BETA2/NeuroD Protein Can Be Transduced Into Cells Due to an Arginine- and Lysine-Rich Sequence

Full-length BETA2/NeuroD cDNA was amplified by PCR using appropriate linker-primers and then subcloned into the Nde I and Xho I sites of pET21b(+) (Novagen, Madison, WI) using a ligation kit (TaKaRa, Tokyo, Japan). On mutant plasmids, sequences before and after each deletion were amplified by PCR using appropriate linker-primers and were then subcloned into the Nde I–HindIII and HindIII−Xho I sites of pET21b(+), respectively. BL21 (DE3) cells containing the expression plasmids were grown at 37°C to an optical density 600 of 0.8. Isopropyl-β-d-thiogalactopyranoside was added to a final concentration of 0.1 mmol/l, and the cells were then incubated for 12 h at 24°C. Cells were sonicated and the supernatants were recovered and applied to a column of Ni-nitrilotriacetic acid agarose (Invitrogen, San Diego, CA). Purified BETA2/NeuroD proteins were conjugated using a Cy5-labeling kit (Amersham Pharmacia, Piscataway, NJ).

Islets were isolated from the pancreases of anesthetized male SD rats (200–250 g; Taconic Farms, Germantown, NY). All animal procedures were approved by the review committee of Kyoto University Graduate School of Medicine. The common bile duct was cannulated and injected with 6 ml cold M199 medium containing 1.5 mg/ml collagenase (Roche Boehringer Mannheim, Indianapolis, IN). The islets were separated on Histopaque 1077 (Sigma, St. Louis, MO) density gradient, hand picked under a dissecting microscope to ensure a pure islet preparation, and used immediately afterward. In these experiments, islets were cultured in RPMI medium with 10% FCS.

Western blotting was conducted in accordance with the method described previously (19). In brief, after being washed three times with PBS and high salt buffer, cells were scraped from the dish and sonicated. Then 10 μg of cell extracts were fractionated by 10% SDS-PAGE and transferred to polyvinylidine fluoride membranes (Immun-Blot PVDF Membrane; Bio-RAD, Hercules, CA) using a transfer buffer containing 20% methanol, 25 mmol/l Tris base, and 192 mmol/l glycine (300 mA, 2 h). After being blocked at room temperature for 1 h in 50 mmol/l Tris-HCl, 150 mmol/l NaCl, and 0.1% Tween-20 (TBST) with 5% nonfat dry milk, the membranes were incubated overnight at 4°C in TBST using 5% nonfat dry milk containing goat anti−BETA2/NeuroD antibody (Santa Cruz Biotechnology, Santa Cruz, CA), mouse anti-actin antibody (1:1,000; Sigma), or horseradish peroxidase−conjugated anti−6 histidine antibody (1:5,000; Invitrogen). For anti−BETA2/NeuroD antibody or anti-actin antibody, the membranes were then incubated for 1 h at room temperature in TBST with 5% nonfat dry milk containing anti-goat or anti-mouse IgG antibody coupled to horseradish peroxidase (1:1,000; Bio-Rad).

For this assay, 2 μg nuclear extract were incubated with 2 μg poly (dI-dC), 25 mmol/l HEPES (pH 7.8), 0.2 mmol/l EDTA, 150 mmol/l KCl, 5 mmol/l dithiothreitol, and 9% glycerol at room temperature. The binding reaction was initiated by adding ³²P-labeled double-stranded oligonucleotide probes. A double-stranded oligonucleotide reproducing the rat insulin gene BETA2/NeuroD−binding region (E-box) and surrounding sequences (ACGTTCTGGCCATCTGCTGAT CCTACGT; E-box in italics) or E-box−mutated sequences (ACGTTCTGGCTTTCCCCTGAT CCTACGT) (Sigma Genosys, Woodlands, TX) was used as a binding probe. In some binding assays, anti−BETA2/NeuroD, anti−6 histidine antibody, and preimmune antisera were added to the reaction mixture 1 h before DNA probes were added.

A rat insulin II promoter-reporter (luciferase) plasmid (1.0 μg) containing 238-bp 5′-flanking sequences of the rat insulin II promoter region (20) or E-box−mutated insulin II promoter-reporter plasmid (1.0 μg) (18) was transfected into cells with LipofectAMINE (Invitrogen) using the conditions recommended by the manufacturer. Then 48 h after the transfection, the cells were harvested and assayed (Promega, Madison, WI).

