Protein Inhibitor of Activated STAT3 (PIAS3) Protein Promotes SUMOylation and Nuclear Sequestration of the Intracellular Domain of ErbB4 Protein*

Background: Mechanisms regulating nuclear translocation of the ErbB4 intracellular domain (ICD) are unclear. Results: PIAS3 interacts with ErbB4, sequesters ICD into the nucleus, and represses its nuclear signaling. Conclusion: PIAS3 is a regulator of subcellular localization and nuclear functions of ErbB4. Significance: A novel mechanism controlling the signaling of an ICD of a receptor tyrosine kinase in the nucleus is described. ErbB4 is a receptor tyrosine kinase implicated in the development and homeostasis of the heart, central nervous system, and mammary gland. Cleavable isoforms of ErbB4 release a soluble intracellular domain (ICD) that can translocate to the nucleus and function as a transcriptional coregulator. In search of regulatory mechanisms of ErbB4 ICD function, we identified PIAS3 as a novel interaction partner of ErbB4 ICD. In keeping with the small ubiquitin-like modifier (SUMO) E3 ligase function of protein inhibitor of activated STAT (PIAS) proteins, we showed that the ErbB4 ICD is modified by SUMO, and that PIAS3 stimulates the SUMOylation. Upon overexpression of PIAS3, the ErbB4 ICD generated from the full-length receptor accumulated into the nucleus in a manner that was dependent on the functional nuclear localization signal of ErbB4. In the nucleus, ErbB4 colocalized with PIAS3 and SUMO-1 in promyelocytic leukemia nuclear bodies, nuclear domains involved in regulation of transcription. Accordingly, PIAS3 overexpression had an effect on the transcriptional coregulatory activity of ErbB4, repressing its ability to coactivate transcription with Yes-associated protein. Finally, knockdown of PIAS3 with siRNA partially rescued the inhibitory effect of the ErbB4 ICD on differentiation of MDA-MB-468 breast cancer and HC11 mammary epithelial cells. Our findings illustrate that PIAS3 is a novel regulator of ErbB4 receptor tyrosine kinase, controlling its nuclear sequestration and function.

ErbB4 is a receptor tyrosine kinase implicated in the development and homeostasis of the heart, central nervous system, and mammary gland. Cleavable isoforms of ErbB4 release a soluble intracellular domain (ICD) that can translocate to the nucleus and function as a transcriptional coregulator. In search of regulatory mechanisms of ErbB4 ICD function, we identified PIAS3 as a novel interaction partner of ErbB4 ICD. In keeping with the small ubiquitin-like modifier (SUMO) E3 ligase function of protein inhibitor of activated STAT (PIAS) proteins, we showed that the ErbB4 ICD is modified by SUMO, and that PIAS3 stimulates the SUMOylation. Upon overexpression of PIAS3, the ErbB4 ICD generated from the full-length receptor accumulated into the nucleus in a manner that was dependent on the functional nuclear localization signal of ErbB4. In the nucleus, ErbB4 colocalized with PIAS3 and SUMO-1 in promyelocytic leukemia nuclear bodies, nuclear domains involved in regulation of transcription. Accordingly, PIAS3 overexpression had an effect on the transcriptional coregulatory activity of ErbB4, repressing its ability to coactivate transcription with Yes-associated protein. Finally, knockdown of PIAS3 with siRNA partially rescued the inhibitory effect of the ErbB4 ICD on differentiation of MDA-MB-468 breast cancer and HC11 mammary epithelial cells. Our findings illustrate that PIAS3 is a novel regulator of ErbB4 receptor tyrosine kinase, controlling its nuclear sequestration and function.
ErbB receptors form the epidermal growth factor receptor/ ErbB subfamily of receptor tyrosine kinases, which includes ErbB1/EGF receptor, ErbB2, ErbB3, and ErbB4. ErbB receptors are activated upon binding of EGF-like ligands and regulate gene transcription through the activation of a multilayered network of intracellular signaling molecules. Signaling via ErbB receptors results in various biological outcomes, including proliferation, differentiation, migration, and survival. Aberrant ErbB signaling has been implicated in the initiation and progression of several human malignancies, and drugs targeting the EGF receptor and ErbB2 are currently in clinical use (1).
