Phosphatidic Acid Increases Epidermal Growth Factor Receptor Expression by Stabilizing mRNA Decay and by Inhibiting Lysosomal and Proteasomal Degradation of the Internalized Receptor

Abstract

Overexpression of epidermal growth factor receptor (EGFR) is one of the frequent mechanisms implicated in cancer progression, and so is the overexpression of the enzyme phospholipase D (PLD) and its reaction product, phosphatidic acid (PA). However, an understanding of how these signaling molecules interact at the level of gene expression is lacking. Catalytically active PLD enhanced expression of EGFR in human breast cancer cells. Overexpression of the PLD2 isoform increased EGFR mRNA and protein expression. It also negated an EGFR downregulation mediated by small interfering RNA targeting EGFR (siEGFR). Several mechanisms contributed to the alteration in EGFR expression. First was the stabilization of EGFR transcripts as PLD2 delayed mRNA decay, which prolonged their half-lives. Second, RNase enzymatic activity was inhibited by PA. Third, protein stabilization also occurred, as indicated by PLD resistance to cycloheximide-induced EGFR protein degradation. Fourth, PA inhibited lysosomal and proteasomal degradation of internalized EGFR. PLD2 and EGFR colocalized at the cell membrane, and JAK3 phosphorylation at Tyr980/Tyr981 followed receptor endocytosis. Further, the presence of PLD2 increased stabilization of intracellular EGFR in large recycling vesicles at ∼15 min of EGF stimulation. Thus, PLD2-mediated production of PA contributed to the control of EGFR exposure to ligand through a multipronged transcriptional and posttranscriptional program during the out-of-control accumulation of EGFR signaling in cancer cells.

INTRODUCTION

Epidermal growth factor receptor (EGFR) is overexpressed in many epithelial tumors, including bladder, kidney, pancreatic, and squamous cell carcinomas (1). In breast cancer, EGFR overexpression is associated with advanced-stage disease and shortened relapse-free survival, which occurs concomitantly with low estrogen receptor expression (2). The cell signaling events after EGFR ligand stimulation and cancer have been greatly studied and require association of the receptor with a number of cytoplasmic tyrosine kinases, as well as activation of the Janus kinase (JAK)/STAT pathway (3). EGFR directly interacts with phospholipase D2 (PLD2) (4–6). Stimulation of EGFR increases cellular PLD activity and the production of phosphatidic acid (PA) in cancer cell lines (7, 8). PA is also the precursor to lysophosphatidic acid (LPA) that is relevant in ovarian cancer (9). PLD2 activity is regulated by the phosphorylation of the kinases EGFR and Janus kinase 3 (JAK3) on the Y296 and Y415 tyrosine residues, respectively (10, 11). JAK3 is a 130-kDa intracellular, nonreceptor tyrosine kinase (12, 13). It can also function as the docking site for other proteins if they have the proper Src homology 2 (SH2) domains.

Tyrosine kinases associated with EGFR, such as Fer/Fes, promote cell motility in a PLD/PA-dependent pathway (14). We have recently found that Fes binds PA and participates in a PLD-induced pathway of myeloid differentiation (15). PLD2 is associated with EGFR signaling by binding to Grb2 at two specific residues (Y169 and Y179) that leads to activation of the growth factor pathway (16–18). PA interacts with Sos during EGF-induced membrane recruitment and Ras activation (6). Production of PA by PLD2 is essential for ligand-induced EGFR nanocluster formation; these clusters are cholesterol dependent and actin independent and induce mitogen-activated protein kinase signal output (19). Internalization of inactive (neither tyrosine-phosphorylated nor ubiquitinated) EGFR is mimicked by PA micelles, is strongly counteracted by PLD2 silencing, and is mediated by clathrin-dependent and -independent pathways (20). EGFR protein and PA-lipid interactions enable a link between EGFR and the Cbl endocytic complex leading to a fusible membrane (21).

In spite of this knowledge of how EGFR regulates PLD activity at short times after addition of the EGF agonist to the cell, little is known about the long-term effects on gene expression, if present. Further, the converse, how information flows from PLD to EGFR, has been explored only slightly. We report for the first time that PLD-PA caused activation of EGFR gene and protein expression through two distinct mechanisms: inhibition of mRNA decay and inhibition of internalized EGFR degradation by lysosomes and the proteasome. These results represent an innovation in EGFR signaling, and if PLD and PA are considered, then this constitutes a novel target for modulating cancer growth.

MATERIALS AND METHODS

Reagents.

