Presenilin-dependent (cid:1) -Secretase Processing Regulates Multiple ERBB4/HER4 Activities*

Transmembrane receptors typically transmit cellular signals following growth factor stimulation by coupling to and activating downstream signaling cascades. Reports of proteolytic processing of cell surface receptors to release an intracellular domain (ICD) has raised the possi-bility of novel signaling mechanisms directly mediated by the receptor ICD. The receptor tyrosine kinase ERBB4/ HER4 (referred to here as ERBB4) undergoes sequential processing by tumor necrosis factor- (cid:2) converting enzyme and presenilin-dependent (cid:1) -secretase to release the ERBB4 ICD (4ICD). Our recent data suggests that regulation of gene expression by the ERBB4 nuclear protein and the proapoptotic activity of ERBB4 involves the (cid:1) -secre-tase release of 4ICD. To determine the role (cid:1) -secretase processing plays in ERBB4 signaling, we generated an ERBB4 allele with the transmembrane residue substitution V673I (ERBB4-V673I). We demonstrate that ERBB4-V673I fails to undergo processing by (cid:1) -secretase but retains normal cell surface signaling activity. In contrast to wild-type ERBB4, however, ERBB4-V673I was excluded from the nuclei of transfected cells and failed to activate STAT5A stimulation of the (cid:3) -casein ERBB4, Notch1, and APP. A , schematic of ERBB4 functional domains. The ERBB4 ectodomain consists of an amino-terminal ligand binding region composed of two cysteine-rich regions ( cys1 , cys2 ). Proteolytic processing by TACE results in ectodomain cleavage. Subsequent presenilin-de- pendent (cid:1) -secretase processing is predicted to occur at Val-673 ( V673 ) and results in membrane release of the 4ICD (residues 673–1309). We have identified functional domains harbored within 4ICD including a NLS (residues 676–684) (15) and a BH3 domain (residues 986–992). 2 B , alignment of transmembrane domains from ERBB4, Notch1, and APP reveals a conserved valine residue ( asterisks ) with similar trans- membrane positions. This valine (Val-1743) is essential for Notch1 processing by (cid:1) -secretase, and we predict that a similar base substitu- tion in ERBB4 (V673I) will abolish (cid:1) -secretase processing of this receptor. APP is processed by (cid:1) -secretase with relaxed specificity at two residues ( underlined ) in addition to the conserved valine. The ERBB4 NLS is located immediately downstream of the transmembrane domain ( underlined ).

Activated single transmembrane cell surface receptors typically transmit extracellular signals through the recruitment of membrane and cytosolic signal transduction proteins. These complex cascades of protein:protein interactions and posttranslational modifications culminate in the nucleus where the activation of specific target genes regulates diverse cellular responses including proliferation, differentiation, migration, and apoptosis. Recent biochemical and genetic evidence suggests that the presenilin-dependent ␥-secretase processing of cell surface receptors contributes to cellular signaling through novel pathways involving an active receptor intracellular domain (ICD) 1 (1,2). For example, the presenilin-dependent ␥-secretase processing of the transmembrane receptor Notch results in release of the Notch ICD (NICD) and subsequent NICD transcription factor activity in the nucleus (3)(4)(5).
The receptor tyrosine kinase ERBB4 also undergoes ␥-secretase processing releasing the ERBB4 intracellular domain (4ICD); however, the contribution of this event to ERBB4 signaling remains to be determined. ERBB4 is a member of the ERBB-family of receptor tyrosine kinases, which also includes the epidermal growth factor receptor, ERBB2/HER2/Neu, and ERBB3. This receptor family controls several cell fate decisions during multiple stages of embryonic and postnatal development (6). ERBB4 activation has been associated with diverse cellular responses including proliferation (7), cell migration (8), and differentiation (9,10).
The biochemical details of ERBB4 proteolytic processing resulting in ERBB4 ectodomain shedding and nuclear accumulation of 4ICD have been elucidated. Activated ERBB4 is first cleaved within the juxtamembrane region through the activity of tumor necrosis factor-␣-converting enzyme (TACE) (12). This cleavage event results in shedding of the 120-kDa ERBB4 ectodomain and membrane association of the m80 ERBB4 transmembrane region and endodomain. Subsequent cleavage of TACE processed ERBB4 by ␥-secretase releases 4ICD from cellular membranes. ERBB4 proteolytic processing results in nuclear accumulation of ERBB4 (13)(14)(15) and our recent biochemical data indicates that 4ICD is the predominant form of nuclear ERBB4 (15).
