Arginine-rich Peptides AN ABUNDANT SOURCE OF MEMBRANE-PERMEABLE PEPTIDES HAVING POTENTIAL AS CARRIERS FOR INTRACELLULAR PROTEIN DELIVERY

A basic peptide derived from human immunodeficiency virus (HIV)-1 Tat protein (positions 48–60) has been reported to have the ability to translocate through the cell membranes and accumulate in the nucleus, the characteristics of which are utilized for the delivery of exogenous proteins into cells. Based on the fluorescence microscopic observations of mouse macrophage RAW264.7 cells, we found that various arginine-rich peptides have a translocation activity very similar to Tat-(48–60). These included such peptides as the D-amino acidand arginine-substituted Tat-(48–60), the RNA-binding peptides derived from virus proteins, such as HIV-1 Rev, and flock house virus coat proteins, and the DNA binding segments of leucine zipper proteins, such as cancer-related proteins c-Fos and c-Jun, and the yeast transcription factor GCN4. These segments have no specific primary and secondary structures in common except that they have several arginine residues in the sequences. Moreover, these peptides were able to be internalized even at 4 °C. These results strongly suggested the possible existence of a common internalization mechanism ubiquitous to arginine-rich peptides, which is not explained by a typical endocytosis. Using (Arg)n (n 5 4–16) peptides, we also demonstrated that there would be an optimal number of arginine residues (n ; 8) for the efficient translocation.

Intraperitoneal injection of the protein resulted in delivery of the protein with ␤-galactosidase activity to various tissues in mice, including the brain. The peptide-mediated approaches would allow the incorporation of peptides containing unnatural amino acids or nonpeptide molecules such as fluorescence probes. These methods would become powerful tools not only for therapeutic purposes as an alternative to gene delivery, but also for the understanding of the mechanisms behind fundamental cellular events, such as signal transduction and gene transcription.
Besides the potential of Tat-(48 -60) as a protein carrier, the internalization mechanism of the peptide attracted our interest. For example, Tat-(48 -60) (GRKKRRQRRRPPQ) is a highly basic and hydrophilic peptide, which contains 6 arginine and 2 lysine residues in its 13 amino acid residues. However, the peptide was reported to be translocated through the cell membranes in 5 min at a concentration of 0.1 M (2). Internalization of the peptide was not inhibited even at 4°C. The peptide is less toxic to cells than other basic membrane-interacting agents. The above features suggested that the internalization mechanism of Tat-(48 -60) was completely different from the typical transmembrane mechanisms reported so far. Questions arise as to whether such an efficient translocation is specific for Tat-(48 -60) and Antennapedia-(43-58) peptides and what is the mechanism of the highly efficient internalization. Based on experiments using synthetic peptides, we suggest the possible presence of a very similar translocation mechanism to Tat-(48 -60) present among the various arginine-rich peptides. We also suggest the possible existence of the optimum chain length of arginine peptides for the internalization.

EXPERIMENTAL PROCEDURES
Peptide Synthesis and Fluorescent Labeling-All the peptides used in this study were chemically synthesized by Fmoc (9-fluorenylmethyloxycarbonyl)-solid-phase peptide synthesis on a Rink amide resin as reported previously (12). Fluorescent labeling of the peptides was conducted by the treatment with 1.5 eq of 5-maleimidofluorescein diacetate (Sigma) in dimethylformamide-methanol (1:2) for 3 h followed by reverse-phase HPLC purification. The fidelity of the products was ascertained by time-of-flight mass spectrometry.
Conjugation of Carbonic Anhydrase with Basic Peptides-Carbonic anhydrase in phosphate-buffered saline (PBS) was simultaneously treated with fluorescein-5(6)-carboxamidocaproic acid N-hydroxysuccinimide ester (Sigma) and N-(6-maleimidocaproyloxy)succinimide ester (Dojin) (15 eq, each) at room temperature for 1 h to introduce the fluorescein and the maleimide function to the protein. After the removal of the unreacted reagents by gel-filtration on a Sephadex G-25 (Amersham Pharmacia Biotech) column, the cysteine of the respective arginine-rich peptides was allowed to react with the maleimide moiety on the above fluorescein-labeled protein at room temperature for 16 h, and then the unreacted peptides were removed by gel-filtration. Based on the molecular weight estimation by SDS-polyacrylamide gel electro-* This work was supported by grants-in-aid for scientific research from the Ministry of Education, Science, Sports and Culture of Japan and in part by the Mochida Memorial Foundation for Medical and Pharmaceutical Research. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Cell Culture-Mouse macrophage RAW264.7 cells were maintained in RPMI 1640 medium with 10% heat-inactivated fetal bovine serum. Cells were grown on 60-mm dishes and incubated at 37°C under 5% CO 2 to ϳ70% confluence. A subculture was performed every 3-4 days.
