Epidermal growth factor receptor juxtamembrane region regulates allosteric tyrosine kinase activation

Abstract

Structural studies of the extracellular and tyrosine kinase domains of the epidermal growth factor receptor (ErbB-1) provide considerable insight into facets of the receptor activation mechanism, but the contributions of other regions of ErbB-1 have not been ascertained. This study demonstrates that the intracellular juxtamembrane (JM) region plays a vital role in the kinase activation mechanism. In the experiments described herein, the entire ErbB-1 intracellular domain (ICD) has been expressed in mammalian cells to explore the significance of the JM region in kinase activity. Deletion of the JM region (ΔJM) results in a severe loss of ICD tyrosine phosphorylation, indicating that this region is required for maximal activity of the tyrosine kinase domain. Coexpression of ΔJM and dimerization-deficient kinase domain ICD mutants revealed that the JM region is indispensable for allosteric kinase activation and productive monomer interactions within a dimer. Studies with the intact receptor confirmed the role of the JM region in kinase activation. Within the JM region, Thr-654 is a known protein kinase C (PKC) phosphorylation site that modulates kinase activity in the context of the intact ErbB-1 receptor; yet, the mechanism is not known. Whereas a T654A mutation promotes increased ICD tyrosine phosphorylation, the phosphomimetic T654D mutant generates a 50% reduction in ICD tyrosine phosphorylation. Similar to the ΔJM mutants, the T654D mutant ICD failed to interact with a wild-type monomer. This study reveals an integral role for the intracellular JM region of ErbB-1 in allosteric kinase activation.

A high level of homology exists among the ErbB receptors in the juxtamembrane (JM) region, which contains ≈37 residues (Fig. 1A). Of note is an abundance of basic residues in the N-terminal portion of this region, which is unique to the ErbB receptor family. Contained within the JM sequence are lysosomal (1, 2) and basolateral (3) sorting motifs, a nuclear localization sequence (4), a calmodulin binding site (5), and protein kinase C (PKC) (6, 7) and mitogen-activated protein kinase (8, 9) phosphorylation sites.

Fig. 1.

Tyrosine phosphorylation of ErbB-1 ICD. (A) Diagram of the ErbB receptors with alignment of JM regions. Basic residues are shaded gray; PKC phosphorylation sites in ErbB-1 (6, 7) and ErbB-2 (32) and a putative PKC phosphorylation site in ErbB-4 are boldface. TM, transmembrane domain; CT, C-terminal domain. (B) Lysates of Cos-7 cells transfected with ErbB-4 or ErbB-1 ICDs were precipitated (IP) with the indicated antibodies followed by Western blotting (WB) with a phosphotyrosine antibody (PY99; Upper). Blots were stripped and reprobed with the indicated antibodies to detect either ErbB-4 ICD or ErbB-1 ICD (Lower). Similar results were obtained when lysates were directly blotted with anti-PY99 (data not shown). (C) Tyrosine phosphorylation of PKC-site mutants. The indicated ErbB-1 ICD constructs were expressed as in B, precipitated with anti-ErbB-1, blotted with anti-PY99 (Upper), stripped, and reprobed with anti-FlagM2 (Lower). Relative pY, tyrosine phosphorylation (pY) relative to wild-type after normalizing for expression levels.

Pretreatment of cells with the phorbol ester 12-O-tetradecanoyl-phorbol-13-acetate (TPA), a PKC agonist, prevents epidermal growth factor (EGF)-induced ErbB-1 tyrosine phosphorylation (10–12). TPA treatment results in PKC phosphorylation of Thr-654 in the JM region (6, 7), and mutation of this residue abrogates TPA inhibition of EGF-induced ErbB-1 tyrosine phosphorylation (13). ErbB-1 ligands also promote phosphorylation of Thr-654 (14). However, the mechanism by which this PKC phosphorylation event promotes kinase inactivation is not understood. Complicating experiments with the intact receptor in cells is the fact that TPA also induces ErbB-1 endocytosis (15) and recycling of the internalized receptor (16).

