Nuclear ErbB4 signaling through H3K9me3 is antagonized by EGFR-activated c-Src

Protein-tyrosine kinases are found in multicellular organisms and have important roles in a variety of the intracellular signal transduction pathways, which are critical in the control of cellular processes, such as cell proliferation, differentiation, gene expression, cell adhesion, and metabolic changes (Hubbard and Till, 2000). Receptor-type tyrosine kinases, a subclass of transmembrane-spanning receptor, transmit signals across the plasma membrane from extracellular milieus to the inside of cells (Ullrich and Schlessinger, 1990). On the other hand, the function of non-receptor-type tyrosine kinases depends upon their intracellular localization (Neet and Hunter, 1996). Although tyrosine phosphorylation is mainly associated with cytoplasmic events, some types of tyrosine kinases and phosphatases localize within the nucleus and nuclear tyrosine phosphorylation may play a role in nuclear events (Cans et al., 2000; Moorhead et al., 2007). We have shown that Lyn, a member of non-receptor-type Src-family tyrosine kinases (SFKs), is imported into and rapidly exported from the nucleus (Ikeda et al., 2008), and nuclear tyrosine phosphorylation by SFKs, Csk and c-Abl has a role in changes of chromatin structure and histone modification (Yamaguchi et al., 2001; Nakayama and Yamaguchi, 2005; Takahashi et al., 2009; Aoyama et al., 2011).

The ErbB receptor tyrosine kinases comprise the four members, epidermal growth factor receptor (EGFR)/ErbB1, HER2/ErbB2, ErbB3 and ErbB4. ErbB-family members are known to play critical roles in regulating cell proliferation, differentiation, apoptosis and migration (Yarden and Sliwkowski, 2001). Each of the receptors is a type I transmembrane protein and is composed of: (1) an extracellular domain that provides a ligand-binding site; (2) a single transmembrane domain; and (3) an intracellular domain that encodes a tyrosine kinase catalytic domain and a C-terminal region. The cytoplasmic domain contains multiple tyrosine residues (Schulze et al., 2005), and the receptor function is regulated by tyrosine phosphorylation. For instance, EGFR signaling is positively regulated by SFKs through tyrosine phosphorylation (Dittmann et al., 2008; Iida et al., 2013).

ErbB4 is known to have a unique signaling feature among the ErbB-family members. Stimulation with NRG-1, a ligand of ErbB4, not only activates downstream signaling molecules close to or on the cytoplasmic face of the plasma membrane but also cleaves the ErbB4 extracellular domain and the intracellular domain by TNF-α-converting enzyme (TACE) and γ-secretase, respectively (Rio et al., 2000; Ni et al., 2001). Cleavage by TACE removes the extracellular domain and leaves the membrane-associated 80-kDa ErbB4 fragment (m80 ErbB4). The m80 ErbB4 fragment is further cleaved by γ-secretase to release the soluble 80-kDa ErbB4 fragment, termed the ErbB4 intracellular domain (4ICD), into the cytoplasm. Then, 4ICD acts as a non-receptor-type tyrosine kinase and can shuttle between the cytoplasm and the nucleus. It is evident that 4ICD can signal directly to the nucleus to regulate transcription (Ni et al., 2001; Carpenter, 2003; Williams et al., 2004; Sardi et al., 2006). Despite the existence of multiple tyrosine phosphorylation sites on 4ICD (Kaushansky et al., 2008), the regulation of 4ICD nuclear signaling through tyrosine phosphorylation is poorly understood.

In this study, we investigate ErbB4-mediated nuclear signaling upon stimulation with NRG-1. Generation of 4ICD increases the levels of H3K9me3, which is antagonized by EGF-activated EGFR, another member of the ErbB family. We further identify activated c-Src downstream of EGFR as a negative regulator of nuclear ErbB4 signaling. Our findings demonstrate that tyrosine phosphorylation of 4ICD by EGFR-activated c-Src plays an inhibitory role in nuclear ErbB4 signaling.

To examine the role of ErbB4 signaling in the nucleus, T47D breast cancer cells were stimulated with NRG-1. We observed that endogenous ErbB4 was activated at 5 min and subsequent 4ICD release was peaked at 30 min after NRG-1 stimulation (Fig. 1A), consistent with previous observations (Ni et al., 2001). NRG-1 stimulation was found to decrease high-intensity areas (red and yellow) and increase low-intensity areas (green and blue) of DNA staining, leading to an increase in nucleus size (Fig. 1B). Upon NRG-1 stimulation, treatment with AG1478, an inhibitor of EGFR and ErbB4 (Egeblad et al., 2001), blocked NRG-1-induced 4ICD release and an increase in nuclear size (supplementary material Fig. S1A,B). Since NRG-1 stimulation did not induce BrdU incorporation (supplementary material Fig. S1C), the increase in nuclear size may be not due to DNA synthesis but rather chromatin structural changes.

