapter and effector proteins. Thus, the active form of the EGFR could reasonably be expected to be a dimer. However, recent studies using single-molecule imaging, image correlation spectroscopy (ICS), fluorescence correlation spectroscopy (FCS), and immunoelectron microscopy (immuno-EM) show that the EGFR is, in fact, nonrandomly organized into oligomers on the plasma membrane (6, 7, 16, 34, 44). ICS measurements estimate that, in the absence of ligand, there are, on average, 2.2 EGFRs per cluster, which increases to 3.7 receptors per cluster upon stimulation (7). Single-molecule tracking experiments also suggest that unliganded EGFRs continually fluctuate between monomers and dimers that are primed for activation (5). Furthermore, the organization of the EGFR is dynamic and clustering of the EGFR increases over time after EGF stimulation (7, 16). However, neither the precise role of EGFR oligomerization in signal transduction nor the mechanisms driving oligomer formation have been resolved.
The organization of the EGFR into oligomers is dependent upon cellular cholesterol. Saffarian et al., using FCS, estimated that 70% of EGFRs exist as monomers, 20% as dimers, and 10% as oligomers (34). However, depletion of cholesterol decreases the percentage of monomeric receptors and increases the proportion of oligomeric receptors. Cholesterol depletion and actin depolymerization also alter the diffusion coefficient of the EGFR and the confinement area size (22). The finding that EGFR membrane organization is dependent upon cholesterol is of particular interest because a number of studies have demonstrated that EGFR activity is negatively regulated by cholesterol (4, 23, 28, 32).
Phospholipase D2 (PLD2) hydrolyzes phosphatidylcholine (PC) to produce choline and phosphatidic acid (PA). PLD2 is localized to the plasma membrane (10), associates with the EGFR (39), and is rapidly activated upon EGF stimulation, leading to increased production of PA (15, 38, 39). A number of lines of evidence suggest that PA is an important mediator of EGFR action. First, exogenous PA is mitogenic when incubated with cells (17, 19, 42, 45). Second, direct interaction with membrane PA regulates the activity of a number of components downstream of the EGFR, including Sos (47) and Raf (12, 13, 30, 31).
In the current study, we used high-resolution spatial analysis techniques to investigate EGFR plasma membrane organization. Using these approaches, we identified PA as the key molecular component responsible for driving EGFR nanocluster formation in response to EGF binding and demonstrated that the level of PLD2 activity regulates the duration of mitogen-activated protein kinase (MAPK) signal output.
MATERIALS AND METHODS
Cell culture.Baby hamster kidney (BHK) cells were maintained in HEPES-buffered Dulbecco's modified Eagle's medium containing 10% heat-inactivated serum supreme (Lonza, Basel, Switzerland). BHK cells were seeded onto either 13-mm glass coverslips for EM and confocal microscopy or 6-cm dishes for biochemical assays and transfected using Lipofectamine reagent (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions as previously described (33). Where indicated, cells were serum starved for approximately 4 h and then treated with 50 ng/ml EGF (Sigma-Aldrich, St. Louis, MO), 100 μM phosphatic acid (1,2-dipalmitoyl-sn-glycero-3-phosphate sodium salt; Sigma-Aldrich, St. Louis, MO), 100 μM phosphatidylserine (1,2-dipalmitoyl-sn-glycero-3-phospho-l-serine sodium salt; Sigma-Aldrich, St. Louis, MO), diacylglycerol (DAG) kinase inhibitor II (R59949; EMD Biosciences), or (±)-propranolol hydrochloride (Sigma-Aldrich, St. Louis, MO) for the indicated times.
Plasmids.The yeast soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) protein Spo20p PA binding domain, described previously (46), was subcloned into a vector containing monomeric red fluorescent protein (mRFP). To increase the cytoplasmic pool of the fusion protein, a nuclear export sequence (NES) was also added.
Western blotting.Cells were washed and subjected to detergent lysis. For analysis of signal transduction, whole-cell lysates were produced (50 mM Tris [pH 7.5], 75 mM NaCl, 25 mM NaF, 5 mM MgCl2, 5 mM EGTA, 1 mM dithiothreitol, 100 μM NaVO4, 1% Nonidet P-40 plus protease inhibitors) and a total of 20 μg of each was immunoblotted with anti-pERK antibody (Cell Signaling Technologies, Danvers, MA). Anti-β-actin and -α-tubulin antibodies were used as loading controls. Signal was detected by enhanced chemiluminescence (SuperSignal; Pierce, Thermo Fisher Scientific, Rockford, IL) and imaged by FluorChemQ (Alpha Innotech, San Leandro, CA). Quantification of intensities was performed using FluorChemQ software.