Similar to Notch, proteolytic processing of ERBB4 may contribute to novel ERBB4 signaling properties. For example, ERBB4 is essential for lactation initiation and milk-gene expression in the mouse mammary gland at parturition (9,16). Moreover, ERBB4 regulates milk-gene expression through the direct association of nuclear 4ICD with a transcription complex at the endogenous ␤-casein promoter, which includes the signal transducer and activator of transcription family member STAT5A (15). Indeed, nuclear accumulation of 4ICD is necessary for STAT5A stimulation of the ␤-casein promoter suggesting that 4ICD regulates gene expression by functioning as a nuclear chap-erone for STAT5A (15). In contrast, we have shown that ERBB4 induces cell-killing of malignant human breast cell lines. We further demonstrate that 4ICD accumulation within mitochondria correlates with apoptosis. Taken together these results support the hypothesis that multiple ERBB4 activities are regulated by proteolytic processing and subsequent membrane release of 4ICD; however, the functional significance of ERBB4 proteolytic processing remains to be confirmed. Here we demonstrate that an ERBB4 allele lacking a transmembrane ␥-secretase cleavage site retains canonical cell surface receptor signaling activity but fails to regulate gene expression and lacks cell-killing activity.

EXPERIMENTAL PROCEDURES
ERBB4 cDNA-The human ERBB4 cDNA used in these experiments has been sequenced in its entirety (17) and represents the JM-a isoform (18). This isoform retains both TACE and ␥-secretase recognition sequences and is therefore an ERBB4 isoform that undergoes complete proteolytic processing at the cell surface following ligand stimulation.
Cell Lines and Transfections-The COS-7 cell line was obtained from ATCC and cultured in Dulbecco's modified Eagle's medium with 10% fetal bovine serum and 2 mM L-glutamine. Because of the cell-killing activity of ERBB4 we performed all functional assays in the MCF-7 human breast cancer cell line MCF-7B, which was modified to stably overexpress BCL-2 (20) and is resistant to ERBB4-induced cell killing. 2 MCF-7B cells were cultured in minimum essential medium containing 10% fetal bovine serum, 1 mM Na-pyruvate, and 2 mM L-glutamine.
Transfections were performed on 2 ϫ 10 5 cells in 35-mm tissue culture dishes or 1.0 ϫ 10 6 cells in 100-mm tissue culture dishes with 2 and 5 g of each transfected DNA, respectively, using FuGENE 6 (Roche Applied Science) transfection reagent as described by the manufacturer. In each experiment transfected cells were harvested at 48 h post-transfection.
Preparation of Membrane and Cytosolic Cellular Fractions-Membrane and cytosolic fractions were isolated from COS-7 cells transfected with vector control, ERBB4-Flag, or ERBB4-V673I-Flag and treated at 2 days post-transfection with 100 ng/ml of tetradecanoylphorbol-13-acetate (TPA) (Sigma) for 30 min at 37°C using a modification (15) of a protocol published elsewhere (13). Cytosolic and membrane fractions were subjected to ERBB4 immunoprecipitation and Western blot analysis.
Deconvolution Microscopy-Transfected MCF-7B cells cultured on coverslips in a 35-mm tissue culture dish were mock-stimulated or stimulated with 50 ng/ml of heregulin ␤1 (referred to here as HRG) in phosphate-buffered saline for 30 min at room temperature. In some experiments cells were pretreated with the ␥-secretase inhibitor Compound E at 10 nM overnight in growth media with 2% fetal bovine serum. Following appropriate treatments, cells were fixed in 4% paraformaldehyde for 10 min, briefly washed in phosphate-buffered saline, counterstained in 500 mM DAPI (Molecular Probes), and coverslipped with Prolong Antifade Media (Molecular Probes). The slides were analyzed on a Leica DMRXA automated upright epifluorescent microscope (Leica Microsystems, Bannockburn, IL) equipped with a 63ϫ (NA 1.4) oil immersion objective and a Sensicam QE CCD digital camera, and quantitation of intranuclear enhanced green fluorescent protein (EGFP) was performed exactly as described elsewhere (15).
