Activation of ErbB4 by the Bifunctional Epidermal Growth Factor Family Hormone Epiregulin Is Regulated by ErbB2*

Epiregulin (EPR) is a recently described member of the epidermal growth factor (EGF) family of peptide growth factors. The ever expanding size of the EGF family has made distinguishing the activities of these hormones paramount. We show here that EPR activates two members of the ErbB family of receptor tyrosine kinases, epidermal growth factor receptor (EGFR) and ErbB4. Therefore by these criteria, EPR is qualitatively similar to another EGF family hormone, betacellulin (BTC). Yet, here we also demonstrate quantitative differences between EPR and BTC. EPR stimulates higher levels of EGFR phosphorylation than does BTC, whereas BTC stimulates higher levels of ErbB4 phosphorylation than does EPR. Moreover, the EPR and BTC dose response curves show that although EGFR is more sensitive to EPR than is ErbB4, ErbB4 is more sensitive to BTC than is EGFR. Finally, ErbB2, which is not activated by EPR when expressed on its own, increases the sensitivity of ErbB4 for activation by EPR. Therefore, these results establish that EPR exhibits novel activities and modes of regulation, which may have significant implications for EPR function in vivo.

EPR was initially purified from the conditioned medium of a tumorigenic clone of NIH3T3 fibroblasts. It competes with EGF for binding to A431 cells, which overexpress EGFR, suggesting that EPR is a ligand for EGFR (17). Since at least one of the EGF family hormones that activates EGFR also activates ErbB4, we wished to evaluate EPR function in a set of cell lines expressing all four ErbB family receptors, both singly and in every pairwise combination.
We demonstrate here that EPR activates not only EGFR, but ErbB4 as well. However, the dose-response curves for BTC and EPR in a cell line expressing both EGFR and ErbB4 are markedly different. Whereas ErbB4 is more responsive to BTC than is EGFR, ErbB4 is less responsive to EPR than is EGFR. Moreover, ErbB2 expression increases saturated ErbB4 phosphorylation in response to EPR and dramatically enhances the sensitivity of ErbB4 for activation by EPR as well. In this respect EPR resembles NRG, which displays a low affinity for ErbB3 that increases in cells where ErbB2 is co-expressed (12).

EXPERIMENTAL PROCEDURES
Cell Lines and Cell Culture-BaF3 is an immortal mouse lymphoblastoid cell line (31). BaF3-derived cell lines expressing combinations of ErbB family receptors have been described previously (14). The ranked order of receptor expression in the double recombinant BaF3 cell lines is as follows. For EGFR expression, BaF3/EGFRϩErbB4 is higher than BaF3/EGFRϩErbB2, which is higher than BaF3/ EGFRϩErbB3. For ErbB2 expression, BaF3/ErbB2ϩErbB4 is equivalent to BaF3/EGFRϩErbB2, both of which are markedly higher than BaF3/ErbB2ϩErbB3. The levels of ErbB3 expression are similiar in the BaF3/EGFRϩErbB3, BaF3/ErbB2ϩErbB3, and BaF3/ErbB3ϩErbB4 cell lines. The levels of ErbB4 expression are similiar in the BaF3/ EGFRϩErbB4, BaF3/NeuϩErbB4, and BaF3/ErbB3ϩErbB4 cell lines (7,14).
CEM is an immortal human T-lymphoblastoid cell line that does not endogenously express EGF receptor, ErbB2, ErbB3, or ErbB4. CEMderived cell lines expressing ErbB4 or ErbB2 and ErbB4 have been described previously (10). Cell culture conditions were as described (10,14).
Growth Factors-Recombinant human EPR was produced in Bacillus brevis. 2  Stimulation and Analysis of Receptor Phosphorylation-The conditions for stimulation of ErbB family receptor tyrosine phosphorylation have been described previously (7,14). The analysis of ErbB family receptor tyrosine phosphorylation by immunoprecipitation and antiphosphotyrosine immunoblotting has been described previously (7,14). Immunoprecipitating antireceptor antibodies were anti-EGFR mouse monoclonal antibody 528 (33), anti-ErbB2 mouse monoclonal antibody TA-1 (OP-39, Calbiochem), anti-ErbB3 rabbit polyclonal antiserum SC-285 (Santa Cruz Biotechnology), and anti-ErbB4 rabbit polyclonal antiserum SC-283 (Santa Cruz Biotechnology). Specificity of antireceptor antibodies has been verified by testing for cross-reactivity (data not shown).
