Platelet-Derived Growth Factor Receptor Association with Na⁺/H⁺ Exchanger Regulatory Factor Potentiates Receptor Activity

Fusion protein preparation and overlays.

Hexahistidine- and S-tagged NHERF fusion proteins, for both full-length NHERF and various NHERF truncations, were created via insertion of PCR products derived from a rabbit NHERF cDNA into pET-30A (Novagen), followed by expression and purification. β₂AR-CT (last 80 amino acids of the human β₂AR) as well as PDGFR-CT (last 45 amino acids of human PDGFR-β) were expressed as glutathione S-transferase (GST) fusion proteins. The GST fusion proteins were expressed using the pGEX-2TK vector (Pharmacia), which places a consensus protein kinase A (PKA) phosphorylation site at the start of each fusion protein. This site was radiolabeled with PKA (Calbiochem) and [³²P]ATP (DuPont-NEN) for experiments where PDGFR-CT–GST was labeled with ³²P, cleaved from the GST with thrombin (Novagen), and then overlaid onto unlabeled PDGFR-CT–GST in the presence and absence of NHERF fusion proteins. For all other experiments, however, the GST fusion proteins were not radiolabeled.

NHERF binding to the receptor-CT–GST fusion proteins was assayed via a far-Western blot overlay technique. The GST fusion proteins (5 μg per lane) were run on sodium dodecyl sulfate (SDS)–4 to 20% polyacrylamide gels (Novex, San Diego, Calif.), blotted, and overlaid with NHERF in 2% milk and 0.1% Tween 20 in phosphate-buffered saline (PBS) (blot buffer) for 1 h at room temperature. The blots were then washed three times with blot buffer, incubated for 1 h at room temperature with a horseradish peroxidase-conjugated anti-S-tag antibody (Novagen) in blot buffer, washed three more times with blot buffer, and visualized via enhanced chemiluminescence. To perform saturation-binding curves, equal amounts of PDGFR-CT–GST and control GST were loaded into multiple alternating lanes on SDS–4 to 20% polyacrylamide gels, and the resultant blots were cut into two-lane strips. The strips were incubated with increasing concentrations of NHERF(1-151) fusion protein, and the amount of binding for each strip was quantified via scanning laser densitometry; results are shown as means ± standard errors of the means (SEM). The specific binding of NHERF(1-151) to PDGFR-CT–GST was defined as the total amount of binding to PDGFR-CT–GST minus the binding of NHERF(1-151) to control GST on the same strip of blot. Binding of PDGFR-CT to NHERF-2, also known as SIP-1, E3KARP, and TKA-1 (18, 37, 51), was also examined via the overlay assay; NHERF-2 fusion proteins were prepared as previously described (18).

Cell culture and transfection.

CHO cells were maintained in Ham's F-12 medium, and COS-7 cells were maintained in Dulbecco's modified Eagle medium (GibcoBRL). Both media were supplemented with 10% fetal calf serum and 1% penicillin (10 U/ml)-streptomycin (10 μg/ml) at 37°C in a humidified 5% CO₂ atmosphere. Transient transfections of CHO cells with hemagglutinin epitope (HA)-tagged rabbit NHERF cDNAs or Flag-tagged human β₂AR, or of COS-7 cells with human PDGFR-β cDNA, were performed using 5 μl of Lipofectamine per μg of cDNA transfected. Transfections were performed over 2 h in 100-mm-diameter plates with a total of 10 μg of cDNA transfected per plate. All cells were serum starved for 12 to 16 h before experimentation by incubation in either Ham's F-12 medium or Dulbecco's modified Eagle medium supplemented with 0.5% bovine serum albumin, 0.5% penicillin-streptomycin, and 10 mM HEPES. All assays described were performed 48 h posttransfection. The level of PDGFR expression was quantified via binding studies with radiolabeled PDGF, following a protocol described previously (41). For the majority of experiments described here, the amount of DNA used in transfection was 4 μg of full-length NHERF cDNA per 100-mm-diameter plate of cells (referred to as “moderate” overexpression of NHERF in Results). For the NHERF truncation constructs NHERF(1-151) and NHERF(1-121), 9 μg of cDNA was transfected per 100-mm-diameter plate, as expression of these truncated proteins was somewhat lower than for full-length NHERF per unit of DNA transfected.