PLD activity assay.PLD activity was measured by transphosphatidylation reaction. BHK cells grown in 35-mm dishes were labeled overnight with [3H]palmitic acid. Cells were serum starved and then incubated with 0.3% 1-butanol in the presence or absence of 50 ng/ml EGF. The reaction was stopped by the addition of cold methanol.
EM and statistical analysis.Plasma membrane sheets were prepared, fixed, and labeled with affinity-purified anti-GFP (green fluorescent protein) antiserum coupled directly to 5-nm gold particles as described previously (14, 25). Digital images of the immunogold-labeled plasma membrane sheets were taken in a transmission electron microscope (JEOL 1011; JEOL, Tokyo, Japan). Intact 1-μm2 areas of the plasma membrane sheet were identified using ImageJ, and the x-y coordinates of the gold particles were determined as described previously (14, 25). Ripley's K function (2, 29) was calculated using the x-y coordinates and then standardized on the 99% confidence interval (CI) for a random pattern (14, 25). Bootstrap tests to examine differences between replicated point patterns were constructed exactly as described previously (9), and statistical significance was evaluated against 1,000 bootstrap samples.
Analysis of endogenous EGFR nanoclusters was performed as described above but with the following changes. The EGFR was detected using an anti-EGFR antibody (Cell Signaling Technologies, Danvers, MA) and a goat anti-rabbit secondary antibody conjugated to 5-nm gold particles (BB International, Cardiff, United Kingdom).
Confocal microscopy.BHK cells grown on 13-mm glass coverslips were fixed using 4% paraformaldehyde and imaged using a Zeiss LSM 510 confocal microscope (Carl Zeiss, Thornwood, NY) with the appropriate GFP filter set.
Fluorescence lifetime imaging-fluorescence resonance energy transfer (FLIM-FRET) microscopy.FLIM-FRET experiments were carried out using a lifetime fluorescence imaging attachment (Lambert Instruments, Leutingewolde, Netherlands) on an inverted microscope (Eclipse Ti; Nikon Instruments Inc., Melville, NY). BHK cells transiently expressing EGFR-GFP (donor), alone or with mRFP-NES Spo20p (acceptor) (using a 1:2 ratio of plasmid DNA), were serum starved, treated with 50 ng/ml EGF for the indicated times, and then fixed using 4% paraformaldehyde. The samples were excited using a sinusoidally modulated 3-W 470-nm light-emitting diode at 40 MHz under epi-illumination. Fluorescein was used as a lifetime reference standard. Cells were imaged with a Plan Apo 60× 1.40 oil objective using an appropriate GFP filter set. The phase and modulation were determined from a set of 12 phase settings using the manufacturer's software. Analysis was performed on a cell-by-cell basis.
RESULTS
The EGFR is organized into nanoclusters on the plasma membrane.The spatial organization of the EGFR was examined in BHK cells in the absence of serum. In order to analyze the intracellular organization, we used an EGFR-GFP fusion construct in which GFP was used to tag the C terminus of the EGFR. Plasma membrane sheets were generated from cells expressing EGFR-GFP and labeled with anti-GFP antibody conjugated directly to 5-nm gold particles. Spatial analysis demonstrated that in the absence of serum, ∼32% ± 1.9% (standard error of the mean [SEM]; n = 15) of the EGFRs exist as monomers and ∼67% ± 1.9% (SEM; n = 15) of the EGFRs were organized into nanoclusters ranging from dimers to high-order oligomers that had an average radius of ∼24 nm (Fig. 1A). We detected an average of 2.3 gold particles per cluster. Accounting for the previously characterized capture ratio of the anti-GFP antibody (24), we estimated that the upper range of EGFR proteins per nanocluster was about six. For the frequency distribution of cluster size, see Fig. S1A in the supplemental material. Next, we analyzed whether the level of EGFR-GFP expression affected the degree of nanoclustering detected in these experiments. We found that there was no significant correlation between EGFR-GFP plasma membrane expression and the degree of nanocluster formation detected (see Fig. S1B in the supplemental material). Finally, we confirmed that endogenous EGFR was organized into similar nanoclusters in A431 cells by using an anti-EGFR primary antibody detected by a secondary antibody conjugated to 5-nm gold particles (see Fig. S2A in the supplemental material).