Luciferase Assay-Transfections were performed on MCF-7B cells as described above with combinations of ERBB4 and STAT5A expression plasmids and a ␤-casein promoter luciferase reporter plasmid (15). Cell lysates were prepared in 500 l of cell culture lysis reagent (Promega), and the luciferase assay was performed on 30 l of lysate using the luciferase assay system (Promega) in a Berthold AutoLumat Plus luminometer. Each sample was performed in duplicate, and the data represent the mean of at least three independent experiments. Statistically significant differences between data sets were determined using the paired Student's t test.
Apoptosis Assay-The COS-7 cell line was transfected in 6-well tissue culture plates with the indicated cDNA fused to EGFP using Fu-GENE 6, and apoptosis was analyzed at 2 days post-transfection. Apoptosis was determined visually by examining cells using an inverted Leica DMIRB fluorescent microscope and calculating the percentage of EGFP-positive cells displaying morphological signs of apoptosis. The percentage of apoptotic cells induced by each treatment minus the percentage of apoptotic cells observed in the EGFP vector control was reported. When indicated, ERBB4-transfected cells were treated with 10 nM Compound E overnight in growth medium containing 2% fetal bovine serum. All samples were prepared in duplicate, and each experiment was repeated at least three times. Significant differences between data sets were determined using the paired Student's t test.
Isolation of Subcellular Fractions-The SKBr3 breast cancer cell line was transfected with pEGFP vector control, ERBB4-EGFP, or ERBB4-V673I-EGFP, and at 40 h post-transfection subcellular fractions including cytosolic, endoplasmic reticulum/endosomal, and mitochondrial were isolated exactly as described previously. 2 Fifty g of each fraction was analyzed for ERBB4/4ICD expression by Western blot.
A Point Mutation (V673I) within the ERBB4 Transmembrane Abolishes ␥-Secretase Processing-We used Western blot analysis of membrane and cytosolic fractions from transiently transfected COS-7 cells to confirm that the ERBB4-V673I base substitution abolished ERBB4 ␥-secretase processing and subsequent cytosolic accumulation of 4ICD. In the absence of TPA, Western blot analysis of ERBB4 immunoprecipitates revealed the membrane-associated 180-kDa holoreceptor and the TACEprocessed ERBB4 m80 fragment (Fig. 2, lane 2). Detection of the m80 fragment in ERBB4-V673I immunoprecipitates indicates that the V673I mutation does not impact TACE processing or ERBB4 ectodomain shedding (Fig. 2, lane 3). The treatment of ERBB4-transfected cells with TPA stimulates ␥-secretase processing of ERBB4 and subsequent accumulation of cytosolic 4ICD (14) (Fig. 2, lane 11). In contrast, TPA treatment of ERBB4-V673I-transfected cells failed to liberate 4ICD from cellular membranes, confirming that this base substitution abolishes ␥-secretase processing of the ERBB4 m80 fragment (Fig. 2, lane 12).
Western blot analysis of lysates prepared from transiently transfected COS-7 cells indicated similar levels of ERBB4 and ERBB4-V673I tyrosine phosphorylation in response to HRG (Fig. 3, IP: ERBB4 and IB: P-Tyr). Likewise, despite lower levels of ERBB4-V673I expression, HRG treatment of cells expressing ERBB4-V673I or ERBB4-activated equivalent levels of Akt and Erk1/2 phosphorylation (Fig. 3). In another experiment, ERBB4-V673I regulated phosphorylation of coexpressed STAT5A at the regulatory Tyr-694 (Fig. 5A). Taken together these results indicate that the residue substitution in ERBB4-V673I does not impact signal transduction mediated by the ERBB4 holoreceptor.
␥-Secretase Processing Is Required for ERBB4 Nuclear Translocation-We have previously demonstrated that ERBB4 nuclear translocation is mediated by a NLS harbored within 4ICD (residues 676 -684) and positioned immediately downstream of the potential ␥-secretase cleavage site (15). Other investigators have demonstrated that pharmacological inhibitors of ␥-secretase abolished ERBB4 nuclear localization (14), presumably because 4ICD was tethered at cellular membranes.