We next examined histone modifications, because the regulation of chromatin structure is known to involve changes of histone modifications, such as methylation and acetylation (Jenuwein and Allis, 2001; Aoyama et al., 2011). Intriguingly, unlike trimethylated histone H3 on lysine 4 (H3K4me3) and acetylated histone H4 on lysine 16 (H4K16ac), the levels of H3K9me3 in T47D cells were increased by NRG-1 stimulation of ErbB4 but decreased by EGF stimulation of EGFR (Fig. 1C; supplementary material Fig. S1D). NRG-1 stimulation was also found to increase H3K9me3 levels in ErbB4-transfected COS-1 cells (supplementary material Fig. S2A). These results suggest that ErbB4 and EGFR, both of which are members of the ErbB-family, have opposing effects on H3K9me3.

4ICD has one functional nuclear localization signal (NLS) and three putative nuclear export signals, and can shuttle between the cytoplasm and the nucleus (Ni et al., 2001; Williams et al., 2004). We asked whether an ErbB4-mediated increase of H3K9me3 involved the release of 4ICD and its nuclear translocation. When treatment with DAPT, a γ-secretase inhibitor, prevented γ-secretase processing of ErbB4 and accumulated the membrane-associated 80-kDa ErbB4 fragment (m80 ErbB4; Fig. 2A) (Määttä et al., 2006; Sardi et al., 2006; Paatero et al., 2012), we examined the effect of 4ICD generation on H3K9me3 levels. Fig. 2B shows that DAPT treatment blocked the NRG-1-induced increase of H3K9me3 levels, suggesting that 4ICD release has a critical role in NRG-1-induced increase of H3K9me3 levels. Next, to examine whether nuclear translocation of 4ICD increased H3K9me3 levels, we constructed NLS-4ICD by linking an additional NLS to the N-terminus of 4ICD (Fig. 2C). When 4ICD and NLS-4ICD were expressed in COS-1 cells, NLS-4ICD was mainly localized in the nucleus and the levels of H3K9me3 were much higher in cells expressing NLS-4ICD than those expressing 4ICD (Fig. 2D). Furthermore, to examine the effect of the kinase activity on H3K9me3, we compared NLS-4ICD with the kinase-dead mutant NLS-4ICD(KD) and an irrelevant NLS-tagged Src tyrosine kinase (NLS-Src; Fig. 2C). Although similar levels of nuclear localization were observed among NLS-4ICD, NLS-4ICD(KD), and NLS-Src (supplementary material Fig. S2B; Fig. 2F), the tyrosine-phosphorylated band pattern induced by NLS-4ICD was somewhat different from that induced by NLS-Src (Fig. 2E). Notably, H3K9me3 levels were much higher in cells expressing NLS-4ICD than those expressing NLS-4ICD(KD) or NLS-Src (Fig. 2F,G). In contrast to H3K9me3, H3K4me3 levels in cells expressing NLS-4ICD were comparable to those in cells expressing NLS-4ICD(KD) or NLS-Src (Fig. 2G). These results suggest that nuclear translocation of 4ICD is critical for an increase of H3K9me3 levels in a manner dependent on 4ICD kinase activity.

ErbB4 activation decreases cell proliferation and induces differentiation, whereas the other ErbB-family members are involved in cell proliferation and survival (Yarden and Sliwkowski, 2001; Sardi et al., 2006). Given that NRG-1 and EGF bind to ErbB4 and EGFR, respectively (Linggi and Carpenter, 2006), expression levels of NRG-1 and EGF are well controlled in developing mammary glands (Schroeder and Lee, 1998). We thus hypothesized that H3K9me3 levels may be controlled by signal balance between ErbB4 and EGFR. To test this hypothesis, cells were co-stimulated with NRG-1 plus EGF. As a result, NRG-1-induced increase of H3K9me3 levels were decreased by concomitant stimulation with EGF (Fig. 3A). In addition, inhibition of cell proliferation by NRG-1 was substantially rescued by EGF stimulation (Fig. 3B). However, the effect of EGF on NRG-1-repressed cell proliferation was diminished after 4 days of stimulation (Fig. 3B), because EGF stimulation decreased the level of EGFR expression in a 4-day stimulation period (supplementary material Fig. S3) (Dikic and Giordano, 2003). Furthermore, concomitant stimulation with EGF and NRG-1 was incapable of inhibiting NRG-1-induced ErbB4 activation and 4ICD release (Fig. 3C). Taken together, these results suggest that EGFR activation blocks intracellular ErbB4 signaling toward an increase of H3K9me3 and growth inhibition.