Ligand-induced dimerization of ErbB-1 is a prerequisite for receptor activation and subsequent tyrosine phosphorylation (17). Recent analysis of the ErbB-1 tyrosine kinase domain (TKD) crystal structure has provided evidence for a novel asymmetric mechanism of TKD dimerization (18), contrary to the previously assumed symmetrical mode of TKD dimerization. The asymmetric mode of dimerization and activation is similar to that of the CDK2/cyclinA complex (19). In the asymmetrical ErbB-1 dimer, the C-lobe of a cyclin-like “donor” monomer contacts residues in the N-lobe of an “acceptor” monomer, resulting in allosteric activation of the acceptor (18). Mutation of key residues at this dimer interface results in the loss of EGF-induced tyrosine phosphorylation of full-length ErbB-1, indicating that the asymmetric dimer mechanism is not limited to the isolated TKD fragment. However, all structural studies of the ErbB-1 kinase have been limited to the TKD proper and have omitted the adjacent JM and C-terminal regions (18, 20). The results described herein show that the JM region is a significant participant in the allosteric kinase activation mechanism.

Results and Discussion

Tyrosine Phosphorylation of the ErbB-1 Intracellular Domain (ICD).

To understand the contribution of the JM region to ErbB-1 receptor activation, a construct containing the entire ErbB-1 ICD (residues 645-1186) was engineered (Fig. 1A). When expressed in Cos-7 cells, the ICD migrated at ≈60 kDa and was tyrosine-phosphorylated (Fig. 1B) in a manner that depended on its intrinsic kinase activity as revealed by mutagenesis of Lys-721, a residue required for ATP binding in the kinase domain (Fig. 1C). The ErbB-4 ICD, which is generated by regulated intramembrane proteolysis of the full-length ErbB-4 receptor (21), is included for comparison and is similarly tyrosine-phosphorylated (Fig. 1B), as reported in ref. 22. The data of Zhang et al. (18) show that high concentrations of purified ErbB-1 TKD promote autoactivation. However, tyrosine phosphorylation of the ICD occurs within a milieu of intracellular proteins, thus avoiding the potential artifact of aggregation-induced activation exhibited by the purified TKD.

Role of Thr-654 Phosphorylation in Kinase Activation.

To test whether modification of Thr-654 influences activity of the ICD, T654A and T654D mutants were produced. Compared with the wild-type ICD, the T654A mutant ICD displays a 2.7-fold increase in tyrosine phosphorylation, whereas tyrosine phosphorylation of the phosphomimetic T654D mutant is reduced by 2-fold (Fig. 1C). These results are consistent with data showing that phosphorylation of ErbB-1 at Thr-654 results in a decrease in tyrosine kinase activity (10). We infer that the T654A effect is due to Thr-654 phosphorylation of the wild type; however, this was not experimentally verified. In addition, the data indicate that Thr-654 modulates activity of the kinase domain within the context of the ICD fragment and that the mechanism does not require the presence other receptor domains or regions.

Based on in vitro interactions between a basic peptide comprising JM residues 645–660 and phospholipid micelles, it has been proposed that an electrostatic interaction may regulate ErbB-1 kinase activation (23). A similar interaction of the ErbB-1 ICD with the plasma membrane could potentially result in membrane-localized aggregation, which might promote constitutive kinase activation of the ICD. However, no decrease in wild-type ICD tyrosine phosphorylation occurred after treatment with 2 μM sphingosine [supporting information (SI) Fig. 6], which prevents the electrostatic interaction of JM peptides with phospholipids (23, 24).

JM Deletions Result in a Significant Loss of Kinase Activity.

The preceding data suggested that phosphorylation of Thr-654 alters kinase activity by either of two mechanisms: (i) phosphorylation at Thr-654 promotes an autoinhibitory interaction of the JM region with the kinase domain or (ii) the JM region positively modulates the kinase domain and phospho-Thr-654 prevents the interaction. To test these possibilities, segments of the JM region were deleted in the ICD: Δ645–662 and Δ645–676 (Fig. 2A). When expressed in cells, the data revealed that tyrosine kinase activity is decreased by ≈95% in vivo and 65% in vitro for the Δ645–662 mutant and ≈95% for the Δ645–676 mutant in vivo and in vitro (Fig. 2 B and C). This result demonstrates that the presence of the JM region is critical for maximal kinase activation. The difference in the level of kinase activity detected for the two JM mutants also suggests that elements in both the N-terminal half of the JM, including the basic cluster and Thr-654, and the C-terminal region participate in regulation of kinase activity. The studies of Zhang et al. (18) identify Gly-672–Ala-677 as a likely element in the C-terminal JM region that participates in kinase activation, particularly Pro-675.