Here we have shown that ERBB4-V673I fails to generate a cytosolic 4ICD fragment (Fig. 2). We therefore reasoned that a mutation of the ␥-secretase cleavage site within ERBB4 would abolish ERBB4 nuclear translocation. In this experiment, we used deconvolution microscopy to visualize nuclear accumulation of ERBB4 proteins fused at the carboxyl terminus with EGFP. Consistent with our recent observations (15), wild-type ERBB4 expressed on the cell surface of transiently transfected MCF-7B cells (Fig. 4A, arrowhead), was translocated to the perinuclear region (Fig. 4B, arrow) and nucleus (Fig. 4, B N, and E) following HRG stimulation. Similar to other reports, HRG stimulation of ERBB4 in the presence of the ␥-secretase inhibitor, Compound E, prevented nuclear translocation of ERBB4 (14,30) and resulted in the retention of ERBB4 at the cell membrane (Fig. 4C, arrowhead) and also within perinuclear structures resembling late endosomes (Fig. 4C, arrow).  4 -6) for 30 min at room temperature, and ERBB4 was immunoprecipitated from cell lysates. Immunoprecipitates (IP) were resolved by SDS-PAGE with a 7.5% resolving gel and transferred to membrane for anti-phosphotyrosine (P-Tyr) and anti-ERBB4 Western blot analysis. In addition, 50 g of total cell lysate was resolved by SDS-PAGE with a 12% resolving gel and transferred to membrane for Western blot analysis of MAPK (P-Erk1/2) and PI3K (P-Akt) pathway activation. Western blot detection of ␣-tubulin was included as a control for protein loading.  (cys1, cys2). Proteolytic processing by TACE results in ectodomain cleavage. Subsequent presenilin-dependent ␥-secretase processing is predicted to occur at Val-673 (V673) and results in membrane release of the 4ICD (residues 673-1309). We have identified functional domains harbored within 4ICD including a NLS (residues 676 -684) (15) and a BH3 domain (residues 986 -992). 2 B, alignment of transmembrane domains from ERBB4, Notch1, and APP reveals a conserved valine residue (asterisks) with similar transmembrane positions. This valine (Val-1743) is essential for Notch1 processing by ␥-secretase, and we predict that a similar base substitution in ERBB4 (V673I) will abolish ␥-secretase processing of this receptor. APP is processed by ␥-secretase with relaxed specificity at two residues (underlined) in addition to the conserved valine. The ERBB4 NLS is located immediately downstream of the transmembrane domain (underlined).

␥-Secretase Processing Regulates ERBB4 Signaling
ERBB4-V673I was also observed at the cell surface following HRG stimulation (Fig. 4D, arrowhead). Significantly, HRGstimulated ERBB4-V673I was excluded from the nucleus (Fig.  4, D and E) and accumulated within perinuclear endosome-like structures (Fig. 4D, arrow) identical to those observed in Compound E-treated cells transfected with wild-type ERBB4 (Fig.  4C, arrow). Taken together these results strongly imply that the mutant ERBB4-V673I fails to undergo ␥-secretase processing at the cell membrane thereby preventing nuclear translocation of 4ICD.
ERBB4-V673I Activates STAT5A but Fails to Stimulate ␤-Casein Promoter Activity-We have previously demonstrated that ERBB4 activates STAT5A (10), and in the developing breast the ERBB4/STAT5A signaling pathway is essential for expression of critical milk genes during lactation (9,16,31). Furthermore, ERBB4 regulates STAT5A transcriptional activation of the ␤-casein promoter by functioning as a STAT5A nuclear chaperone (15). We therefore predict that ERBB4-V673I, which remains tethered to cellular membranes, thereby preventing nuclear accumulation of 4ICD, would lack the ability to induce STAT5A stimulation of a bovine ␤-casein promoter fused to luciferase. Co-transfection of STAT5A with ERBB4 or ERBB4-V673I resulted in significant levels of STAT5A phosphorylation at the regulatory Tyr-694 (Fig. 5A). In concordance with our previous observations (15), co-transfection of STAT5A and ERBB4 resulted in significant stimulation of the ␤-casein promoter leading to luciferase expression (Fig. 5B). In contrast, ERBB4-V673I failed to induce STAT5A stimulation of the ␤-casein promoter (Fig. 5B). The inability of ERBB4-V673I to cooperate with STAT5A and stimulate gene expression provides additional evidence that ␥-secretase processing of ERBB4 is required for nuclear translocation of the functionally active 4ICD. Furthermore, these results corroborate our recent observations demonstrating that 4ICD nuclear localization is necessary to induce STAT5A stimulation of gene expression (15).