Unlike other ErbB receptors, ErbB4 is expressed as four isoforms that are generated by alternative mRNA splicing. Two of the isoforms differ structurally in the extracellular juxtamembrane region (JM-a 3 and JM-b), and two in the intracellular cytoplasmic domain (CYT-1 and CYT-2) (2, 3). The ErbB4 JM-a isoform was the first receptor tyrosine kinase found to employ a signaling paradigm involving regulated intramembrane proteolysis (RIP) and generation of a soluble intracellular domain (ICD). Upon ligand binding, ErbB4 JM-a is cleaved by tumor necrosis factor-␣-converting enzyme, which triggers a second cleavage by ␥-secretase activity (4,5). The released ErbB4 ICD can translocate into the nucleus, where it associates with transcriptional regulators including yes-associated protein (YAP) (6), STAT5 (7), estrogen receptor (8), Eto2 (9), TAB2-N-CoR complex (10), AP-2 (11), and HIF-1␣ (12), and modifies their activity. In vivo, the soluble ErbB4 ICD has been shown to regulate astrogenesis in the developing brain (10) as well as mammary epithelial cell differentiation and proliferation (13). In addition to nuclear localization and regulation of transcriptional events, the ErbB4 ICD has been reported to localize in the cytosol in mitochondria, where it regulates apoptosis (14). In clinical breast cancer samples, membrane, cytosolic, and nuclear localization of ErbB4 has been described. Interestingly, nuclear localization of ErbB4 epitope is more frequent in malignant cells than in morphologically normal cells (15), and nuclear ErbB4 immunoreactivity has been suggested to associate with different disease outcome in cancer when compared with membranous ErbB4 (16,17). Although these observations suggest that subcellular localization of the ErbB4 ICD has a role in determining the biological outcome of ErbB4 ICD signaling, little is known about how the subcellular localization and functions of ErbB4 ICD are regulated.
Members of the PIAS (protein inhibitor of activated STAT) family (PIAS1, PIASx␣, PIASx␤, PIAS3, and PIASy) interact with and modulate functions of a number of proteins, especially transcriptional regulators. PIAS proteins can function as E3 ligases in the enzymatic reaction that posttranslationally attaches SUMO (small ubiquitin-related modifier) to target proteins. Indeed, most interaction partners of PIAS proteins are also modified by SUMO (18). Many known SUMO targets are nuclear proteins involved in transcriptional regulation. SUMOylation has been reported to regulate several biological processes by altering the activity, stability, or subcellular localization of its targets (19). In addition, PIAS proteins can also regulate their interaction partners in a manner that is independent of their SP-RING finger, a domain that is central for their ability to promote SUMOylation. For example, PIAS proteins can affect the functions of the interacting proteins by altering their subcellular or subnuclear localization (18).
Here, we demonstrate that ErbB4 selectively interacts with members of the PIAS protein family, and show that ErbB4 is modified by SUMO. Our results implicate a role for one of the PIAS proteins, PIAS3, in sequestration of the ErbB4 ICD into the nucleus, repression of its transcriptional coregulatory activity, and in ErbB4 ICD-mediated inhibition of mammary epithelial cell differentiation. Taken together, our data reveal novel mechanisms that control nuclear localization and function of ErbB4 RTK.
Coimmunoprecipitation of ErbB4 and PIAS3-COS-7 cells were transiently transfected with indicated expression plasmids and lysed 24 h after transfection. Coimmunoprecipitation was carried out as described previously (28). Precipitates were analyzed by Western blotting using anti-FLAG and anti-ErbB4 (sc-283, Santa Cruz Biotechnology) antibodies.
