Cell-Surface Interactions on Arginine-Rich Cell-Penetrating Peptides

Sep 14, 2017 - The Evolution of DNA-Templated Synthesis as a Tool for Materials Discovery. Accounts of Chemical Research. O'Reilly, Turberfield, and W...
8 downloads 4 Views 1MB Size
Article Cite This: Acc. Chem. Res. 2017, 50, 2449-2456

pubs.acs.org/accounts

Cell-Surface Interactions on Arginine-Rich Cell-Penetrating Peptides Allow for Multiplex Modes of Internalization Published as part of the Accounts of Chemical Research special issue “Chemical Biology of Peptides”. Shiroh Futaki*,† and Ikuhiko Nakase‡ †

Institute for Chemical Research, Kyoto University, Uji, Kyoto 611-0011, Japan Nanoscience and Nanotechnology Research Center, Research Organization for the 21st Century, Osaka Prefecture University, Naka-ku, Sakai, Osaka 599-8570, Japan



CONSPECTUS: One of the recent hot topics in peptide-related chemical biology research is the potential of cell-penetrating peptides (CPPs). Owing to their ability to deliver exogenous molecules into cells easily and effectively, their flexible design that allows transporters to comprise various chemical structures and functions, and their potential in chemical and cell biology studies and clinical applications, CPPs have been attracting enormous interest among researchers in related fields. Consequently, publications on CPPs have increased significantly. Although there are many types of CPPs with different physicochemical properties and applications, arginine-rich CPPs, which include the human immunodeficiency virus type 1 (HIV-1) TAT peptide and oligoarginines, are among the most extensively employed and studied. Previous studies demonstrated the importance of the guanidino group in arginine, which confers flexibility in transporter design. Therefore, in addition to peptides, various transporters rich in guanidino groups, which do not necessarily share specific chemical and three-dimensional structures, have been developed. Typically, cell-penetrating transporters have 6−12 guanidino groups. Since the pKa of the guanidino group in arginine is approximately 12.5, these molecules are highly basic and hydrophilic. Our group is interested in why these cationic molecules can penetrate cells. Understanding their mechanism of action should lead to the rational design of intracellular delivery systems that have high efficacy. Additionally, novel cellular uptake mechanisms may be elucidated during the course of these studies. Therefore, our group is trying to understand the basic aspects underlying the ability of these peptides to penetrate cells. Regarding the delivery of biopharmaceuticals including proteins and nucleic acids, achieving efficient and effective delivery to target organs and cells is one of the biggest challenges. Furthermore, when the target sites of these drug molecules are within cells, effective cell penetration becomes another obstacle. Cells are surrounded by a membrane that separates the inside of the cell from its outside. This barrier function is critical for keeping cellular contents inside cells, and without this, cells cannot function. Therefore, understanding the mechanism of action of CPPs is necessary to overcome these obstacles and will allow us not only to improve CPP-mediated delivery but also to create other types of intracellular delivery systems. In this Account, we summarize the current knowledge on the mechanisms of internalization of arginine-rich CPPs, from the viewpoints of both direct cell-membrane penetration (i.e., physicochemical aspects) and endocytic uptake (i.e., physiological aspects), and discuss the implications of this knowledge. We also discussed loosening of lipid packing as a factor to promote direct cell-membrane penetration.



INTRODUCTION: HIV-1 Tat PEPTIDE (TAT) AND ITS CELL-PENETRATION ABILITY

conjugation of the TAT with bioactive peptides was used successfully to deliver bioactive peptides with cell cycle modification and apoptosis activities into cells.4,10 The possibility of an in vivo application of this approach was demonstrated using a β-galactosidase (120 kDa) fusion protein.10 We initially thought that the TAT sequence could have special characteristics, including the use of a specific receptor, that allow for membrane translocation. Therefore, we synthesized TAT analogs, namely, R9-TAT and D-TAT (Table