Fluorescein isothiocyanate (FITC)−conjugated BETA2/NeuroD PTD peptide was synthesized by Sigma Genosys-Japan. The peptide was purified by preparative reversed-phase high-performance liquid chromatography, was confirmed as >95% pure as analyzed by high-performance liquid chromatography, and had the expected amino acid composition and mass spectra.

Cervix-derived HeLa cells were maintained in Dulbecco’s modified Eagle’s medium supplemented with 10% FCS and penicillin-streptomycin. Cells were treated with 1 μmol/l BETA2/NeuroD PTD for the specified time and then washed with fresh medium. To evaluate endocytosis, cells were incubated with 1 μmol/l FITC−BETA2/NeuroD PTD and 10 μmol/l FM 4–64 (Molecular Probes, Eugene, OR), a marker of endocytosis. Treated living cells were analyzed by conventional fluorescence microscopy and confocal microscopy.

HeLa cells were transfected with a plasmid expressing a dominant-negative mutant of dynamin-1 (DynDN), Discosoma sp. red fluorescent protein (DsRed), as previously described (21). After 24 h, cells were washed and treated with 1 μmol/l FITC−BETA2/NeuroD PTD in complete medium. After cells were incubated overnight, the FITC and DsRed signals were examined.

To measure the mechanism of BETA2/NeuroD internalization, HeLa cells were plated and cultured in Dulbecco’s modified Eagle’s medium with 10% FBS. Cells were then pretreated for 10 or 30 min in serum-free medium with 5 mmol/l amiloride (Sigma) or 10 μmol/l cytochalasin D (Sigma), respectively. After the BETA2/NeuroD PTD was added, cells were maintained for 1 h in the presence of inhibitors, washed several times, and incubated for 1 h in complete medium.

To test whether the purified BETA2/NeuroD protein can be transduced into cells, cervix-derived HeLa, liver-derived HepG2, β-cell−derived HIT-T15 cells, and primary rat islets and duct cells were treated with Cy5-conjugated BETA2/NeuroD protein. Then 6 h after this treatment, BETA2/NeuroD was observed as a fluorescence signal in primary islets, duct cells, HIT-T15, HepG2 (Fig. 1B), and HeLa cell lines (Fig. 2C1). Although it has previously been shown that some homeoprotein transcription factors can enter a large number of cell types (9,20–22), our results showed that the BETA2/NeuroD protein, a bHLH transcription factor, is delivered into cells without requiring any additional sequences.

To clarify the location of the PTD of BETA2/NeuroD, we made several Cy5-conjugated deletion proteins (Fig. 2A and B). Full-length BETA2/NeuroD protein and deletion BETA2/NeuroD (dNeuroD)-2, -4, -7, -8, and -9, but not the other dNeuroDs, could be transduced into HeLa cells (Fig. 2C and D). Both dNeuroD-8 and -9 proteins could be transduced into cells, but the Cy5 signals were weaker than that with full-length protein treatment (Fig. 2C). These data show that the 34 amino acids of the putative BETA2/NeuroD PTD are necessary for the BETA2/NeuroD protein to be efficiently transduced into cells (Fig. 3A). As we and others have recently shown, the sequences of BETA2/NeuroD protein are rich in arginine and lysine (10–12).

To test the transduction efficiency of the PTD of BETA2/NeuroD, we made enhanced green fluorescent protein (EGFP)−fused PTDs. EGFP with 11 arginine (11R) PTD, the most effective protein transduction domain (10,11), and EGFP without PTD were used as positive and negative controls, respectively (Figs. 3A and B). HeLa cells were treated with these EGFP proteins for 6 h and analyzed by Western blotting (Fig. 3C) and confocal microscopy (Fig. 3D). The transduction efficiency of the full-length BETA2/NeuroD PTD was similar to that of the 11R PTD, whereas the efficiency of the partial sequence of the PTD was about half that of the 11R PTD (Figs. 3C and D). These data show that the transduction efficiency of BETA2/NeuroD PTD is as strong as that of 11R PTD.

To test the effectiveness of transduced BETA2/NeuroD, HeLa cells were treated with the BETA2/NeuroD protein at several concentrations for 2 h, washed with PBS, incubated with conventional medium, and analyzed by Western blotting (Fig. 4A). The BETA2/NeuroD protein was transduced in a dose-dependent manner up to 1 μmol/l. To determine the stability of the BETA2/NeuroD protein, HeLa cells were treated with 1 μmol/l BETA2/NeuroD for 2 h, washed with PBS, incubated with conventional medium for 1–46 h, and analyzed by Western blotting. The half-life of BETA2/NeuroD was ∼5 h (Fig. 4B). In addition, the transduced BETA2/NeuroD could be detected using an antibody to the histidine tag included in this synthesized protein.