It is known that the binding of EGF to EGFR activates SFKs, ERK1/2, and PI3K (supplementary material Fig. S1D; Yarden and Sliwkowski, 2001). To ask whether activation of SFKs, ERK1/2, or PI3K is required for EGFR-mediated inhibition of ErbB4 signaling, we treated cells with PP2 (SFK inhibitor), U0126 (MEK1/2 inhibitor), and wortmannin (PI3K inhibitor). In sharp contrast to U0126 and wortmannin, PP2 treatment increased H3K9me3 levels upon co-stimulation with NRG-1 plus EGF (supplementary material Fig. S4A; Fig. 3D). A similar effect of SU6656 (another SFK inhibitor; Blake et al., 2000) was seen upon co-stimulation with NRG-1 plus EGF (supplementary material Fig. S4B). To scrutinize the role of SFKs (e.g. c-Src, c-Yes, Fyn and Lyn) in EGFR-mediated inhibition of ErbB4 signaling, we used embryonic fibroblast SYF cells, which are genetically deficient in expression of c-Src, c-Yes and Fyn and are undetectable in expression of Lyn (Klinghoffer et al., 1999), and the add-back version of SYF cells expressing active c-Src (SYF/c-Src cells) (supplementary material Fig. S5A) (Takahashi et al., 2009). ErbB4 activation by NRG-1 increased H3K9me3 levels in ErbB4-transfected SYF cells but could not increase those in ErbB4-transfected SYF/c-Src cells (Fig. 3E, top and middle). Note that high levels of H3K9me3 were observed in SYF cells exhibiting increased levels of nuclear ErbB4 (supplementary material Fig. S5C) and that in SYF/c-Src cells, NRG-1 stimulation did not increase the levels of nuclear ErbB4 and H3K9me3 (Fig. 3E, bottom; supplementary material Fig. S5C). To further examine whether the kinase activity of c-Src affected nuclear translocation of 4ICD, COS-1 cells were co-transfected with 4ICD and c-Src. Although c-Src was mainly localized to the cytoplasm (supplementary material Fig. S5D) (Kasahara et al., 2007a), c-Src expression inhibited nuclear localization of 4ICD and SU6656 treatment recovered the level of 4ICD in the nucleus (Fig. 3F), suggesting the inhibitory role of c-Src kinase activity in nuclear translocation of 4ICD.

Since phosphorylation of EGFR by activated SFKs changes subcellular localization and functions of EGFR (Stover et al., 1995; Dittmann et al., 2008; Iida et al., 2012), we examined whether 4ICD was phosphorylated by c-Src. When COS-1 cells were co-transfected with 4ICD and c-Src, expression of c-Src increased the levels of tyrosine phosphorylation of 4ICD and SU6656 treatment obviously inhibited c-Src-induced 4ICD phosphorylation (Fig. 4A, bottom). Intriguingly, 4ICD was co-immunoprecipitated with c-Src in a manner dependent on c-Src kinase activity (Fig. 4A, bottom). We next examined the role of Tyr950 (highly conserved among all ErbB-family members) and Tyr1056 (unique to ErbB4; Carpenter, 2003) on 4ICD in association of 4ICD with c-Src and nuclear translocation of 4ICD. Mutation at Tyr950 and Tyr1056 to Phe (Y950F/Y1056F) greatly decreased association of 4ICD with c-Src (Fig. 4B,C), and this Y950F/Y1056F mutant failed to induce c-Src-mediated inhibition of the nuclear translocation of 4ICD (Fig. 4D). However, it was difficult to evaluate the contribution of Y950F and Y1056F to the c-Src-mediated 4ICD phosphorylation sites, because ErbB4 has at least 19 potential sites of tyrosine phosphorylation (Kaushansky et al., 2008). Nonetheless, high accumulation of Y950F/Y1056F in the nucleus was found to induce high levels of H3K9me3 (Fig. 4E). These results suggest that phosphorylation of Tyr950 and Tyr1056 on 4ICD by c-Src mediates the association of 4ICD with c-Src, thereby inhibiting nuclear translocation of 4ICD. In other words, unphosphorylated 4ICD at Tyr950 and Tyr1056 translocates into the nucleus and increases H3K9me3 levels.

The hTERT is involved in cell proliferation and differentiation (Poole et al., 2001), and silencing of the hTERT gene involves repressive histone modifications, i.e. H3K9me3 and H3K27me3, in its promoter region (Zinn et al., 2007; Chen et al., 2012). To examine how the signaling pathways via ErbB4 and EGFR affected gene expression of hTERT, we analyzed the expression levels of the hTERT mRNA in T47D cells stimulated with NRG-1, EGF, or NRG-1 plus EGF or left untreated for 4 days. Semi-quantitative RT-PCR analyses showed that the levels of hTERT gene expression were dramatically decreased in cells stimulated with NRG-1 but co-stimulation with NRG-1 plus EGF failed to increase the levels of hTERT gene expression (Fig. 5A). The results could be explained by a decrease in expression of EGFR on day 4 after EGF stimulation (supplementary material Fig. S3). We thus examined the effects of ErbB4 and EGFR on hTERT gene expression after 30 min of stimulation, because stimulation with EGF for 30 min did not decrease the levels of EGFR (supplementary material Fig. S1D) and inhibited NRG-1-induced increase of H3K9me3 levels (Fig. 3A,D; supplementary material Fig. S4B). It is worthy of note that NRG-1 stimulation for 30 min actually decreased hTERT gene expression, and that co-stimulation with NRG-1 plus EGF for 30 min could restore the levels of hTERT gene expression to those seen in unstimulated cells (Fig. 5B). In the presence of SU6656, co-stimulation with NRG-1 plus EGF for 30 min again decreased the levels of hTERT gene expression. Previous studies reported that hTERT gene expression is decreased by breast cancer 1 (BRCA1) (Li et al., 2002) and prolonged stimulation with NRG-1 increases the expression levels of BRCA1 through the ErbB4-JNK pathway (Muraoka-Cook et al., 2006). Given that NRG-1 alone and NRG-1 plus EGF consistently increased BRCA1 gene expression on day 4 after stimulation (supplementary material Fig. S6), the increased levels of BRCA1 may be involved in repression of hTERT gene expression after 4-day stimulation (Fig. 5A). However, NRG-1 stimulation for 30 min did not increase the expression levels of BRCA1 (Fig. 5B), suggesting that an increase in expression of BRCA1 is dispensable for NRG-1-induced repression of hTERT gene expression that is inhibited by EGF-activated Src.

