Protogenin (PRTG) is a transmembrane protein of immunoglobulin superfamily, which has multiple roles in embryogenesis as a receptor or an adhesion molecule. In this study, we present sequential proteolytic cleavage of PRTG. The cleavage first occurs at the extracellular domain, then at the interface of the transmembrane and the intracellular domain by γ-secretase, which results in the release of the intracellular domain of PRTG (PRTG-ICD). PRTG-ICD contains putative nuclear localization signal (NLS) at its N-terminal, and translocates to the nucleus in cultured cells and in the neuroepithelial cells of chick embryos. We propose that the PRTG-ICD is cleaved by γ-secretase and translocates to the nucleus, which is potentially implicated in signaling for neural differentiation and in cell adhesion mediated by PRTG.
Protogenin (PRTG) is a transmembrane protein of the immunoglobulin (Ig) superfamily. Its extracellular domain is composed of four Ig domains and five fibronectin III domains, which is structurally related to Deleted in Colorectal Cancer (DCC) and Neogenin (Toyoda et al. 2005). prtg gene is conserved among vertebrates (Toyoda et al. 2005; Vesque et al. 2006). Recent studies revealed that PRTG has multiple roles during embryogenesis including neural, mesodermal, and tooth development. PRTG in neural progenitors functions as a receptor to prevent precocious neuronal differentiation through ligand-binding (Wong et al. 2010). In mesodermal lineage, PRTG serves as a homophilic adhesion molecule to support ingression and epithelialization of the paraxial mesoderm to the somite (Ito et al. 2011). In early tooth germ, PRTG is required for the differentiation of the inner enamel epithelial cells in the mouse molar (Takahashi et al. 2010). Although these studies emphasize the importance of PRTG during development of various tissues, it remains elusive how the signaling of PRTG is transmitted within the cell to achieve these functions.
Proteolytic processing of certain transmembrane molecules is a crucial step to mediate extracellular signaling from the membrane to the nucleus. Substrates for such proteases include Notch (Mumm et al. 2000), CD44 (Okamoto et al. 2001), ErbB4 (Ni et al. 2001), N-, and E-cadherins (Marambaud et al. 2002, 2003; Ferber et al. 2008), whose proteolytic cleavage results in the release of bioactive signaling fragments. In the first step, cleavage occurs at the extracellular domain, called ectodomain-shedding, by the enzyme like metalloprotease. Subsequently, cleavage by γ-secretase releases the intracellular domain (ICD), which eventually translocates to the nucleus. Some ICDs are known to catalyze other proteins and act as transcriptional regulators. The Notch ICD (NICD) interacts with the transcriptional co-factor (CSL), and activates target genes such as HES family (Mumm & Kopan 2000; Fortini 2002). The CD44 ICD can potentiate the transcriptional co-activator CBP (CREB-binding protein) and activate promoters that contain a TPA-responsive element (TRE), which upregulates its own expression (Okamoto et al. 2001). N-cadherin ICD binds CBP and promotes its proteasomal degradation to inhibit CREB (cAMP-responsive element binding protein)-dependent transcription (Marambaud et al. 2003).
It was reported that DCC and Neogenin, close relatives of PRTG, are processed by γ-secretase to release ICD domain, which may be implicated in transcriptional regulation in the nucleus (Taniguchi et al. 2003; Goldschneider et al. 2008). Considering the conformational analogy, we suspected that PRTG might also be the substrate for proteolytic cleavages. In this study, we investigated proteolytic features of PRTG and the subcellular localization of its processed products in vitro and in vivo. We present evidence that PRTG is cleaved by γ-secretase, leading to release of the intracellular domain (PRTG-ICD) that is sorted to the nucleus. The results suggest that proteolytic cleavage is potentially implicated in physiological function of the PRTG signaling.