Dulbecco's modified Eagle's medium (DMEM) was from Mediatech (Manassas, VA); Opti-MEM, Lipofectamine Plus reagent, and Lipofectamine 2000 were from Invitrogen (Carlsbad, CA); TransIT-2020 transfection reagent was from Mirus (Madison, WI); primers and 6-carboxyfluorescein (FAM)-labeled probes for quantitative PCR (qPCR) were from Applied Biosystems (Foster City, CA); [³H]butanol was from American Radiolabeled Chemicals (St. Louis, MO); [γ-³²P]ATP was from Perkin-Elmer (Waltham, MA); enhanced chemiluminescence (ECL) reagent was from GE Healthcare (Piscataway, NJ); and EGF was from Peprotech (Rocky Hill, NJ). 5-Fluoro-2-indolyl des-chlorohalopemide (FIPI) was from Cayman Chemical; bortezomib (Velcade) and E-64 were from Sigma-Aldrich. The plasmids used in this experiment were as follows (all carrying human open reading frames [ORFs]): pcDNA3.1-mycPLD2-WT (where WT is wild type), pcDNA3.1-mycPLD2-K758R, pcDNA3.1-mycPLD2-Y415F, and pCMV6-XL4-EGFR. The PLD2 plasmids used were myc tagged, which contributed to a slight increase in the molecular weight of PLD2, as detected using Western blot analyses. MTLn3 cells were a generous gift from Jeffrey E. Segall (Albert Einstein College of Medicine), and MDA-MB-231, MCF-7, and COS-7 cells were obtained from the American Type Culture Collection. COS-7, MCF-7, and MDA-MB-231 cells were cultured in DMEM supplemented with 10% (vol/vol) fetal bovine serum (FBS), while the MTLn3 cells were cultured in alpha minimal essential medium (α-MEM) supplemented with 10% (vol/vol) fetal bovine serum. Anti-JAK3, anti-myc, anti-PLD2, and rabbit anti-EGFR antibodies were from Cell Signaling (Danvers, MA); donkey anti-rabbit tetramethyl rhodamine isothiocyanate (TRITC)-conjugated IgG antibodies were from Santa Cruz (Santa Cruz, CA).

Lipid preparation.

Lipids from Avanti Polar Lipids (Alabaster, AL) were prepared from powder in stock buffer of phosphate-buffered saline (PBS)–0.5% bovine serum albumin (BSA) (50 mg of BSA per 10 ml of 1× PBS), pH = 7.2, with a final concentration of lipids of 1 mM. This solution was sonicated on ice (at the medium setting) once for 4 sand then kept on ice for 4 s; this cycle was repeated two more times, and then the lipids were extruded (Avanti Polar Lipids, Alabaster, AL). Lipids were kept on ice, overlaid with N₂ in the tubes, tightly capped, and stored at 4°C protected from light in a desiccator. An intermediate dilution (10 μM) was prepared on the day of the experiment in Hanks balanced salt solution (HBSS)-HEPES (0.24 g of HEPES/100-ml bottle of HBSS) and 0.5% BSA at pH 7.35. Lipids were added (dropwise) to the cells (30 μl per 1 ml of cells) for a final concentration of 300 nM unless otherwise indicated. Cells were treated with increasing concentrations (as indicated in the figures) of 1,2,-dioleyl phosphatidic acid (DOPA) for 4 to 5 h. Our laboratory has shown previously that this form of PA is cell soluble (22). Posttreatment cells were harvested and subjected to immunoblot analysis.

Gene overexpression and silencing.

The protocol for overexpression involved transfection of plasmid DNAs into COS-7, MDA-MB-231, MTLn3, or MCF-7 cells using TransIT-2020 (Mirus, Madison, WI). Appropriate amounts of DNA were mixed with the transfection reagent in Opti-MEM (Invitrogen) in sterile glass test tubes and incubated for 15 to 30 min at room temperature. The transfection mixture was added to cells in complete medium and incubated for 48 h at 37°C. Silencer small interfering RNAs (siRNAs) used to silence EGFR, PLD2, and JAK3 were from Life Technologies (siRNAs 103549, 12068, and 144522, respectively; Carlsbad, CA) or from Origene (SR30004, Trilencer-27 universal scrambled negative-control siRNA duplex; Rockville, MD). The silencing effect was highly effective for each of the gene-specific siRNAs and varied only ±15% within each target. To initiate transfection, the siRNA was mixed with Opti-MEM and siQuest transfection reagent and incubated at room temperature for 15 min before being added to the cells. TransIT-2020 transfection reagent was used for all other transfections (4 days).

Real-time RT-qPCR.

Total RNA was isolated from COS-7, MDA-MB-231, and MCF-7 cells with an RNeasy minikit (Qiagen, Valencia, CA). RNA concentrations were determined, and equal amounts of RNA (0.5 μg) were used for analyses. Reverse transcription-quantitative PCR (RT-qPCR) was performed according to published technical details (23).

In vitro PLD assay.

PLD activity (transphosphatidylation) in cell sonicates was measured in liposomes of short-chain phosphocholine (PC), 1,2-dioctanoyl-sn-glycero-3-phosphocholine (PC8), and [³H]butanol as previously described (24–26). Approximately 50 μg of cell sonicates was added to microcentrifuge Eppendorf tubes containing the following assay mix (120-μl final volume): 3.5 mM PC8 phospholipid, 1 mM PIP2, 75 mM HEPES, pH 7.9, and 2.3 μCi (4 mM) of [³H]butanol. The mixture was incubated for 20 min at 30°C, and the reaction was stopped by the addition of 300 μl of ice-cold chloroform-methanol (1:2) and 70 μl of 1% perchloric acid. Lipids were extracted and dried for thin-layer chromatography (TLC). TLC lanes that migrated as authentic phosphatidyl butanol (PBut) were scraped, dissolved in 3 ml of ScintiVerse II scintillation mixture, and counted. Background counts (boiled samples) were subtracted from those of experimental samples. For some experiments, liposomes were made with 1,2-dimirystoyl-sn-glycero-3-phosphocholine or 1,2-diarachidonoyl-sn-glycero-3-phosphocholine.

Immunofluorescence microscopy.