To further substantiate transcriptional repression of hTERT by nuclear 4ICD, the levels of hTERT gene expression were examined in the presence or absence of DAPT. Treatment with DAPT blocked NRG-1-induced repression of hTERT gene expression (Fig. 5C). We then analyzed the levels of hTERT gene expression in NLS-4ICD-expressing HeLa S3 cells where not only increased levels of H3K9me3 but also decreased levels of H3K27me3 were detected by immunostaining and western blotting (Fig. 5D–F; supplementary material Fig. S7). Fig. 5G showed that the levels of hTERT gene expression were decreased upon NLS-4ICD expression. Taken together, these results suggest that repression of hTERT gene expression through nuclear ErbB4 signaling is negatively regulated by EGFR-activated Src.

In the present study, we show that ErbB4 activation by NRG-1 induces an increase of H3K9me3 levels and transcriptional repression of hTERT. NRG-1 stimulation activates ErbB4 and releases 4ICD from the cytoplasm to the nucleus. 4ICD-mediated nuclear tyrosine phosphorylation is involved in increased H3K9me3 levels, repression of hTERT gene expression, and inhibition of cell proliferation. We further show that co-activation of ErbB4 and EGFR by NRG-1 and EGF, respectively, blocks 4ICD nuclear signaling. It is of note that EGFR activation does not inhibit ErbB4 activation and 4ICD release. Rather, c-Src activated by EGFR inhibits nuclear translocation of 4ICD in a manner dependent on c-Src kinase activity. We can illustrate a model of the mechanism for the antagonizing effect of EGF to NRG-1-induced signaling toward histone modification and gene expression as well (Fig. 6A,B). The two tyrosine residues, Tyr950 and Tyr1056, on 4ICD are critical for association of 4ICD with c-Src, leading to inhibition of nuclear translocation of 4ICD and subsequent blockade of 4ICD nuclear signaling. Regulation of ErbB4-mediated nuclear signaling by c-Src-mediated tyrosine phosphorylation plays an important role in changes of histone modification and gene expression. It is therefore intriguing to note that the two ErbB-family members ErbB4 and EGFR have opposing effects on nuclear signaling.

It is well known that non-receptor-type tyrosine kinases act as co-transducers of transmembrane signals emanating from a variety of receptors, including receptor-type tyrosine kinases (Neet and Hunter, 1996). c-Src is a signal transducer of EGFR, and EGFR-activated c-Src enhances EGFR signaling through phosphorylation of tyrosine residues in the cytoplasmic region of EGFR (Biscardi et al., 1999; Dittmann et al., 2008). In addition to c-Src, the other SFK members c-Yes and Lyn are involved in nuclear translocation of EGFR and the binding of EGFR to the promoter region of target genes, such as B-Myb and iNOS (Iida et al., 2012), suggesting that SFKs are positive regulators of EGFR signaling. In sharp contrast, our results indicate that activated c-Src antagonizes nuclear ErbB4 signaling. NRG-1-induced increase of H3K9me3 is decreased by concomitant stimulation with EGF, and treatment with PP2 or SU6656 increases H3K9me3 levels upon co-stimulation with NRG-1 plus EGF (Fig. 3D; supplementary material Fig. S4). Furthermore, NRG-1 stimulation increases H3K9me3 levels in SYF cells transfected with ErbB4, whereas it cannot increase H3K9me3 levels in c-Src-expressing SYF cells (SYF/c-Src) transfected with ErbB4 (Fig. 3E). Importantly, hTERT gene expression is repressed by NRG-1 stimulation through 4ICD release, which is blocked by EGF stimulation through Src kinase activity (Fig. 5B,C). These results indicate that c-Src inhibits the nuclear ErbB4 signaling pathway for an increase of H3K9me3 and repression of hTERT gene expression.