In Situ Proximity Ligation Assay (PLA)-MCF-7 cells cultured on coverslips were transfected with control siRNA (Ambion) or siRNAs targeting PIAS3 at a final concentration of 30 nM. Cells were starved without serum overnight, treated with 50 ng/ml neuregulin-1 (NRG-1-␤1 EGF domain, R&D Systems) for 15 min, and fixed with methanol. Samples were incubated with anti-ErbB4 (HFR-1, Abcam) and anti-PIAS3 (ab22856) for 90 min. Proximity ligation was carried out with the Duolink II in situ PLA kit (Olink Biosciences) according to the instructions of the manufacturer. PLA signals were detected with an Axiovert 200 M confocal microscope with LSM510 Meta (Zeiss). Two confocal images of each sample were acquired with a ϫ63 objective, and all signals were counted in five sections separated by 1.5 m to quantify the number of interactions per cell. The interactions were classified as cytosolic or nuclear depending on their colocalization with DAPI.
Cell Fractionation-Cytosolic and nuclear fractions were prepared with a NE-PER kit (Thermo Scientific) as described previously (20) and analyzed by Western blotting with anti-HA, anti-FLAG, anti-ErbB4 (E200, Abcam), and anti-PIAS3 (sc-46682) as indicated. Antibodies against Lamin B (sc-6217, Santa Cruz Biotechnology) and Mek1/2 (Cell Signaling) were used to control the fractionation.
GAL4 Transactivation Assay-HEK 293 cells were cotransfected with ErbB4 ICD2 fused to the GAL4 DNA-binding domain (ErbB4 ICD2-GAL4) and a Firefly luciferase reporter construct pFR-luc (Stratagene) driven by GAL4 binding sites with or without Omni-YAP2 and FLAG-tagged PIAS proteins. A cotransfected Renilla luciferase construct pTK-RL (Promega) was used as an internal control. Luciferase activity was measured and analyzed as described previously (30). Expression of PIAS proteins and YAP2 were controlled by Western blotting with anti-FLAG and anti-Omni (sc-7270, Santa Cruz Biotechnology), respectively, and loading with anti-HSP90 (Calbiochem).
Differentiation of MDA-MB-468 and HC11 cells in Threedimensional Cultures-Stable MDA-MB-468 cell lines expressing ErbB4 JM-a CYT-2 have been described (31). HC11 cells were infected with retroviruses encoding pBABE-puroErbB4JM-aCYT-2 or pBABE-puroErbB4JM-bCYT-2 to generate stable cell lines (31). The protocol for three-dimensional cultures of HC11 cells was modified from the protocol described for MDA-MB-468 cells by Tvorogov et al. (31). Briefly, MDA-MB-468 transfectants were suspended into Matrigel (BD Biosciences) on duplicate 96-well plates, supplemented with or without 50 ng/ml NRG-1 and 5 M GSI IX, and maintained at 37°C for 6 days. HC11 cells were suspended into growth factor-reduced Matrigel (BD Biosciences), supplemented with 50 ng/ml NRG-1, and maintained at 37°C for 14 or 15 days as indicated in the figures. For siRNA knockdown experiments, cells were transfected with siRNAs targeting PIAS3 or PML or with negative control siRNAs and suspended into Matrigel 24 h after siRNA transfection. One hundred colonies from three independent views through the whole thickness of Matrigel were counted and classified as undifferentiated colonies or differentiated acini on the basis of their morphology (31).
Statistical Analyses-Student's t test was used for all statistical analyses.

