Back signaling by the Nrg-1 intracellular domain

Abstract

Transmembrane isoforms of neuregulin-1 (Nrg-1), ligands for erbB receptors, include an extracellular domain with an EGF-like sequence and a highly conserved intracellular domain (ICD) of unknown function. In this paper, we demonstrate that transmembrane isoforms of Nrg-1 are bidirectional signaling molecules in neurons. The stimuli for Nrg-1 back signaling include binding of erbB receptor dimers to the extracellular domain of Nrg-1 and neuronal depolarization. These stimuli elicit proteolytic release and translocation of the ICD of Nrg-1 to the nucleus. Once in the nucleus, the Nrg-1 ICD represses expression of several regulators of apoptosis, resulting in decreased neuronal cell death in vitro. Thus, regulated proteolytic processing of Nrg-1 results in retrograde signaling that appears to mediate contact and activity-dependent survival of Nrg-1-expressing neurons.

Figure 1.

Interaction between CRD-Nrg-1 and erbB receptors enhances survival of CRD-Nrg-1–expressing neurons. (A) Spiral ganglion neurons are lost after genetic disruption of CRD-Nrg-1. Spiral ganglia from wild-type (left) or CRD-Nrg-1^−/− mutant (right) E16.5 mouse embryos were hybridized to cRNA probes for either erbB4 (top) or GAP43 (to identify neurons). Based on 3-D reconstructions of serial sections, there was a 90% decrease in spiral ganglion neurons (GAP43 mRNA positive; SG) in mutant embryos. The erbB4-expressing cochlear epithelium (Co) was not measurably affected in the mutants. (B and C) Dispersed neurons were maintained in culture for 2 d and treated with soluble erbB2 + erbB4 (erbB2:B4), the CRD-Nrg-1 ECD (Nrg-ECD), a mixture of soluble erbB2:4 and Nrg-ECD, or staurosporine with or without soluble erbB2:B4. Apoptotic cells were visualized by staining nuclei with bisbenzimide (B, arrows). The percentage of total nuclei (±SEM) that appeared apoptotic was quantified in 10 fields from each treatment group in three independent experiments. Where indicated, values differed significantly (P < 0.01) from the untreated control group (*) or from the staurosporine-treated group (**).

Figure 2.

Interaction with erbB receptors, or depolarization, target Nrg-1-ICD to the nucleus in primary neurons. (A) Dispersed E16 spiral ganglion neurons were maintained in vitro for 3 d and stained with antibodies recognizing the ICD of the “a” form of Nrg-1 (red), or the ECD of all Nrg-1β isoforms (green). Nuclei were stained with DAPI (blue). 15 min before fixation and staining, neuronal cultures were either untreated (control) or treated with soluble erbB2:B4 (serbB2:B4). Fluorescent images in 1-μm sections were collected with a two-photon microscope. Under control conditions, intracellular and ECDs colocalize in soma and in processes. After treatment, colocalization is lost in neurites, both domains appear in clusters, and clusters of ICD staining are clearly present in the nucleus. The top six images are from 1-μm optical sections collected from the middle of the nuclei, whereas the bottom two images were collected at a level below the nuclei to emphasize the distribution of Nrg-1 in the processes. (B) The percentage of neuronal nuclei showing staining with the antibody recognizing the Nrg-1-ICD was quantified after a 15-min treatment with nothing (control), soluble erbB2 (which does not bind to Nrg-1), soluble erbB2 + erbB4 (erbB2:B4), or 50 mM KCl. Only cells showing clearly outlined nuclei were included in the analyses and at least 50 cells were scored per field. The data are plotted as the mean of the counts from two experiments. (C) Spiral ganglion neurons from six E13.5 embryos were maintained in culture overnight before treatment for 15 min with nothing (control), soluble erbB2 (B2), soluble erbB2 + erbB4 (B2:B4), or 50 mM KCl. Nuclear extracts (18 μg protein/lane) were resolved by SDS-PAGE, and Nrg-1-ICD was detected by probing immunoblots with the ICD antibody (sc-348). After stripping, the filters were reprobed sequentially with antibodies recognizing histone H1 (H1, nuclear marker) or the translation initiation factor eIF5 (eIF5, cytoplasmic marker). The anti–Nrg-1-ICD antibody recognized a protein of ∼50 kD. Nuclei from cells treated with soluble erbB2:B4 or 50 mM KCl had significantly elevated levels of this 50-kD band. (D) Spiral ganglia neurons were treated for 15 min with soluble erbB2 + erbB4 and were either fixed and stained with antibody recognizing the Nrg-1-ICD (left) or lysed and analyzed by immunoblotting (right). Where indicated (+peptide, +pept), the primary antibody was preincubated with immunizing peptide before staining cells or probing filters. Preincubation eliminated staining of cells and staining of the 50-kD protein on immunoblots (+pept).