MCF-7 cells were transfected with 1.5 μg of the PA sensor and/or 1.5 μg of PLD2-WT or PLD2 with a K758R substitution (PLD2-K758R) and were grown on glass coverslips inside six-well tissue culture plates. At 2 days posttransfection, cells were fixed with 4% paraformaldehyde, permeabilized in 0.1% Triton X-100 in phosphate-buffered saline (PBS), and blocked in 10% fetal calf serum (FCS) in 0.1% Triton X-100 in PBS. For detection of myc-tagged PLD2, cells were incubated with anti-myc-tetramethyl rhodamine isothiocyanate (TRITC) primary antibody conjugate overnight at 4°C. Nuclei were stained using a 1:2,000 dilution of 4,6-diamidino-2-phenylindole (DAPI) in PBS for 5 min at room temperature. Coverslips were mounted onto glass microscope slides using VectaShield mounting medium and imaged using a fluorescein isothiocyanate (FITC) filter for the PA sensor and a TRITC filter for PLD2 (DAPI was also included to visualize the nucleus). Cells were examined on a Nikon Eclipse 50i fluorescence microscope and photographed (Lumenera camera; Lumenera Corp., Ottawa, Ontario, Canada) by consecutive exposures using green, red, and blue filters with a 100× (numerical aperture, 0.9) oil objective. The intensity of green light (∼520 nm) observed in the fluorescence microcopy is correlated to the amount of PA present in the cell.

Actinomycin D and cycloheximide treatments.

MCF-7 cells were cultured according to the procedures outlined above at 37°C in their respective media. Cells were treated with 5 μM cycloheximide- or actinomycin D-containing medium to inhibit global protein synthesis and extracted at time points of 0, 15, 30, 60, and 90 min. EGFR and α-actin protein expression was detected by Western blotting with polyclonal anti-EGFR and antiactin antibodies. The resulting blots were then analyzed by Gel Logic software, and the resulting intensities were plotted using Sigma plot.

Measurement of RNase activity.

An RNase-Alert QC system kit (catalog no. AM1966; Ambion, Austin, TX) was used to detect RNase A activity in cell lysates at a 1:100 dilution as the positive control. RNase activity was measured by fluorometric endpoint (at 15 min) in the presence of single-stranded RNA, as indicated by DiNitto et al. (27) and according to the manufacturer's protocol (28).

Inhibition of protein degradation.

Clathrin-mediated endocytosis (CME) of EGFR was followed by internalization, degradation, and/or recycling of the receptor (29–33). Cells were incubated first with cycloheximide to prevent new synthesis of proteins even in the presence of EGF ligand. Degradation of existing proteins (including EGFR) in lysosomes and the proteasome was inhibited for 30 min in the presence or absence of PA; samples were then incubated with EGF, and images were taken at 7 to 14 min after ligand addition to ensure import and formation of early endosomes. Fluorescence images were analyzed by ImageJ (34–36). Bortezomib (or PS-341; B-[(1R)-3-methyl-1-[[(2S)-1-oxo-3-phenyl-2-[(2-pyrazinyl-carbonyl)amino]propyl]amino]butyl]-boronic acid) is a specific inhibitor of proteasome-mediated intracellular proteolysis of long-lived proteins, with a proteolysis 50% inhibitory concentration (IC₅₀) of <0.1 μM (37, 38). Bortezomib was incubated with cells for 30 min (37, 38). The lysosome inhibitor E-64 [N-(N-(l-3-trans-carboxyoxirane-2-carbonyl)-l-leucyl)agmatine or l-trans-epoxysuccinyl-leucylamide-(4-guanidino)butane] was incubated with cells for 30 min (39, 40). The PA phosphatase inhibitor d-propranolol (100 μM) was also incubated with cells for 30 min (41).

Statistical analysis.

Data are presented as means plus SEMs. The difference between means was assessed by a single-factor analysis of variance (ANOVA) test. A probability value P of <0.05 indicated a significant difference.

RESULTS

PLD and its enzymatic reaction product, PA, upregulated EGFR expression.

To determine the effect of PLD2 on EGFR, we enhanced the levels of PLD2 in cells by transient overexpression of PLD2. Transfection experiments that overexpressed PLD2 resulted in increased EGFR protein mass, as assessed by immunoblotting (Fig. 1A, left group of bands). However, the inverse experiment (i.e., increasing levels of EGFR to investigate if PLD2 mass would change) did not result in any significant effect on PLD2 protein mass (Fig. 1A, right group of bands). A schematic is shown in Fig. 1B indicating the directionality of effects of one molecule on the expression of the other. As EGFR did not cause any change in PLD protein expression, we continued to investigate the effect of PLD2 overexpression on EGFR gene expression. Figure 1C presents evidence by immunofluorescence imaging of cells overexpressing both EGFR and PLD2 that these two proteins colocalized in the cells at all times tested after stimulation with the ligand EGF. Using RNA isolated from cells that overexpressed PLD2-WT, we found that PLD2 overexpression increased the levels of EGFR mRNA transcripts based on qPCR analysis (Fig. 1D). Figure 1E shows that during PLD2 overexpression, the EGFR gene amplification curve was shifted to the left, which suggests that PLD2 expression had a positive effect on EGFR RNA transcripts. Positive controls for PLD2 overexpression are shown in Fig. 1F and G for gene expression and Western blotting, respectively. These results indicate that PLD2 regulated EGFR at the transcriptional level and provide the first evidence that PLD2 positively regulates EGFR transcription. Figure 1H indicates that the endogenous levels of EGFR varied with the type of cell considered, with COS-7 and MCF-7 cell lines having the lowest expression and the aggressive breast cancer cells MDA-MB-231 and MTLn3 having the highest expression. In accordance with this finding, we used COS-7 or MCF-7 cells if the goal of the experiment was to increase the amount of EGFR expression, or we used MDA-MB-231 or MTLn3 cells if the goal of the experiment was to decrease or silence the amount of EGFR expression.