The EGFR is organized into nanoclusters on the plasma membrane. (A) Plasma membrane sheets generated from BHK cells expressing EGFR-GFP were labeled with anti-GFP antibody conjugated to 5-nm gold particles. The spatial distribution of the gold labeling was analyzed using Ripley's K function. Maximum L(r) − r values above the 99% CI for complete spatial randomness indicate clustering at the value of r (supr). Univariate K functions are weighted means (n = 15) standardized on the 99% CI. (B) BHK cells expressing EGFR-GFP were left untreated or treated with either 1% methyl-β-cyclodextrin for 60 min or 1 μM latrunculin A for 5 min. Plasma membrane sheets were labeled with anti-GFP antibody conjugated to 5-nm gold particles. Univariate K functions are weighted means (n ≥ 6) standardized on the 99% CI. Significant differences from the control EGFR pattern were assessed using bootstrap tests. Treatment with latrunculin A did not significantly alter EGFR nanocluster formation (P = 1). However, treatment with methyl-β-cyclodextrin significantly decreased EGFR nanoclustering (P = 0.001).
To determine the membrane components required to maintain EGFR nanoclusters in the absence of serum, we treated serum-starved cells expressing EGFR-GFP with either methyl-β-cyclodextrin to deplete cellular cholesterol or latrunculin A to depolymerize the actin cytoskeleton. Plasma membrane sheets derived from these cells were immunogold labeled with anti-GFP antibody. Spatial analysis of the resulting gold pattern demonstrated that depletion of cellular cholesterol significantly decreased EGFR nanoclustering (Fig. 1B). However, treatment with latrunculin A did not have a significant effect (Fig. 1B).
EGFR nanocluster formation is regulated by EGF.Next, we investigated whether the formation of EGFR nanoclusters is functionally relevant to EGF-induced signal transduction. We stimulated serum-starved BHK cells expressing EGFR-GFP with EGF and examined the spatial organization of the EGFR over time. We observed a significant increase in EGFR nanocluster formation after EGF stimulation (Fig. 2A). The highest levels of EGFR nanoclustering were detected following 10 to 20 min of EGF treatment. These data suggested that EGFR nanocluster formation occurred in response to receptor activation. A similar increase in EGFR nanoclustering upon EGF stimulation was also detected in HEK 293 cells expressing EGFR-GFP (see Fig. S2B in the supplemental material). During the course of this analysis, we noticed that the radius of the EGFR nanoclusters increased upon ligand stimulation. Our analysis demonstrated that following EGF stimulation for 20 min, the average radius of an EGFR nanocluster had increased to 30 nm. The percentage of EGFR in clusters had also increased from 67% ± 1.9% (SEM; n = 15) to ∼83% ± 2.1% (SEM; n = 12). The percentage of monomers decreased from ∼32% to 17%, and the percentage of receptors present in high-order oligomers increased. For the frequency distribution of cluster size, see Fig. S1B in the supplemental material. EGFR nanoclusters had ∼3.4 gold particles per cluster. Accounting for the anti-GFP antibody capture ratio, we estimated that there could be as many as nine proteins per cluster following EGF stimulation.
EGFR nanocluster formation is induced by EGF stimulation. BHK cells expressing EGFR-GFP were serum starved and then stimulated with 50 ng/ml EGF for the indicated times. Plasma membrane sheets were labeled with anti-GFP antibody conjugated to 5-nm gold particles. Univariate K functions are weighted means (n ≥ 10) standardized on the 99% CI. Significant differences from the control unstimulated EGFR pattern were assessed using bootstrap tests. (A) Stimulation with EGF resulted in a significant increase in EGFR nanocluster formation at all time points (P = 0.001). (B) Stimulation with EGF for 10 and 20 min led to a significant increase in EGFR nanoclustering compared to that obtained with the unstimulated control (P = 0.003 and P = 0.001, respectively). However, EGFR nanoclustering was not significantly different from unstimulated EGFR nanoclustering following stimulation with EGF for 40 and 60 min (both P = 1).