Peptide Internalization and Visualization-For each assay, 4 ϫ 10 4 /ml cells were pelleted on a eight-well Lab-Tek-II chamber slide (Nalge Nunc) (250 l/well) and cultured for 16 h. After complete adhesion, the culture medium was exchanged. The cells were incubated at 37°C for 3 h with the fresh medium (250 l) containing fluoresceinlabeled peptides or proteins. The concentrations of the peptides and proteins were adjusted before addition to the cell based on their fluorescent intensity. Cells were washed three times with PBS, fixed with acetone-methanol (1:1) for 2 min at room temperature, washed three times with PBS again, and then mounted in fluorescent mounting medium containing 15 mM NaN 3 (Dako). The distribution of fluoresceinlabeled peptides was analyzed on a Zeiss Axioskop fluorescence microscope using a 100ϫ oil immersion lens.
Confocal Microscopy-Cells were grown, incubated with proteins, and fixed basically as described above. Cells were then treated with PBS containing 5 M propidium iodide (200 l) at room temperature for 30 min, washed four times with PBS, and mounted in glycerol:PBS (9:1) containing 1% p-phenylenediamine dihydrochloride. Data were obtained using a confocal scanning laser microscope MRC 1024 (Bio-Rad) equipped with a 60ϫ oil immersion lens or LSM 510 (Zeiss) equipped with a 40ϫ lens.

Uptake of Tat-(48 -60) Analogs by the Macrophage Cell-To
obtain insight into the translocation mechanisms of the Tat-(48 -60) peptide, Tat-(48 -60), its D-amino acid-substituted analog (D-Tat) and arginine-substituted analog (R 9 -Tat), where residues corresponding to positions 49 -57 were replaced with arginine, were synthesized ( Fig. 1a). An extra cysteine amide was incorporated into the C terminus of each peptide for the fluorescent labeling. The peptides corresponding to nuclear localization sequences (NLS) derived from simian virus 40 (13) and nucleoplasmin (14) were also synthesized as references. Treatment of the peptides with 5-maleimidofluorescein diacetate gave the corresponding fluorescein-labeled peptides. Internalization of the peptides was monitored by fluorescence microscopic observation after a 3-h incubation of the peptides with mouse macrophage RAW 264.7 cells at 37°C. As a result, D-Tat and R 9 -Tat were internalized into the cell as efficiently as the Tat-(48 -60) peptides, and localization into both the cytoplasm and nucleus was observed (Fig. 2). A similar internalization of the D-amino acid analog of Tat was reported by Huq et al. (15) using a linear peptide corresponding to residues 37-72. These results would contradict the idea that a specific receptor may play a crucial role in the translocation of the Tat-(48 -60) peptide. On the other hand, the simian virus 40derived and nucleoplasmin-derived peptides showed a much lower degree of internalization. These NLS-derived peptides are rich in lysine. The above results suggested that arginine residues would play an important role in the translocation.
Translocation of Various Arginine-rich Nucleic Acid-binding Peptides through the Macrophage Cell Membranes-Argininerich basic segments are used by a variety of RNA-binding proteins to recognize specific RNA structures (16). If arginine in the sequence plays an important role in the translocation, peptides corresponding to these RNA-binding segments may translocate through the cell membranes. To test this hypothesis, 10 arginine-rich RNA-binding peptides bearing a C-terminal Gly-Cys-amide (Fig. 1b) were similarly prepared, fluorescein-labeled, and applied to the macrophage cells.
To our surprise, all the peptides other than the human U2AF-(142-153) peptide translocated through the cell membranes and accumulated in the cytoplasm and nucleus (Fig. 3). As judged from the fluorescent intensity, the efficiency of incorporation into the cells showed a tendency to correspond to the number of arginine residues in the sequence. Internalization activity of the HIV-1 Rev-(34 -50), FHV coat-(35-49), HTLV-II Rex- (4 -16), and BMV Gag-(7-25) peptides, which have more than seven arginine residues in their sequences, were comparable with that of the Tat-(48 -60) peptide. Fluorescence was observed in the cells as early as 5 min after the addition of these peptides ( in the cells treated with the former peptides (0.1 M) was judged not to be less than that in those treated with the latter peptides (10 M). The P22 N-(14 -30) and cowpea chlorotic mottle virus Gag-(7-25) peptides that have six arginine residues showed a moderate degree of translocation. HIV-1 Tat-(48 -60) is reported to translocate through the cell membranes and accumulate in the nucleus, especially the nucleolus (2). A similar tendency was observed with the above peptides. Not only the RNA-binding peptides but also the DNA-binding peptides corresponding to the basic leucine zipper segments derived from cancer-related proteins, c-Fos and c-Jun, and the yeast transcription factor, GCN4, which were also rich in arginine (Fig. 1c), were internalized into the cells with almost the same efficiency as that of Tat-(48 -60) (Fig. 4).