FIG 1.

Effect of PLD on EGFR expression levels. (A) EGFR protein levels were examined after transfection with PLD2. Additionally, PLD2 protein levels were examined after transfection with EGFR. Cells were transfected with increasing concentrations of EGFR, and posttransfection, the effect of EGFR on endogenous PLD2 was examined. (B) Schematic showing the effect of PLD2 on EGFR expression and vice versa. (C) Double transfection (EGFR+PLD2) showing the accumulation of EGFR (TRITC labeled) and PLD2 (FITC labeled) at similar subcellular structures. Shown are representative images of five fields for each condition, with similar results. (D) Cells were transfected with increasing concentrations of PLD2, and EGFR gene expression was measured by qPCR. (E) A representative image retrieved from qPCR showing that in the presence of PLD2, the EGFR cycle number is shifted by 2-fold. DR, Dark Reader. (F and G) Results for controls of gene expression (F) and results for controls of protein expression (G). For the Western blots (WB) shown in panels A and G, experiments were performed for four independent sets in total. (H) Analysis of EGFR basal expression in different cell types. Experiments presented in this figure were performed in triplicate for at least three independent sets in total (n = 9). Results are means ± SEMs and are expressed in terms of gene expression (*, P < 0.05, by ANOVA of increases between sample and control values).

Downregulating EGFR initiated a compensatory mechanism through PLD.

To further confirm that PLD2 had a positive effect on EGFR, MDA-MB-231 cells were silenced with increasing concentrations of siRNA specific for PLD2. Under these conditions, EGFR was downregulated for both gene expression and protein mass (Fig. 2A and B). Silencing controls for PLD2 are shown in Fig. 2C and D. We next silenced EGFR, followed by overexpression of PLD2 in the same sets of cells. Cell lysates were prepared and used for SDS-PAGE, and subsequent Western blot analyses were used to measure protein expression levels of both EGFR and PLD2. Overexpression of wild-type PLD2 was able to reverse/rescue the silencing effect on EGFR protein expression of an EGFR-specific siRNA (Fig. 2E, middle panels) compared to the mock-transfected control (Fig. 2E, siEGFR). This set of experiments indicated that downregulating EGFR provided a signal as a compensatory mechanism that helped drive PLD2, which in turn aided in increasing EGFR expression levels. This is, to our knowledge, the first report indicating such a phenomenon. Interestingly, overexpression of the lipase-inactive mutant of PLD2, PLD2-K758R, failed to rescue EGFR (Fig. 2E, right panels), which highlights the potential importance of the product of the lipase reaction, PA. Additionally, we performed a similar experiment with the cytosolic tyrosine kinase, JAK3, which is directly related to EGFR signaling and activation, and we found a similar trend, whereby PLD2-WT but not PLD2-KR overexpression reversed JAK3 gene silencing (Fig. 2F). Positive controls for EGFR and PLD gene expression are shown in Fig. 2G and H. The differential effects of PLD2-WT and PLD2-K758R mutant overexpression on EGFR gene expression following EGFR silencing shown in the immunoblots (Fig. 2E) were also evidenced in similarly treated samples that were used instead for RNA isolation and subsequent qPCR (Fig. 2G); in these samples PLD2 wild-type overexpression increased the expression of EGFR transcripts, while overexpression of the PLD2-KR mutant did not. These data point to an effect of PLD mediated by its reaction product, phosphatidic acid (PA).

FIG 2.

Silencing and lipase-dead mutant. (A to E) Cells were silenced with two concentrations of small interfering RNA targeting PLD2 (siPLD2), and qPCR analysis was performed to determine the effect of silencing on EGFR (A) or PLD2 (C) on gene expression. In parallel, Western blotting was performed to determine the effect of silencing on EGFR (B) or PLD2 (D) protein mass. Panels E and F present the effect of lipase-inactive PLD2-K758R (PLD2-KR) on EGFR expression after EGFR or JAK3 silencing. Cells were silenced with two different concentrations of small interfering RNA targeting EGFR (siEGFR, E) or JAK3 (siJAK3, F) followed by overexpression of either PLD2-WT or PLD2-K758R. Posttransfection, EGFR levels were examined by Western blotting. For Western blots in panels B, D, E, and F, experiments were performed for at least five independent sets in total (n = 5). (G and H) Quantitative PCR analysis was performed using the same cell conditions to measure mRNA levels of EGFR (G) or PLD2 (positive control of plasmid expression) (G and H). Experiments were performed in triplicate for at least three independent sets in total (n = 9). Results are means ± SEMs and are expressed in terms of gene expression (*, P < 0.05, by ANOVA of increases between sample and control values; #, P < 0.05, by ANOVA of decreases between sample and control values).

PLD2-driven upregulation of EGFR was mediated by PA.