Following activation, EGFR undergoes endocytosis, leading to the downregulation of receptor activity. Therefore, we investigated how EGFR nanoclustering is regulated over longer time periods. Cells expressing EGFR-GFP were serum starved and then stimulated with EGF for up to 60 min. Using spatial analysis, we found that the level of EGFR nanoclustering had returned to basal levels after 40 min of EGF stimulation (Fig. 2B). The EGFR is known to be clustered in clathrin-coated pits following ligand binding. Therefore, it was conceivable that the increased formation of EGFR nanoclusters, in response to ligand binding, occurred as a consequence of clathrin-coated pit formation. However, while we were able to detect the clustering of EGFR into clathrin-coated pits by our immuno-EM technique, we were also able to detect a significant proportion of EGFR nanoclusters that existed in morphologically smooth regions of the plasma membrane (see Fig. S3 in the supplemental material). These data suggest that clathrin-coated pit formation is not a prerequisite for EGFR nanocluster formation and that additional mechanisms may also regulate EGFR nanocluster formation.
EGFR nanocluster formation is dependent on receptor TK (RTK) activity.To determine directly whether EGFR nanocluster formation requires RTK activity, we utilized the TK inhibitor AG1478. Treatment with 100 nM AG1478 was sufficient to inhibit the production of ppERK in response to EGF stimulation (Fig. 3A). Spatial analysis of plasma membrane sheets generated under the same conditions demonstrated that cells treated with AG1478 had a lower basal level of EGFR nanoclustering than control untreated cells (Fig. 3B). Importantly, our data also demonstrated that EGF stimulation failed to induce EGFR nanoclustering in those cells that had been pretreated with AG1478 (Fig. 3B). Together, these data demonstrate that formation of EGFR nanoclusters in response to EGF requires RTK activity.
Induction of EGFR nanoclustering following ligand binding requires RTK activity. (A) BHK cells were serum starved and then pretreated with 100 nM AG1478 for 10 min as indicated, followed by EGF stimulation in the presence or absence of AG1478 for 10 min as indicated. Whole-cell lysates were generated, and 20 μg of lysate was blotted for ppERK. Actin was used as a loading control. The graph represents the mean of three independent experiments (±SEM). Representative blots are shown. (B) BHK cells expressing EGFR-GFP were serum starved and treated with AG1478 for 10 min as indicated prior to stimulation with 50 ng/ml EGF for 10 min. Plasma membrane sheets were labeled with anti-GFP antibody conjugated to 5-nm gold particles. Univariate K functions are weighted means (n ≥ 13) standardized on the 99% CI. Significant differences from the control unstimulated EGFR pattern were assessed using bootstrap tests. Treatment with AG1478 led to a significant reduction in EGFR nanoclustering in the presence and absence of EGF (both P = 0.001).
PLD2 activity regulates EGFR nanocluster formation.The data in Fig. 3 suggest that formation of EGFR nanoclusters requires RTK activity. Therefore, we examined what factors immediately downstream of RTK activation are involved in regulating EGFR nanocluster formation. PLD2 converts PC into PA by removal of the bulky choline head group and is activated by EGFR TK activity (18, 39). We confirmed that PLD activity was increased upon EGF stimulation (Fig. 4) and that PLD2 was localized to the plasma membrane in BHK cells (data not shown). To determine whether PLD2 lipase activity is necessary for the formation of EGFR nanoclusters, we utilized the PLD2 K758R mutant that is deficient in lipase activity and acts in a dominant interfering manner (40, 41). Cells expressing EGFR-GFP alone or in the presence of PLD2 K758R were serum starved and then stimulated with EGF for 10 min. Expression of PLD2 K758R significantly decreased the basal level of EGFR nanoclustering detected in unstimulated cells (Fig. 5A) in a manner similar to that of treatment with the TK inhibitor AG1478 (Fig. 3B). Stimulation with EGF resulted in a significant increase in the number of EGFR nanoclusters in cells expressing EGFR-GFP alone. However, in cells coexpressing EGFR-GFP and PLD2 K758R, EGF stimulation led to only a small increase in EGFR nanocluster formation that did not reach basal levels (Fig. 5A). In contrast, the expression of dominant negative PLD1 (PLD1 K898R) did not decrease the basal level of EGFR nanoclustering, and upon EGF stimulation, the levels of EGFR nanoclustering were similar in the presence and absence of PLD1 K898R (data not shown). We analyzed the effect of PLD2 K758R on EGFR signal transduction and found that expression of PLD2 K758R significantly reduced ppERK production in response to EGF (Fig. 5D). These data suggest that PLD2 activity is required for stimulus-dependent EGFR nanocluster formation and consequently signal output.