Although Src kinase activity does not affect 4ICD cleavage (Fig. 3C), c-Src inhibits nuclear accumulation of 4ICD in a manner dependent on its kinase activity (Fig. 3E,F; supplementary material Fig. S5C). Mutational analyses reveal that Tyr950 and Tyr1056 on 4ICD are involved in association of 4ICD with c-Src and inhibition of the nuclear translocation of 4ICD (Fig. 4A–D). Furthermore, expression of NLS-4ICD greatly increases H3K9me3 levels and represses hTERT gene expression (Fig. 2C–G; Fig. 5D,E,G), suggesting that nuclear accumulation of 4ICD has a critical role in NRG-1-induced increase of H3K9me3 levels and gene repression of hTERT.

Why does c-Src have opposing effects on ErbB4 and EGFR signaling? Although c-Src inhibits the nuclear translocation of 4ICD through phosphorylation of Tyr950 and Tyr1056 on 4ICD (Fig. 3F; Fig. 4D), nuclear accumulation of EGFR is increased by c-Src (Dittmann et al., 2008). Phosphorylation of Y1101 on EGFR by the SFK members c-Yes and Lyn promotes nuclear translocation of EGFR (Iida et al., 2012). The ErbB-family members show high homology in the kinase domain (59–81% identity), whereas their divergent C-terminal regions are not conserved (11–25% identity) (Schulze et al., 2005). Given that Y1056 on ErbB4 and Y1101 on EGFR are located in their C-terminal regions, Y1056 unique to ErbB4 is not conserved among ErbB members but Y950 present in the ErbB4 kinase domain is conserved among all of them. We assume that the nuclear-cytoplasmic distribution of ErbB-family members is largely controlled by phosphorylation of the particular tyrosine residues on their cytoplasmic C-terminal domains. In addition, we have shown that the trafficking pathway of c-Src is different from those of c-Yes and Lyn, and differential trafficking of SFKs is specified by the state of palmitoylation in the Src homology 4 domain (Kasahara et al., 2004; Sato et al., 2009; Obata et al., 2010). These results raise the intriguing possibility that the role of each ErbB member could be modulated by individual SFK member-specific tyrosine phosphorylation of their cytoplasmic C-terminal domains.

Despite the involvement of nuclear 4ICD in gene expression (Fig. 5D) (Williams et al., 2004; Sardi et al., 2006; Zhu et al., 2006), it is still unclear how tyrosine phosphorylation by nuclear ErbB4 is involved in gene expression through histone modifications. Recent studies show that H3K9me3 is controlled by histone methyltransferases and histone demethylases (Kouzarides, 2007). Although 4ICD is known to interact with Krab-associated protein 1 (KAP1), which can associate with the H3K9-methyltransferase SETDB1, the interaction of 4ICD with KAP1 is independent of its kinase activity and KAP1 phosphorylation by 4ICD has not been detected thus far (Schultz et al., 2002; Gilmore-Hebert et al., 2010). A small increase of H3K9me3 levels in cells expressing NLS-4ICD(KD) may be explained by the interaction of 4ICD with KAP1 (Fig. 2F,G). In addition to an increase of H3K9me3, a decrease of H3K27me3 was observed upon NLS-4ICD expression (Fig. 5F; supplementary material Fig. S7). Recent studies showed that demethylation of H3K27me3 by JMJD3 is promoted through its interaction with estrogen receptor α (Agger et al., 2007; Svotelis et al., 2011). 4ICD is also known to interact with STAT5 in lactational gene expression (Williams et al., 2004; Zhu et al., 2006). It is therefore likely that a decrease of H3K27me3 by NLS-4ICD (Fig. 5F), which might involve JMJD3 and STAT5, leads to cell differentiation, whereas an increase of H3K27me3 by the histone methyltransferase Ezh2 that is recruited by STAT5 was reported to repress Igk recombination in proliferative pre-B cells (Mandal et al., 2011).

The expression levels of the ErbB members and their ligands are dynamically changed in mammary gland development. During pregnancy where differentiation and development of mammary glands actively take place (Robinson et al., 1995), transcriptional levels of ErbB4 and NRG-1 are increased and in return those of EGF are repressed (Schroeder and Lee, 1998). On lactation after pregnancy, transcriptional levels of NRG-1 are contrariwise repressed and those of EGF are increased (Schroeder and Lee, 1998). Furthermore, impaired differentiation and lactational failure were observed in ErbB4-deficient mice (Long et al., 2003). Given that overexpression of c-Src inhibits mammary epithelial cell differentiation (Jehn et al., 1992), we assume that the interplay of the NRG-1-ErbB4 signaling with the EGF-EGFR-c-Src signaling is important for the regulation of cell proliferation and differentiation. In addition, cell proliferation involves hTERT activity but low-level expression of hTERT relates to cell differentiation (Smith et al., 2003). High-level expression of hTERT, involved in cell transformation and carcinogenesis (Elenbaas et al., 2001), has been shown to correlate with poor differentiation and poor prognosis for a number of cancers, including neuroblastoma, acute myelogenous leukemia, and breast cancer (Poole et al., 2001). Taken together with our findings that nuclear ErbB4 signaling is involved in repression of hTERT gene expression (Fig. 5), these results suggest that ErbB4-mediated nuclear signaling regulates histone modifications and gene expression for differentiation.