Cells were also pharmacologically inhibited by the small-molecule inhibitor of PLD activity, 5-fluoro-2-indolyl des-chlorohalopemide (FIPI). MDA-MB-231 cells were first mock or PLD2 transfected; next at 48 h posttransfection cells were incubated in the presence of increasing concentrations of FIPI, and then RNA was isolated from these cells. Figure 3A indicates that a large portion of the PLD2-driven effect on EGFR expression was to a large extent dependent on the activity of the lipase. For comparison, Fig. 3C shows the positive-control experiment that measures the negative effect on PLD2 gene expression following FIPI treatment. These data, like the data presented in Fig. 2, suggest that PLD2 or its enzymatic reaction product, PA, might be involved with PLD2's effect on EGFR. Taking into account that PA can also be synthesized in the cell by the phospholipase C (PLC)-diacylglycerol kinase (DAGK) combination, we measured the effect of increasing concentrations of the DAGK inhibitor R23095 on EGFR gene expression in COS-7 cells following PLD-WT overexpression. The results shown in Fig. 3B indicate that a small portion of the PLD2-driven effect on EGFR expression was dependent on the activity of DAGK, while this DAGK inhibitor had no effect on PLD2 gene expression (Fig. 3D). Overall, EGFR expression was affected to a very large extent by PLD. To directly confirm that the PLD2-mediated effect on EGFR was via PA, COS-7 or breast cancer (MCF-7 or MDA-MB-231) cells were treated with increasing concentrations of a cell-permeable PA, 1,2-dioleoyl-sn-glycero-3-phosphate (DOPA) (24, 42), ranging from 0 to 3 μM for 4 h, which resulted in an incremental increase in EGFR protein mass (Fig. 3E to G) in these three different cell lines. This effect was particularly evident in both the COS-7 and MCF-7 cells, which had low basal levels of EGFR expression. Thus, the product of the PLD enzymatic reaction, PA, was able to mimic the positive effect on EGFR expression first seen earlier here as a result of the overexpression of constitutively active PLD2 in cells.

FIG 3.

Effect of PLD inhibitors and direct treatment with PA on EGFR expression. (A and C) Effect of the PLD-inhibitor FIPI on EGFR or PLD2 gene expression. (B and D) Effect of DAGK-inhibitor (DAGKin) R59022 on EGFR or PLD2 gene expression. Positive controls indicating expression of PLD2 after PLD2 plasmid transfection are the three bars on the right in panels C and D. Experiments were performed in triplicate for at least three independent sets in total (n = 9). Results are means ± SEMs and are expressed in terms of gene expression (*, P < 0.05, by ANOVA of increases between sample and control values; #, P < 0.05, by ANOVA of decreases between sample and control values). (E to G) Effect of direct incubation of cells with PA on EGFR protein expression. COS-7, MCF-7, or MDA cells were treated with increasing concentrations of cell-permeable PA (DOPA), and the effect on EGFR levels was examined by Western blotting. Shown are representative blots of (n = 3) for each with similar results.

PLD2 and PA stabilized EGFR mRNA.

After demonstrating that PLD/PA caused a general increase in EGF gene expression, we wondered if another (or a complementary) reason existed that could account for the effect of PLD on EGFR which could possibly manifest itself as an effect on RNA decay. The assumption was that if it did, then this would help explain why mRNA transcripts measured by qPCR were elevated in PA/PLD treatments. To study mRNA decay, we subjected MDA-MB-231 (mock-treated or overexpressing PLD2-WT or PLD2-KR) cells to the transcription inhibitor actinomycin for 0 to 90 min, followed by qPCR or Western blot analyses. In mock-transfected cells, a typical decay curve of actinomycin treatment for EGFR gene expression was observed (Fig. 4A, gray bars). However, the effect of actinomycin was delayed significantly in cells overexpressing PLD2-WT (red bars) but not PLD2-KR (blue bars) (Fig. 4A for gene expression and 4B for protein mass). Densitometry results of the Western blots similar to the ones shown in Fig. 4B are presented in Fig. 4C, and the results from MCF-7 cells are shown in Fig. 4D. Even though MCF-7 cells have a low basal level of mRNA EGFR expression, that low level of EGFR expression was also affected by actinomycin and PLD in the same fashion as in MDA-MB-231 cells.

FIG 4.

Effect of PLD2-WT or PLD2-KR on EGFR mRNA decay ascertained by treating cells with the transcription inhibitor actinomycin. (A) Effect of time of incubation of cells with actinomycin on EGFR gene expression in control or PLD2-WT- or PLD2-K758-transfected cells. Experiments were performed in triplicate for at least three independent sets in total (n = 9). (B) Western blotting analyses were performed to determine the effect of PLD2 on EGFR at the level of transcription in MDA-MB-231 cells. For the Western blots, experiments were performed for four independent sets in total (n = 4). (C and D) Densitometry of blots from an actinomycin chase experiment in MDA-MB-231 or MCF-7 cells after mock treatment or in PLD2-WT- or PLD2-K758-transfected cells. #, P < 0.05, by ANOVA of decreases between sample and control values.

Calculated EGFR half-life (t_1/2) data for Fig. 4 for actinomycin-treated MDA-MB-231 cells were as follows: in mock-transfected cells, t_1/2 of 28 min; PLD2-WT-expressing cells, t_1/2 of 41 min; PLD2-K758-expressing cells, t_1/2 of 25 min. For actinomycin-treated MCF-7 cells, the EGFR half-life values were as follows: in mock-transfected cells, t_1/2 of 29 min; PLD2-WT-expressing cells, t_1/2 of 56 min; PLD2-K758 expressing cells, t_1/2 of 39 min. It can be concluded that the half-life of EGFR transcripts was almost doubled for PLD2-WT-expressing cells (both MDAs and MCF-7s) in comparison with levels in both mock-treated and PLD2-K758R-expressing cells. These results indicate that the overexpression of catalytically active PLD2 that can produce PA delayed and/or prevented mRNA degradation and/or stabilized mRNA decay.