Cytosolic/Mitochondrial 4ICD Is Necessary and Sufficient for ERBB4 Apoptotic Activity-We have demonstrated that ERBB4 induces apoptosis of malignant cell lines. Here we have determined the apoptotic activity of ERBB4-V673I fused to EGFP in transiently transfected COS-7 cells. Consistent with our unpublished observations, 2 ectopic ERBB4 and 4ICD induced significant levels of COS-7 cell killing (Fig. 6B). Interestingly, ERBB4 lacking an intact NLS (ERBB4muNLS) harbored apoptotic activity equivalent to wild-type ERBB4 (Fig.  6B). This result indicates that nuclear translocation of 4ICD is MCF-7B cells were co-transfected with the bovine ␤-casein promoter fused to luciferase and plasmids expressing the indicated cDNAs. Cell lysates were prepared at 2 days post-transfection, and luciferase activity was determined using standard methods. Results are reported as the -fold increase in luciferase activity relative to ␤-casein promoter luciferase co-transfected with empty vector controls. Each treatment was performed in duplicate, and the entire experiment was repeated three times.
␥-Secretase Processing Regulates ERBB4 Signaling dispensable for 4ICD apoptotic activity. The treatment of ERBB4-transfected cells with the ␥-secretase inhibitor, Compound E, dramatically reduced ERBB4 cell-killing activity (Fig.  6B). Moreover, apoptosis was essentially abolished in cells transfected with ERBB4-V673I, clearly demonstrating that proteolytic processing of ERBB4 by ␥-secretase is necessary for ERBB4 apoptotic activity. (Fig. 6B). Ectopic overexpression of ERBB4 and ERBB4-V673I results in the generation of the membrane-bound TACE catalyzed ERBB4 cleavage product, m80. The m80 ERBB4 product, however, lacks apoptotic activity as demonstrated by the lack of cell killing induced by ERBB4 in the presence of Compound E or ERBB4-V673I. Equivalent levels of ERBB4 and ERBB4-V673I expression (Fig.  6A) indicates that cell-killing activity of ERBB4 is regulated by proteolytic processing. Interestingly, the apoptotic activity of 4ICD was significantly greater (p Ͻ 0.01) than the ERBB4 holoreceptor suggesting that proteolytic processing of ERBB4 to release 4ICD represents the limiting step during ERBB4induced cell killing.
Our unpublished results indicate that mitochondrial accumulation of 4ICD regulates ERBB4 cell-killing activity. 2 We therefore performed an experiment to determine whether the mutation in ERBB4-V673I prevents mitochondrial accumulation of 4ICD. Cell fractionation of ERBB4 transfected SKBr3 breast cancer cells indicates 4ICD accumulation within endosomal/endoplasmic reticulum (E) and mitochondrial (M) fractions (Fig. 6C). Although ERBB4-V673I and m80 were detected within the endosomal/endoplasmic reticulum fractions of ERBB4-V673I-transfected cells, we failed to detect mitochondrial accumulation of 4ICD (Fig. 6C). These results support the contention that mitochondrial accumulation of 4ICD is essential for ERBB4 apoptotic activity and provides a mechanistic explanation for the lack of ERBB4-V673I cell-killing activity. Taken together, these data indicate that membrane-associated ERBB4 fails to activate apoptosis, and cytosolic/mitochondrial accumulation of 4ICD is both necessary and sufficient for ERBB4 cell-killing activity. DISCUSSION A canonical transmembrane signal transduction pathway involves ligand stimulation of a cell surface receptor followed by recruitment and activation of downstream signaling molecules. A cascade of protein:protein interactions or phosphorylation events or both culminate in the cell nucleus where modulated gene expression elicits specific cellular responses. The transmembrane receptor ERBB4 undergoes proteolytic processing at the cell membrane to release the 4ICD through the sequential activities of TACE (12) and presenilin-dependent ␥-secretase (13,14). Although the treatment of cells with pharmacological inhibitors of ␥-secretase results in membrane retention of 4ICD (13,14,32), the functional consequence of disrupted ERBB4 proteolytic processing remains to be established. We have employed a genetic approach to investigate the contribution of ERBB4 ␥-secretase processing to recently described ERBB4-signaling activities including activation of gene expression and induction of apoptosis.