RESULTS
ErbB4 Interacts with PIAS3 and PIASy and Is Modified by SUMO-Ligand-induced RIP of ErbB4 is of biological significance and results in nuclear translocation of the soluble ErbB4 ICD, which has transcriptional coregulatory activity (5, 10). PIAS proteins have been implicated as regulators of a variety of proteins, particularly transcription factors (18). To address the putative interactions between ErbB4 ICD and PIAS proteins, the GST-tagged ICD of CYT-2 type (ICD2) or C-or N-terminal deletion of ICD2 (ICD2-⌬C or ICD2-⌬N, respectively) (Fig. 1, A and B) were purified from bacteria and incubated together with lysate from COS-7 cells expressing FLAG-tagged PIAS proteins. PIAS3 and PIASy interacted with ICD2-GST and ICD2-⌬C-GST containing the ErbB4 tyrosine kinase domain but not with ICD2-⌬N-GST containing the C-terminal tail of ErbB4 (Fig. 1C). In contrast, no interaction between ErbB4 and PIAS1 or PIASx␣ was observed (Fig. 1C). To control the integrity of both deletion constructs, GST pull-down assays were carried out with STAT5a and WW domain-containing oxidoreductase (Wwox), the expected interaction partners of ICD-⌬C and ICD-⌬N, respectively (30, 32) (Fig. 1C). Reciprocal coimmunoprecipitation analyses with either anti-ErbB4 (Fig. 1D) or anti-FLAG (data not shown) confirmed the interaction between transiently expressed FLAG-tagged PIAS3 and ErbB4 ICD of CYT-1 type (ICD1) in COS-7 cells. Finally, the interaction between endogenous PIAS3 and ErbB4 in MCF-7 breast cancer cells was analyzed using PLA, which allows for detection of endogenous protein complexes in situ (33). Positive PLA signals (red dots) were detected in both the cytosol and the nucleus (Fig. 1, E and F). Knockdown of PIAS3 with siRNA significantly reduced the number of signals in the nucleus, controlling the specificity of observed interactions (Fig. 1, E and F).
Most interaction partners of PIAS proteins are modified by SUMO, and PIAS proteins can function as SUMO E3 ligases (18). This prompted us to examine whether ErbB4 could be SUMOylated. COS-7 cells expressing ErbB4 ICD2 with or without His-or GFP-tagged SUMO-1 were lysed in the presence of N-ethylmaleimide to inhibit the activity of SUMO isopeptidases. In addition to the unmodified ErbB4 ICD2 (ϳ80 kDa), Western blot analyses revealed higher molecular weight species of ErbB4, corresponding to the size of ErbB4 ICD2 modified with His-or GFP-tagged SUMO-1 (ϳ90 kDa and 120 kDa, respectively) (Fig. 1G). Similar results showing ErbB4 ICD SUMOylation were obtained when ErbB4 ICD2 and SUMO-1 or SUMO-3 were overexpressed in HEK 293T, WM-266 -4, or MCF-7 cells (supplemental Fig. S2). In accordance with the requirement of intact SP-RING for the SUMO E3 ligase function of PIAS proteins, overexpression of PIAS3, but not a SP-RING-disrupted mutant of PIAS3 (C299S), further increased SUMOylation of the ErbB4 ICD (Fig. 1H). Overexpression of PIAS3 could also promote SUMOylation of ErbB4 ICD in MCF-7 cells (supplemental Fig. S2D), whereas PIAS3 knockdown with siRNAs inhibited ICD SUMOylation (supplemental Fig. S2E). These data show H, COS-7 cells were transfected with HA-tagged ErbB4 ICD2 and the indicated FLAG-tagged PIAS3 and His-tagged SUMO-1 constructs. Lysates prepared as in G were immunoprecipitated with anti-ErbB4 (HFR-1) followed by Western blotting with anti-HA. Expression of PIAS3 was controlled by Western blotting with anti-FLAG. The band marked with an asterisk presumably represents another SUMO modification site.
that ErbB4 interacts with PIAS3 and that PIAS3 can stimulate the SUMOylation of ErbB4.
PIAS3 Promotes Accumulation of the Nuclear ErbB4 ICD-To visualize ErbB4 in cells expressing PIAS proteins, COS-7 cells were transfected with constructs encoding PIAS1, PIASx␣, PIAS3, or PIASy together with full-length, HA-tagged ErbB4 JM-a CYT-2. Notably, coexpression of PIAS3 and PIASy, but not PIAS1 or PIASx␣, resulted in a primarily nuclear stain-ing pattern of ErbB4 ( Fig. 2A). PIAS protein expression levels were controlled by Western blotting (Fig. 2B). In accordance with the increased nuclear staining of ErbB4 upon PIAS3 overexpression, Western blot analysis demonstrated an increased protein level of ErbB4 ICD2 in the nuclear fraction when PIAS3 was coexpressed (Fig. 2C). Inhibition of the RIP of the fulllength receptor at the cell membrane by the ␥-secretase inhibitor IX (GSI IX) efficiently abolished nuclear ErbB4 staining  (HFR-1, red). The nuclei were stained with DAPI (blue). The cells were visualized by confocal microscopy with a ϫ40 objective. G, MCF-7 cells transfected with the indicated control siRNA (Ambion) or siRNAs targeting PIAS3 were subjected to subcellular fractionation. Nuclear and cytosolic fractions were analyzed by Western blotting with anti-ErbB4 (E200), anti-PIAS3 (sc-46682), anti-lamin B, and anti-Mek1/2. Loading was controlled by reblotting with anti-actin.