Some experiments, as indicated in the text, were performed on AP1-CHO cells stably transfected with either wild-type (WT) β₂AR or mutant L413A β₂AR or and PKA⁻ β₂AR in the vector pBK-CMV. The stable transfections were performed in the same way as the transient transfections, but the cells were then selected for 2 to 3 weeks by the addition of G418 (1 mg/ml) to the F-12–calf serum medium. Cellular expression of WT and mutant β₂AR was confirmed by radioligand binding of ¹²⁵I-cyanopindolol. The relative levels of expression of the WT, L413A, and PKA⁻ β₂AR variants were comparable for the three stable cell lines (0.50, 0.55, and 0.34 pmol of receptor per mg of protein, respectively).

Phospho-ERK assay.

Serum-starved cells were either unstimulated or treated with PDGF-BB homodimer (Calbiochem) at 37°C. To generate the PDGF-BB dose-response series, the exposure time for each dose was 5 min. For single-dose agonist stimulations, 200 pM PDGF-BB was applied for 5 min. Agonist-stimulated cell monolayers were lysed in Laemmli sample buffer, lysates resolved by SDS-polyacrylamide gel electrophoresis (PAGE) and transferred to polyvinylidene difluoride (PVDF) membrane (NEN Life Sciences) by electroblotting. Activated extracellular signal-regulated kinase (ERK) was detected using anti-phospho-ERK antibodies (New England Biolabs), visualized by enhanced chemiluminescence (Amersham), and quantified by scanning laser densitometry. Blots were then stripped and reprobed with an anti-ERK1/2 antibody (Upstate Biotechnology, Inc.) in order to normalize the amount of phosphorylated ERK to the amount of total ERK.

PDGFR immunoprecipitation and covalent cross-linking.

After either no stimulation or stimulation with 200 pM PDGF-BB, cell monolayers in 100-mm-diameter plates were washed at 4°C with PBS followed by lysis in radioimmunoprecipitation assay (RIPA) buffer. To covalently cross-link extracellular domains of the PDGFR-β, 2 mM BS³ (Pierce Laboratories) was applied to the washed monolayers in PBS supplemented with 10 mM HEPES and slowly agitated at 4°C for 30 min. Cross-linking was terminated by aspiration of the BS³-containing PBS solution followed by two washes of the monolayers with PBS containing 0.1 M Tris. After such cross-linking, monolayers were lysed with RIPA buffer as described previously (32). For cross-linking or non-cross-linking experiments, cell lysates were then clarified by centrifugation and normalized for protein content. Immunoprecipitation was performed with 5 μg of an anti-PDGFR-β rabbit polyclonal antibody (Upstate Biotechnology) plus 50 μl of a 50% slurry of protein G-plus-protein A–agarose (Calbiochem), which was agitated for 4 h at 4°C. Immunocomplexes were washed twice with ice-cold RIPA buffer, washed once with ice-cold PBS, denatured in Laemmli sample buffer, resolved by SDS-PAGE, and then transferred to a PVDF membrane. Tyrosine phosphorylation of the PDGFR immunocomplexes was detected using a horseradish peroxidase-conjugated antiphosphotyrosine monoclonal antibody, PY20H (Signal Transduction Laboratories). Nonspecific antibody interaction with the membrane was prevented by incubation of the PVDF membrane in 4% bovine serum albumin–Tris-buffered saline (0.05% Tween 20, 0.05% NP-40, 150 mM NaCl, 10 mM Tris). Immunocomplexes on PVDF membranes were visualized by enhanced chemiluminescence and quantified by scanning laser densitometry. Blots were then stripped and reprobed with the anti-PDGFR antibody in order to normalize the amount of phosphorylated receptor to the amount of total receptor.

Coimmunoprecipitation of PDGFR and NHERF.