Figure 2.

Interaction with erbB receptors, or depolarization, target Nrg-1-ICD to the nucleus in primary neurons. (A) Dispersed E16 spiral ganglion neurons were maintained in vitro for 3 d and stained with antibodies recognizing the ICD of the “a” form of Nrg-1 (red), or the ECD of all Nrg-1β isoforms (green). Nuclei were stained with DAPI (blue). 15 min before fixation and staining, neuronal cultures were either untreated (control) or treated with soluble erbB2:B4 (serbB2:B4). Fluorescent images in 1-μm sections were collected with a two-photon microscope. Under control conditions, intracellular and ECDs colocalize in soma and in processes. After treatment, colocalization is lost in neurites, both domains appear in clusters, and clusters of ICD staining are clearly present in the nucleus. The top six images are from 1-μm optical sections collected from the middle of the nuclei, whereas the bottom two images were collected at a level below the nuclei to emphasize the distribution of Nrg-1 in the processes. (B) The percentage of neuronal nuclei showing staining with the antibody recognizing the Nrg-1-ICD was quantified after a 15-min treatment with nothing (control), soluble erbB2 (which does not bind to Nrg-1), soluble erbB2 + erbB4 (erbB2:B4), or 50 mM KCl. Only cells showing clearly outlined nuclei were included in the analyses and at least 50 cells were scored per field. The data are plotted as the mean of the counts from two experiments. (C) Spiral ganglion neurons from six E13.5 embryos were maintained in culture overnight before treatment for 15 min with nothing (control), soluble erbB2 (B2), soluble erbB2 + erbB4 (B2:B4), or 50 mM KCl. Nuclear extracts (18 μg protein/lane) were resolved by SDS-PAGE, and Nrg-1-ICD was detected by probing immunoblots with the ICD antibody (sc-348). After stripping, the filters were reprobed sequentially with antibodies recognizing histone H1 (H1, nuclear marker) or the translation initiation factor eIF5 (eIF5, cytoplasmic marker). The anti–Nrg-1-ICD antibody recognized a protein of ∼50 kD. Nuclei from cells treated with soluble erbB2:B4 or 50 mM KCl had significantly elevated levels of this 50-kD band. (D) Spiral ganglia neurons were treated for 15 min with soluble erbB2 + erbB4 and were either fixed and stained with antibody recognizing the Nrg-1-ICD (left) or lysed and analyzed by immunoblotting (right). Where indicated (+peptide, +pept), the primary antibody was preincubated with immunizing peptide before staining cells or probing filters. Preincubation eliminated staining of cells and staining of the 50-kD protein on immunoblots (+pept).

Figure 2.

Interaction with erbB receptors, or depolarization, target Nrg-1-ICD to the nucleus in primary neurons. (A) Dispersed E16 spiral ganglion neurons were maintained in vitro for 3 d and stained with antibodies recognizing the ICD of the “a” form of Nrg-1 (red), or the ECD of all Nrg-1β isoforms (green). Nuclei were stained with DAPI (blue). 15 min before fixation and staining, neuronal cultures were either untreated (control) or treated with soluble erbB2:B4 (serbB2:B4). Fluorescent images in 1-μm sections were collected with a two-photon microscope. Under control conditions, intracellular and ECDs colocalize in soma and in processes. After treatment, colocalization is lost in neurites, both domains appear in clusters, and clusters of ICD staining are clearly present in the nucleus. The top six images are from 1-μm optical sections collected from the middle of the nuclei, whereas the bottom two images were collected at a level below the nuclei to emphasize the distribution of Nrg-1 in the processes. (B) The percentage of neuronal nuclei showing staining with the antibody recognizing the Nrg-1-ICD was quantified after a 15-min treatment with nothing (control), soluble erbB2 (which does not bind to Nrg-1), soluble erbB2 + erbB4 (erbB2:B4), or 50 mM KCl. Only cells showing clearly outlined nuclei were included in the analyses and at least 50 cells were scored per field. The data are plotted as the mean of the counts from two experiments. (C) Spiral ganglion neurons from six E13.5 embryos were maintained in culture overnight before treatment for 15 min with nothing (control), soluble erbB2 (B2), soluble erbB2 + erbB4 (B2:B4), or 50 mM KCl. Nuclear extracts (18 μg protein/lane) were resolved by SDS-PAGE, and Nrg-1-ICD was detected by probing immunoblots with the ICD antibody (sc-348). After stripping, the filters were reprobed sequentially with antibodies recognizing histone H1 (H1, nuclear marker) or the translation initiation factor eIF5 (eIF5, cytoplasmic marker). The anti–Nrg-1-ICD antibody recognized a protein of ∼50 kD. Nuclei from cells treated with soluble erbB2:B4 or 50 mM KCl had significantly elevated levels of this 50-kD band. (D) Spiral ganglia neurons were treated for 15 min with soluble erbB2 + erbB4 and were either fixed and stained with antibody recognizing the Nrg-1-ICD (left) or lysed and analyzed by immunoblotting (right). Where indicated (+peptide, +pept), the primary antibody was preincubated with immunizing peptide before staining cells or probing filters. Preincubation eliminated staining of cells and staining of the 50-kD protein on immunoblots (+pept).