PA increased EGFR transcripts through a mechanism involving inhibition of RNase A.

We next hypothesized that PA/PLD delayed mRNA degradation because PA affected RNase activity. RNases are crucial for maintaining a cellular pool of matured RNA molecules. For example, RNase A (RNase A family, 1 enzyme) is a well-studied, 13.7-kDa protein that cleaves internal phosphodiester RNA bonds of single-stranded RNAs on the 3′-side of pyrimidine bases (43). It prefers unpaired poly(C) as a substrate and hydrolyzes RNA through a 2′,3′-cyclic nucleotide intermediate (43). We measured RNase activity in cell lysates either from cells that overexpressed PLD (for 48 h) or from cells that were directly incubated with cell-permeable PA for 4 h. Figure 5A shows that RNase activity was inhibited with PLD2 or PA in all cases considered. The results of downregulating intracellular PA by treating cells with FIPI (a dual PLD1/PLD2 inhibitor) or NFOT (N-[2-[1-(3-fluorophenyl)-4-oxo-1,3,-8-triazaspiro[4.5]dec-8-yl]ethyl]-2-naphthalenecarboxamide; a PLD2-specific inhibitor) are presented in Fig. 5B. A significant effect on both EGFR gene expression and RNase activity was observed, albeit inversely proportionally. Whereas PA inhibition led to a significant decrease in EGFR expression, RNase activity was significantly increased, providing further proof that PA increased EGFR gene expression through a mechanism involving inhibition of RNase A activity and a resulting increase in mRNA stability. Other signaling molecules that are associated with EGFR signaling (Rac and JAK3) were targeted by FIPI (Fig. 5C), whereas mTOR and CSK, more distant to EGFR signaling, were not affected by the PLD inhibitor FIPI.

FIG 5.

Effects of PLD inhibitors on RNase activity. (A) Either MCF-7 or MDA-MB-231 cells were transfected with PLD2 plasmids or incubated for 4 h with DOPA. Cell extracts were prepared and assayed for RNase activity in vitro. RU, relative units. (B) Effect of the PLD inhibitors FIPI (dual PLD1/PLD2 inhibitor) and NFOT (specific PLD2 inhibitor) on EGFR gene expression and on RNase activity. (C) Effect of the PLD inhibitor FIPI on the expression of other signaling molecules. #, P < 0.05, by ANOVA of decreases between sample and control values.

PLD2 and its product, PA, increased EGFR protein mass by inhibiting protein degradation.

Data presented in Fig. 4 and 5 demonstrated that PLD/PA upregulated gene expression. We next asked if PLD/PA would also have a posttranslational, direct effect on protein mass, which could also contribute to the accumulation of EGFR observed in Western blots from the experiment(s) shown in Fig. 1A (for PLD) or Fig. 3E to G (for PA). The first series of experiments designed to investigate this matter are presented in Fig. 6. To study protein stability, we subjected cells (mock-treated or overexpressing PLD2-WT or PLD2-KR) to the translational inhibitor cycloheximide for 0 to 90 min, followed by qPCR and Western blot analyses. In control cells, a typical decay pattern of EGFR protein was found in both untransfected (Fig. 6A) and mock-transfected cells (Fig. 6B, left panels). However, in cells overexpressing PLD2-WT but not PLD2-KR, the effect of cycloheximide on EGFR protein mass was significantly delayed. Calculated EGFR half-lives were as follows: in mock-transfected cells, t_1/2 of 31 min; PLD2-WT-expressing cells, t_1/2 of 47 min; PLD2-K758-expressing cells, t_1/2 of 34 min. Further, data in both Fig. 6B and C show the same trend in the effect of PLD2-WT protein overexpression on EGFR protein mass in two different breast cancer cell lines MDA-MB-231 and MCF-7. Results shown in Fig. 6E indicate that cycloheximide treatment did not affect EGFR gene expression (except at times >60 min, and even then, the inhibition was very small). Therefore, posttranslational events contributed to the positive effect of PLD2 on EGFR protein expression, in addition to the transcriptional effects already demonstrated in Fig. 2 to 4.

FIG 6.

Effect of PLD2-WT or PLD2-KR on EGFR protein mass stability. (A to D) Treatment with the translation inhibitor cycloheximide was used for the study of EGFR protein decay. (B) Western blotting analyses from MDA-MB-231 cell lysates were performed to determine the effect of PLD2 on EGFR at the level of translation. (C and D) Densitometry was performed of blots from a cycloheximide chase experiment in MDA-MB-231 (C) or MCF-7 (D) cells after mock treatment or in PLD2-WT- or PLD2-K758-transfected cells. (E) The effect of time of incubation of MDA-MB-231 cells with cycloheximide on EGFR gene expression in control (Mock) or PLD2-WT- or PLD2-K758-transfected cells was analyzed. Experiments were performed in triplicate for three independent sets in total (n = 9). (F) Inhibition of EGFR protein proteolysis is rescued by phosphatidic acid. (G to I) MDA-MB-231 cells were incubated with the proteasome inhibitor E-64 (1 μM), with the lysosome inhibitor bortezomib (100 nM), or with the PA phosphatase inhibitor propranolol (100 μM) in the absence or presence of 300 nM PA for 4 h at 37°C. After this time, 5 μM cycloheximide was added to all conditions to inhibit new protein synthesis. Time zero of the x axis represents the time of addition of cycloheximide. Samples were harvested at the times indicated, and lysates were prepared. EGFR protein mass was measured after Western blotting of samples taken under the indicated conditions. The insets in panels F to I show Western blot analyses measuring the effect of the lipid PA or various proteasome or PA phosphatase inhibitors. For all Western blots in panels A, B, and F to I, experiments were performed for three independent sets in total (n = 3). #, P < 0.05, by ANOVA of decreases between sample and control values.