The current flux of cell-penetrating peptide (CPP) research (see refs 1−5 for examples) was derived from the serendipitous finding of the membrane translocation ability of the HIV-1 Tat protein, which is a transcription regulator protein of HIV-1 and is not involved in HIV infection to host cells.6,7 Following studies have demonstrated the importance of a short segment corresponding to the RNA binding segment (amino acids 48− 60) of the protein for membrane translocation8,9 (in this Account, this peptide segment was denoted TAT to discriminate it from the Tat protein). Tandem alignment or © 2017 American Chemical Society

Received: May 2, 2017 Published: September 14, 2017 2449

DOI: 10.1021/acs.accounts.7b00221 Acc. Chem. Res. 2017, 50, 2449−2456

Article

Accounts of Chemical Research Table 1. Membrane-Permeable Arginine-Rich Peptidesa peptide HIV-1 Tat-(48−60) D-Tat

R9-Tat HIV-1 Rev-(34−50) FHV Coat-(35−49) BMV Gag-(7−25) HTLV-II Rex-(4−16) CCMV Gag-(7−25) P22 N-(14−30) λ N-(1−22) ϕ21 N-(12−29) yeast PRP6-(129−144) human U2AF-(142−153) a

sequence

translocation efficiency

Tat and Related Peptides GRKKRRQRRRPPQ grkkrrqrrrppq GRRRRRRRRRPPQ Arginine-Rich RNA-Binding Peptides TRQARRNRRRRWRERQR RRRRNRTRRNRRRVR KMTRAQRRAAARRNRWTAR TRRQRTRRARRNR KLTRAQRRAAARKNKRNTR NAKTRRHERRRKLAIER MDAQTRRRERRAEKQAQWKAAN TAKTRYKARRAELIAERR TRRNKRNRIQEQLNRK SQMTRQARRLYV

+++ +++ +++ +++ +++ +++ +++ ++ ++ + + + −

Lowercase letters represent D-amino acids.

1).11 In the former peptide, all of the amino acids in TAT, except arginine, were substituted for arginine. In the latter peptide, all of the amino acids except glycine were substituted for D-amino acids. However, no significant reduction in the internalization efficacy of these peptides was observed. Since the TAT segment corresponded to the RNA-binding domain of Tat protein, we constructed several RNA-binding segments from other origins to evaluate their membrane permeability. To our surprise, internalization of the majority of these peptides was observed, and the efficacy correlated with the number of arginine residues in the segments (Table 1). Therefore, these results show that arginine-rich peptides have ubiquitous membrane permeability.11,12 Wender and co-workers also found the role of arginine residues in membrane permeation.3,13,14 Using their strong backgrounds in organic synthesis and molecular design, they demonstrated the membrane permeability of various molecules rich in guanidino groups and concluded that it is the presence of several guanidino groups, but not peptide structure, that plays a crucial role in membrane permeability. At the early stages of CPP studies, the membrane permeation ability of TAT was thought to occur through direct cell membrane penetration without using endocytosis, since TAT internalization was observed at 4 °C (an experimental condition used to inhibit endocytosis). However, it was reported later that the fixation used for microscopic observation of TAT-treated cells yielded considerable artifacts. This was due to high extent of membrane adsorption of TAT peptide resulting in relocalization of the peptide within cells (especially into nucleus).15 Further studies using live cells demonstrated that endocytosis does play a considerable role for the internalization of TAT. In fact, TAT and other arginine-rich CPPs enter cells either by directly passing through the plasma membrane or by endocytosis, depending on the physicochemical properties of cargo molecules attached to the CPPs and the administration conditions.16−19 We describe our understanding of both pathways and discuss the strategies to utilize these in the following section.

arginine can form bidentate hydrogen bonds and electrostatic interactions with sulfate, phosphate, and carboxylate moieties (Figure 1). These functional groups can be seen on cell surface

Figure 1. Arginine can form ion-pair and hydrogen bonding interactions with phosphate, sulfate, and carboxylate.