PLD activity is elevated by EGF. BHK cells were serum starved and then stimulated with EGF for 1 min, and the production of phosphatidylbutanol was measured. Bars represent the relative proportion of phosphatidylbutanol to total lipid (n = 4) ± the standard deviation. Statistical significance was determined by t test (**, P = 0.002). The graph is representative of four independent experiments.
PLD2 activity is required for ligand-induced EGFR nanocluster formation. BHK cells expressing EGFR-GFP alone (B and C) or in combination with PLD2 K758R (A) were serum starved and treated as follows: (A) 50 ng/ml EGF for 10 min where indicated, (B) 200 μM propranolol for 30 min, followed by 50 ng/ml EGF for 10 min where indicated (C), or 50 ng/ml EGF for 10 min in the presence or absence of 1 μM R59949 where indicated. Plasma membrane sheets were generated and labeled with anti-GFP antibody conjugated to 5-nm gold particles. Univariate K functions are weighted means (n ≥ 6) standardized on the 99% CI. Significant differences from the control unstimulated EGFR pattern were assessed using bootstrap tests. (A) Expression of PLD2 K758R significantly decreased the basal level of EGFR nanoclustering (P = 0.001). (B) Treatment with 200 μM propranolol led to a significant increase in EGFR nanocluster formation (P = 0.001). (D and E) Whole lysates were extracted from cells treated as indicated for panels A and B, and 20 μg of each lysate was blotted for ppERK. Actin was used as a loading control. The graphs represent the mean of three independent experiments (±SEM). Significant differences from the control were assessed by t test. (D) The expression of PLD2 K758R led to a significant reduction in ppERK following EGF stimulation compared to those cells expressing EGFR alone (*, P = 0.02). (E) Treatment with 200 μM propranolol in the absence of EGF increased basal ppERK levels, although the increase did not reach significance (P = 0.2).
PA can be converted into DAG through the action of PA phosphohydrolase (PAP) (3). Propranolol, a β-adrenergic receptor antagonist, can inhibit PAP activity when used at high concentrations. Incubation with a high dose of propranolol (200 μM) is sufficient to elevate cellular PA levels, with a maximal increase detected following ∼30 min of treatment (21). Therefore, we investigated whether the production of PA or its subsequent conversion into DAG is important for EGFR nanoclustering. Serum-starved cells expressing EGFR-GFP were treated with 200 μM propranolol, and plasma membrane sheets were generated. Spatial analysis demonstrated that propranolol treatment led to an increase in the basal level of EGFR nanocluster formation (Fig. 5B). These data suggested that the increased level of PA, through the inhibition of PAP activity, was sufficient to promote EGFR nanoclustering in the absence of a stimulus. It was important to confirm that propranolol was mediating its effect on EGFR nanoclustering via inhibition of PAP activity and not through antagonism of β-adrenergic receptor activity. To investigate this possibility, we examined the effect of lower doses of propranolol on EGFR nanoclustering. Our data demonstrate that basal nanoclustering of EGFR is not altered by treatment with 2 μM and 20 μM propranolol, doses that are sufficient to completely block β-adrenergic receptor activity (see Fig. S4 in the supplemental material). These data suggest that the elevated nanoclustering detected following treatment with 200 μM propranolol is indeed due to the specific function of propranolol as a PAP inhibitor. Thus, inhibiting the conversion of PA to DAG does not inhibit EGFR nanoclustering. Interestingly, treatment with a high dose of propranolol led to a small increase in ppERK production in serum-starved cells (Fig. 5E), consistent with increased EGFR nanoclustering.
PA can also be generated from DAG via the action of the DAG kinase (DGK). To determine whether the production of PA by DGK is also important for EGFR nanocluster formation in response to EGF, we stimulated cells with EGF in the presence or absence of R59949, a specific DGK inhibitor. In the presence of R59949, there was a slight reduction in basal EGFR nanoclustering in serum-starved cells. However, treatment with R59949 did not inhibit EGF-stimulated nanocluster formation (Fig. 5C). These data suggest that PA production by DGK has only a minor role in regulating EGFR nanocluster formation.