Immunoblot autoradiographs were digitized using a Hewlett-Packard 3p flatbed scanner set for 600 dpi resolution and controlled by Hewlett-Packard Deskscan II for Macintosh software. Images were cropped using Adobe Photoshop, and the band intensity was quantified using NIH Image software. Net receptor activation was calculated by subtracting the amount of tyrosine phosphorylation observed in samples from mock-stimulated cells.

RESULTS
EPR Activates EGFR-We first sought to identify which ErbB family receptors are activated by EPR when the receptors are expressed individually. We previously developed a panel of cell lines based on the mouse BaF3 hematopoietic cell line that expresses the four ErbB family receptors, both singly and in every pairwise combination. Hence, we incubated the BaF3 cell lines ectopically expressing EGFR, ErbB2, or ErbB4 with EPR. EPR, like BTC, stimulated EGFR tyrosine phosphorylation, consistent with published results suggesting that EPR binds EGFR (17) (Fig. 1, EGFR panel; compare lanes E and B with M). However, EPR did not stimulate phosphorylation of ErbB2 or ErbB4 (Fig. 1, ErbB2 and ErbB4 panels; compare lanes E and M). In contrast, the positive control NRG␤ stimulated ErbB2 tyrosine phosphorylation and BTC stimulated ErbB4 phosphorylation. The ErbB2 phosphorylation observed in BaF3/ErbB2 cells stimulated with NRG is the result of transmodulation of ErbB2 by the NRG receptor ErbB3, which is endogenously expressed at low levels in BaF3 cells (14). Neither EPR nor any of the other EGF family ligands tested to date stimulated ErbB3 tyrosine phosphorylation in BaF3 cells expressing only ErbB3 (data not shown) (7,8,14). However, since ErbB3 lacks tyrosine kinase activity (18), these experiments do not rule out EPR binding to ErbB3.
Since EPR activates EGFR, we next determined whether EPR activates the other three ErbB family receptors in trans via EGFR. A panel of BaF3 cell lines ectopically expressing EGFR together with one of the other three ErbB family receptors was stimulated with EPR. EPR activated the EGFR in all three cell lines ( Fig. 2A; compare E ␣1 lanes with the M ␣1 lanes). Both EPR (E lanes) and the positive control TGF-␣ (T lanes) strongly activated ErbB2 in the cell line co-expressing EGFRϩErbB2 ( Fig. 2A, EGFRϩErbB2 panel; ␣2 lanes). In contrast, neither EPR nor TGF-␣ activated ErbB3 or ErbB4 ( Fig. 2A, EGFRϩErbB3 and EGFRϩErbB4 panels; ␣3 or ␣4 lanes). This is consistent with the conclusion that ErbB2 is a preferential target for transmodulation by the EGFR compared with the other ErbB family receptors (7, 8, 19 -21). However, higher concentrations of EPR than those used for these experiments did stimulate ErbB4 phosphorylation in the EGFRϩErbB4 cell line (see below; Fig. 6A).
EPR Activates ErbB4 in CEM Cells-BTC activates both ErbB4 and EGFR when expressed individually (7). We tested whether EPR behaves like BTC and also activates ErbB4 when expressed alone using derivatives of the CEM human T-lymphoblastoid cell line that ectopically expresses ErbB4 alone or both ErbB2 and ErbB4 (10). EPR activated ErbB4 in CEM cells expressing ErbB4 alone and both receptors in CEM cells expressing ErbB2 and ErbB4 together ( Fig. 3; compare E lanes with M lanes). In experiments done in parallel using identical growth factor concentrations, EPR did not activate ErbB4 in BaF3 cells expressing ErbB4 alone (also see Fig. 1) but stimulated ErbB2 and ErbB4 phosphorylation in BaF3 cells expressing both ErbB2 and ErbB4 (data not shown). It is unclear why EPR failed to activate ErbB4 in the BaF3 cells expressing ErbB4 alone. Nonetheless, because EPR activates EGFR as well as ErbB4, EPR resembles BTC, which also activates these receptors (7).