The insulin-enhancer element E-box is the binding site for BETA2/NeuroD and plays an important role in the regulation of insulin gene transcription (4,17,18). We therefore used a gel shift assay to test whether the transduced BETA2/NeuroD protein binds to the insulin-enhancer element. HeLa cells were treated with 1 μmol/l BETA2/NeuroD protein, dNeuroD-6 (PTD-deleted BETA2/NeuroD protein), and EGFP with BETA2/NeuroD PTD (EGFP-PTD) for 24 h. The nuclear extract of HIT-T15 cells was used as a positive control. The E-box binding complex, but not the mutated E-box, was observed in nuclear extracts from treated HeLa cells. The specificity of the band was confirmed by its supershift with BETA2/NeuroD antibody and anti−6 histidine antibody but not by preimmune serum (Fig. 5A).

To test whether transduced BETA2/NeuroD protein can activate insulin gene transcription, we examined its effect on insulin promoter activity using a luciferase assay. HeLa cells were transfected with an insulin promoter−luciferase plasmid and allowed to grow for 24 h; they were then treated with 1 μmol/l BETA2/NeuroD protein, dNeuroD-6, and EGFP-PTD for an additional 24 h, after which a luciferase assay was performed. Insulin promoter activity was increased by treatment with BETA2/NeuroD protein in HeLa cells (Fig. 5B) but not by treatment with dNeuroD-6 or EGFP-PTD. Mutated insulin promoter activity was not significantly changed by any treatment. Taken together, these results suggest that exogenous BETA2/NeuroD protein can be transduced into cells and their nuclei, bind to the E-box, and activate the insulin promoter.

To investigate the mechanism of BETA2/NeuroD protein transduction, HeLa cells were treated with FITC-conjugated BETA2/NeuroD PTD and FM 4–64, a marker of endocytosis, for 30 min and then washed with fresh media. The PTD fluorescence signal appeared as punctate within living, unfixed HeLa cells. This punctate PTD mainly co-localized with FM 4–64 in living cells (Fig. 6A). After an additional 90 min, the PTD was homogeneously distributed in the cytosol and nuclei of some cells, whereas the FM dye remained in the endosome (Fig. 6A). These data show that the cellular uptake of BETA2/NeuroD PTD occurs through electrostatic interaction with the plasma membrane, subsequent endocytosis, and escape from the endosome with eventual homogeneous distribution in the cytoplasm.

It was then determined whether the cellular uptake of BETA2/NeuroD occurs through a specific endocytic pathway. Multiple forms of endocytosis, including phagocytosis and clathrin- and caveolar-mediated endocytosis, require dynamin GTPase activity for vesicle formation at the cell surface. The expression of DynDN effectively blocks these endocytic pathways (21,23). We transfected cells with the DynDN-DsRed plasmid, treated the cells with FITC-conjugated BETA2/NeuroD PTD 24 h after transfection, and analyzed FITC signals at 48 h. The expression of DynDN did not block BETA2/NeuroD PTD uptake and recombination (Fig. 6B). These observations show that the BETA2/NeuroD internalization process does not involve caveolar- or clathrin-mediated endocytosis.

Recently, it was reported that PTD-mediated cellular entry occurs by macropinocytosis (7,21). Macropinocytosis is a rapid, lipid raft−dependent, and receptor-independent form of endocytosis. It requires actin membrane protrusions that envelope vesicles, termed macropinosomes (23). To examine the involvement of macropinocytosis in transduction, cells were pretreated with amiloride, a specific inhibitor of the Na⁺/H⁺ exchange required for macropinocytosis (24), or cytochalasin D, an F-actin elongation inhibitor (25). Pretreated cells were then incubated with BETA2/NeuroD PTD for 1 h and washed with complete medium. Cell treatment with either macropinosome inhibitor resulted in the reduction of BETA2/NeuroD PTD transduction into vesicles with no cytotoxicity (Fig. 7A). Taken together, these observations showing rapid transduction into cells, dynamin-1 independence, and uptake inhibition by amiloride and cytochalasin D, indicate that BETA2/NeuroD PTD−mediated cellular entry occurs by lipid raft−mediated macropinocytosis.

Bafilomycin A1 is a highly potent and selective inhibitor of vacuolar H⁺ ATPases and inhibits endosomal acidification (26). The effect of bafilomycin A1 on the intracellular localization of the peptide was tested: 30 min before the PTD peptide was added, cells were washed once and then 300 nmol/l bafilomycin was added to the medium. After a 2-h incubation with BETA2/NeuroD PTD, no cytoplasmic fluorescence was present and only vesicular staining was observed (Fig. 7B). These data show that the release of PTD into the cytosol occurs by a mechanism dependent on endosomal acidification and preserved endosomal integrity.