CHO cells transfected with 4 μg of cDNA encoding WT PDGFR-β or both the PDGFR and 2 μg of HA-tagged NHERF were stimulated with 200 pM PDGF for 5 min as previously described, lysed, and clarified by centrifugation. HA-NHERF was immunoprecipitated by addition of a 20:1 dilution of HA affinity matrix slurry (BAbCo) with agitation at 4°C for 4 h. Anti-HA immunocomplexes were washed twice in RIPA lysis buffer and once with ice-cold PBS. Whole-cell lysates and HA affinity matrix immunocomplexes were transferred to PVDF membranes for immunodetection of HA-NHERF and the PDGFR. The PDGFR was detected with a 1:1,000 dilution of rabbit anti-PDGFR polyclonal antibody (Upstate Biotechnology) and a 1:7,000 dilution of a horseradish peroxidase-conjugated anti-rabbit polyclonal antibody (Jackson Immunochemicals). HA-tagged NHERF was detected with a 1:4,000 dilution of mouse anti-HA monoclonal antibody 12CA5 (Boehringer Mannheim) and a 1:5,000 dilution of a horseradish peroxidase-conjugated anti-mouse polyclonal antibody (Jackson Immunochemicals).

Creation of L1106A PDGFR-β full-length and CT fusion protein.

WT human PDGFR-β cDNA in the pUC13 vector was the generous gift of A. Kazlauskas. The PDGFR-β cDNA was subcloned into the pBK-CMV vector (Stratagene) using EcoRI and XbaI restriction sites. The L1106A mutant was created using custom-designed oligonucleotides containing the desired mutation in the last codon (CTG to GCG) with flanking NcoI and XbaI sites. The 380-bp fragment was gel purified, restriction enzyme digested, and then subcloned into the pUC13 vector containing the WT PDGFR-β cDNA, which was digested with NcoI and XbaI. The full-length mutant gene was subcloned into pBK-CMV using the EcoRI and XbaI sites. The single point mutation was confirmed by ABI sequencing. When either WT or L1106A PDGFR was overexpressed in COS-7 cells, cells were stimulated with 40 pM PDGF-BB (fivefold lower than for experiments done with endogenous receptors in CHO cells) to avoid reaching maximum levels of receptor autophosphorylation or ERK activation. To create a GST fusion protein corresponding to the CT of L1106A PDGFR, oligonucleotide primers bracketing the final 135 nucleotides of the mutant receptor and containing a 5′ BamHI site and a 3′ EcoRI site were used in a 25-cycle PCR using the full-length mutant receptor as template. The PCR product was gel purified, enzyme digested, and subcloned into the expression vector pGEX-2T, using the BamHI and EcoRI restriction sites. The point mutation was confirmed by ABI sequencing.

PDGFR-CT binds NHERF with high affinity.

The CTs of both the α and β forms of the PDGFR end in DSFL, a sequence consistent with the D(S/T)XL motif preferred for binding by NHERF PDZ domains (17, 18, 47). It has therefore been proposed (18) that the PDGFR-CT might bind with high affinity to NHERF. This possibility was tested and confirmed in the blot overlay experiments shown in Fig. 1A.

NHERF(1-151), a truncated version of full-length NHERF that encompasses the first PDZ domain of NHERF, has previously been shown to bind the β₂AR-CT (17, 18). NHERF(1-151) binds as well to immobilized PDGFR-CT (the carboxyl-terminal 45 amino acids of the PDGFR-β, expressed as a GST fusion protein) as to immobilized β₂AR-CT. In the reverse overlay experiment, PDGFR-CT binds avidly to both immobilized full-length NHERF and immobilized NHERF(1-151) but only weakly to immobilized NHERF(152-358), which encompasses the second NHERF PDZ domain (Fig. 1B). The estimated K_D for the interaction of NHERF(1-151) with PDGFR-CT is 26 nM (Fig. 1C), a value similar to that estimated previously for the affinity of the interaction of NHERF(1-151) with β₂AR-CT (K_D = 18 nM) (18). The affinity of the second NHERF PDZ domain for PDGFR-CT is too low to accurately measure in saturation binding overlay studies. The interaction between NHERF and the PDGFR was also assessed in CHO cells expressing WT PDGFR and HA-tagged NHERF. PDGFR was detected in anti-HA-NHERF immunoprecipitates in an agonist-independent manner (Fig. 1D).

NHERF potentiates cellular PDGFR activity.