Materials and methods
pCAG-PRTG contains chick full-length PRTG (1-1187aa), which is transcribed by cytomegalovirus enhancer and chicken β-actin promoter. pCAG-PRTGdECD contains construct where extracellular Ig and fibronectin III domains of PRTG (28-920aa) were deleted (Fig. 1A). A valine residue within the transmembrane domain of PRTGdECD was mutated in PRTGdECD-V947L (V947 to L) and PRTGdECD-V945G (V945 to G). pCAG-PRTGdECDmutNLS contains construct where putative nuclear localizing signal (NLS) was mutated (964-969aa, RSKARK to ASAAAA). pCAG-V945PRTG-ICD or pCAG-PRTG-ICD contains intracellular region of PRTG C-terminal side from valine 945 (945-1187aa) or immediately C-terminal of the transmembrane domain (964-1187aa, Fig. 1A). N-terminal six residue of putative NLS (RSKARK) was deleted in pCAG-PRTG-ICDdNLS. Open reading frame of each construct contains c-myc tag at C-terminal.
Transfection of 293T and immunoblotting
293T cells were transfected using TransFast transfection reagent according to the manufacture’s protocol (Promega). 24 h after transfection, the cells were treated with 2 μmol/L DAPT (N-[N-(3,5-Difluorophenacetyl)-L-alanyl]-S-phenylglycine t-butyl ester, Cayman Chemical) and/or 10 μmol/L epoxomicin (Peptide Institute) for an additional 24 h. Cell lysates were prepared in lysis buffer (50 mM Tris pH7.4, 0.15 M NaCl, 1 mM ethylenediaminetetraacetic acid [EDTA], 1% Triton X-100) containing protease inhibitor cocktail (P8340; Sigma) and analyzed by immunoblotting using gradient sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) (5–20% gel) with rabbit anti-PRTG antibody raised against intracellular domain of chick PRTG (Ito et al. 2011) or mouse anti-myc monoclonal antibody (9E10; DSHB). The signals of immunoblot were detected using enhanced chemiluminescence (ECL)-plus western blotting detection reagent (GE) and analyzed with LAS-4000mini (Fuji). For subcellular localization of PRTG proteins, 293T cells were cultured on Lab-Tek II CC2 chamber-slide (Nunc), and detected with a confocal laser scanning microscope (FV300, Olympus).
In ovo electroporation
For transfection to the neural tube of chick embryos, pCAG-PRTG-ICD or pCAG-PRTG-ICDdNLS (4 μg/μL) was mixed with pCAG-EGFP (4 μg/μL) and injected into the central canal at stage 10 (Hamburger & Hamilton 1951). Using a pair of electrodes (0.5 mm diameter, 1.0 mm length, and 4 mm distance between the electrodes), rectangular pulse of 25 V, 50 mseconds/s was charged four times by the electroporator (CUY21EDIT, BEX, Japan; Watanabe & Nakamura 2000). Transfected embryos were fixed 30 h after electroporation. Cryosections were immunostained with mouse anti-myc monoclonal antibody (9E10; DSHB). Fluorescent images for subcellular localization of the proteins were captured with a cooled CCD digital camera (ORCA-ER, Hamamatsu) or a confocal laser scanning microscope (FV300, Olympus). The staining intensity of each cell in the fluorescent image was measured with FluoView software (Olympus).
PRTG is a substrate for proteolytic cleavage
Protogenin is a transmembrane protein structurally related with DCC and Neogenin (Fig. 1A, Toyoda et al. 2005). Prompted by the reports that ectodomain-shedding and γ-secretase processing occur in DCC and Neogenin (Taniguchi et al. 2003; Goldschneider et al. 2008), we examined if PRTG receives such proteolytic cleavages. For this purpose, we transiently expressed chick full length PRTG in 293T cell lines and checked if PRTG is cleaved by immunoblot using an antibody against intracellular domain of chick PRTG. Chick PRTG is composed of 1187 amino acids and its predicted molecular weight is 128 kD. PRTG in 293T cells was detected as bands more than 150 kD in western blot (Fig. 2, lane 1), as in chick embryos (Ito et al. 2011). It is consistent with the fact that mouse PRTG is highly glycosylated (Takahashi et al. 2010). Treatment with DAPT, a potent inhibitor of γ-secretase, produced several small fragments between 35–43 kD (lane 2, circles). These fragments correspond to α-fragments of DCC and Neogenin produced by the cleavage of metalloprotease as ectodomain-shedding (Galko & Tessier-Lavigne 2000; Goldschneider et al. 2008). The production of these putative α-fragments of PRTG after DAPT treatment indicates that PRTG is a substrate for γ-secretase.