components, including proteoglycans [membrane proteins that contain repeats of disaccharides bearing sulfate (glycosaminoglycan, GAG)], phospholipids, sialic acids, and so on.3 This interaction is thought to promote cell surface accumulation of CPPs, leading to their internalization. Since these interactions are not very strong, usually 6−12 arginine residues are needed to cause significant accumulation on cell surfaces and subsequent internalization. Rothbard et al. synthesized octamers of mono- and dimethylated arginines and used them in experiments that elegantly demonstrated the importance of hydrogen bonding.20 Methylation should increase the basicity of the guanidino function; however, the cellular uptake of methylated arginine octamers was lower than the uptake of D-octaarginine (r8) with intact guanidine function. The same group also demonstrated that membrane potential was important for cellular uptake.20 The voltage is usually lower inside a cell than outside it. It is reasonable from an energetic point of view that positively charged arginine-rich CPPs are driven by the membrane potential to the inside of the cell, where the voltage is lower. Additionally, the translocation of positively charged argininerich CPPs through hydrophobic membrane cores should be energetically unfavorable. Guanidino groups in these molecules



DIRECT MEMBRANE PENETRATION: PHYSICOCHEMICAL ASPECTS As described above, the importance of guanidino groups to CPP function has been advocated. The guanidino group on 2450

DOI: 10.1021/acs.accounts.7b00221 Acc. Chem. Res. 2017, 50, 2449−2456

Article

Accounts of Chemical Research

Figure 2. (A) Although the guanidino group of R8 is cationic and hydrophilic, other parts of the molecule are poorly hydrophilic. (B) Curved membranes should have more lipid packing defects than planar membranes, facilitating the interaction of R8 backbone with lipid core areas and the eventual translocation.

segment led to comparable or higher cellular uptake than with R8.27 We also found that the total hydrophobicity of the peptide conjugates was an important determinant of the internalization efficacy and of obtaining the desired bioactivity of cargo molecules. It has also been suggested that the ability of various CPPs to induce membrane curvature mediates membrane translocation.28−30 It has been proposed that the HIV-1 TAT peptide forms transient pores in membranes by inducing saddle-splay curvature, where the potential contribution of bidentate hydrogen bonding to curvature formation plays an important role.28 Thus, the coadministration of curvature-inducing peptides may also promote the translocation of arginine-rich CPPs. Therefore, we studied the effect of a curvature-inducing peptide on R8 membrane translocation.31 Epsin-1 is an accessory protein that is involved in clathrin-coated pit formation and is a representative curvature-inducing protein.32 We found that the N-terminal 18-residue peptide of epsin-1 (EpN18) had curvature-inducing ability.31 The addition of EpN18 elicited direct membrane penetration of R8, as expected; however, no significant direct penetration of R8 in the absence of EpN18 was observed under the applied conditions. Therefore, a practical approach to accelerate direct membrane penetration of arginine-rich CPPs may involve the use of other peptides or agents to induce membrane curvature or morphological alterations. Very recently, we proposed loosening of lipid packing as a key factor that governs the membrane translocation of CPPs.33 Arginine-rich CPPs have been considered, in general, to be highly cationic and hydrophilic; the guanidino groups of arginine-rich CPPs form hydrogen bonds and ion-pairs with lipid head groups, assisting in membrane binding. Upon membrane translocation of CPPs, the peptide backbone has to cross the lipid core area. The methylene groups in arginine side chains and the peptide backbone are poorly hydrophilic. We thus considered the possible involvement of hydrophobic interactions between the less hydrophilic peptide backbones and the lipid core, which may lead to eventual translocation (Figure 2). Using a polarity-sensitive di-4-ANEPPDHQ dye, lipid packing on live cell membranes was analyzed. There was significant loosening of lipid packing in conditions that stimulate direct translocation conditions [e.g., cholesterol depletion (treatment with methyl-β-cyclodextrin), counter-