EPR Stimulates EGFR More Than ErbB4 and EGFR Is More Sensitive Than ErbB4 to EPR-Since BTC and EPR can activate EGFR and ErbB4 when the receptors are expressed individually, we measured BTC and EPR stimulation of EGFR and ErbB4 phosphorylation in BaF3 cells that express both receptors together (Fig. 4). BTC stimulated saturated levels of EGFR phosphorylation at a concentration of 10 ng/ml, whereas BTC stimulated saturated levels of ErbB4 phosphorylation at a concentration of 25 ng/ml. Therefore, in subsequent experiments we have assumed that 100 ng/ml BTC stimulates saturated levels of EGFR and ErbB4 phosphorylation.
At a concentration of 4000 ng/ml, EPR stimulates slightly greater levels of EGFR phosphorylation than does BTC (Fig. 4).
In contrast, 4000 ng/ml EPR stimulates much lower levels of ErbB4 phosphorylation than does BTC (Fig. 4). This suggests that EPR stimulates EGFR more than ErbB4 and that EGFR is more sensitive than ErbB4 to EPR. We investigated this possibility by comparing EGFR and ErbB4 phosphorylation in BaF3/EGFR and CEM/ErbB4 cells stimulated with increasing concentrations of EPR (Fig. 5A). At saturation, EPR stimulated a slightly higher level of EGFR phosphorylation (1.7-fold) than did BTC, whereas EPR stimulated a much lower level of ErbB4 phosphorylation (0.3-fold) than did BTC (Fig. 5C). We next compared the dose sensitivity of EGFR and ErbB4 to EPR stimulation by plotting receptor phosphorylation relative to the maximal amounts of receptor phosphorylation stimulated by EPR (Fig. 5D) to identify the EPR concentrations required for half-maximal receptor phosphorylation (Table I). Half-maximal EGFR phosphorylation occurred at an EPR concentration of approximately 380 ng/ml ( Fig. 5D; Table I). In contrast, half-maximal ErbB4 phosphorylation required about a 4-fold higher concentration of EPR with half-maximal activation occurring at an EPR concentration of approximately 1790 ng/ml ( Fig. 5D; Table I).

ErbB2 Expression Increases ErbB4 Activation by EPR and Sensitivity to EPR-The affinity of NRG for cells expressing
ErbB3 is greater when these cells also express ErbB2 (12). Hence, we evaluated the possibility that ErbB2 modulates ErbB4 activation by EPR by stimulating CEM/ErbB4 and CEM/ErbB2ϩErbB4 cells with increasing concentrations of EPR (Fig. 5, A and B). Relative to the BTC positive control, EPR stimulated 2-fold higher levels of ErbB4 phosphorylation in the ErbB2ϩErbB4 cell line than in cells expressing ErbB4 alone (Fig. 5C). Therefore, ErbB2, which is not activated by EPR when expressed by itself (Fig. 1), doubles the magnitude of ErbB4 activation by EPR at saturation. We next examined the effects of ErbB2 expression on the sensitivity of ErbB4 to EPR (Fig. 5D). Half-maximal ErbB4 phosphorylation occurred at an EPR concentration of approximately 1790 ng/ml in the cell line expressing ErbB4 alone (Fig. 5D; Table I), but occurred at an EPR concentration of approximately 630 ng/ml ( Fig. 5D (10) were stimulated with 100 ng/ml BTC (B), 100 ng/ml recombinant NRG␤ (N), 1000 ng/ml EPR (E), or were mock stimulated with phosphate-buffered saline (M) as described previously (14). ErbB2 (␣2) or ErbB4 (␣4) was immunoprecipitated from lysed cells using specific antireceptor antibodies and separated by SDS-PAGE as described previously (7,14). Tyrosinephosphorylated ErbB family receptors were detected and visualized by immunoblotting using the 4G10 monoclonal antiphosphotyrosine antibody as described previously (7,14).  (14) were stimulated with 10, 25, 50, 75, or 100 ng/ml BTC, 4000 ng/ml EPR, or mock stimulated with phosphate-buffered saline as described previously (7,8,14). EGFR or ErbB4 was immunoprecipitated from lysed cells using specific antireceptor antibodies and separated by SDS-PAGE as described previously (7,14). Tyrosine-phosphorylated ErbB family receptors were detected and visualized by immunoblotting using the 4G10 monoclonal antiphosphotyrosine antibody as described previously (7,14).