We next examined whether NHERF might alter PDGFR signaling in cells. As shown in Fig. 2A, transient overexpression of HA-tagged NHERF in CHO cells led to a potentiation of ERK phosphorylation induced by PDGF stimulation of endogenous PDGFR. This potentiation was not due to increased levels of cellular PDGFR, as assessed by radiolabeled ligand binding studies (data not shown). The magnitude of the potentiation was highly sensitive to the level of cellular HA-NHERF overexpression: excessive levels of NHERF overexpression failed to yield the potentiation of PDGF-induced ERK phosphorylation observed at more moderate levels (Fig. 2B).

Moderate overexpression of NHERF not only enhanced PDGF-induced ERK activation but also potentiated agonist-induced autophosphorylation of endogenous PDGF receptors (Fig. 2C). This potentiation was similar in magnitude to the NHERF-induced potentiation of PDGF-stimulated ERK phosphorylation and, like the potentiation of ERK phosphorylation, was blunted at the highest levels of NHERF overexpression (data not shown). As an additional measure of the effect of NHERF overexpression on PDGFR function, both basal and PDGF-stimulated PDGFR dimerization were assessed via nonhydrolyzable receptor cross-linking. As was seen for both PDGF-induced receptor autophosphorylation and subsequent ERK activation, both basal and PDGF-induced levels of PDGFR dimerization were promoted severalfold by moderate overexpression of NHERF (Fig. 2D).

The cell lines used in this study, CHO and COS-7, contain significant levels of endogenous NHERF (17). To test the possibility that PDGFR activity in these cells is potentiated by endogenous NHERF, we prepared a point mutant of the PDGFR, L1106A, in which the terminal leucine of the PDGFR polypeptide was changed to alanine. A similar mutation of β₂AR-CT, L413A, abolishes NHERF binding (17). The L1106A point mutation in the PDGFR carboxyl-terminal tail prevented NHERF binding in vitro when the mutant PDGFR-CT was expressed as a GST fusion protein (Fig. 3A). The full-length L1106A mutant and WT PDGFR were next expressed separately in COS-7 cells, which contain significantly lower levels of endogenous PDGFR than CHO cells (data not shown). In this context, L1106A PDGFR was impaired relative to the similarly overexpressed WT PDGFR with respect to PDGF-induced receptor autophosphorylation (Fig. 3B) and downstream activation of ERK (Fig. 3C).

NHERF oligomerizes when bound to PDGFR-CT.

We next explored the mechanism by which NHERF might potentiate PDGFR activation and signaling events. We hypothesized that NHERF might act as an adapter protein to facilitate a closer spatial association of two PDGFR monomers, thereby reducing the energy required to dimerize the PDGFR and activate its intrinsic kinase activity. This hypothesis was investigated by examining the ability of NHERF to facilitate the physical association of PDGFR-CTs in blot overlay studies. Radiolabeled PDGFR-CT did not detectably bind to nitrocellulose-immobilized PDGFR-CT in overlay assays, although it bound very well to immobilized NHERF in the same experiments (Fig. 4A and B). When these studies were repeated in the presence of a molar excess of NHERF in the overlay solution, radiolabeled PDGFR-CT bound to the immobilized PDGFR-CT (Fig. 4C). NHERF(1-151), the fusion protein encompassing the first PDZ domain, was also able to bridge radiolabeled PDGFR-CT to immobilized PDGFR-CT (data not shown). This suggests that the second PDZ domain, which as shown earlier binds only very weakly to the PDGFR-CT, is not required for NHERF to facilitate the association of PDGFR-CTs. Since crystallographic studies of peptides complexed with PDZ domains suggest that it is unlikely that a single PDZ domain can bind multiple carboxyl-terminal tails (6, 13), the ability of NHERF(1-151) to cross-link PDGFR-CTs in the overlay studies was somewhat surprising.

Since some individual PDZ domains can oligomerize (3, 4, 23, 44, 50), we examined the possibility that NHERF could oligomerize to facilitate PDGFR-CT association. Initial experiments revealed that overlaid NHERF did not bind detectably to immobilized NHERF in overlay assays, although it bound very well to immobilized PDGFR-CT (Fig. 4D). To explore whether binding of PDGFR-CT to NHERF might enhance NHERF-NHERF association, we repeated the NHERF overlay experiments in the presence of a molar excess of PDGFR-CT in the overlay solution. Under these conditions, overlaid NHERF bound very well to the immobilized NHERF (Fig. 4E).

NHERF(1-121) can act as a dominant negative of cellular PDGFR activity.