The α-fragments are cleaved by γ-secretase to release γ-fragment, which are subsequently degraded by the proteasome (Taniguchi et al. 2003; Goldschneider et al. 2008). To know if γ-fragment is produced from PRTG, we treated the transfected cells with a proteasome inhibitor, epoxomicin. After treatment with epoxomicin, we detected a smaller fragment about 32 kD, which corresponds well to γ-fragment (Fig. 2, lane 3, asterisk).
Next we searched for the position of the first cleavage, which produces putative α-fragments. The size of putative α-fragments suggests that the first cleavage sites are between the fibronectin III domain and the transmembrane domain. Alignment of amino acid sequences of PRTG of various species from mammals and birds showed conserved sequences between the fibronectin III and the transmembrane domain (Fig. 1B, underlined). We wondered if the first cleavage site for ectodomain-shedding is within this conserved domain (Fig. 1B, underlined), and expressed the deleted PRTG (Fig. 1A, PRTGdECD), which contains the transmembrane and intracellular domain of PRTG from a site within the conserved sequence (Fig. 1A,B, arrowhead). Then we compared the size of PRTGdECD with that of putative α-fragments by immunoblotting. The size of PRTGdECD was about 35 kD, comparable to the smallest α-fragment of PRTG (Fig. 2, lane 2, 4). Since treatment of PRTGdECD with DAPT, an inhibitor of γ-secretase, did not generate smaller fragment (lane 5), we concluded that one of the cleavage is in the conserved domain.
We then checked if γ-secretase cleaves putative α-fragments to release γ-fragment. Since the size of PRTGdECD is comparable to the smallest α-fragment of PRTG, we used PRTGdECD as a putative α-fragment. Epoxomicin treatment of PRTGdECD generated 32 kD fragment, which corresponds to γ-fragment (Fig. 2, lane 6, asterisk). If this is a bona fide γ-fragment, inhibition of γ-secretase activity must cancel the generation of this fragment. As we expected, the fragment of 32 kD was not detected after adding DAPT (lane 7). The results were first obtained using the anti-PRTG antibody, which was raised against its intracellular domain (lane 4–7), and were confirmed by using the antibody against myc epitope located at C-terminal (lane 8–11). The results of immunoblotting of PRTG after treatment with protease inhibitors suggested that PRTG receives dual proteolytic processing; the first cleavage produce putative α-fragments, and the second cleavage releases γ-fragment.
The intracellular domain of PRTG translocates to the nucleus
Next, we were interested in the second cleavage site of PRTG by γ-secretase to produce γ-fragment. γ-secretase cleavage site of Notch-1, p75 and Syndecan-3 is the valine residue within the transmembrane domain (Fig. 1C; Kopan & Ilagan 2004). Other substrates, such as E-cadherin and presumably DCC, are cleaved at the interface of the transmembrane and the intracellular domain (Fig. 1C; Marambaud et al. 2002; Taniguchi et al. 2003). The size of PRTG-ICD protein immediate C-terminal side of the transmembrane domain was equivalent to that of γ-fragment revealed by immunoblotting with both anti-PRTG and with anti-c-myc (Fig. 3B, lane 2, 3; lane 5, 6) antibodies. The result indicates that the γ-secretase cleavage site of PRTG is at the interface of the transmembrane and the intracellular domain, as that of E-cadherin and DCC. But we noticed two conserved valine residues within the transmembrane domain of PRTG (Fig. 1B), so we checked whether these valine residues are involved in cleavage by γ-secretase, and expressed two mutation constructs (PPRTGdECD V945 to G, V947 to L) in 293T cells. Since γ-fragment was generated from either mutant protein after epoxomicin treatment (Fig. 3A, lane 3, 4), we regarded these residues as not being the cleavage sites. In addition, γ-fragment observed after the treatment with epoxomicin on PRTGdECD (Fig. 3B, lane 2, asterisk) was slightly smaller than V945-PRTG-ICD protein containing the C-terminal side of PRTG from valine 945 (V945-PRTG-ICD; lane 1). These data suggest that PRTG is cleaved by γ-secretase at the interface of the transmembrane and the intracellular domain to release PRTG-ICD.