ErbB4 homodimers.
The EPR and BTC Dose-Response Curves Are Different in Cells Expressing EGFR and ErbB4 -EPR resembles BTC in its ability to activate either EGFR or ErbB4 when expressed individually (7). Yet, at saturation EPR stimulated almost 2-fold more EGFR phosphorylation than BTC, whereas BTC activated about 3-fold more ErbB4 phosphorylation than did EPR (Fig. 5, A and C). This suggested that BTC and EPR are functionally distinct. Hence, we compared EGFR and ErbB4 phosphorylation following stimulation with increasing concentrations of BTC or EPR in a BaF3 cell line that expresses both EGFR and ErbB4 (Fig. 6A).
We first compared the magnitude of receptor phosphorylation stimulated by EPR and BTC by plotting receptor phosphorylation relative to the maximal phosphorylation stimulated by BTC (Fig. 6, B and C). In agreement with results presented above (Fig. 5, A and C), EPR stimulated higher saturated levels of EGFR phosphorylation than BTC, whereas BTC activated greater ErbB4 phosphorylation than did EPR (Fig. 6, A and C). However, the magnitude of these differences was much less in the EGFRϩErbB4 cell line compared with the differences in phosphorylation that we observed between the cell lines expressing EGFR and ErbB4 individually (Fig. 5, A and C).
Next, we compared the sensitivities of EGFR and ErbB4 with BTC and EPR by identifying the growth factor concentrations required for half-maximal receptor phosphorylation. Half-maximal EGFR activation occurred at a BTC concentration of approximately 35 ng/ml, whereas half-maximal ErbB4 activation occurred at a BTC concentration of approximately 5 ng/ml ( Fig.  6B; Table I). In contrast, half-maximal EGFR activation occurred at an EPR concentration of approximately 320 ng/ml, whereas half-maximal ErbB4 activation occurred at an EPR concentration of approximately 790 ng/ml ( Fig. 6C; Table I).

FIG. 5. EPR dose response in BaF3 cells expressing EGFR or in CEM cells expressing ErbB4 alone or both ErbB2 and ErbB4.
A and B, BaF3/EGFR, CEM/ErbB4, or CEM/ErbB2ϩErbB4 cells were stimulated with 100 ng/ml BTC (BTC), increasing concentrations of epiregulin as indicated or were mock stimulated with phosphate-buffered saline (Mock) as described previously (7,14). EGFR, ErbB2, or ErbB4 was immunoprecipitated as indicated or appropriate using specific antireceptor antibodies and separated by SDS-PAGE as described previously (14). Tyrosine-phosphorylated ErbB family receptors were detected and visualized by immunoblotting using the 4G10 monoclonal antiphosphotyrosine antibody as described previously (7,14). C and D, antiphosphotyrosine immunoblot images were scanned on a Hewlett-Packard ScanJet 3p flatbed scanner set for 600 dpi optical resolution. Images were cropped using Adobe Photoshop and receptor tyrosine phosphorylation was quantified using NIH Image. Net receptor tyrosine phosphorylation was calculated by subtracting the receptor tyrosine phosphorylation exhibited by mock stimulated cells. Tyrosine phosphorylation was either expressed relative to the tyrosine phosphorylation stimulated by 100 ng/ml BTC (C) or relative to the maximal receptor tyrosine phosphorylation stimulated by EPR (D). ", EGFR; q, ErbB4; ⅜, 2ϩ4 Anti-4. This suggests that ErbB4 is 7-fold more sensitive to BTC than is EGFR, whereas EGFR is more than 2-fold more sensitive to EPR than is ErbB4. Finally, these results illustrate that EGFR expression, like ErbB2 expression, shifts the EPR dose-response curve in cells expressing ErbB4. Half-maximal ErbB4 phosphorylation in a CEM cell line expressing ErbB4 alone occurs at an EPR concentration of 1790 ng/ml ( Fig. 5D; Table I). In contrast, halfmaximal ErbB4 phosphorylation in BaF3 cells expressing both EGFR and ErbB4 occurs at 790 ng/ml ( Fig. 6C; Table I).