Figure 2.

Interaction with erbB receptors, or depolarization, target Nrg-1-ICD to the nucleus in primary neurons. (A) Dispersed E16 spiral ganglion neurons were maintained in vitro for 3 d and stained with antibodies recognizing the ICD of the “a” form of Nrg-1 (red), or the ECD of all Nrg-1β isoforms (green). Nuclei were stained with DAPI (blue). 15 min before fixation and staining, neuronal cultures were either untreated (control) or treated with soluble erbB2:B4 (serbB2:B4). Fluorescent images in 1-μm sections were collected with a two-photon microscope. Under control conditions, intracellular and ECDs colocalize in soma and in processes. After treatment, colocalization is lost in neurites, both domains appear in clusters, and clusters of ICD staining are clearly present in the nucleus. The top six images are from 1-μm optical sections collected from the middle of the nuclei, whereas the bottom two images were collected at a level below the nuclei to emphasize the distribution of Nrg-1 in the processes. (B) The percentage of neuronal nuclei showing staining with the antibody recognizing the Nrg-1-ICD was quantified after a 15-min treatment with nothing (control), soluble erbB2 (which does not bind to Nrg-1), soluble erbB2 + erbB4 (erbB2:B4), or 50 mM KCl. Only cells showing clearly outlined nuclei were included in the analyses and at least 50 cells were scored per field. The data are plotted as the mean of the counts from two experiments. (C) Spiral ganglion neurons from six E13.5 embryos were maintained in culture overnight before treatment for 15 min with nothing (control), soluble erbB2 (B2), soluble erbB2 + erbB4 (B2:B4), or 50 mM KCl. Nuclear extracts (18 μg protein/lane) were resolved by SDS-PAGE, and Nrg-1-ICD was detected by probing immunoblots with the ICD antibody (sc-348). After stripping, the filters were reprobed sequentially with antibodies recognizing histone H1 (H1, nuclear marker) or the translation initiation factor eIF5 (eIF5, cytoplasmic marker). The anti–Nrg-1-ICD antibody recognized a protein of ∼50 kD. Nuclei from cells treated with soluble erbB2:B4 or 50 mM KCl had significantly elevated levels of this 50-kD band. (D) Spiral ganglia neurons were treated for 15 min with soluble erbB2 + erbB4 and were either fixed and stained with antibody recognizing the Nrg-1-ICD (left) or lysed and analyzed by immunoblotting (right). Where indicated (+peptide, +pept), the primary antibody was preincubated with immunizing peptide before staining cells or probing filters. Preincubation eliminated staining of cells and staining of the 50-kD protein on immunoblots (+pept).

Figure 3.

Treatments that target Nrg-1-ICD to the nucleus alter gene expression. E13.5 spiral ganglion neurons maintained in culture overnight were untreated (control) or stimulated with soluble erbB2:B4 or 50 mM KCl for 2 h. Total RNA was isolated and the relative levels of Bcl-X_L, RIP, BAK, p19^INK4, Oct-3, and actin mRNAs were determined by RT-PCR. Amplified products were resolved on agarose gels and visualized with ethidium bromide. Arrows indicate amplified products whose identities were verified by DNA sequencing (the additional bands in the treated BAK lanes and the untreated Oct-3 lanes were nonspecific amplification products). This experiment was repeated with soluble erbB2:B4 that had been preincubated with the ECD of CRD-Nrg-1 (right).