After γ-secretase cleavage, ICDs of γ-secretase substrates listed in Fig. 1C, are sorted to the nucleus. We asked whether PRTG-ICD, released by γ-secretase cleavage, is localized to the nucleus. After transfection in 293T cells, full-length PRTG was localized on the plasma membrane as previously reported (Fig. 3C, top; Takahashi et al. 2010; Ito et al. 2011). PRTGdECD, deletion of the extracellular domain, was observed on the plasma membrane and in the cytoplasm (Fig. 3C, middle). Interestingly, PRTG-ICD in 293T cells was detected preferentially in the nucleus (Fig. 3C, bottom; see also Fig. 4C), indicating PRTG-ICD is localized to the nucleus.
Nuclear localization signal for the intracellular domain of PRTG
Near the N-terminal of PRTG-ICD, we noticed that six residues enriched in basic amino acids are conserved among species (Fig. 1B; RSKARK, double underlined). Since N-terminal motifs enriched in basic amino acids (K, R) are suggested to function as nuclear localization signal (NLS; Cokol et al. 2000; Goldschneider et al. 2008), we wondered if the six residues (RSKARK) are involved in nuclear translocation of PRTG-ICD. We deleted the six residues from PRTG-ICD (PRTG-ICDdNLS) and transfected it or PRTG-ICD in 293T cells. Confocal images show that PRTG-ICD was localized mainly in the nuclei (Fig. 4B, left panels), while PRTG-ICDdNLS was localized both in nuclei and cytoplasm (Fig. 4B, right panels). Then, we quantified subcellular localization of PRTG-ICD and PRTG-ICDdNLS. We classified the cells into three categories according to the ratio of localization of the protein in Figure 4C; Nuc, protein was exclusively in nuclei; Nuc>Cyto, nuclear localization is more than cytoplasmic; Nuc=Cyto, nuclear localization is equivalent to cytoplasmic. The quantification confirmed that PRTG-ICD was localized exclusively within nuclei (Fig. 4C, upper), but that PRTG-ICDdNLS was localized in the cytoplasm as well as in the nucleus (Fig. 4C, lower). These results indicate that N-terminal six residues of PRTG-ICD are important for nuclear localization. The possibility that the conserved six residues are the cleavage site for γ-secretase is eliminated by showing that PRTGdECD-mutNLS where the six residues (RSKARK) were converted to (ASAAAA) generated γ-fragment after epoxomicin treatment in 293T cells (Fig. 4A, lane 2).
PRTG-ICD is localized in the nucleus in neuroepithelial cells
It has been previously shown that PRTG protein is expressed in the neuroepithelial cells of mice and chicks (Wong et al. 2010; Ito et al. 2011). We finally asked the localization of PRTG-ICD, which corresponds to C-terminal product of PRTG after γ-secretase processing, within the neuroepithelial cells. We performed in ovo electroporation toward the neural tube to transfect a construct pCAG-PRTG-ICD or pCAG-PRTG-ICDdNLS. pCAG-EGFP was also electroporated to monitor the transfection. Green fluorescent protein (GFP) fluorescence was found in the neural tube and in migrating neural crest cells (Fig. 5A–C, L–N). In the neural tube, PRTG-ICD was localized preferentially in the nuclei (Fig. 5B–G). On the other hand, PRTG-ICD in neural crest cells was localized in the cytoplasm (Fig. 5B–C, H–K). PRTG-ICDdNLS was localized in the cytoplasm both in the neural tube (Fig. 5M–R) and the neural crest cells (Fig. 5M–N, S–V). These results indicated that PRTG-ICD translocates to the nuclei in the neuroepithelial cells but not in neural crest cells, and that six residues at the N-terminal region of PRTG-ICD are indispensable for its nuclear localization both in vitro and in vivo.
