ARTICLE pubs.acs.org/Langmuir
Physicochemical Mechanism for the Enhanced Ability of Lipid Membrane Penetration of Polyarginine Yuki Takechi,† Haruka Yoshii,† Masafumi Tanaka,‡ Toru Kawakami,§ Saburo Aimoto,§ and Hiroyuki Saito*,† †
Institute of Health Biosciences and Graduate School of Pharmaceutical Sciences, The University of Tokushima, Tokushima 770-8505, Japan ‡ Department of Biophysical Chemistry, Kobe Pharmaceutical University, Kobe, Japan § Institute for Protein Research, Osaka University, Osaka, Japan ABSTRACT: Arginine-rich, cell-penetrating peptides (e.g., Tatpeptide, penetratin, and polyarginine) are used to carry therapeutic molecules such as oligonucleotides, DNA, peptides, and proteins across cell membranes. Two types of processes are being considered to cross the cell membranes: one is an endocytic pathway, and another is an energy-independent, nonendocytic pathway. However, the latter is still not known in detail. Here, we studied the effects of the chain length of polyarginine on its interaction with an anionic phospholipid large unilamellar vesicle (LUV) or a giant vesicle using poly-Larginine composed of 69 (PLA69), 293 (PLA293), or 554 (PLA554) arginine residues, together with octaarginine (R8). ζ-potential measurements confirmed that polyarginine binds to LUV via electrostatic interactions. Circular dichroism analysis demonstrated that the transition from the random coil to the R-helix structure upon binding to LUV occurred for PLA293 and PLA554, whereas no structural change was observed for PLA69 and R8. Fluorescence studies using membrane probes revealed that the binding of polyarginine to LUV affects the hydration and packing of the membrane interface region, in which the degree of membrane insertion is greater for the longer polyarginine. Isothermal titration calorimetry measurements demonstrated that although the binding affinity (i.e., the Gibbs free energy of binding) per arginine residue is similar among all polyarginines the contribution of enthalpy to the energetics of binding of polyarginine increases with increasing polymer chain length. In addition, confocal laser scanning microscopy showed that all polyarginines penetrate across giant vesicle membranes, and the order of the amount of membrane penetration is R8 ≈ PLA69 < PLA293 ≈ PLA554. These results suggest that the formation of R-helical structure upon lipid binding drives the insertion of polyarginine into the membrane interior, which appears to enhance the membrane penetration of polyarginine.
’ INTRODUCTION Recently, biotechnology such as proteomics and genomics has achieved many new potential therapeutic molecules including oligonucleotides, plasmids, peptides, and proteins. Generally, these compounds are difficult to use directly as a medicine because of their poor ability to permeate cell membranes. One of the most potent strategies for delivering such poorly permeating molecules into cells is to use cell-penetrating peptides (CPPs) containing high arginine sequences (e.g., Tat-peptide, penetratin, pVEC, and polyarginine). Although CPPs are promising because of their ability to carry the various cargo across cell membranes in vitro and in vivo or even in the clinical trial,1,2 the cell entry mechanism of CPPs is still not fully understood. Two types of mechanisms are being considered for crossing cell membranes: one is an endocytic pathway and another is an energyindependent, nonendocytic pathway consisting of physical membrane penetration.1 The cellular uptake mechanism of CPPs is thought to be mainly favored for endocytosis in which 98% of the delivered cargo becomes biologically inactive because it is trapped in endosome compartments.2 A first step in the internalization of CPPs is their binding to the membrane phospholipids and/or r 2011 American Chemical Society
proteoglycans such as heparan sulfate at the cell surface.35 Thus, to design peptide sequences that have a more rational cellpenetration pathway, knowledge from a physicochemical point of view of the interactions of CPPs with lipid membranes57 and proteoglycans4,8,9 is of fundamental importance. The roles of the arginine residue in CPP sequences have been studied in detail.1012 Arginine polymers enter cells more efficiently compared to lysine, ornithine, or histidine polymers with similar chain lengths.13 Also, it was reported that the penetratin analogue in which arginine residues were substituted for lysines exhibited no cellular uptake at all.14 These results suggest that the guanidine moiety attached to the side chain of arginine is a key structure in the membrane permeation of peptides. There exist three distinct steps for the translocation of argininerich CPPs through lipid bilayer membranes: positively charged CPPs (1) bind to the membrane surface, (2) translocate into the inner membrane over a potential barrier of the hydrophobic core Received: March 11, 2011 Revised: April 7, 2011 Published: April 28, 2011 7099