PIAS3-ErbB4 Interaction
(87% of untreated cells and 40% of GSI IX-treated cells were scored positive for nuclear staining, n ϭ 100, Fig. 2D), indicating that overexpression of PIAS3 resulted in accumulation of the ICD and not the full-length ErbB4 receptor into the nucleus. Consistently, the nuclear accumulation of a mutant ErbB4 JM-a CYT-2 not susceptible to RIP (V675A) (34) was reduced significantly compared with wild-type ErbB4 (Fig. 2E and supplemental Fig. S3).
ErbB4 has been shown to contain a tripartite NLS at the juxtamembrane intracellular region (7,26). To further characterize the role of PIAS3 in increased nuclear localization of the ErbB4 ICD, an NLSI/II mutant of full-length ErbB4 JM-a CYT-2, with a disrupted first (I) and second (II) part of the tripartite NLS, was generated. Compared with cells coexpressing wild-type ErbB4 and PIAS3, significantly less nuclear ErbB4 staining was detected in cells coexpressing the ErbB4 NLSI/II mutant together with PIAS3 ( Fig. 2E and supplemental Fig. S3), implying that the functional NLS is critical for PIAS3 induced nuclear accumulation of the ErbB4 ICD.
The requirement of PIAS3 on the localization of endogenous ErbB4 was also assessed. ErbB4 staining was detected in the nuclei of MCF-7 breast cancer cells when cells were stimulated with an ErbB4 ligand NRG-1, and nuclear export was inhibited by leptomycin B treatment (supplemental Fig. S4). Consistent with our data on the nuclear accumulation of ErbB4 upon PIAS3 overexpression, knockdown of PIAS3 with siRNA resulted in reduced nuclear staining of ErbB4 (Fig. 2F). In addition, a Western analysis of nuclear fraction of MCF-7 cells demonstrated a decrease in ErbB4 ICD protein level upon PIAS3 knockdown with siRNAs (Fig. 2G). Taken together, these data demonstrate that PIAS3 sequesters the soluble ErbB4 ICD into the nucleus in a manner that depends on ␥-secretase cleavage and the functional nuclear localization signal of ErbB4.

ErbB4 Colocalizes with PIAS3, SUMO-1, and PML in Nuclear
Bodies-Several transcriptional regulators, as well as SUMOylated proteins localize in promyelocytic leukemia (PML) nuclear bodies (35). To test whether this would be the case for ErbB4, full-length HA-tagged ErbB4 JM-a CYT-2 was coexpressed together with FLAG-PIAS3, PML-3, and GFP-tagged SUMO-1 in COS-7 cells, and the proteins were visualized by immunofluorescence and confocal microscopy. In the presence of PIAS3, ErbB4 colocalized with SUMO-1 and PIAS3 (Fig. 3A) in PML-immunoreactive nuclear bodies (B). However, SUMO E3 ligase activity of PIAS3 was not necessary for efficient recruitment of ErbB4 into the PML bodies because overexpression of the C299S mutant of PIAS3 could also induce colocalization of ErbB4 with SUMO-1 (Fig. 3A) and PML (B). In addition to COS-7 cells, stable transfectants of HC11 mouse mammary epithelial cells expressing cleavable ErbB4 JM-a CYT-2 were analyzed for colocalization of ErbB4 with endogenous PIAS3 and PML. In a quantitative colocalization analysis, ϳ11% of the PIAS3 signal and 19% of the PML signal were also positive for ErbB4 (supplemental Fig. S5).