A valine residue within the Notch1 transmembrane was shown to be essential for ␥-secretase processing of this receptor (27). We predicted that a similarly positioned ERBB4 transmembrane base substitution (V673I) would abolish ␥-secretase processing of ERBB4. Several lines of experimental evidence confirmed that the mutant receptor, ERBB4-V673I, was no longer a substrate for ␥-secretase processing. For example, the introduction of the V673I mutation into full-length ERBB4 completely abolished TPA-induced accumulation of 4ICD in cytosolic cell fractions and prevented mitochondrial accumulation of 4ICD in cells ectopically expressing ERBB4-V673I. Furthermore, we show that ERBB4-V673I failed to translocate to the nucleus and accumulated within cytosolic endosome-like structures. Similar endosomal ERBB4 accumulations were observed in cells transfected with wild-type ERBB4 and treated with the ␥-secretase inhibitor Compound E.
Using the ERBB4-V673I mutant, we investigated the impact of impaired ERBB4 proteolytic processing on the ERBB4 regulation of gene expression. We have recently identified the in vivo contribution of ERBB4 signaling to STAT5A stimulation of the ␤-casein and whey acidic protein genes during lactation (9,31) and the ␤-casein promoter in a luciferase reporter assay (15). Interestingly, ERBB4 functions as a STAT5A nuclear chaperone and ERBB4-induced STAT5A stimulation of the ␤-casein promoter required an intact ERBB4 NLS harbored Each sample was prepared in duplicate and the entire experiment was performed three times. Asterisks indicate statistically significant differences from ERBB4-EGFP (*, p Ͻ 0.01; **, p Ͻ 0.05; ***, p Ͻ 0.005). C, SKBr3 breast cancer cells transfected with the indicated cDNAs were fractionated into cytosol-rich S100 (C), an endoplasmic reticulum fraction containing endosomes (E), and mitochondria (M) purified on a 10/1.5 M sucrose step gradient. Fifty g of each fraction was analyzed by Western blot using antibodies directed against ERBB4 and the mitochondrial membrane protein TOM40. within 4ICD (15). These results implied that ERBB4 processing and subsequent nuclear translocation of 4ICD was necessary for ERBB4/STAT5A activation of the ␤-casein promoter. In support of this contention, co-expression of ERBB4-V673I and STAT5A resulted in phosphorylation of STAT5A at the regulatory Tyr-694, but ERBB4-V673I failed to induce STAT5A stimulation of the ␤-casein promoter. Taken together our results support a model of ERBB4/STAT5A stimulation of gene expression where ␥-secretase processing of ERBB4 at the cell surface is a critical event regulating the release and subsequent nuclear translocation of 4ICD (Fig. 7). Interestingly, ERBB4 and Notch1 appear to employ similar molecular mechanisms while regulating gene expression. Modulation of transcriptional activation by ERBB4 and Notch1 first requires receptor ectodomain cleavage followed by transmembrane cleavage of the processed receptor by ␥-secretase, nuclear translocation of 4ICD and the NICD, respectively, and association with a DNA-binding protein at the target gene promoter. STAT5A appears to provide the DNA binding activity for 4ICD association with the ␤-casein promoter (15), whereas Notch1 interacts with members of the CSL (CBF1, SuH, Lag-1) family of DNA-binding proteins (28,33).