The findings described above demonstrate that NHERF can oligomerize, but that it does so efficiently only when one of its PDZ domains is actively engaged with a carboxyl-terminal ligand such as the PDGFR-CT. To further characterize the phenomenon of CT-induced NHERF oligomerization, truncations of NHERF were examined for the ability to bind each other. Immobilized full-length NHERF, NHERF(1-151), and NHERF(152-358) all bound to NHERF(1-151) overlaid in the presence of a molar excess of PDGFR-CT (Fig. 5A). Further successive truncations of NHERF(1-151) shed light on the determinants of NHERF PDZ domain oligomerization. Truncation of 30 amino acids from the amino-terminal side resulted in loss of PDGFR-CT binding as well as loss of NHERF-NHERF oligomerization, consistent with previous reports of the importance of the amino-terminal portions of PDZ domains in binding to carboxyl-terminal ligands (4, 50). Removal of 30 or 50 residues from the carboxyl-terminal end, resulting in NHERF(1-121) and NHERF(1-101), respectively, eliminated NHERF-NHERF oligomerization without altering PDGFR-CT binding. Removal of 70 amino acids, resulting in NHERF(1-81), prevented both PDGFR-CT binding and NHERF-NHERF oligomerization.

The in vitro association data with the truncated mutants suggested that NHERF(1-121) might be a useful reagent for testing the hypothesis that NHERF potentiation of PDGFR activity in cells depends in part on CT-induced NHERF oligomerization, since NHERF(1-121) would bind to the same intracellular partners as endogenous NHERF without being able to facilitate NHERF oligomerization. We therefore examined the effects of overexpression of full-length NHERF, NHERF(1-151), or NHERF(1-121) on autophosphorylation of endogenous PDGFR or overexpressed epidermal growth factor (EGF) receptor (EGFR) in CHO cells. EGF-induced EGFR autophosphorylation was unaffected by overexpression of NHERF or the truncated versions of NHERF (Fig. 5B), revealing that NHERF does not have a general effect on cellular RTK activity. PDGF-induced autophosphorylation of endogenous PDGFR, as described above, was consistently enhanced twofold by overexpression of full-length NHERF, while overexpression of NHERF(1-151) had no significant effect. Overexpression of NHERF(1-121), in contrast, decreased PDGF-induced autophosphorylation of endogenous PDGFR by nearly 50% (Fig. 5C).

Activation of β₂AR can influence cellular PDGFR activity in a NHERF-dependent manner.

It has previously been shown that NHERF binds in an agonist-promoted fashion to the β₂AR and that this association in cells imparts to the β₂AR the ability to regulate Na⁺/H⁺ exchanger type 3 activity (17). One might therefore infer that the β₂AR would also be able to inhibit the activity of the cellular PDGFR in an agonist-dependent fashion, since the two receptors should compete for binding to cellular NHERF. This simplistic prediction needs to be tempered, however, by the potential for the existence of a NHERF-independent stimulatory coupling between the β₂AR and PDGFR. It has been shown for many G-protein-coupled receptors that agonist activation can induce cellular mitogenic signals that are mediated via transactivation of an RTK. Transactivation of growth factor receptors, such as the PDGFR (21, 29) or EGFR (11, 12, 28, 32, 46), has often been reported to be a step necessary for mitogenic signaling by G-protein-coupled receptors. If stimulation of the β₂AR leads to such a transactivation of the PDGFR, this obviously would complicate studies aimed at examining the potential for NHERF-mediated interactions between the two receptors. Thus, before examining whether the β₂AR can inhibit PDGFR function by competing for binding to intracellular NHERF, we investigated the potential for β₂AR-mediated transactivation of the PDGFR. These studies were carried out with CHO cells, which express fairly high levels of endogenous PDGFR.