We also noticed that the amount of PRTG-ICD, assessed by anti-myc staining, in the neuroepithelial cells was considerably smaller than that in the neural crest cells (compare Fig. 5E and 5I; Fig. 5F is a intensified image of Fig. 5E). We calculated average staining intensity per cell on the images of anti-myc staining (Fig. 5E, I; P, T) normalized with GFP staining intensity (Fig. 5D, H; O, S) and then compared them between the neuroepithelial cells and the neural crest cells. For PRTG-ICD, the staining intensity in the neuroepithelial cells was half of that in the neural crest cells (ratio of average signal intensity per cell = 0.45), while it was comparable for PRTG-ICDdNLS (ratio = 0.89). The results suggested that the amount of PRTG-ICD is reduced in the neuroepithelial cells after the translocation to the nuclei.
In the present study, we showed that PRTG receives two proteolytic cleavages. First cleavage occurs between the fibronectin III and the transmembrane domain for ectodomain-shedding, leaving putative α-fragments containing the transmembrane and the intracellular domain (Fig. 6, ectodomain-shedding). Second processing by γ-secretase cleaves at the interface of the transmembrane and the intracellular domain to release the C-terminal intracellular domain, PRTG-ICD (Fig. 6, γ-secretase cleavage). PRTG-ICD is localized to the nucleus in 293T cells, and the putative nuclear localization signal at N-terminal region of PRTG-ICD is important for translocation into the nucleus. Nuclear translocation of PRTG-ICD occurred in the neuroepithelial cells, but not in the neural crest cells. The amount of PRTG-ICD after nuclear translocation reduced in the neuroepithelial cells. Based on these results, the feature of these proteolytic cleavages and its role in PRTG signaling are suggested, as discussed below.
Many γ-secretase substrates are subjected to the ectodomain-shedding, in which extracellular domain is cleaved by metalloprotease, prior to γ-secretase cleavage. We showed that one of the first cleavage sites is within the conserved domain between the fibronectin III and the transmembrane domain (Fig. 1B, underlined). Since several putative α-fragments of similar sizes are produced by the first cleavage (Fig. 2A, lane 2), multiple cleavage sites should be located close to the conserved domain. Suspected protease for these cleavages is α-secretase/TACE/ADAM17, which cleaves Notch1 and Neogenin for ectodomain-shedding. However, treatment with TAPI-1, a specific inhibitor of α-secretase/TACE/ADAM17 did not prevent accumulation of putative α-fragments (data not shown), that the identity of the protease responsible for the first cleavage is uncertain.
We have reported that prtg mRNA is expressed in the neural tube but not in the neural crest (see Fig. 9 of Toyoda et al. 2005). Similar expression pattern for PRTG protein was observed by immunostaining using anti-PRTG antibody (data not shown). In the present study, we demonstrated that PRTG-ICD translocated to the nuclei in the neuroepithelial cells, but that in the neural crest cells it stayed in the cytoplasm instead of translocating to the nuclei. The result suggests that the neuroepithelial cells expressing endogenous PRTG may possess intrinsic nuclear sorting machinery depending on NLS of PRTG, but that neural crest cells do not possess it. In response to extracellular stimuli, like ligand-binding, endogenous PRTG may be processed by dual cleavages, then released PRTG-ICD, where NLS exists, may be subjected to nuclear translocation by virtue of this sorting machinery.
The amount of PRTG-ICD in the neuroepithelial cells was considerably smaller than that in the neural crest cells (Fig. 5E, I). Since various ICDs released from DCC, syndecan-3, nectin-1α and p75 by γ-secretase cleavage are degraded (Kim et al. 2002; Jung et al. 2003; Schulz et al. 2003; Taniguchi et al. 2003), PRTG-ICD is supposed to be degraded in the neuroepithelial cells. Considering that PRTG-ICD in the cytoplasm of the neural crest cells was not reduced (Fig. 5I), we suspected that the degradation of PRTG-ICD might occur in the nucleus after translocation. PRTG-ICD in the nucleus may be unstable due to proteasomal degradation so that γ-fragment could not be detected without proteasome inhibitor (Fig. 2). To support this speculation, endogenous PRTG in the neuroepithelial cells was observed on the plasma membrane but not in the nucleus (data not shown). PRTG in the neuroepithelial cells may be processed by γ-secretase upon ligand-binding, and released PRTG-ICD may be degraded quickly in the nuclei. Nuclear proteasome system may function to accelerate the turnover of ICDs to control its protein level (von Mikecz 2006). In contrast, when we transfected PRTG-ICD in neuroepithelial cells, because exogenous PRTG-ICD was so abundant, the protein was retained in the nuclei.