EGF stimulation generates PA production in EGFR nanoclusters.To confirm that PA was produced in a spatially relevant manner, we analyzed PA production in response to EGF by means of a GFP-tagged PA binding probe (PA binding domain of Spo20p) described previously (46). In cells that had been serum starved, the PA binding probe was predominantly cytosolic (Fig. 6 A). However, following EGF stimulation, we detected a significant enrichment of GFP fluorescence at the plasma membrane. Membrane recruitment of GFP-Spo20p was detected after as little as 1 min of EGF stimulation (Fig. 6A). GFP-Spo20p plasma membrane recruitment was also detected by immuno-EM. Spatial analysis demonstrated that the PA probe was organized into nanoclusters on the plasma membrane following 10 min of EGF stimulation (Fig. 6B). To confirm that the production of PA was localized to EGFR nanoclusters, we performed FLIM-FRET experiments with cells expressing EGFR-GFP (donor) alone or in the presence of mRFP NES Spo20p (acceptor). The cells were serum starved and then stimulated with EGF, and changes in the fluorescent lifetime of the GFP fluorophore were measured. Stimulation with EGF led to a rapid and significant reduction in the fluorescent lifetime of GFP, indicating that the PA probe was recruited to the EGFR nanoclusters over the same time course in which increased EGFR nanoclustering was detected (Fig. 6C).
EGF stimulates PA production at the plasma membrane. (A) BHK cells expressing GFP NES Spo20p were serum starved and then stimulated with 50 ng/ml EGF. Cells were fixed and imaged by confocal microscopy. (B) Plasma membrane sheets were generated from BHK cells expressing GFP NES Spo20p that had been serum starved and then stimulated with EGF for the indicated times. Plasma membrane sheets were labeled with anti-GFP antibody conjugated to 5-nm gold particles. (Left panel) The L(r) − r curve represents the clustering of GFP NES Spo20p after 10 min of EGF stimulation. (Right panel) The graph represents the mean number of gold particles/μm2 (±SEM). (C) BHK cells expressing EGFR-GFP alone or in combination with mRFP NES Spo20p were serum starved and then stimulated with 50 ng/ml EGF for the indicated times. Cells were imaged in the frequency domain using a wide-field FLIM-FRET microscope. Data points represent the mean fluorescence lifetime of GFP (±SEM). The green dashed line represents the lifetime of EGFR-GFP in the absence of mRFP NES Spo20p. Representative fluorescence lifetime images are shown.
PA directly regulates EGFR nanocluster formation.To demonstrate that PA was the critical mediator of EGFR nanocluster formation, we investigated whether addition of exogenous PA is sufficient to stimulate EGFR nanocluster formation. BHK cells expressing EGFR-GFP were serum starved and then incubated with PA over a time course. Spatial analysis revealed that treatment with PA alone was sufficient to increase EGFR nanoclustering over the same time course as EGF stimulation (Fig. 7A). We detected a similar increase in EGFR nanoclustering upon PA stimulation in HEK 293 cells expressing EGFR-GFP (see Fig. S2B in the supplemental material).
EGFR nanocluster formation is induced by treatment with exogenous PA but not PS. BHK cells expressing EGFR-GFP were serum starved and then stimulated with either (A) 100 μM PA or (B) 100 μM PS for the indicated times. Plasma membrane sheets were labeled with anti-GFP antibody conjugated to 5-nm gold particles. Univariate K functions are weighted means (n ≥ 14) standardized on the 99% CI. Significant differences from the control unstimulated EGFR pattern were assessed using bootstrap tests. (A) Treatment with PA led to a significant increase in EGFR nanoclustering at all time points (P = 0.001). (B) Incubation with PS did not have a significant impact on EGFR nanoclustering over time.
In control experiments, treatment of cells with exogenous PS did not alter EGFR nanocluster formation (Fig. 7B). Therefore, we conclude that the increased EGFR nanoclustering was specific to the action of exogenous PA and not a consequence of perturbing the plasma membrane in general.
PA regulates EGFR nanocluster formation downstream of RTK activity.The data in Fig. 1 to 7 demonstrate that upon ligand binding, the EGFR promotes the activation of PLD2, resulting in an increase in the local concentration of PA that induces EGFR nanocluster formation. Based on these data, we hypothesized that if production of PA was the critical factor required for EGFR nanocluster formation, the inhibition of RTK activity would not impede the ability of exogenous PA to induce EGFR nanocluster formation. To address this hypothesis, we pretreated cells expressing EGFR-GFP with the TK inhibitor AG1478 and then incubated them with PA. Figure 8A shows that stimulation with PA in the presence of AG1478 led to a substantial increase in EGFR nanoclustering. Indeed, the level of EGFR nanocluster formation was similar to that induced by treatment with PA in the absence of AG1478. Taken together, these data suggest that PA is a potent regulator of EGFR spatial organization and acts downstream of RTK activation.