Fig. 2.

Tyrosine phosphorylation of JM deletion mutants. (A) JM residues in ErbB-1. Basic residues are shaded and Thr-654 is boldface as in Fig. 1. Arrows denote the N-terminal residue of the two JM deletions in the ICD, Δ645–662 and Δ645–676. The asterisk indicates the N-terminal residue of the TKD construct used by Zhang et al. (18) and Stamos et al. (20) for crystallography. (B) Indicated ICD constructs were transiently transfected into Cos-7 cells. Cell lysates were precipitated with anti-FlagM2 and blotted with anti-phosphotyrosine (PY99) or anti-FlagM2. (C) In vitro kinase assay comparing phosphorylation of ΔJM mutants with wild-type and K721R ICDs. After expression in Cos-7 cells, lysis, and precipitation with anti-FlagM2, the immunoprecipitates were incubated with [γ-³²P]ATP (Upper) as described in Materials and Methods. Lysates were probed with anti-FlagM2 to confirm equal expression (Lower).

Contribution of the JM Region to Allosteric Kinase Activation.

To address whether the JM region contributes to allosteric kinase activation, point mutations were engineered into the ICD at residues required for contacts at the asymmetric kinase dimer interface, I682Q in the N-lobe or V924R in the C-lobe of separate monomers (18). The C-lobe V924R mutant can only act as an acceptor monomer, and the N-lobe I682Q mutant can only function as a donor to allosterically activate the acceptor monomer (18). An experimental advantage of using these mutations is that neither the N-lobe nor the C-lobe point mutants homodimerize; thus, any observed kinase activation is a result of asymmetrical dimerization of these mutants when expressed on separate monomers.

Compared with the wild-type ICD, the I682Q and V924R ICD mutants were inactive when expressed alone (Fig. 3A, lanes 3 and 4). When coexpressed with a kinase-deficient K721R donor ICD, the V924R C-lobe acceptor mutant ICD was active (Fig. 3A, lane 6). Coexpression of a C-lobe V924R acceptor ICD with an N-lobe I682Q donor ICD also resulted in kinase activation (Fig. 3A, lane 5). This level of activation is ≈20% of that obtained with wild-type ICD monomers (lane 1), which is comparable to reported results with the isolated TKD monomers (18). One explanation for the decreased phosphorylation detected when I682Q and V924R ICDs were coexpressed is that these mutants can only act as donors or acceptors, respectively, whereas both the N- and C-lobes of wild-type ICDs are able to participate in a dimer. These results indicate that the critical contacts between the N- and C-lobes of monomers previously shown to be required for allosteric kinase activation in the isolated TKD and the intact receptor (18) are preserved in the context of the ICD.

Fig. 3.

Ability of JM mutants to act as acceptor or donor monomers. (A Left) Cartoon of ICD mutants and predicted associations in coexpression experiments. Corresponding data lanes (A Right) are listed in brackets. Circles, V924R C-lobe mutation; stars, I682Q N-lobe mutation; ×, K721R mutation; green text, acceptor monomer; blue text, donor monomer. (A Right) Acceptor (V924R) and donor (I682Q) monomers were expressed alone (lanes 3 and 4) or coexpressed (lanes 5 and 6). Lysates were precipitated with anti-FlagM2 and blotted with either anti-PY99 or anti-FlagM2. (B Upper) Cartoon of potential interactions of donor (I682Q) and acceptor (V924R) monomers with ΔJM mutants containing additional N-lobe (I682Q) or C-lobe (V924R) mutations. (B Lower) ΔJM mutants having N-lobe or C-lobe mutations were coexpressed with acceptor (V924R) or donor (I682Q) monomers with intact JM regions.