Furthermore, our recent studies showed that nuclear tyrosine phosphorylation is in fact involved in chromatin structural changes and histone modifications (Yamaguchi et al., 2001; Nakayama and Yamaguchi, 2005; Takahashi et al., 2009; Aoyama et al., 2011). It was also reported that histone H3 is tyrosine-phosphorylated by Janus kinase 2, a non-receptor-type tyrosine kinase (Dawson et al., 2009). These results provide evidence that nuclear tyrosine phosphorylation by nuclear-localized tyrosine kinases, including 4ICD, has important roles in histone modifications and gene expression for proliferation and differentiation.

In conclusion, we show for the first time that nuclear ErbB4 signaling is involved in an increase of H3K9me3 and transcriptional repression of hTERT, both of which are antagonized by EGFR-activated c-Src, and the signal balance between ErbB4 and EGFR may be important for cell proliferation and differentiation. It would be interesting to determine how c-Src inhibits nuclear translocation of 4ICD and to identify substrates downstream of 4ICD. Further studies will contribute to a better understanding of the regulation of nuclear ErbB4 signaling through tyrosine phosphorylation.

cDNA encoding human ErbB4 CYT-1 subcloned into the pcDNA3.1-Zeo (+) (Ogiso et al., 2002) was gifted from S. Yokoyama (RIKEN, Yokohama). To construct the ErbB4 intracellular domain (4ICD, residues 676–1308), the Kozak sequence was created by PCR using ErbB4 as a template with 5′-GTGGTCGACGCCACCATGGTTTATGTTAGAAGGAAGAGCATCAAAAAG-3′ (sense), and 5′-GCGACTAGTCGGAGCTGACA-3′ (antisense). The SalI−XbaI fragment of the PCR product was subcloned into the XhoI-XbaI site of pcDNA4/TO (Invitrogen). FLAG-HA-NLS (Takahashi et al., 2009; Aoyama et al., 2011)-fused 4ICD (NLS-4ICD) was generated as follows. 4ICD was amplified by PCR using ErbB4 as a template with 5′-GCTGGTCGACTTAGAAGGAAGAGCATCAAAAAGAAAAGAGCCTTG-3′ (sense), and 5′-GCGACTAGTCGGAGCTGACA-3′ (antisense), and the SalI–XbaI fragment of the PCR product was introduced into the XhoI–XbaI site of pcDNA4/TO-NLS-Lyn-HA (Takahashi et al., 2009), resulting in the insertion sequence VLDPAQWRPRL between the NLS and 4ICD. NLS-4ICD was amplified by PCR using 5′-GAAGTCGACATGTCTTCTGATGATGAAGCTACTGC-3′ (sense) and 5′-GGAGCTGACACGGAAGATCAACGGCGCCGGCGCAAC-3′ (antisense). The SalI–NotI fragment of the PCR product was introduced into the XhoI–NotI site of pOZ-FHN (Nakatani and Ogryzko, 2003; provided by A. Iwama, Chiba University). The sequence LDGGYP was inserted between the FLAG and HA epitopes, and the sequence GGLA was inserted between the HA epitope and the NLS. The resulting FHN-NLS-4ICD was subcloned into pcDNA4/TO. The Lys→Arg mutation at position 751 in the ATP-binding site [kinase-dead (KD), K751R] was generated by site-directed mutagenesis using FHN-NLS-4ICD as a template with 5′-GGGACCAGTTGTCTCATTAAGAATTCTAATAGCCACAGGAATCTTCAC-3′ (sense) and 5′-GTGAAGATTCCTGTGGCTATTAGAATTCTTAATGAGACAACTGGTCCC-3′ (antisense). The Tyr→Phe mutation at position 1056 (4ICD-Y1056F) was created by PCR using 4ICD as a template with 5′-GACACAGCCCTCCGCCGGCCTTCACCCCCATGTCAGGAAACCAGTTTG-3′ (sense) and 5′-CTGACATGGGGGTGAAGGCCGGCGGAGGGCTGTGTCCAATTTCACTCC-3′ (antisense). The Tyr→Phe mutation at positions 950/1056 (4ICD-Y950F/Y1056F) was created by PCR using 4ICD-Y1056F as a template with 5′-TCTGCACTATTGACGTCTTCATGGTCATGGTCAAATGTTGGATGATTG-3′ (sense) and 5′-TTTGACCATGACCATGAAGACGTCAATAGTGCAGATGGGAGGCTGAGG-3′ (antisense). The NLS was added to the N-terminus of Src-HA was subcloned into pcDNA4/TO (Takahashi et al., 2009). To construct pcDNA4/TOneoR/NLS-4ICD, the XbaI–XbaI fragment of pcDNA4/TO/NLS-4ICD was subcloned into pcDNA4/TOneoR (Nakayama et al., 2009). pcDNA4/TOneoR/c-Src-HA (c-Src) was constructed as follows. The HindIII–XhoI fragment of pcDNA3/c-Src-HA (provided by S. A. Laporte, McGill University) was subcloned into pcDNA4/TOneoR. As a mutation was found at position Leu175, the NotI–KpnI fragment was replaced with the NotI–KpnI fragment of human wild-type c-Src (Bjorge et al., 1995) (provided by D. J. Fujita, University of Calgary).