EPR Activates ErbB Family Receptor Coupling to IL3 Independence-Although EPR and BTC stimulate qualitatively identical patterns of receptor phosphorylation, these hormones are quantitatively distinct. One possible mechanism is that EPR and BTC stimulate EGFR and ErbB4 tyrosine phosphorylation at different sites. This would account for the higher levels of EGFR activation by EPR compared with BTC and the higher levels of ErbB4 activation by BTC compared with EPR. Moreover, this would also enable these hormones to couple to distinct receptor effectors and physiologic responses. Therefore, we compared EPR and BTC induction of receptor coupling with physiologic responses. BaF3 cells require interleukin-3 (IL3) for survival and for proliferation. Activation of either EGFR or ErbB2 permits survival of BaF3 cells in the absence of IL3 (7,14). However, ErbB4 activation by either NRG or BTC is not coupled to IL3-independent survival (7,14). In BaF3 cells expressing both ErbB2 and ErbB4 together, activation by either BTC or NRG induces IL3-independent survival, presumably through ErbB2 transmodulation by ErbB4 (7,14).
EPR, like BTC (7), induces IL3-independent survival in BaF3 cells expressing EGFR, but not in vector control BaF3 cells or cells expressing ErbB2 (Fig. 7). (The IL3-independent response of BaF3 cells expressing ErbB2 to NRG is the result of ErbB2 transmodulation by endogenous ErbB3 in these cells (14).) EPR, BTC, and NRG all induced IL3 independence in cells co-expressing ErbB2 and ErbB4 (Fig. 7). This implies that BTC and NRG are functionally equivalent. However, the response to BTC and NRG is greater than the response to EPR, which may reflect subtle functional differences between BTC and EPR. DISCUSSION We previously demonstrated that the EGF family of peptide growth factors can be divided into three distinct functional groups (8) (Fig. 8). The first group consists of EGF, TGF-␣, and amphiregulin. These hormones bind and activate only the EGFR, but they can activate the other three ErbB family receptors in trans via heterodimerization with the EGFR. The second group consists of NRG and NRG2, which bind ErbB3 and ErbB4 and transmodulate EGFR and Neu via the binding receptors. The third group consists of BTC, which binds and activates both EGFR and ErbB4. Recent data suggests that heparin-binding EGF-like growth factor may also bind and activate EGFR and ErbB4, which would make heparin-binding EGF-like growth factor a member of this group as well (22).
Although EPR activates both EGFR and ErbB4, the interactions of EPR with these two receptors appear to be quite different. Compared with BTC, EPR stimulates higher levels of EGFR phosphorylation and lower levels of ErbB4 phosphorylation. Whereas both EPR and BTC stimulate EGFR and ErbB4 homodimerization and signaling, the geometry of the receptor dimers induced by EPR and BTC may be subtly different. The alignment of the kinase domain of one receptor molecule of a receptor homodimer with the autophosphorylation site of the other receptor molecule following EPR stimulation could be different from this alignment following BTC stimulation, affecting the cross-phosphorylation within receptor dimers. Al-FIG. 6. EPR and BTC dose response in BaF3 cells expressing both EGFR and ErbB4. A, BaF3/EGFRϩErbB4 cells were stimulated with increasing concentrations of betacellulin or epiregulin or were mock stimulated with phosphate-buffered saline (Mock) as described previously (7,14). EGFR or ErbB4 was immunoprecipitated as indicated using specific antireceptor antibodies and separated by SDS-PAGE as described previously (7,14). Tyrosine phosphorylated ErbB family receptors were detected and visualized by immunoblotting using the 4G10 monoclonal antiphosphotyrosine antibody as described previously (7,14). B and C, antiphosphotyrosine immunoblot images were scanned on a Hewlett-Packard ScanJet 3p flatbed scanner set for 600 dpi optical resolution. Images were cropped using Adobe Photoshop and receptor tyrosine phosphorylation was quantified using NIH Image. Net receptor tyrosine phosphorylation was calculated by subtracting the receptor tyrosine phosphorylation exhibited by mock stimulated cells. Tyrosine phosphorylation stimulated by BTC (B) or EPR (C) was expressed relative to the maximal tyrosine phosphorylation stimulated by BTC. ", EGFR; Ⅺ, ErbB4.