It is plausible that released PRTG-ICD has some biological functions in vivo, as manifested in other ICDs released by γ-secretase. The intracellular domain of Notch (NICD) is a best-characterized among ICD proteins released by γ-secretase cleavage. NICD translocates to the nucleus and interacts with CSL (C-promoter-binding factor RBP-Jκ/Suppressor-of-Hairless/Lag1), which activates target genes of Notch (Mumm & Kopan 2000). Similarly, ICD of N-cadherin and CD44 after γ-secretase cleavage cooperate with the transcriptional co-activator CBP (CREB-binding protein) and regulates transcription (Okamoto et al. 2001; Marambaud et al. 2003). ICD of DCC and Neogenin, closely related proteins of PRTG, also have potential transcriptional activities revealed by the reporter assay (Taniguchi et al. 2003; Goldschneider et al. 2008). Thus PRTG-ICD, sorted to the nucleus, is potentially involved in some transcriptional regulations. It was previously reported that in neuroepithelial cells, loss of PRTG resulted in precocious neuronal differentiation (Wong et al. 2010). One possibility is that PRTG-ICD might modulate transcription to prevent neuronal differentiation, as is the case of NICD in Notch signaling. In this scenario, the binding of a ligand for PRTG receptor may trigger proteolysis of PRTG, and released PRTG-ICD can translocate to the nucleus to regulate transcription for maintaining neural progenitors. On the other hand, transfection of PRTG-ICD in the neural tube did not affect neuronal differentiation (data not shown). Considering that knockdown of PRTG has increased neuronal differentiation, but simple PRTG-ICD may not modulate transcription, PRTG-ICD may need other cooperative factor(s) to modulate transcription.
Besides the potential activity of released PRTG-ICD in transcription, proteolysis of full-length PRTG may modulate its biological function of cell adhesion. In the previous paper we showed that PRTG contributed to the formation of paraxial mesoderm as a homophilic cell adhesion molecule (Ito et al. 2011). Proteolytic cleavages of PRTG may prevent or disrupt homophilic trans-interaction of PRTG between adjacent cells by eliminating extracellular domain of PRTG. Actually, in the case of E-cadherin, E-cadherin-mediated cell–cell adhesion is diminished by the ectodomain-shedding (Lochter et al. 1997; Noe et al. 2001). Moreover, γ-secretase cleavage dissociates the adhesion complex of E-cadherin, β-catenin and α-catenin from the cytoskeleton, thus promoting disassembly of this complex required for cell adhesion (Marambaud et al. 2002). Therefore, ectodomain-shedding and subsequent γ-secretase cleavage of PRTG may dissociate homophilic association of extracellular PRTG and PRTG-mediated adhesion complex. Different from Notch signaling in which sequential cleavages occur upon ligand-binding, the proteolytic cleavages of adhesion molecules such as E-cadherin, N-cadherin, and CD44 are not necessarily evoked by ligand-binding, but stimulated by some extracellular or intracellular cues depending on biological contexts (Kopan & Ilagan 2004). Since homophilic adhesion of PRTG on the cell membrane is important to maintain adhesion level between epiblast cells and between paraxial mesodermal cells, excess or lack of PRTG may impair successive ingression of epiblast cells and disorganizes the epithelial structure of the somites (Ito et al. 2011). It is thus tempting to speculate that proteolytic cleavage of PRTG may play a role in modulating the level of adhesion among the epiblast and paraxial mesodermal cells.
In conclusion, our study has shown that PRTG receives two proteolytic cleavages, ectodomain-shedding and γ-secretase cleavage, to generate the intracellular domain of PRTG (PRTG-ICD), which translocates into the nucleus (Fig. 6). We propose that nuclear PRTG-ICD is potentially implicated in transcriptional regulation of neural differentiation, and that proteolytic cleavage of full-length PRTG may modulate cell adhesion mediated by PRTG for formation of paraxial mesoderm.
This study was supported by Grants-in-Aid for Scientific Research (C) from the Japan Society for the Promotion of Science (JSPS) to Y.W.