HIV-1 Tat-(48 -60) was reported to induce little toxicity to HeLa cells (2). Using R 9 -Tat, HIV-1 Rev-(34 -50), and FHV coat-(35-49) peptides as representatives of the above argininerich peptides, cytotoxicity of the peptides was investigated. Determined by the MTT assay, the above peptides did not show a significant cytotoxicity to the macrophage cells during the treatment with a peptide (10 M) for 24 h. At 100 M, cell viability of the cells treated with R 9 -Tat became 70%, whereas viability of those treated with other peptides as well as HIV-1 Tat-(48 -60) was still greater than 95%. These results suggested that many of the arginine-rich peptides can be of low cytotoxicity as reported for the HIV-1 Tat-(48 -60) peptide.

Consideration of the Translocation Mechanism of the Arginine-rich Peptides-
The above experiments showed that a variety of arginine-rich RNA/DNA-binding peptides were able to translocate through the cell membranes. Little homology in these sequences was observed, except that they all have 5-11 arginine residues. Moreover, the D-amino acid substituted Rev-(34 -50) peptide (1 M) was internalized as efficiently as the L-peptide in 3 h (data not shown). Circular dichroism (CD) spectra of the HIV-1 Tat-(48 -60), R 9 -Tat, and FHV coat-(35-49) peptides in methanol were suggestive of their not having a significant secondary structure (Fig. 5), whereas the HIV-1 Rev-(34 -50) peptide showed a spectrum typical of an ␣-helical peptide. The U2AF peptide, which was only slightly internalized into the cell, showed a spectrum very similar to that of the FHV coat-(35-49) peptide. These results were suggestive of the absence of even a common secondary structure in the membrane-permeable peptides. When the cells were incubated with a peptide (1 M) at 4°C for 30 min, no significant decrease in fluorescent intensity in the cell was observed using the HIV-1 Rev-(34 -50), and FHV coat-(35-49) peptides (Fig. 6). These results suggested that typical endocytosis pathways so far established would not play a crucial role in the translocation of these arginine-rich peptides.
We next focused on the question whether the entry of arginine-rich peptides into the cells is one-way or not. The cells were treated with the HIV-1 Rev-(34 -50) peptide (1 M) for 3 h, then the medium was exchanged with a fresh one not containing the peptide. The fluorescence intensity from the cells 1 h later was almost comparable with or only slightly less than that of the cells just before the medium exchange. However, a substantial decrease in the fluorescence intensity was recognized in the cells 6 h later, and complete disappearance of the fluorescence was observed 24 h later. To examine if the above results were due to the leakage of the peptide from the cells, the medium was analyzed by an HPLC equipped with a fluorescence spectrophotometer. No peak was detected at the retention time corresponding to the peptide; however, peaks were observed that eluted at positions identical with those of the peptide treated with trypsin (data not shown). Therefore, we concluded that the decrease in fluorescence intensity of the cells mainly resulted from the degradation of the peptides, and not from the leakage of the intact peptide. The question whether the ingested peptide had a certain effect on the cell growth was also examined. The above HIV-1 Rev-(34 -50)treated cells were harvested 24 h later and counted. The cell number for the peptide-treated cells was comparable with that for the control cells (without peptide treatment). Thus, the peptide-ingesting cells were judged to remain viable to divide with little effect by the peptide. It would be plausible that the peptide evenly distributes in each of the daughter cells upon  -(34 -50) (a), P22 N-(14 -30) (b), and N-(1-22) (c) (10 M each) for 3 h. The cells were incubated with the peptide (1 M) for 30 min at 4°C or at 37°C. In the former case, the cells were preincubated at 4°C for 1 h before addition of the peptide. All the following procedures were also conducted at 4°C until the completion of the fixation. cell division, since significant differences in the florescence intensity were not observed among the adjoining cells 6 h later. Considering the doubling time of the cell, which was estimated to be about 18 h, a certain amount of cells must have divided within the 6 h. If the peptides would preferentially stay in one of the daughter cells upon cell division, a certain discrepancy in the fluorescence intensity will be observed among the adjacent cells. However, further study will be necessary to adequately address this question.