PIAS3 Represses Nuclear Functions of the ErbB4 ICD-Because PIAS proteins often regulate the transcriptional activity of their interaction partners (18), the effect of PIAS3 overexpression on transcriptional coregulatory activity of ErbB4 was examined. One system described previously to measure the transcriptional coregulatory activity of ErbB4 is a transactiva-

PIAS3-ErbB4 Interaction
tion assay where ErbB4 ICD fused to the GAL4 DNA-binding domain (ErbB4 ICD2-GAL4) and YAP coactivate a luciferase gene driven by GAL4 binding sites (6). Consistently, we observed a robust increase in luciferase activity when ErbB4 ICD2-GAL4 was expressed together with YAP2 in HEK 293 cells (Fig. 5). Coexpression of PIAS3 in this system decreased the luciferase activity by ϳ70%, whereas PIASx␣ or PIASy could not repress ICD-YAP2-mediated coactivation. Interestingly, disrupting the PIAS3 SP-RING domain with the C299S point mutation partially removed the repressive effect of PIAS3 (Fig. 5). These data suggest specificity for PIAS3 in this system and a role for its SUMO E3 ligase activity.
We and others have previously utilized three-dimensional cultures of MDA-MB-468 human breast cancer (31) and HC11 mouse mammary epithelial cells (13,34) as models to address the role of ErbB4 isoforms in mammary differentiation. In a reconstituted basement membrane matrix (Matrigel), mammary epithelial cells are able to differentiate morphologically and functionally, forming small, organized colonies that resemble mammary acini, whereas transformed cells grow in large, disorganized colonies (36). To characterize the function of soluble ErbB4 ICD2 in three-dimensional growth of MDA-MB-468 human breast cancer cells, stable transfectants expressing ErbB4 JM-a CYT-2 or an empty vector (31) were embedded in Matrigel and treated with or without GSI IX to inhibit ErbB4 cleavage (Fig. 6A). Roughly half of cells expressing empty vector but only ϳ25% of cells expressing ErbB4 grew in small spheres resembling mammary acini (Fig. 6B). When ErbB4 cleavage was inhibited by GSI IX, cells expressing ErbB4 showed a shift

PIAS3-ErbB4 Interaction
toward a more differentiated phenotype (Fig. 6B), indicating that the soluble ErbB4 ICD2 had an inhibitory effect on differentiation, consistent with previous findings (13). To address the effect of PIAS3 and PML on ErbB4 ICD function in this system, MDA-MB-468 cells expressing ErbB4 JM-a CYT-2 and the vector control cells were transfected with either non-targeting siRNA or two independent siRNAs targeting PIAS3 or PML. Knockdown of PIAS3 (Fig. 6C) or PML (D) had a negative effect on the basal differentiation of the vector control cells. However, knockdown of PIAS3 or PML in cells expressing ErbB4 increased the formation of differentiated acini (Fig. 6, C and D).
To further characterize the functional significance of ErbB4 regulation by PIAS3, stable transfectants of HC11 mouse mammary epithelial cells expressing cleavable ErbB4 JM-a CYT-2, non-cleavable JM-b CYT-2, or an empty vector were generated (Fig. 6E) and analyzed for their three-dimensional growth in Matrigel. Similar to our observations with MDA-MB-468 cells, HC11 cells expressing ErbB4 JM-a CYT-2 demonstrated less differentiation than the vector control cells (Fig. 6F). In contrast, cells expressing non-cleavable ErbB4 JM-b CYT-2 did not differ from control cells in their ability to form differentiated acini (Fig. 6F). Next, HC11 vector control cells and cells expressing ErbB4 JM-a CYT-2 were transfected with non-targeting siRNA or two independent siRNA oligonucleotides targeting PIAS3. Again, knockdown of PIAS3 could partially rescue the reduced differentiation of the cells expressing cleavable ErbB4 JM-a CYT-2 (Fig. 6G). Together with the role of PIAS3 as a regulator of nuclear localization of ErbB4 ICD (Fig. 2), these data prompted us to hypothesize that the outcome of ErbB4 JM-a CYT-2 signaling in MDA-MB-468 and HC11 cells is associated with the subcellular distribution of the soluble ICD2. Indeed, confocal immunofluorescence analyses of HC11 transfectants revealed reduced nuclear ErbB4 staining upon PIAS3 knockdown (Fig. 6, H and I). Taken together, these data indicate that PIAS3 and PML have repressive effects on nuclear signaling functions of ErbB4 ICD.