Figure 4.

Nuclear translocation of Nrg-1-ICD–GFP fusions can be visualized in living cells. (A) The schematic illustration at the top shows the NRG-1βa-GFP chimeric used in this work. ECD, extracellular domain; TM, transmembrane domain; NLS1, the putative nuclear localization sequence (B); ICD, intracellular domain; GFP, green fluorescent protein. Intracellular movement of NRG-1βa-GFP was followed in live cells (HEK 293 cells) by two-photon microscopy. Images (1 μm from the middle of the nucleus) were collected at various intervals (min) after treatment with soluble erbB2 + erbB4. The arrows in the enlargement point out puncta of Nrg-1-GFP that have entered the nucleus. (B) HEK293 cells were transfected with plasmids encoding either an intact Nrg-1-ICD fused to GFP (top) or an Nrg-1-ICD lacking the putative NLS fused to GFP (bottom). The subcellular localization of the fusion proteins was followed by conventional fluorescence microscopy. At left are phase images. In the middle, the green fluorescence signal is shown (arrows point to positive cells), and at the right, both green fluorescence and DAPI staining are shown. (C) Cytoplasmic (Cyto) and nuclear (Nuc) fractions were prepared from mock-transfected or NRG-1βa-HA (HA, influenza virus hemagglutinin-derived epitope added to COOH terminus of Nrg-1)–transfected HEK293T cells. Proteins were analyzed by immunoblotting using antibodies recognizing the HA epitope. NRG-1βa-HA– transfected cells were treated for 15 min with erbB2 (32 μg/ml) or erbB2:4 (32 μg/ml). In addition to a doublet of nonspecific bands, proteins of >100 kD (full-length and aggregated NRG-1βa) and 50 kD (ICD-HA) were detected. The 50-kD band enriched in the nuclear fraction was only seen in cells treated with soluble erbB2:B4.

Figure 5.

Treatment of transfected cells with soluble erbB receptors stimulates Nrg-1-ICD cleavage and translocation to a transcriptionally active compartment. (A) Schematic illustrations showing organization of the Gal4-Nrg-1 fusion proteins used. Gal4, DNA-binding domain; VP16, activation domain from the Herpes virus VP16 transcription factor. (B) Nrg-1βa-Gal4-VP16, Nrg-ICD-Gal4_DBD, Nrg-ICD_ΔNLS-Gal4_DBD, Nrg-1-ICD, or Gal4_DBD-VP16_AD expression plasmids or empty vector (control) were cotransfected into HEK293T cells with a reporter plasmid containing four copies of the Gal4 UAS fused to the luciferase coding region. Luciferase activities were measured 48 h after transfection and values normalized to the control levels (±SEM). Membrane tethering (Nrg-1-Gal4-VP16) or exclusion from the nucleus (Nrg-1-ICD_ΔNLS-Gal4_DBD) reduced the Nrg-1-ICD-Gal4 transcriptional activity by over 10-fold. (C) The NRG-1βa-Gal4-VP16–expressing plasmid was cotransfected into HEK293T cells with the Gal4-UAS–luciferase reporter. Cells were treated with one of three γ-secretase inhibitors or an inactive analogue of these inhibitors from 24–48 h after transfection. For the final 8 h before measuring, luciferase activity cells also were treated with soluble erbB2, soluble erbB2 + erbB4 (erbB2:B4), erbB2:B4 preincubated with the ECD of CRD-Nrg-1 (NRG-ECD), or CRD-Nrg-1 ECD alone. Luciferase levels (±SEM) were normalized to control (untreated cells cotransfected with reporter and Nrg-1-Gal4-VP16 plasmids = 1).

Figure 6.

Bidirectional signaling by transmembrane Nrg-1. Both forward and back signaling result from interactions between erbB receptors (blue) and membrane-tethered Nrg-1 (green and red). Interaction (steps 1 and 2) results in activation of erbB receptor tyrosine kinases and subsequent induction of target genes (step 6) in erbB-expressing cells, as well as intramembranous (and possibly extracellular) cleavage of Nrg-1 (steps 3 and 4). The released Nrg-1-ICD (red) translocates from neurites to cell bodies (step 5a), and then to the nucleus, possibly with other proteins (step 5b), where it regulates target gene expression (step 5c). See the Discussion for further details of this model.