The β₂AR has previously been shown to activate ERK in HEK-293 and COS-7 cells via a G_i/G_βγ-dependent pathway (10, 32), with switching of the G-protein coupling of the β₂AR from G_s to G_i occurring via PKA phosphorylation of the receptor (10). CHO cells transfected with WT β₂AR exhibited robust ERK activation in response to stimulation with the β-adrenergic agonist isoproterenol (Fig. 6A, first and second bars). We next examined whether this response involved RTK transactivation. CHO cells are known to express little or no endogenous EGFR (30), and indeed the β₂AR-mediated activation of ERK reported here was completely insensitive to preincubation with the specific EGFR antagonist AG1478 (Fig. 6A, third bar). In contrast, the specific PDGFR antagonist AG1295 potently inhibited isoproterenol-induced increases in ERK phosphorylation (Fig. 6A, fourth bar), revealing an essential role for PDGFR kinase activity in β₂AR mitogenic signaling in CHO cells. The isoproterenol-mediated ERK activation in CHO cells, like that in HEK-293 cells, was attenuated by pretreatment of the cells with pertussis toxin, by overexpression of the G_βγ sequestrant βARK-ct, and by preincubation with the specific PKA inhibitor H89 (Fig. 6A, final three bars). Thus, the β₂AR in CHO cells employs a G_i/G_βγ/PKA-dependent pathway to stimulate ERK via PDGFR transactivation.

We next examined whether, in addition to this G-protein-dependent transactivation of the PDGFR by the β₂AR, there might also be a simultaneous inhibitory form of cross talk between the two receptors due to competition for binding to endogenous NHERF. For these studies, we used a point-mutated β₂AR that would be unable to sequester NHERF from the PDGFR. This mutant has the final residue of the WT β₂AR-CT, Leu413, changed to alanine. This mutant receptor (L413A β₂AR) does not bind NHERF yet is identical to the WT β₂AR with respect to agonist binding and G-protein coupling (17). As would be expected if β₂AR binding of NHERF leads to diminished PDGFR activity, removing the ability of the β₂AR to bind endogenous NHERF resulted in an enhanced capacity of L413A β₂AR to activate ERK compared to WT β₂AR (Fig. 6B, first three bars). These data suggest that mutation of the final residue of the β₂AR alleviates a negative regulatory effect of WT β₂AR activation on PDGFR function. If this regulatory effect is indeed due to competition for NHERF binding, then it should also be attenuated by overexpression of NHERF. This prediction is borne out by the observation that NHERF overexpression enhances the ability of the WT β₂AR to activate ERK via PDGFR transactivation (Fig. 6B, fourth bar).

To more definitively establish the presence of two opposing forms of β₂AR cross talk to the PDGFR, we studied the effect of β₂AR prestimulation on PDGF-induced ERK activation. It might be predicted that in cells stably expressing WT β₂AR, a combination of G-protein-mediated transactivation and NHERF-mediated inhibition of PDGFR signaling may occur. The data shown in Fig. 7A demonstrate that isoproterenol stimulation of WT β₂AR before determination of the level of ERK activation at various PDGF concentrations did not cause a significant alteration in PDGFR signaling to ERK. In contrast, prestimulation of L413A β₂AR, which cannot sequester NHERF from the PDGFR, resulted in a significant enhancement of subsequent PDGF-mediated ERK activation (Fig. 7B). Thus, L413A β₂AR retains the capacity to transactivate the PDGFR and in fact transactivates the PDGFR more strongly than does WT β₂AR. Conversely, prestimulation of the mutant receptor PKA⁻ β₂AR, which can sequester NHERF but which lacks PKA phosphorylation sites and thus cannot engage the G_i/G_βγ-dependent transactivation pathway (10), yielded no PDGFR transactivation. Instead, isoproterenol prestimulation of PKA⁻ β₂AR resulted in an attenuation of subsequent PDGF-mediated ERK activation (Fig. 7C).

Since prestimulation of WT β₂AR exhibited little effect on PDGFR function in the experiments described above, it is likely that the G-protein-mediated transactivation and NHERF-mediated inhibition roughly negated each other under our experimental conditions. One might predict, however, that under conditions of impaired PKA activation, PDGFR transactivation by WT β₂AR would be suppressed and NHERF-mediated inhibition would predominate. To mimic conditions where PKA is inactive or unavailable, and PDGFR transactivation by the β₂AR is thus minimized, we performed experiments in which cells were incubated with the PKA inhibitor H89. Pretreatment with H89 blocked isoproterenol-induced increases in ERK phosphorylation, consistent with previous findings (10), but did not attenuate PDGF-induced activation of ERK (data not shown). However, isoproterenol prestimulation of the WT β₂AR in H89-treated cells profoundly inhibited PDGF-induced ERK activation (Fig. 7D).