form hydrogen bonding and ion-pair interactions with water molecules and solutes, and these interactions are difficult to form when CPPs enter the membrane. The presence of counteranions in membranes may compensate for these energy losses during translocation, by providing alternative sources of ion-pairs and hydrogen bonds. Sakai and Matile showed that oligoarginines could be partitioned into the chloroform phase from the aqueous phase in the presence of phosphatidylglycerol.21 Rothbard et al. also reported that r8 could be extracted into the octanol phase in the presence of sodium laurate.20 These results suggested the possibility that the translocation of oligoarginines can be facilitated in the presence of hydrophobic counteranions in the membranes. Based on this hypothesis, we collaborated with Sakai and Matile to screen several hydrophobic counteranions and found that the membrane translocation of octaarginine (R8), a frequently employed oligoarginine CPP, was considerably increased in the presence of pyrenebutyrate (PyB).22,23 A significant influx of oligoarginines into cells was observed in 1 min, and peptide localization throughout the cytosol was observed for almost all of the cells in 5 min. This method was applicable to the cytosolic delivery of small oligoarginine-tagged proteins. Internalization of an enhanced green fluorescent protein (EGFP)-tagged fusion protein into rat primary cultured neurons occurred within a few minutes in the presence of PyB. This method was also successfully used to deliver N15-labeled ubiquitin into cells to measure the in-cell nuclear magnetic resonance (NMR) of the protein.24 Internalization of R8 in the presence of PyB was also observed when cells were exposed to a temperature of 4 °C; this showed that direct penetration through cell membranes did not require endocytosis.23 The attachment of hydrophobic segments to arginine-rich CPPs may also increase interactions with the cell surface and promote direct penetration. We studied the membrane translocation ability of N-terminally acylated R8 and found that the attachment of a hexanoyl moiety significantly enhanced the internalization efficacy of R8.25 Similarly, the translocation of arginine-rich peptides was accelerated by the addition of a short hydrophobic peptide segment [penetration accelerating sequence, Pas (FFLIPKG)].26 Significant cytosolic localization of Pas-attached arginine-rich CPPs was observed within 5 min after the addition of the peptides to the medium. Further studies demonstrated that the attachment of an FFFFG 2451

DOI: 10.1021/acs.accounts.7b00221 Acc. Chem. Res. 2017, 50, 2449−2456

Article

Accounts of Chemical Research anion (PyB) treatment, and treatment with curvature inducing peptides (EpN18)]. Loosening of lipid packing should allow greater interaction between peptide backbones and lipid acyl chains to stimulate peptide translocation. In addition, PyB was also found to have an ability to induce membrane curvature. As described above, membrane curvature has been considered as a possible factor promoting membrane translocation of CPPs. However, this was attributed to transient pore formation on the membranes upon membrane phase alteration; no possible contribution of loosening of membrane packing as a means to facilitate greater peptide−membrane interaction and translocation was considered. The findings about the concept of loosening of lipid packing should add new insight on argininerich CPP entry and may have implications for membrane interactions with other bioactive molecules. To summarize, factors that stimulate direct cell membrane penetration of CPPs include cell surface adsorption of the peptides, lowered energy barriers allowing arginine-rich CPPs to pass through or to have greater interaction with hydrophobic membrane cores (e.g., counteranions, curvature, and loosening of lipid packing), and driving forces that stimulate translocation (e.g., membrane potential and cell surface concentration of peptides). By simultaneously employing these factors, argininerich CPPs should attain direct translocation through cell membranes.

Figure 3. Arginine-rich CPP interactions with cell-surface proteoglycans lead to actin organization, lamellipodia formation, and macropinocytic uptake.