ternatively, ligand-induced changes in the conformation of the receptor kinase domains might be different when the receptors are activated by BTC and EPR. Therefore, BTC and EPR may differentially stimulate receptor kinase activity. In either scenario, BTC and EPR could stimulate receptor autophosphorylation on different tyrosine residues, which could also be reflected in differences in gross levels of receptor phosphorylation. In this manner BTC and EPR could differentially modulate receptor coupling to signaling effectors and physiologic responses.
Another difference between EPR and BTC is that while EGFR is much more sensitive than ErbB4 to EPR, EGFR is less sensitive than ErbB4 to BTC. This suggests that although the affinity of EPR for EGFR is higher than the affinity for ErbB4, the affinity of BTC for EGFR is lower than the affinity for ErbB4. This too suggests that BTC and EPR have distinct biological functions, even in cells with identical patterns of ErbB family receptor expression.
Another important aspect of EPR function is the observation that the sensitivity of ErbB4 for EPR and the magnitude of ErbB4 activation by EPR can be modulated by the expression of other ErbB family receptors. EGFR expression increases the sensitivity of ErbB4 for EPR, suggesting that EGFR-ErbB4 heterodimers have a higher affinity for EPR than ErbB4-ErbB4 homodimers (Fig. 5, A and D; Fig. 6, A and C; Table I). Of course an alternative explanation is that the increased ErbB4 sensitivity in the presence of EGFR is due solely to EPRinduced transphosphorylation of ErbB4 by EGFR.
ErbB2 also increases the sensitivity of ErbB4 for EPR (Fig. 5, A-D; Table I). Because EPR does not activate ErbB2 in cells devoid of EGFR or ErbB4 (Figs. 1 and 7), the mechanism for the increased sensitivity of ErbB4 for EPR may be that ErbB2-ErbB4 heterodimers have a higher affinity for EPR than do ErbB4-ErbB4 homodimers. These results resemble observations made with NRG, which does not bind to ErbB2, binds with low affinity to cells expressing ErbB3, and binds with higher affinity to cells that express both ErbB2 and ErbB3 (12). ErbB2 expression also increases the magnitude of ErbB4 activation by EPR.
These observations that ErbB4 activation by EPR can be influenced by EGFR or ErbB2 is consistent with existing models for receptor heterodimerization and transmodulation. It has been proposed that receptor heterodimerization is mediated through low affinity hormone-receptor interactions and heterotypic receptor-receptor contacts, after which there is crossphosphorylation by the receptor kinase domains (23). It is possible that EGFR and ErbB2 are favored over ErbB4 for dimerization with ErbB4 in the presence of EPR. Therefore, there would be greater ErbB4 dimerization in cells expressing EGFR and ErbB4 or Neu and ErbB4 than in cells expressing ErbB4 alone. This may account for the increased sensitivity of ErbB4 for EPR in the presence of EGFR or ErbB2. It is also possible that ErbB2 is a better kinase for ErbB4 than ErbB4 itself. Consequently, ErbB2 may cross-phosphorylate more ErbB4 tyrosine residues in receptor heterodimers than ErbB4 would in receptor homodimers. Similarly, ErbB2 may phospho- FIG. 7. EPR stimulation of IL3-independent responses in BaF3 cells expressing various ErbB family receptors. The IL3-independent responses of BaF3/LXSN (vector control), BaF3/EGFR, BaF3/ErbB2, and BaF3/ErbB2ϩErbB4 cells to EPR stimulation were assayed as described earlier (7,14). Briefly, cells were seeded in duplicate or triplicate at an initial density of 100 ϫ 10 3 cells/ml in medium lacking IL3 (u), containing IL3 (^), or lacking IL3 but supplemented with 10 ng/ml EPR (d), 10 ng/ml BTC (z), 10 ng/ml synthetic NRG␤ (_) or 10 ng/ml TGF-␣ (s). After seeding, samples were taken every 24 h, and the viable cell density was calculated by staining cells with trypan blue and counting them in a hemocytometer. Samples were taken until the viable cells reached a saturation density. The mean and standard error densities for three to seven trials are shown. NT, not tested. Unpublished work has demonstrated that 10 ng/ml BTC induces saturated amounts of IL3 independence in a variety of BaF3 cell lines, whereas 10 ng/ml EPR induces saturated amounts of IL3 independence in the BaF3/EGFR cell line. rylate the same tyrosine residues as ErbB4 to a greater extent than does ErbB4. Either of these last two possibilities would account for the increased tyrosine phosphorylation of ErbB4 by EPR in the presence of ErbB2.