The following antibodies were used: phosphotyrosine (pTyr, 4G10; Upstate Biotechnology Inc.; provided by T. Tamura, National Institute of Infection Diseases, Tokyo, and T. Yoshimoto, Tokyo Medical University) (Tamura et al., 2000), HA (12CA5), ErbB4 (C18; Santa Cruz Biotechnology, and HFR1; Abcam), EGFR (provided by M. N. Fukuda, Sanford-Burnham Medical Research Institute), phosphor-p44/42 MAP kinase Thr²⁰²/Tyr²⁰⁴ (pERK1/2, E10; New England Biolabs), ERK2 (C-14; Santa Cruz Biotechnology), Src[Y⁴¹⁶] (phospho-Src family; BioSource), Src (GD11; Millipore), H3K4me3 (ab8580; Abcam), H3K9me3 (ab8898; Abcam), H4K16ac (Sigma-Aldrich), H3K27me3 (ab6002; Abcam), lamin A/C (N-18; Santa Cruz Biotechnology), and actin (clone C4; CHEMICON International). Horseradish peroxidase (HRP)-conjugated F(ab′)₂ fragment of anti-mouse IgG, anti-rabbit IgG or anti-goat IgG antibodies were purchased from Amersham Biosciences. TRITC-anti-rabbit IgG or anti-mouse IgG (Fc specific), Alexa Fluor 488 anti-mouse IgG or anti-rabbit IgG, and Alexa Fluor 647 anti-rabbit IgG secondary antibodies were from BioSource International, Sigma-Aldrich and Invitrogen.

COS-1 (Japanese Collection of Research Bioresources, Osaka), T47D (provided by M. Tagawa, Chiba Cancer Center Research Institute), and SYF cells (Klinghoffer et al., 1999) were cultured in Iscove's modified Dulbecco's medium containing 5% fetal bovine serum (FBS). SYF cells stably expressing c-Src-wt (SYF/c-Src) were prepared (Takahashi et al., 2009). Cells were transiently transfected with plasmid DNA using polyethylenimine (Fukumoto et al., 2010). For growth factor stimulation, T47D cells were starved under low serum conditions [0.05% bovine serum (BS)] for 24 hours, or SYF and SYF/c-Src cells transfected with vector or ErbB4 were cultured for 8 hours and then starved for 16 hours. Starved cells were left untreated or treated with 20 ng/ml NRG-1 (Neuregulin β-1; R&D Systems), 20 ng/ml EGF, or 20 ng/ml NRG-1 plus 20 ng/ml EGF for 30 min under low serum conditions (0.05% BS). A HeLa S3 cell clone expressing NLS-4ICD was generated in an inducible manner (HeLa S3/TR/NLS-4ICD), because we could not establish a cell line stably expressing NLS-4ICD. HeLa S3/TR (clone A3f5) cells, which stably express the tetracycline repressor (TR) (Aoyama et al., 2011), were transfected with pcDNA4/TOneo/NLS-4ICD, and cell clones inducibly expressing NLS-4ICD were selected in 500 µg/ml G418. Expression of NLS-4ICD was induced for 8 hours by 2 µg/ml doxycycline, a tetracycline derivative.

The following inhibitors were used: 20 µM AG1478, 10 µM PP2, 10 µM SU6656, 20 µM U0126, 1 or 25 µM DAPT, and 100 nM wortmannin.

Immunofluorescence staining was performed as described (Yamaguchi and Fukuda, 1995; Tada et al., 1999; Nakayama and Yamaguchi, 2005; Kasahara et al., 2007b; Ikeda et al., 2008; Takahashi et al., 2009; Nakayama et al., 2012). In brief, cells were washed in warmed phosphate-buffered saline (PBS) and fixed in 100% methanol for 5 min at −20°C (Ni et al., 2001; Aoyama et al., 2011). Fixed cells were permeabilized and blocked in PBS containing 0.1% saponin and 3% BSA for 20 min at room temperature, and then incubated with a primary and a secondary antibody for 1 hour each at room temperature. For DNA staining, cells were subsequently treated with 200 µg/ml RNase A for 30 min and 20 µg/ml propidium iodide (PI) or 20 nM TOPRO-3 for 30 min at room temperature. After washing with PBS containing 0.1% saponin, cells were mounted with ProLong antifade reagent (Molecular Probes). The resulting red emission of TOPRO-3-stained nuclei is pseudo-colored as blue. Confocal and Nomarski differential-interference-contrast (DIC) images were obtained at room temperature using a Fluoview FV500 confocal laser scanning microscope with a 40× 1.00 NA or a 60× 1.00 NA water-immersion objective (Olympus, Tokyo) and an LSM510 laser scanning microscope with 40× 0.75 NA or 63× 1.40 NA oil immersion objective (Carl Zeiss). One planar (xy) section slice (0.8- or 2.0-µm thickness) was shown in most experiments. Composite figures were prepared using Photoshop 11.0 and Illustrator 14.0 software (Adobe). For quantification of histone modifications, confocal images were obtained using a Fluoview FV500 confocal laser scanning microscope, and then fluorescence intensities of immunostaining were measured using ImageJ software (National Institutes of Health). Since two or three independent experiments gave similar results, a representative experiment was shown.