DISCUSSION
ErbB4 is a receptor tyrosine kinase that has been demonstrated to undergo RIP upon ligand binding (5). Functionally, ErbB4 ICD plays critical roles in regulating astrogenesis in the developing brain (10) and epithelial differentiation in the mammary gland (13). ErbB4 ICD localizes in the nucleus but also in the cytosol (15) and exhibits differential functional outcomes depending on its localization (14,16,37). Although previous studies have reported nuclear localization and putative nuclear export signals within the ErbB4 sequence (5,7,15,26), the factors determining subcellular distribution of ErbB4 ICD have remained largely unknown. Here, we characterized PIAS3 as a novel interaction partner and regulator of nuclear localization of ErbB4 ICD. PIAS3 also stimulated posttranslational SUMO modification of ErbB4. Interaction of ErbB4 ICD by PIAS3 had functional consequences, demonstrated by inhibition of ErbB4 ICD nuclear signaling functions.
The nuclear ErbB4 ICD interacts with a number of transcriptional regulators and modulates their activity (6 -12). PIAS proteins and SUMOylation both regulate transcriptional processes (18 -19), and ErbB4 ICD was found to both interact with and become SUMOylated by PIAS3. PIAS3, but not a SP-RING mutant incapable of SUMOylating ErbB4, repressed the tran-scriptional coregulatory activity of ErbB4 ICD in the GAL4 assay where ErbB4 and YAP2 coactivate transcription. In contrast, both the wild type and the SP-RING mutant of PIAS3 were equally efficient in promoting the localization of ErbB4 in PML nuclear bodies, nuclear subdomains that are considered to have a central role in transcriptional regulation (35). It is interesting to note that N-CoR and YAP2, transcriptional regulators that associate with ErbB4, have also been reported to localize in PML bodies (38,39). PML bodies could thus regulate nuclear signaling of the ErbB4 ICD by providing a site for formation of transcriptional regulatory complexes, including ErbB4. Although the functional SP-RING domain of PIAS3 was not necessary in promoting colocalization of ErbB4 with PML, the fraction of ErbB4 in PML bodies may still be SUMOylated, as many other proteins are (35). Interestingly, our colocalization experiments indicated that PIAS3 also partially localizes in PML bodies. Future experiments will elucidate the functional significance of PIAS3 promoted SUMOylation versus nuclear sequestration for ErbB4 signaling.