The results in Fig. 2 suggest that the JM region may contribute to contacts at the asymmetric dimer interface. Using these mutants, the capacity of the ΔJM (deleted JM region) mutants to act as donors or acceptors in allosteric kinase activation was assessed. To test the ability of ΔJM mutants to act as acceptors, cells were cotransfected with either a ΔJM mutant also containing a C-lobe V924R mutation as acceptor monomer and an N-lobe I682Q single mutant donor (Fig. 3B, lanes 3 and 8). As a positive control, a V924R acceptor monomer (with an intact JM region) was coexpressed with an I682Q donor (Fig. 3B, lane 1). The data in Fig. 3B show that, compared with the positive control (lane 1), deletion of residues 645–662 decreases the acceptor function by ≈80% (lane 3), and deletion of residues 645–676 completely abrogates the capacity of that ICD to act as an acceptor monomer (lane 8).

Subsequently, the capacity of the ΔJM mutants to function as donor monomers was examined by using a C-lobe V924R mutant as an acceptor monomer. To make certain that the ΔJM mutants possess donor functions only, an N-lobe I682Q mutation was engineered into each ΔJM construct. Coexpression of a V924R acceptor mutant with either ΔJM donor mutant resulted in tyrosine phosphorylation of the acceptor V924R monomer (Fig. 3B, lanes 5 and 10), although at levels lower than the positive control; e.g., an I682Q donor with an intact JM region (lane 1). The Δ645–662 mutant demonstrated 40% of wild-type donor activity (lane 5), whereas the Δ645–676 mutant had 20% of wild-type activity (lane 10). This result indicates that the ΔJM mutants can still act as donors but at a lower efficiency.

It is possible that the JM region facilitates but is not necessary for dimerization. However, the kinase activity of the Δ645–676 mutant ICD was not rescued when a kinase-deficient K721R donor was overexpressed relative to the ΔJM construct (data not shown), indicating that an intact JM region must be present in an acceptor monomer and cannot be overcome by a concentration-dependent increase in the interaction of acceptor and donor monomers. These data lead to the conclusion that the JM region is essential for the formation of an activated kinase acceptor.

Capacity of JM Mutants to Associate with a Wild-Type ICD.

Because the ΔJM mutants are not competent, particularly as acceptors, in tyrosine kinase activation, the following experiments were performed to assess their capacity to interact with wild-type ICDs. This interaction of ΔJM and wild-type ICDs follows the strategy used in Fig. 3. Cells were cotransfected with Myc-tagged wild-type and Flag-tagged wild-type or Flag-tagged ΔJM ICD constructs. Subsequently, lysates were precipitated with anti-Myc and blotted with anti-Flag. As a control for nonspecific tag recognition, cells were singly transfected with the Myc- or Flag-tagged constructs (Fig. 4A, lanes 1–4). A low level of cross-reaction of the Flag epitope with the Myc antibody was detected (lanes 2–4), and this background was subtracted from subsequent coprecipitation assays (lanes 5–7). The results in Fig. 4A show that interaction of Δ645–662-Flag ICD with wild-type-Myc ICD was reduced by 50% compared with wild-type-Flag ICD (lanes 5 and 6), whereas the data in lane 7 show that the Δ645–676-Flag ICD interaction with wild-type-Myc ICD was reduced by 90%. The same results were achieved when transfected lysates were precipitated with anti-Flag and blotted with anti-Myc (data not shown). These data demonstrate that deletion of the JM region attenuates the physical association of the N-lobe of an acceptor with the C-lobe of a donor. The data in Fig. 4A, lane 7, indicate that the longer deletion is significantly impaired in its capacity to interact as an acceptor or donor with the wild-type ICD compared with the shorter deletion. That the longer deletion does retain some capacity to interact with the wild-type ICD is indicated by the detection of activation in vivo by phosphotyrosine blotting (Fig. 3B, lane 10).

Fig. 4.