We hypothesized that the reason for this prolonged half-life of EGFR mass or stabilization of EGFR protein was due to an alteration of protein degradation after activation by its ligand, EGF. This is based on the fact that EGFR is known to be internalized after EGF stimulation in a clathrin-mediated mechanism and part of it is degraded intracellularly by lysosomes or by the proteasome (31). To analyze EGFR protein degradation by lysosomes, proteasomes, or PA phosphatases, we used three different inhibitors specific to lysosomes, proteasomes, or PA phosphatases in the absence or presence of PA. First, we used E-64, an irreversible inhibitor of cysteine proteinases that inhibits degradation of autophagic cargo in autophagolysosomes (39, 40). Second, for inhibition of the proteasome, we used bortezomib, a specific inhibitor of proteasome-mediated intracellular proteolysis of long-lived proteins (37, 44). Third, we used the PA phosphatase inhibitor d-propranolol, which induces empty or inactive EGF to be internalized and renders the receptor inaccessible to external stimuli (41). As indicated by the data shown in Fig. 6F, treatment of cells with cycloheximide alone (dashed line) yielded a typical curve of protein decay from 0 to 90 min as no new protein was synthesized. More interestingly, the negative effect of cycloheximide treatment on EGFR protein mass was partially abrogated by the presence of PA (solid line). Under these same experimental conditions, the lysosome inhibitor E-64 alone (Fig. 6G, dashed line) and the proteasome inhibitor bortezomib alone (Fig. 6H, dashed line) prevented the initial sharp loss of protein and stabilized protein decay compared to results with the control samples shown in Fig. 6F. The combination of E-64 or bortezomib with PA further contributed to increased EGFR protein mass (Fig. 6G and H, solid lines). Additionally, the negative effect of the PA phosphatase inhibitor d-propranolol (dashed line) on EGFR protein mass shown in Fig. 6I was reversed in the presence of PA (solid line). These data indicate that PA had an inhibitory effect on protein degradation that occurred at the level of lysosomes and the proteasome, resulting in an overall stabilization of EGFR protein mass.

PLD localized with EGFR during and after EGF stimulation of cells.

We further hypothesized that the prolonged half-life of EGFR mass might also be due to changes in localization of EGFR after activation by its ligand, EGF. This is based on the fact that it is already known that PLD/PA regulates membrane and vesicular trafficking in cells. A tight colocalization of PLD2 and EGFR was found earlier in cells in vivo by immunofluorescence microscopy (Fig. 1C). Figure 7A shows a pattern of endogenous EGFR around the cell periphery that was found intracellularly upon incubation of cells with EGF following its internalization at 7 min (Fig. 7A). At 15 min of EGF treatment, most labeling was found in larger organelles. A similar pattern was observed in cells overexpressing EGFR (Fig. 7B), with the main difference being that the amount of overall immunofluorescence was larger. This positive effect on EGFR immunofluorescence was even more pronounced in cells that overexpressed PLD2-WT (Fig. 7C), an effect that is even more evident in the zoomed-in images presented in Fig. 7D. Thus, EGFR immunofluorescence was increased in the presence of overexpressed PLD2 after EGF stimulation, which also suggests that PLD2 or its reaction product, PA, mediated this positive effect on EGFR accumulation in the cell.

FIG 7.

Intracellular colocalization of PLD and EGFR during endocytosis. (A) Endogenous levels of EGFR in MTLn3 cells mock transfected after the indicated times of incubation with 3 nM EGFR. (B and C) MTLn3 cells were transfected with EGFR (B) or PLD2 (C) and stimulated with 10 nM EGF for 0 to 15 min. EGFR- or PLD2-detected fluorescence corresponds to FITC labeling in each case. (D to F) Further magnification of selected images shown in panels A to C (indicated by *, **, or *** in the original panels) to indicate the level of endocytic recycling. A ring around the cellular membrane contains most of the staining under basal conditions; small vesicles are apparent in cells not expressing PLD2, whereas a larger fluorescent signal was evident in much larger organelles/vesicles (>1 μm) in PLD2-overexpressing cells at 15 min of stimulation with EGF. O/E, overexpressed.

The prolonged half-life of EGFR mass following PLD2 overexpression was also evidenced by changes in the fate of a cell signaling protein that is directly associated with EGFR biology, JAK3, which we observed earlier was affected by PLD overexpression (Fig. 2F) as well as by inhibition of PLD using the small-molecule inhibitor FIPI (Fig. 5C). Upon transphosphorylation of EGFR induced by EGF binding, one of the many proteins that are phosphorylated by the activated receptor is JAK3. Therefore, we followed JAK3 phosphorylation at Tyr980/981 as a readout of EGFR activation in MTLn3 cells. Figure 8 indicates that the low level of phospho-JAK3 seen in mock-treated cells (Fig. 8A) was substantially elevated in PLD2-WT-transfected cells (Fig. 8B) but not in cells transfected with a JAK3-deficient binding mutant of PLD2, PLD2-Y415F (Fig. 8C) (10). These data indicate that EGF-mediated phosphorylation of JAK3 was also affected by PLD2 overexpression, which was also one more readout of the level of EGFR activation and subsequent downstream signaling.