ErbB4 has an important role in the mammary gland, regulating lobuloalveolar differentiation during lactation (40 -42). Studies that employ ErbB4 constructs with a mutated ␥-secretase cleavage site, mutated nuclear localization signal, or ectopic expression of ErbB4 ICD suggest a role for the soluble ErbB4 ICD in promoting mammary epithelial cell differentiation (7,34,43). It is notable that these studies specifically describe the effects of the CYT-1 isoform of ErbB4. Interestingly, in a recent study that compared the effects of ectopic expression of the ErbB4 ICD of the CYT-1 and CYT-2 types, CYT-1 promoted but CYT-2 suppressed mammary gland differentiation both in vitro and in vivo (13). Here, we studied the differentiation of MDA-MB-468 breast cancer and HC11 mammary epithelial cells expressing either an empty vector or ErbB4 JM-a CYT-2 in three-dimensional Matrigel cultures. Expression of the cleavable JM-a CYT-2 isoform had inhibitory effects on differentiation. However, when the formation of soluble ErbB4 ICD was inhibited either chemically with a ␥-secretase inhibitor or genetically by expressing the non-cleavable JM-b isoform, the number of differentiated acinus-like structures increased. siRNA-mediated knockdown of PIAS3 or PML resulted in increased differentiation of cells expressing ErbB4, ). *, p Ͻ 0.05. Densitometric data of PIAS3 protein levels relative to control knockdown as detected by Western blotting are indicated (protein (%)). D, MDA-MB-468 cells expressing ErbB4 JM-a CYT-2 and the vector control cells were treated with non-targeting (Qiagen) or the indicated PML-specific siRNAs for 24 h, suspended into Matrigel, and cultured for 6 days. The columns represent the percentage of acini of all colonies in four independent experiments (mean Ϯ S.D.). *, p Ͻ 0.05. Densitometric data of PML protein levels relative to control knockdown as detected by Western blotting are indicated (protein (%)). E, HC11 cells expressing ErbB4 JM-a CYT-2 or JM-b CYT-2 and the vector control cells were treated for 15 min with 50 ng/ml NRG-1 and analyzed by Western blotting with anti-ErbB4 (E200). The membrane was reprobed with anti-actin to control loading. F, HC11 cells expressing the indicated constructs were suspended into growth factor-reduced Matrigel and cultured for 14 days in the presence of 50 ng/ml NRG-1. The columns represent the percentage of acini of all colonies in three independent experiments (mean Ϯ S.D.). *, p Ͻ 0.05. G, HC11 cells expressing ErbB4 JM-a CYT-2 and the vector control cells were treated with non-targeting (Ambion) or the indicated PIAS3-specific siRNAs for 24 h, suspended into Matrigel, and cultured for 15 days. The columns represent the percentage of acini of all colonies in two independent experiments (mean Ϯ S.D.). *, p Ͻ 0.05. Densitometric data of PIAS3 protein levels relative to control knockdown as detected by Western blotting are indicated (protein (%)). H, representative confocal images (visualized with a ϫ63 objective) of HC11 cells expressing ErbB4 JM-a CYT-2 treated with non-targeting (Ambion) or PIAS3-specific siRNAs as in G. Cells were stained with anti-ErbB4 (HFR-1, green) and anti-PIAS3 (ab22856, red). The nuclei were stained with DAPI (blue). I, Quantification of cytosolic and nuclear ErbB4 immunoreactivity (mean Ϯ S.D.). *, p Ͻ 0.05. Densitometric data of PIAS3 protein levels relative to control knockdown as detected by Western blotting are indicated (protein (%)).
likely reflecting altered subcellular localization of the ErbB4 ICD. Although these data suggest a role for PML bodies in regulating ErbB4 function, it is unclear whether the sequestration of ErbB4 into PML bodies is in fact necessary for PIAS3 to affect nuclear signaling of ErbB4. Together, these findings suggest that the outcome of ErbB4 signaling in MDA-MB-468 and HC11 cells is dependent on the subcellular distribution of soluble ErbB4 ICD2.
Nuclear localization of the ErbB4 epitope has been detected by immunohistochemical analyses in breast cancer in several studies (15)(16)44). Compared with malignant cells, nuclear ErbB4 is detected in only a small proportion of cells representing normal morphology (15). Interestingly, the prognostic significance of nuclear ErbB4 immunoreactivity in breast cancer cells also seems to be different from the prognostic significance of ErbB4 at the cell surface or in the cytosol (16,17). The role of PIAS3 in cancer has not been extensively studied, but increased PIAS3 expression in breast cancer has been reported (45,46). Therefore, it is interesting to speculate whether the increased expression of PIAS3 in breast cancer could provide a mechanism for nuclear accumulation of ErbB4 ICD with functional consequences on ErbB4 signaling.
In conclusion, we have demonstrated that PIAS3 is a novel interaction partner of ErbB4 that regulates the SUMOylation, nuclear localization, and functions of the ErbB4 ICD. Our findings highlight the importance of ErbB4 proteolytic cleavage and subcellular distribution of the ErbB4 ICD in determining the outcome of ErbB4 signaling.