We next addressed the potential involvement of the Golgi complex in the cellular trafficking of BETA2/NeuroD PTD. Brefeldin A interferes with the integrity of the Golgi and trans-Golgi network and affects the uptake of the TAT fusion protein (27). In this study, cells were washed once and 20 μmol/l brefeldin A was added 30 min before the peptide was introduced. After a 2-h incubation with the peptide, the cellular fluorescence of BETA2/NeuroD PTD was reduced (Fig. 7B), similar to what was seen with bafilomycin A1. Taken together, these data suggest that the BETA2/NeuroD protein penetrated cells by macropinocytosis and entered the cytoplasm by means of retrograde transport (Fig. 8).

PTDs are a series of small domains that have been shown to cross biological membranes efficiently, independently of transporters or specific receptors, and promote the delivery of full-length proteins in a rapid and concentration-dependent fashion into a large number of cells. These PTDs are arginine- and lysine-rich sequences, especially the TAT PTD. We and others have shown that proteins with fused polyarginine or polylysine sequences can permeate several cell types even more efficiently (10–12,28). It has also been shown that >40% of lysine- and arginine-rich peptides within 11–16 amino acids can permeate cells (29). BETA2/NeuroD has arginine- and lysine-rich sequences in its own sequence and the arginine and lysine content is 53% over 34 amino acids.

BETA2/NeuroD is selectively expressed in the developing endocrine pancreas, the small intestine, and the nervous system (2) and plays an important role in maintaining the differentiation of endocrine cells, proper islet morphogenesis, and normal cell function, especially β-cell function, by regulating the insulin gene (4,5). Mice with an inactivated BETA2/NeuroD gene die of severe diabetes caused by a major reduction in the number of β-cells and a lack of proper islet formation (4). BETA2/NeuroD strongly stimulates DNA-binding capacity by heterodimerization with class A bHLH proteins, such as the ubiquitous E47. The PTD of BETA2/NeuroD includes the DNA-binding region of the bHLH factor. Thus the 34 amino acids of BETA2/NeuroD PTD are extremely important in several other aspects of its function.

Because the DNA-binding domain is conserved between different bHLH proteins (Fig. 8B, underlined), most, but not all, bHLH proteins could be transduced into cells. However, transduction efficiency is extremely different between each bHLH protein because sequences before the DNA binding domain are quite different. Neurogenin 3 and myogenic determination factor could be transduced into cells, but the efficiency is less than that of the BETA2/NeuroD protein (H.N., M.M., unpublished observations). E47 is hardly transduced into cells.

Our study showed that BETA2/NeuroD protein has a native PTD and, without modification, can permeate several cell types and enhance insulin promoter activity in these cells. In the insulin promoter, E-boxes, corresponding to the consensus CANNTG, are the binding site of heterodimer E47/BETA2 and are juxtaposed to AT-rich elements (A-box) that bind pancreatic duodenal homeobox 1 (PDX-1). The synergistic activation of cis-elements is observed between BETA2 and PDX-1 (30,31). Moreover, these two proteins are important in pancreatic development (5,32). We have reported that the PDX-1 protein can transduce cells because of its own Antennapedia-like protein and that PDX-1 protein induces insulin expression in pancreatic ductal cells, thought to be islet progenitor cells (20). The exogenous BETA2/NeuroD protein also induces insulin expression in pancreatic ductal progenitor cells (H.N., S.M., unpublished observations). The BETA2/NeuroD gene has E-boxes, and the BETA2/NeuroD protein could stimulate its own transcription (33). Furthermore, the pdx-1 gene has A-boxes, and the PDX-1 protein could positively autoregulate its expression (34). Once the BETA2/NeuroD and PDX-1 proteins are transduced into progenitor cells, endogenous BETA2/NeuroD and pdx-1 gene transcription are amplified by these proteins, stimulate insulin transcription, and may facilitate their differentiation into insulin-producing cells. Our finding of a functional PTD in transcription factor BETA2/NeuroD suggests that exogenous BETA2/NeuroD could be a novel approach for generating surrogate β-cells without requiring gene transfer technology.

H.N. is a Research Fellow of the Japan Society for the Promotion of Science, Kyoto University Graduate School of Medicine.

We thank Dr. Susumu Seino (Kobe University) and Kotaro Minami (Kyoto University) for materials.