glycosaminoglycans of proteoglycans strengthens its interaction with the receptor.41 To increase our understanding of the endocytic uptake methods of arginine-rich CPPs, it will be necessary to identify the cellular uptake receptor for oligoarginine molecules. This information will allow us to design more effective delivery systems. With that goal in mind, we used a photo-cross-linking approach42 and used diazirine as the photo-cross-linking agent. Photoradiation of diazirine yields nitrene, which rapidly crosslinks with surrounding molecules. For the CPP, a 12-mer arginine (R12) peptide was employed because it has high cellular uptake efficacy. The photo-cross-linked product that we obtained was myosin-9 (nonmuscle myosin heavy chain IIA). Since myosin-9 was reported to couple with the chemokine receptor CXCR4,43 we knocked down CXCR4 using siRNA. This caused a significant reduction in the cellular uptake of R12, which suggests that CXCR4 is a cellular uptake receptor of R12. Induction of macropinocytosis was confirmed by stimulation with R12. CXCR4 colocalized with 70-kDa dextran, a macropinosome marker, and R12, indicating CXCR4 internalization. If CXCR4 induces macropinocytosis, the intrinsic ligand stromal cell-derived factor-1α (SDF1α) may also induce macropinocytosis. We found that stimulation with SDF1α caused a significant increase in the cellular uptake of extracellular fluid via macropinocytosis. It is known that CXCR4 serves as a coreceptor for HIV. HIV-1 gp120 is a glycoprotein that is expressed on the viral envelope surface.44 This protein is involved in HIV interaction with CXCR4, which mediates virus entry into cells. We hypothesized that stimulation of CXCR4 with gp120 would also induce macropinocytosis and later confirmed that this did indeed occur. Macropinocytosis also led to the internalization of CXCR4. Considering that HIV infection is accomplished on the cell surface, the internalization of CXCR4 should have an inhibitory effect. Thus, CXCR4mediated HIV infection may be accomplished in competition with membrane fusion with CXCR4 internalization. Interestingly, although R8 and TAT peptides are CPPs that are more popularly employed and known to induce macropinocytosis,38 CXCR4 knockdown did not affect the cellular uptake of these peptides, and thus CXCR4 was not responsible for R8 and TAT uptake. It has been reported that CXCR4 is expressed on the cell surface in a complex with syndecan-4.



ENDOCYTOSIS-MEDIATED UPTAKE: PHYSIOLOGICAL ASPECTS Macropinocytosis is a typical form of endocytosis involved in internalization of arginine-rich CPPs.12 The involvement of macropinocytosis in the cellular uptake of TAT and oligoarginines was first proposed by the group of Steven Dowdy and our group.34,35 Macropinocytosis is an actin-driven fluid-phase endocytosis.36 In contrast to conventional clathrinmediated endocytosis, macropinocytosis involves peptide−cell membrane interaction-induced actin reorganization, which leads to the formation of membrane structures thrusting from the cell surface, and the subsequent fusion of these membranes to form large endosomes called macropinosomes. The diameter of macropinosomes has been reported to often exceed 1 μm, which is significantly larger than the diameter of those produced by clathrin-mediated endocytosis (∼120 nm) and caveola endocytosis (∼80 nm) and may be suitable for the cellular uptake of macromolecules and nanoparticles.36 Macropinocytosis is induced by external stimuli and in particular growth factors such as epidermal growth factor (EGF), although macropinocytosis is common in some cancer cells and macrophages.37 The interaction of arginine-rich CPPs with cell surface molecules leads to activation of Rac protein.38 This is followed by the induction of actin reorganization, which leads to the formation of veil-like membrane structures (i.e., lamellipodia) and membrane ruffling (Figure 3).38 The fusion of these ruffled membranes leads to the formation of large vesicular compartments called macropinosomes. Cell-surface proteoglycans play crucial roles in the activation of Rac protein and actin reorganization.15,38−40 In proteoglycan-deficient cells, macropinocytosis is not significantly observed.38 It is possible that proteoglycans per se play a receptor role to activate macropinocytosis. Alternatively, it is possible that proteoglycans work as co-receptors to increase the affinity to the actual receptors. This is the case in the activation of the fibroblast growth factor receptor (FGFR), where the interaction of FGF with 2452

DOI: 10.1021/acs.accounts.7b00221 Acc. Chem. Res. 2017, 50, 2449−2456

Article

Accounts of Chemical Research

Figure 4. A photoaffinity probe containing a diazobenzene linker can be cleaved by Na2S2O4 treatment.46

employed this pathway for intracellular delivery of extracellular vesicles (EVs, exosomes).50,51 EVs are 30−200 nm diameter vesicles that are involved in cell-to-cell communication.51 Increasing efforts are being made to use EVs as intracellular delivery vectors because they have pharmaceutically advantageous properties, including nonimmunogenicity. However, the repulsion of EVs with negatively charged cell membranes and their size limitations have hampered their use for drug delivery. Therefore, we modified the exosomal membrane using stearylated R8 peptides.52 Macropinocytosis induction by the R8 modification significantly enhanced the efficacy of EV uptake. Using saporin (a ribosome-inactivating protein) as a model bioactive protein, we demonstrated the effectiveness of EV-based intracellular delivery.