As this manuscript was being prepared for submission, it was reported that radiolabeled EPR can be cross-linked to EGFR and ErbB4 in human breast tumor cell lines but not to ErbB2 or ErbB3. Furthermore, EPR stimulated high levels of EGFR and ErbB4 tyrosine phosphorylation and more modest levels of ErbB2 and ErbB3 tyrosine phosphorylation (24). Because the cell lines used in these studies express at least two and in some cases all four ErbB family receptors, some caution must be used in interpreting these results. Nonetheless, these data are entirely consistent with our findings that EGFR and ErbB4 are the receptors for EPR.
To date there have been only a few clues to EPR function. EPR transcripts are not detected in normal adult mouse liver, kidney, brain, spleen, testis, or skeletal muscles. However, low levels of EPR transcripts are detectable in adult mouse lung, smooth muscle, and heart, whereas more robust EPR transcription is observed in whole embryo RNA samples from 7-day-old mouse embryos (25). 3 This implies that EPR plays a significant role in early mammalian development but only a limited role in adult tissues.
Additional hints to EPR function arise from our data suggesting that EPR is a ligand for both EGFR and ErbB4. In most contexts EGFR activation is coupled to cellular DNA synthesis and proliferation. In contrast, there is mounting evidence that activated ErbB4 is coupled to growth inhibition, differentiation, and possibly tumor suppression. NRG, a ligand for ErbB3 and ErbB4, inhibits the proliferation and stimulates the differentiation of a number of human breast tumor cell lines (26), whereas NRG implants stimulate the differentiation of the mouse mammary epithelium in vivo (27). BTC stimulates the differentiation of pancreatic AR42J cells into insulin-secreting cells, but EGF and TGF-␣ do not (28). Finally, agonistic anti-ErbB4 antibodies stimulate the differentiation of human breast tumor cell lines (29), and ErbB4 overexpression in breast cancer patients correlates with progesterone receptor expression, which is a marker for longer disease-free survival and better overall prognosis (30). Because EPR is a ligand for both EGFR and ErbB4, EPR may act as a proliferative agent in cells expressing EGFR and may act as a differentiation agent in cells that express ErbB4.
Furthermore, because EPR activation of ErbB4 is regulated by ErbB2 and activated ErbB2 appears to couple to mitogenesis and cell proliferation, the effects of EPR on cells expressing ErbB4 may be tightly linked to a balance of ErbB4 and ErbB2 expression. In cells having relatively low levels of ErbB2, EPR may have little effect because it fails to bind to ErbB4, and in cells having moderate levels of ErbB2 and high levels of ErbB4, EPR may act as a differentiation agent and inhibit cell proliferation, because the relatively high levels of ErbB4 signaling may overcome the effects of ErbB2 signaling. Finally, in cells having relatively high levels of ErbB2 relative to ErbB4, EPR may stimulate such high levels of ErbB2 signaling that the effects of ErbB4 signaling are overcome, and cell proliferation is stimulated. In sum, our data suggests that the physiologic response to EPR will be dictated by relative levels of EGFR, ErbB2, and ErbB4 expression and not just the absolute level of expression of any single ErbB family receptor.