FIG 8.

Phospho-JAK3 (Tyr980/981) as a readout of EGFR activation. (A) Endogenous levels of phospho-JAK3 (pJAK3) in mock-transfected MTLn3 cells. (B and C) MTLn3 cells were transfected with PLD2-WT or with the PLD2-Y415F mutant (which abrogates JAK3 phosphorylation of PLD2 at Y415) and stimulated with EGF for 0 to 15 min. Phospho-JAK3 protein was detected using FITC labeling of anti-phospho-JAK3 antibody. Shown are representative images of five fields for each condition, with similar results. O/E, overexpressed.

DISCUSSION

We report here for the first time that PLD2 via its lipase product PA affected EGFR expression directly and demonstrate the specific molecular mechanisms involved. PLD2-WT expression upregulated EGFR and JAK3 expression even when both kinases were silenced using siRNAs. We have shown in the past that when PLD2-WT was transfected in the cell, EGFR was able to phosphorylate PLD2-WT (10, 11). As PLD2 must be phosphorylated by an upstream tyrosine kinase, which contributes to PLD generating its reaction product, PA, we believe that JAK3 fulfills this requirement (Fig. 8) (10, 11) and that this phosphorylated PLD2 would then be able to affect the downstream expression of EGFR. This created an activation loop where the phosphorylation of PLD2 and subsequent PA production caused EGFR protein expression to be enhanced. The multipronged effects of PLD/PA could be summarized as follows: (i) from the point of view of gene expression, PA is a mitogen that increased EGFR gene transcription, and PA stabilized mRNA by preventing RNase cleavage of new transcripts and (ii) from the point of view of protein expression, PA impeded full degradation of endocytosed EGFR by lysosomes and the proteasome. The combined effect of this multipronged mechanism was a sustained increase in EGFR gene and protein mass in the cell when EGFR was in the presence of PA or PLD. In summary, PLD and PA had profound effects on EGFR protein stabilization and localization, by inhibiting protein degradation by lysosomes and the proteasome, in addition to the stabilization of gene expression. All of these actions combined explain why an increase in EGFR protein mass was observed with PA/PLD in the experiments shown in Fig. 1 to 3.

This novel upregulation of EGFR by intervention of PLD/PA has the potential of making breast cancer cells even more reliant on EGFR signaling. Many types of breast cancer cells are very responsive to epidermal growth factor (EGF) as a chemoattractant, including rat MTLn3 cells and human MDA-MB-231 and MCF-7 cells (45). The number of EGF receptors per cell is the highest in MDA-MB-231 cells and is found to be ∼700,000. The second largest is MTLn3 cells, followed by MCF-7 cells, with the lowest EGFR level at ∼3,000 to 6,000 per cell (46–48). MDA-MB-231 and MTLn3 are two forms of highly aggressive breast cancer cells, whereas MCF-7 cells are a low-aggressive form of breast cancer that has low levels of PLD activity (10, 49). High levels of PLD2 in these cell lines make them somewhat more mobile but still less invasive than MDA-MB-231 breast cancer cells. At any event, PLD and PA would in all cases activate EGFR signaling by increasing the protein mass of available receptor.

We propose a model (Fig. 9) that explains the upregulation of EGFR by PA/PLD2 that includes transcriptional effects (namely, inhibition of mRNA decay by RNase activity), posttranscriptional effects (namely, protein stabilization), and increased EGFR levels in the cell: EGF upon binding to its receptor promoted PLD activity and the formation of PA (Fig. 9, step a); EGFR was internalized along with PA existing in the cellular membrane (step b); part of this internalized EGFR (in early endosomes) was sorted out and degraded or recycled (step c). Lysosomes and the proteasome were the main ways by which degradation occurred. Our data indicated that this protein degradation was less likely to occur in the presence of PA. PA in the nucleus acted as a mitogen, enhancing the transcription of the EGFR gene (step e). mRNA transcripts (step f) eventually were translated into new receptor proteins (step g). Other transcripts were basally degraded; mRNA degradation normally occurred via RNase A (step h). But PLD and PA also acted as inhibitors of RNase A activity, stabilizing the presence of transcripts. Note that PA had a dual function: in some cases it activated (endocytosis and nuclear expression) and in some other cases it inhibited (RNase degradation of mRNA transcripts and degradation of protein by the proteasome and lysosomes) EGFR expression, but the final outcome was to facilitate at any given point the amount of gene and/or protein expression.

FIG 9.

Model of how PLD/PA upregulates EGFR expression at the transcriptional and posttranscriptional levels. The key events (steps a to i) are explained in the text (see Discussion). PA has a dual function, in some cases activating (CME, recycling, and nuclear expression) and in others inhibiting (RNase degradation of mRNA transcripts and degradation of protein by the proteasome and lysosomes) EGFR expression, but the final outcome is to facilitate at any given point the amount of gene and/or protein expression.

Thus, PLD2 appears to be one of the key players in the signaling pathway leading to EGFR upregulation in cancer cells. The current study was performed in breast cancer cell lines, suggesting a key role of PLD2 in maintaining the levels of EGFR and its possible use as a target. Patients with lung cancer with known EGFR mutations who are treated with EGFR inhibitors sometimes present an initial response that is encouraging, but eventually they relapse after developing resistance (14, 50). Therefore, identifying new targets for molecular-targeted therapies and developing combinational therapies may lead to improved cancer patient outcomes.