One possible idea for the difference may be that R12 can interact with both CXCR4 and syndecan-4 to better stabilize the complex structure, whereas the shorter R8 and TAT have less stabilizing effect. However, details should be clarified in the future. Since R8 and TAT are more popularly employed than R12, identification of endocytic receptors of these peptides may potentially lead to finding more ubiquitous receptors for CPP uptake. Therefore, we used the same approach to identify R8 receptors. Eventually, we identified a protein called lanthionine synthetase component C-like protein 1 (LanCL1),45 but this was found to be a cytosolic protein and not a membrane one. It is possible we did not find an R8 receptor due to nonspecific adsorption of cellular proteins on affinity beads. The amount of photo-cross-linked products was much lower than the number of nonspecific proteins; this made isolation of these proteins more difficult. Therefore, we designed an alternative photoaffinity probe consisting of an R 8 peptide bearing a diazobenzene moiety (Figure 4).46 Diazobenzene can be cleaved by treatment with reducing agents such as Na2S2O4, which leads to the selective detachment of photo-cross-linked products from beads also bearing nonspecific binding proteins. Approximately 20 proteins were identified as possible products cross-linking with R8. siRNA knockdown of these proteins resulted in a decrease in R8 uptake only in the case of syndecan4 heparan sulfate proteoglycan (a glycosaminoglycan-containing membrane-associated protein). Cell biological studies using endocytic inhibitors revealed that syndecan-4 serves as an endogenous cell-surface receptor for R8 uptake via clathrinmediated endocytosis. Using a fusion protein of Cre recombinase and R8 , we also demonstrated that this syndecan-4-dependent clathrin-endocytic pathway was responsible for the induction of Cre bioactivity in cells. Although the identified receptor was for clathrin endocytosis and not macropinocytosis, this was the first example of an internalized CPP-tagged protein that retained its expected bioactivity inside the cell. Our group also reported the possible contribution of syndecan-4 to R8-induced macropinocytosis,47 suggesting the multiple roles of syndecan-4 for R8 uptake. More detailed studies are needed to elucidate a clearer view of the mechanisms in R8-mediated macropinocytosis. It was reported that macropinocytosis was involved in the uptake of not only CPPs but also other macromolecules and in the infection of various viruses.48,49 Macropinocytosis is now regarded as a major cellular uptake pathway that is involved in the uptake of large macromolecules and viruses. Considering that macropinocytosis is involved in the effective uptake of biomacromolecules, approaches have been developed to utilize this pathway for drug delivery. For example, our group



USE OF ADAPTABLE INTERNALIZATION METHODS OF ARGININE-RICH CPPS TO THE APPLIED CONDITIONS As stated above, the cellular uptake of arginine-rich CPPs involves direct membrane penetration and endocytosis. These two uptake pathways are not discrete and their contributions vary depending on the conditions. Therefore, mechanistic studies on the uptake of arginine-rich CPPs are both challenging and fascinating. In studies of internalization routes for conventional receptor-mediated endocytosis, knockdown of the receptor completely blocks the pathway. However, argininerich CPPs can find alternative internalization routes when one pathway is blocked. We have shown that direct membrane translocation of arginine-rich CPPs is preferentially observed when the peptide concentration exceeds a threshold, and the endocytic uptake is dominant below the concentration.17−19 It is possible that high accumulation of CPPs on the cell surface would lead to transient membrane structural perturbations, as is the case for the membrane translocation of antimicrobial peptides. Amphiphilic helical antimicrobial peptides form rather stable pores or rupture the membranes through peptide−membrane interaction. In contrast, arginine-rich CPPs are highly cationic and do not have helical structure. Therefore, membrane translocation may be accomplished via a pulse-like method without yielding stable pores in membranes that cause eventual cell toxicity. To promote aggregate formation on the membranes, a certain peptide concentration may be needed. Membrane potential also serves as a driving force for peptide permeation through cell membranes. On the other hand, the inability for stable pore formation on membranes is accompanied by a significant cargo-size-dependence in direct membrane translocation in CPP-mediated delivery. When the cargo is relatively small (