Applicability of the Arginine-rich Peptides to the Intracellular Protein Delivery-To examine the applicability of the above basic peptides as protein carriers, we prepared basic peptide-protein conjugates. Carbonic anhydrase (29 kDa) was selected as a model protein. Basic peptide-carbonic anhydrase conjugates were prepared using N-(6-maleimidocaproyloxy)succinimide ester (EMCS) as a cross-linking agent (17) (Fig. 7A). A fluorescein moiety was introduced into the protein using the fluorescein-5(6)-carboxamidocaproic acid Nhydroxysuccinimide ester simultaneously with EMCS. As judged from the SDS-polyacrylamide gel electrophoresis of the conjugates, one to two molecules of the basic peptide and fluorescein moiety were introduced into a molecule of carbonic anhydrase, respectively. Carbonic anhydrase was successfully delivered into the cells with the help of the HIV-1 Rev-(34 -50), FHV coat-(35-49), and R 9 -Tat peptides as efficiently as with the HIV-1 Tat-(48 -60) peptide (Fig. 7B). Accumulation of the conjugates in the cytosol and nucleus was also observed by fluorescence microscopy of the cells without fixation (Fig. 7B). Confocal microscopic analysis of these conjugates demonstrated both cytoplasmic and nuclear localization and not just attachment to the cellular membranes (Fig. 7C). On the other hand, fluorescein-labeled pro- tein without a carrier peptide was located in a limited part of the cytosol (Fig. 7C). This result suggested that the protein was captured in the endosomes and was not able to be released into the cytosol. Myoglobin (17 kDa) was also introduced into the cell with the help of these carrier peptides (data not shown).
Effect of the Length of Arginine Chain on the Internalization-The above data strongly suggested the importance of arginine residues in the internalization. The possible existence of the unique internalization mechanism common in these arginine-rich peptides was also suggested. We then examined the effect of the number of arginine residues in the sequences. For simplification, peptides that are composed of 4 -16 residues of arginine were prepared (Fig. 1d). To their C termini, the Gly-Cys-amide segment was also attached for the fluorescein labeling. These results are shown in Fig. 8A. Considerable difference was recognized on the translocation efficiency and intracellular localization among these peptides. R 4 showed extremely low translocation activity. R 6 and R 8 exhibited the maximum internalization and accumulation in the nucleus. What is interesting is that the degree of internalization decreased as the chain length further increased. For the R 16 , internalization of the peptide was not significant. The same kind of difference was recognized in the experiments using the conjugates of carbonic anhydrase with the arginine peptides (Fig. 8B). A similar tendency was observed on the protein delivery using R 8 and R 16 as the carrier molecules. Based on the confocal laser microscopic observations, the R 8 -carbonic anhydrase conjugate was efficiently internalized into the macrophage cells and accumulation in the nucleus was observed as was seen in the case of the HIV-1 Rev-(34 -50) conjugate. In contrast, the R 16 -conjugate seemed to mainly reside on the cell membranes after a 3-h incubation with the conjugate, but significant accumulation in the nucleus was not observed.

DISCUSSION
In this report, we have shown that not only Tat-(48 -60) but also various arginine-rich peptides were able to translocate through the mouse macrophage membranes. These peptides include the D-and arginine-substituted HIV-1 Tat-(48 -60) analogs, RNA-binding peptides derived from proteins, such as HIV-1 Rev, HTLV-II Rex, BMV Gag, and FHV coat proteins, and the DNA-binding segments from c-Fos, c-Jun, and the GCN4 proteins. There seems a common or very similar mechanism for the internalization among these peptides. The mechanism is explained neither by adsorptive-mediated endocytosis nor by receptor-mediated endocytosis because the peptides were internalized by the cell at 4°C, and there seemed little homology both in the primary and secondary structures among these membrane-permeable peptides except that they have several arginine residues in the sequences. These results strongly suggest the possible presence of the common and undefined internalization mechanisms ubiquitously laying among the arginine-rich basic peptides. As one more new find-ing concerning the features of the internalization, we have shown that the number of arginine residues has a significant influence on the method of internalization and that there seems to be an optimal number of arginine residues for the internalization. There still remain questions why such efficient translocation is possible for the arginine-rich peptides. Possible hydrogen bond formation of arginine with lipid phosphates (18) or interaction with extracellular matrices such as heparan sulfate (19) may be involved in the initial step during the mechanism. However, as was seen in the case of the R 16 peptide, it is not enough to explain the mechanism only by considering adsorption of peptides on the membranes.
Tat-(48 -60) has been reported to carry various proteins into the cells not only into cultured cells but also into the various organs of a living mouse (4). As the arginine-rich peptides examined here seem to have a similar ability as carriers of proteins, further study of the arginine-based peptides may result in finding peptides penetrating to some specific cells by themselves or with the help of other address peptides.
The results obtained here not only shed light on the possible presence of new types of ubiquitous transmembrane mechanisms for the arginine-rich peptides, but also on the development of novel carrier molecules for the intracellular protein delivery.