Failure of ΔJM and T654D ICD mutants to interact with wild-type. (A) Flag-tagged wild-type and ΔJM mutant ICDs were transfected individually or cotransfected with wild-type Myc ICD as described in Materials and Methods. Lysates were precleared on protein G-Sepharose beads, precipitated with anti-Myc, and blotted with anti-FlagM2. The blot was stripped and reprobed with anti-EGFR. Lysates were also blotted with anti-EGFR. (B) T654D ICD-Flag was transfected with or without wild-type ICD-Myc as in A, and precleared lysates were precipitated with anti-FlagM2 followed by blotting with anti-Myc. Expression was assessed as in A.

The data in Fig. 1C show that the T654D mutation in the JM region abrogates kinase activation of the ICD. In a coprecipitation experiment with Myc-tagged wild-type ICD, this Flag-tagged phosphomimetic mutant displayed a significantly decreased interaction with the wild-type ICD (Fig. 4B, lanes 5 and 7), indicating that phosphorylation at Thr-654 inhibits monomer:monomer associations.

JM Mutations in the Intact Receptor.

The role of the JM region in EGF-induced receptor activation was tested by introducing the previously described ΔJM mutations in the full-length molecule and transiently expressing the constructs in NIH 3T3 cells. To aid in the insertion and/or stability of the transmembrane domain, the stop transfer sequence of residues R645–R647, which immediately follows the transmembrane sequence, were retained in these ΔJM mutants. Initially, the I682Q and V924R mutants were compared with the intact wild-type receptor. The results, shown in Fig. 5A, indicate that coexpression of these mutants results in EGF-mediated receptor activation similar to wild-type, whereas either mutant expressed alone was not activated by EGF treatment.

Fig. 5.

Role of JM region in EGF-stimulated allosteric kinase activation. (A) NIH 3T3 cells were transiently transfected with the indicated constructs. Forty hours later, cells were starved for 60 min in serum-free media and treated with EGF (50 ng/ml) for 5 min. Lysates were immediately immunoblotted with either phosphosite-specific ErbB-1 PY1173 (Upper) or ErbB-1 (Lower). (B and C) Experiments assessing the ability of JM deletions to act either as donors (B) or acceptors (C). The exposure time for C was longer than that in B to detect EGF-dependent tyrosine phosphorylation in lanes 5–8.

Analysis of the donor function of the ΔJM mutants (Fig. 5B) shows that the Δ648–662 mutant (lanes 7 and 8) promoted EGF-dependent ErbB-1 tyrosine phosphorylation as effectively as the control (lanes 3 and 4). However, the longer ΔJM mutant (Δ648–676) was unable to mediate EGF-dependent ErbB-1 activation (lanes 11 and 12). When the ΔJM mutants were tested for acceptor function (Fig. 5C), both mutants (lanes 7 and 8, and lanes 11 and 12) were significantly impaired compared with the control (lanes 3 and 4). The fact that the Δ648–662 mutant retains donor activity equivalent to the control indicates that cell-surface expression is not compromised. This deletion mutant exhibited cell-surface expression in semiquantitative assays using Texas red-conjugated EGF and by cell sorting with anti-Flag (data not shown). However, similar assays indicated that cell-surface expression was significantly impaired in the Δ648–676 mutant. Therefore, no clear conclusion can be drawn from this mutant.

These results support the conclusion that the JM region is required for maximal receptor activation and that this region has a greater role in the acceptor function than donor function in the allosteric model of kinase activation. Previous studies using peptides representing the polybasic JM region of ErbB-1 identified an acidic region in the C-lobe of the TKD that interacts with a polybasic JM peptide (25), and deletion of the polybasic region results in an inactive receptor (26), consistent with our findings.

The study of Zhang et al. (18), which includes residues 672–682 of the JM region, showed that Pro-675 just within the JM region of an acceptor TKD interacts in with Val-956 in the C-lobe of a donor TKD, and that a P675G TKD mutant has diminished activity in vitro. This proline residue is retained in the shorter ΔJM (Δ645–662 ICD and Δ648–662 ErbB-1) mutants but is deleted in the longer ΔJM mutants. Because the shorter ΔJM mutants retain Pro-675 but have significantly diminished kinase activity as acceptors, residues defined by this deletion are additionally implicated in the JM control of kinase activation.