In addition to a role in gene regulation, ERBB4 functions as a proapoptotic protein, inducing cell death of breast cancer cells. 2 Interestingly, independent expression of 4ICD was sufficient to mediate ERBB4 cell-killing activities with malignant cell specificity. 2 Here we have shown that ␥-secretase processing of ERBB4, generating cytosolic 4ICD, is not only sufficient but also necessary to mediate ERBB4 apoptotic activity. Indeed, the unprocessed ERBB4 receptor, ERBB4-V673I, failed to accumulate 4ICD within mitochondria and completely lacked cell-killing activity. In addition, an ERBB4 receptor that undergoes ␥-secretase processing but lacks an intact NLS, positioned within 4ICD (15), induces apoptosis at levels equivalent to wild-type ERBB4, indicating that nuclear translocation of 4ICD is dispensable for 4ICD apoptotic activity. Taken together these results support a model of ERBB4-induced apoptosis where TACE and ␥-secretase processing of activated ERBB4 at the cell surface is necessary to release the proapoptotic BH3-only protein 4ICD (Fig. 7).
An active ERBB4 intrinsic tyrosine kinase is required for both ERBB4 nuclear localization with subsequent regulation of gene expression and ERBB4-induced apoptosis. Kinase activity is dispensable, however, for nuclear translocation and the cellkilling activity of independently expressed 4ICD (15). 2 Although we have shown that ERBB4-V673I activates downstream signal transduction pathways, this mutant receptor failed to induce apoptosis or regulate STAT5A stimulation of the ␤-casein promoter. These results imply that canonical signaling pathways regulated by the holoreceptor are dispensable for these ERBB4 activities. Moreover, our results imply that the intrinsic kinase activity of ERBB4 contributes to 4ICD signaling, in part, by regulating proteolytic processing of ERBB4. Indeed, an interaction between 4ICD and presenilin 1, which harbors ␥-secretase activity, has been reported (32). Likewise, presenilin 1 directly interacts with the substrates Notch1 (36), APP (37), and ␤-catenin (38,39). Furthermore, Notch1 actively recruits presenilin 1 to the cell surface establishing the proteolytic complex (4). We are currently investigating the ability of activated ERBB4 to promote the formation of stable proteolytic complexes containing presenilin 1 and the ERBB4 substrate. Alternatively, ERBB4 kinase activity may stimulate signaling cascades that regulate proteolytic processing at the cell surface. For example, enhanced TACE activity is associated with cellular growth stimulation and intracellular kinase activity (11,40).
In summary, we have generated the ERBB4 allele, ERBB4-V673I, with a mutated ␥-secretase cleavage site that effectively abolishes two ERBB4 functional activities, 4ICD nuclear translocation with subsequent regulation of gene expression and the independent proapoptotic activity of 4ICD. Interestingly, ERBB4-V673I retains canonical signal transduction activities; however, signal transduction through the holoreceptor appears to be dispensable for both ERBB4 regulation of gene expression and ERBB4-induced apoptosis. Our results represent the first demonstration of a physiological function for ERBB4 proteolytic cleavage and underscore the essential contributions of transmembrane receptor proteolytic processing to novel signal transduction mechanisms. Growth factor (HRG) stimulation of ERBB4 results in activation of the downstream signaling pathways including MAPK and PI3K/Akt. The physiological impact of ERBB4 regulation of these signaling pathways remains to be determined. ERBB4 activation also results in sequential proteolytic processing of the receptor first by TACE followed by ␥-secretase cleavage. ERBB4 cleavage by ␥-secretase requires a transmembrane valine positioned at 673 and results in release of the ERBB4 intracellular domain (4ICD). Recruitment of STAT5A by 4ICD results in nuclear co-translocation of the two proteins, mediated by the 4ICD NLS (residues 676 -684) and subsequent activation of STAT5A target genes. The influence of 4ICD transactivation activity (TA) on STAT5A-induced gene expression remains to be determined (?). In the malignant cell, mitochondrial accumulation of 4ICD induces apoptosis through the activity of an intrinsic BH3 domain and proapoptotic members of the BCL-2 family. ERBB4mediated activation of gene expression and induction of apoptosis requires ␥-secretase processing of ERBB4 at the cell surface to release 4ICD; however, ERBB4 activates the MAPK and PI3K/Akt signal transduction pathways independent of ␥-secretase processing.