PA production regulates EGFR nanoclustering downstream of RTK activation. (A) BHK cells expressing EGFR-GFP were serum starved and treated with 100 nM AG1478 for 10 min as indicated prior to stimulation with 100 μM PA for 10 min. (B) BHK cells expressing EGFR-GFP in the presence or absence of PLD2 were serum starved and then stimulated with 50 ng/ml EGF for the indicated times. (C) BHK cells expressing EGFR-GFP were serum starved and then stimulated with 50 ng/ml EGF in combination with 100 μM PA for the indicated time. Plasma membrane sheets were generated and labeled with anti-GFP antibody conjugated to 5-nm gold particles. Univariate K functions are weighted means (n ≥ 9) standardized on the 99% CI. Significant differences from the control unstimulated EGF receptor pattern were assessed using bootstrap tests. (A) Treatment with AG1478 led to a significant reduction in EGFR nanoclustering (P = 0.041). However, cotreatment with PA reversed the effect of AG1478 treatment, leading to a level of EGFR nanoclustering that was not significantly different from that obtained by treatment with PA alone (P = 0.961).
Our data suggest that the production of PA via the action of PLD2 promotes the formation of EGFR nanoclusters. We next asked how prolonged production of PA would affect the long-term regulation of EGFR nanocluster formation. Serum-starved BHK cells that expressed EGFR-GFP alone or in the presence of PLD2 were stimulated with EGF. Figure 8B shows that in cells expressing EGFR-GFP alone, EGFR nanoclustering returned to basal levels after 60 min of EGF stimulation. In contrast, overexpression of PLD2 maintained a high level of EGFR nanoclustering after 60 min of EGF stimulation (Fig. 8B). A similar prolongation of EGFR nanoclustering was observed in cells cotreated with EGF and PA (Fig. 8C). These results suggest that the continued production of PA by PLD2 is required to maintain EGFR nanoclusters.
PLD2 modulates signal output in response to EGF.Since EGFR nanocluster formation is sustained in cells overexpressing PLD2, we examined the impact of altering PLD2 activity on EGF-induced MAPK output. Figure 9 shows that ectopic expression of PLD2 changes the time course of ppERK generation in response to EGF. The peak response to EGF was detected earlier, and there was sustained signal output at later time points.
PLD2 modulates EGF-induced signal transduction. BHK cells expressing EGFR-GFP in the presence or absence of PLD2 were serum starved and then stimulated with 50 ng/ml EGF for the indicated times. Whole-cell lysates were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis and Western blotting for ppERK and α-tubulin. A representative blot is shown. Bars represent the mean fold increase in ppERK compared to control unstimulated cells (± SEM; n = 3).
DISCUSSION
The data presented herein show that in quiescent cells a large percentage of EGFRs are organized into nanoclusters that range is size from dimers to higher-order oligomers and have an average radius of ∼24 nm. Our finding that the EGFR is organized into oligomers on the plasma membrane is consistent with previous studies using other high-resolution imaging techniques (7, 16). Here we extend these studies and show for the first time that EGFR nanocluster formation, in response to ligand binding, is driven by receptor activation and downstream second-messenger lipid production by PLD2.
In quiescent cells, the EGFR resides in cholesterol-dependent nanoclusters. However, our data suggest that following ligand binding, the lipid properties of the EGFR nanocluster change. We propose the following model to describe EGF-induced EGFR nanocluster formation. Ligand binding to the EGFR induces dimerization and activation of the TK domain. The TK domain then autophosphorylates a number of tyrosine residues, which act as docking sites for downstream adaptors and effectors. PLD2 is activated by the EGFR, leading to the hydrolysis of plasma membrane PC to produce PA and choline, resulting in the production of a localized patch of membrane enriched in PA. The production of PA may also be facilitated by the conversion of DAG to PA by DGK. The number of receptors within a cluster increases to about nine upon ligand binding. It is possible that PA promotes the recruitment of additional receptors to the nanocluster or, alternatively, stabilizes the interaction of receptors within the nanocluster. The precise molecular mechanism by which PA drives EGFR nanocluster formation is unclear. However, the juxtamembrane domain of the EGFR (residues 645 to 682) contains a cluster of basic residues that bind to acidic lipids in the plasma membrane via electrostatic interactions (20, 35, 37). There is evidence from studies on KcsA, a bacterial potassium channel, that interaction with PA can stabilize protein structure and regulate function (8, 26). Therefore, we speculate that following receptor activation, the basic residues of the juxtamembrane domain interact with newly generated PA, forming a proteo-lipid complex. In this respect, the formation of the EGFR nanocluster shares similarities with the mechanistic properties underlying the formation of Ras nanoclusters (1). Therefore, we propose that these proteo-lipid complexes form the individual units of the EGFR nanocluster.