DNA synthesis was measured by incorporation of bromodeoxyuridine (BrdU; Roche) into the genomic DNA during the S phase (DNA replication) of the cell cycle. T47D cells were starved under low serum conditions (0.05% BS) for 24 hours, and then treated with NRG-1 or EGF in the presence of BrdU for 24 hours. Cells were fixed with ethanol and 50 mM glycine, pH 2.0, for 45 min at room temperature and denatured in 4 N HCl for 15 min. Subsequent immunodetection of BrdU was accomplished, according to the manufacturer's instructions.

Cell lysates were prepared in SDS-PAGE sample buffer or Triton X-100 lysis buffer (20 mM HEPES, pH 7.8, 5% glycerol, 1% Triton X-100, 5 mM EDTA, 50 mM NaF, 20 mM β-glycerophosphate, 50 µg/ml aprotinin, 100 µM leupeptin, 25 µM pepstatin, 10 mM Na₃VO₄, and 1 mM PMSF), and subjected to SDS-PAGE and electrotransferred onto polyvinylidene difluoride membrane (Millipore). Immunodetection was performed by enhanced chemiluminescence (Amersham Biosciences) as described (Yamaguchi et al., 2001; Matsuda et al., 2006; Kasahara et al., 2007a; Kuga et al., 2007; Ikeda et al., 2008; Sato et al., 2009; Nakayama et al., 2012). Sequential reprobing of membranes with a variety of antibodies was performed after the complete removal of primary antibodies from membranes in stripping buffer or inactivation of HRP by 0.1% NaN₃, according to the manufacturer's instructions. Results were analyzed using a ChemiDoc XRSPlus image analyzer (Bio-Rad), and quantitated with Quantity One software (Bio-Rad). Immunoprecipitation was performed using antibody-precoated protein-G beads, as described (Mera et al., 1999; Yamaguchi et al., 2001; Ikeda et al., 2008; Obata et al., 2010). Composite figures were prepared using Photoshop 11.0 and Illustrator 14.0 software (Adobe).

Total RNAs were isolated from T47D cells with the ISOGEN reagent (Nippon Gene, Tokyo), and cDNAs were synthesized from 1 µg of each RNA preparation using the PrimeScript RT reagent Kit (TakaraBio, Shiga) as described (Kikuchi et al., 2010; Aoyama et al., 2011). To avoid saturation of PCR products, conditions of PCR were optimized before semi-quantitative RT-PCR was carried out. The primers used for PCR are as follows: human telomerase reverse transcriptase (hTERT), 5′-CGGAAGAGTGTCTGGAGCAA-3′ (sense) and 5′-GGATGAAGCGGAGTCTGGA-3′ (antisense) (Nakamura et al., 1997); breast cancer 1 (BRCA1), 5′-TTGCGGGAGGAAAATGGGTAGTTA-3′ (sense) and 5′-TGTGCCAAGGGTGAATGATGAAAG-3′ (antisense) (Ju et al., 2007); glyceraldehyde 3-phosphate dehydrogenase (GAPDH), 5′-ACCACAGTCCATGCCATCAC-3′ (sense) and 5′-TCCACCACCCTGTTGCTGTA-3′ (antisense) (Kikuchi et al., 2010; Aoyama et al., 2011). The sizes of PCR products are 146, 292 and 452 bp, respectively. Amplification was carried out using an MJ mini thermal cycler (Bio-Rad) with Ex Taq DNA polymerase (TakaraBio, Shiga) under the following conditions: For hTERT, initial heating at 95°C for 1 min, followed by 35 cycles of denaturation at 95°C for 30 sec, annealing at 56°C for 30 sec and extension at 72°C for 30 sec. For BRCA1 and GAPDH, initial heating at 94°C for 2 min, followed by 27 or 25 cycles of denaturation at 94°C for 30 sec, annealing at 53°C or 56°C for 30 sec and extension at 72°C for 1 min. The products of RT-PCR were electrophoresed on a 2.0 or 3.0% agarose gel containing ethidium bromide. The density of each amplified fragment was quantified with ChemiDoc XRSPlus and Quantity One software.

We thank S. Yokoyama, D. J. Fujita, S. A. Laporte, T. Tamura, T. Yoshimoto, M. Tagawa, A. Iwama and M. N. Fukuda for their invaluable plasmids, antibodies and cell lines.