The data presented herein indicate that the JM region is essential for effective allosteric activation of ErbB-1 kinase activity by facilitating TKD monomer interactions. Phosphorylation of Thr-654 within the JM region attenuates the capacity of the JM region to mediate activation. Although high-resolution structures of the ErbB-1 kinase domain do not include this region, the crystal structure of Kit, a type III tyrosine kinase receptor, reveals extensive cis interactions between the JM region and the N-lobe of the kinase domain (27). This JM region of Kit, which contains two autophosphorylation tyrosine residues, functions as an autoinhibitory domain to lock the protein in a closed conformation (27, 28). When these JM tyrosine residues are phosphorylated, the autoinhibition is relieved and the kinase domain adopts an active conformation (27). Thus, there is precedent for extensive JM sequence interactions with a receptor TKD.

Materials and Methods

Construction of ErbB-1 ICD and Mutants.

All primer sequences can be found in SI Table 1. Wild-type and ΔJM mutant ICDs were completely sequenced to ensure that no secondary mutations occurred during PCR amplification. Point mutants arising from wild-type or ΔJM ICD templates (T654A, T654D, I682Q, or V924R) were sequenced at the region of mutation.

PCR was used to amplify the entire ICD of ErbB-1 (residues 645-1186) from ErbB-1-EGFP (gift of Alexander Sorkin, University of Colorado Health Sciences Center, Denver). The forward primer included a HindIII restriction site, Kozak sequence, and start codon; the reverse primer retained the stop codon immediately followed by an XbaI site. The ErbB-1 ICD PCR product was cloned into a pFlag CMV 5.1+ vector (Sigma). Site-directed mutagenesis (QuikChange; Stratagene) was used to remove the stop codon and allow coding of the Flag epitope at the C terminus, and the entire cDNA of ErbB-1 ICD-Flag was subjected to sequencing. The Flag epitope did not influence ErbB-1 ICD expression or tyrosine phosphorylation (data not shown), and all experiments herein use the ErbB-1 ICD-Flag construct unless otherwise noted. All point mutations (T654A, T654D, K721R, I682Q, and V924R) were engineered into the ErbB-1 ICD-Flag construct by site-directed mutagenesis.

The sites for JM deletions were chosen based on the predicted structure of the JM region bound to micelles (29). Using ErbB-1 ICD-Flag as the template, forward primers were designed that incorporate a HindIII restriction site, a Kozak sequence, and a start codon immediately preceding the desired first residue (E663 or Q677). PCR-amplified cDNAs were digested with HindIII and ClaI and ligated into a similarly digested ErbB-1 ICD-Flag.

To create a Myc-tagged ErbB-1 ICD, the ErbB-1 ICD cDNA was shuttled from the ErbB-1 ICD-pFlag CMV 5.1+ vector into the pcDNA4-Myc/His A vector (Invitrogen). To prevent coding of the 6× His-tag at the C terminus, a premature stop codon was inserted between Myc and 6× His by site-directed mutagenesis. Next, ErbB-1 ICD cDNA was shuttled from ErbB-1-Flag into the pcDNA4-Myc after digestion of both plasmids with HindIII and XbaI.

Mutations and JM Deletions in Full-Length ErbB-1.

Point mutations (I682Q, K721R, and V924R) were introduced into Flag-ErbB-1 (generously provided by Anthony W. Burgess, Ludwig Institute, Melbourne) by site-directed mutagenesis. The strategy for introducing internal deletions into Flag-ErbB-1 was as follows. First, an EcoRV cut site was introduced by site-directed mutagenesis at nucleotides that correspond to H648. Second, EcoRV cut sites were added to this parental construct at nucleotides corresponding to either R662 or N676. Digestion with EcoRV and ligation generated either Δ648–662 or Δ648–676 Flag-ErbB-1. This cloning approach resulted in the addition of two residues at the deletion (Glu and Iso). Primer sequences are listed in SI Table 1.

Cell Culture and Transient Expression of ErbB-1 Constructs.