The organization of EGFR into PA-enriched nanoclusters has important implications for signal transmission. First, a number of studies have demonstrated lateral propagation of receptor activation upon ligand binding (16, 27, 36, 43). These studies suggest that the active EGFR dimer is transient, and following dissociation, the active receptor is able to interact with and activate additional EGFR proteins that are not bound by ligand. Thus, the organization of the EGFR into nanoclusters will facilitate the activation of unliganded receptors within the cluster. Furthermore, diffusion of receptors into and out of clusters will enhance signal propagation across the plasma membrane.
Second, the localized patch of PA will facilitate the recruitment of a number of proteins involved in MAPK pathway activation. Specifically, the guanine nucleotide exchange factor Sos is recruited to the plasma membrane via a PA binding domain (47). Sos stimulates GTP loading of Ras, which recruits Raf-1 from the cytosol to the plasma membrane via interaction with the Ras binding domain of Raf-1, but once Raf-1 is at the plasma membrane, its activity is regulated via interaction with PS and PA (12, 13). Taken together, the data suggest that EGFR activation stimulates the formation of an autonomous lipid-based signaling domain that is designed to selectively recruit a set of specific signaling molecules.
The activation of PLD2 and production of PA in response to EGF may serve yet a third purpose in regulating EGFR function. PLD2 has a role in regulating EGFR endocytosis and signal output via the MAPK pathway (38). The conical shape of PA favors negative membrane curvature, and thus the generation of a patch of PA may promote endocytic vesicle formation and facilitate EGFR endocytosis via the clathrin-mediated endocytic pathway. However, our EM data show that EGFR nanoclusters do not require preexisting clathrin-coated pits in order to assemble. Therefore, we speculate that the formation of clathrin-coated pits may occur as a consequence of EGFR nanocluster formation.
We have demonstrated that EGFR nanoclusters are transient and stimulated by ligand binding. Furthermore, EGFR nanocluster dynamics are critically linked to the duration and magnitude of EGF-induced signal transduction. A key parameter here is PLD2 activity, which is closely linked to EGFR clustering. This is a particularly intriguing observation, since PLD2 is overexpressed in some human tumors (11). Although PLD2 overexpression per se is not sufficient to increase EGFR nanoclustering, following EGF stimulation, overexpression of PLD2 prevents EGFR nanocluster levels from returning to the basal state and increases EGFR signal output. In human tumors, overexpression of PLD2 may therefore lead to increased signal transduction via the Ras/Raf/MEK/ERK pathway, at least in part, by maintaining inappropriately high levels of EGFR nanoclustering.
In summary, we show a novel role for PA in promoting nanoclustering of a transmembrane TK receptor. Generation of PA occurs in the immediate vicinity of the activated EGFR as a result of activation of PLD2, leading to local remodeling of the lipid bilayer. The generation of autonomous lipid based signaling platforms may represent a general principle for ligand-induced signal transduction via cell surface RTKs.
ACKNOWLEDGMENTS
J.F.H. is the present incumbent of the Fondren Chair in cellular signaling. The work presented herein was supported by awards R01GM066717 (J.F.H.) and RO1GM071475 (G.D.) from the National Institute of General Medical Sciences.
We thank Michael Frohman for the gift of the PLD1 and PLD2 constructs.
The content of this report is solely our responsibility and does not necessarily represent the official views of the National Institute of General Medical Sciences or the National Institutes of Health.
FOOTNOTES
- Received 16 December 2009.
- Returned for modification 15 February 2010.
- Accepted 16 May 2010.
- Accepted manuscript posted online 1 June 2010.
- Copyright © 2010 American Society for Microbiology