Cos-7 and NIH 3T3 cells were routinely maintained in 10% FBS-supplemented DMEM at 37°C with 5% C₂. For ICD experiments, 2 μg of DNA was transiently transfected into a 60-mm dish of Cos-7 cells at ≈80% confluency by using Lipofectamine 2000 (Invitrogen) per the manufacturer's instructions. For Figs. 3 and 4, 1 μg of each construct was transfected into each 60-mm dish of Cos-7 cells. Full-length ErbB-1 constructs were transiently transfected into NIH 3T3 cells as described in ref. 18. The ratio of DNA:FuGENE 6 was 1.5 μg of DNA:4.5 μl of FuGENE 6 per well.

Immunoprecipitation and Blotting.

For ICD experiments, transfected cells were harvested 24 h posttransfection, lysed with TGH buffer [1% Triton X-100, 10% glycerol, 50 mM Hepes (pH 7.2), 100 mM NaCl, 5 mM NaF, 1 mM Na₃VO₄, Complete Mini protease inhibitor mixture tablet (Roche)] as described in ref. 30. Lysates were precipitated with anti-FlagM2 (Sigma), anti-EGFR (sc-1005; Santa Cruz Biotechnology), or anti-ErbB-4 prepared as described in ref. 31. Immunoprecipitates were subjected to immunoblotting with PY99 (Santa Cruz Biotechnology). To confirm equal expression, blots were stripped with 2% SDS/100 mM 2-mercaptoethanol/62.5 mM Tris·HCl (pH 6.8) for 20 min at 60°C and reprobed with anti-FlagM2, anti-EGFR (Upstate Biotechnology), or anti-ErbB-4. Densitometric analysis was performed by using NIH ImageJ software. All results were normalized for expression levels and quantitated relative to wild type.

For coimmunoprecipitation assays, Cos-7 cells were lysed 24 h posttransfection with TGH and preincubated with protein G-Sepharose beads (Invitrogen) to eliminate nonspecific binding. Next, precleared lysates were precipitated with anti-FlagM2 and blotted with anti-Myc (9E10; Santa Cruz Biotechnology) or precipitated with anti-Myc and blotted with anti-Flag. Blots were then stripped and reprobed with anti-EGFR. Whole-cell lysates were also blotted with anti-EGFR to confirm expression. Densitometric analysis was performed as for phosphotyrosine blots. After correcting for ErbB-1 expression levels (Fig. A Lower and B Lower), the background binding of the Myc epitope to the FlagM2 antibody (Fig. A Upper and B Upper) was subtracted from each corresponding coprecipitation lane, such that lane 2 was subtracted from lane 5 and lane 3 was subtracted from lane 6; no nonspecific binding to protein G-Sepharose beads was detected (data not shown). Results are expressed relative to wild-type ErbB-1 ICD-Flag cotransfected with ErbB-1 ICD-Myc.

For expression of full-length ErbB-1, transfected NIH 3T3 cells were starved 40 h posttransfection for 60 min with serum-free DMEM and treated for 5 min with 50 ng/ml EGF (R&D Systems) at 37°C. Lysis was performed with TGH buffer as described above, and 100 μg of each lysate was directly subjected to immunoblotting with either EGFR pY1173 (Santa Cruz Biotechnology) or EGFR after separation by SDS/PAGE.

In Vitro Kinase Assay.

In vitro kinase assays were performed essentially as described in ref. 22. Briefly, cells transiently transfected with 2 μg of each ICD were lysed, and ICDs were precipitated with anti-FlagM2. Immunoprecipitates were washed twice with TGH buffer and once with kinase assay buffer [25 mM Hepes (pH 7.4), 10 mM MgCl₂, 2.5 mM MnCl₂, 50 μM Na₃VO₄, 0.5 mM DTT] and resuspended with 100 μl of kinase assay buffer. Next, 10 μl of immunoprecipitated protein was incubated for 15 min at room temperature with 25 μM ATP containing 62 μCi/ml [γ-³²P]ATP in a total reaction volume of 40 μl. The reaction was quenched by addition of Laemmli buffer, boiled for 10 min, and separated by SDS/PAGE.