dx.doi.org/10.1021/la200917y | Langmuir 2011, 27, 7099–7107
Langmuir portion of lipid bilayer, and (3) pass into the inner aqueous phase of the bilayer. In these steps, it is especially unclear how charged arginine residues can cross over the hydrophobic barrier of membranes associated with the Born energy.15,16 If the radius of the guanidinium ion of arginine is taken to be 2.5 Å, then the Born energy becomes about 32 kcal/mol. This energy cost is likely to be too high for the charged CPPs to be transported into membranes. In this regard, the ability of the guanidino group to form strong bidentate hydrogen bonds with the phosphate group of the phospholipid molecule17 seems to play a role in membrane translocation. In fact, conjugates of poly- and oligoarginines with amphiphilic anions, such as aliphatic acids, sulfates, or phosphates, are preferred to transfer into hydrophobic solvents such as octanol18,19 and chloroform.20,21 Because of its ability to adapt to different environments by anion binding, polyarginine is called a “molecular chameleon”.22 Indeed, the structural flexibility of CPPs appears to be a crucial property in the interaction of peptides with membranes.23,24 Despite the large interest in such “arginine magic”,18,25 only a few articles are available that dealing with defined model systems to elucidate the detailed translocation mechanism of arginine polymer into lipid membranes. There are some reports about the polyargininemembrane interaction, but they focused only on the binding mechanism as the first step in membrane translocation.2628 Because it is reported that none of the CPPs are able to translocate across the membranes of large unilamellar vesicle (LUV) whereas they rapidly traverse the giant vesicles,6,29 it seems that a choice of the model system is crucial in evaluating the ability of polyarginine to penetrate lipid membranes. In this study, we examined the effects of the chain length of polyarginine on its interaction with LUV or giant vesicles composed of anionic phospholipids by spectroscopic and thermodynamic measurements. From the relationship of the membrane penetration ability with the structural change in polyarginine and lipid membranes, it is suggested that the secondary structural change of polyarginine from a random coil to an R-helix upon lipid binding induces the perturbation of the membrane structure, facilitating the membrane translocation of polyarginine.
’ EXPERIMENTAL PROCEDURES Materials. Poly-L-arginine (PLA) hydrochloride with an average degree of polymerization determined by viscosity measurements of 69, 293, or 554 (and a molecular weight of 13 300, 56 400, or 106 800, respectively) and soybean phospholipid (SBPL) were purchased from Sigma-Aldrich (Japan). SBPL contains phosphatidylcholine (40%), phosphatidylethanolamine (30%), phosphatidic acid (15%), phosphatidylinositol (4%), cardiolipin (5%), and others.30 Diphenyl-1,3,5-hexatriene (DPH), 1-[4-(trimethylamino]phenyl]-6-phenylhexa-1,3,5-triene (TMADPH), and dansyl-PE were purchased from Molecular Probes (Eugene, OR). 2,4-Bis-(N,N0 -di(carboxymethyl)aminomethyl)fluorescein (calcein) and 5-(and 6-)-carboxyfluorescein (FAM) succinimidyl ester were purchased from Invitrogen (Carisbad, CA). All other reagents were special grade and used without further purification. Octaarginine (R8) and FAMlabeled R8 were synthesized using Fmoc chemistry as described.31 The peptide purity was verified by analytical HPLC (>97%) and mass spectrometry. FAM Labeling Procedure. PLA was labeled with FAM according to the protocol below. A 10 mg/mL PLA solution in 10 mM sodium bicarbonate buffer (pH 9.0) was coupled with FAM overnight. The resultant FAM-PLA was separated from the free FAM by exclusion chromatography using a Sephadex G-25 column by eluting in 10 mM Tris-HCl buffer (pH 7.4).
ARTICLE
Liposome Preparation. SBPL was dissolved in chloroform in a round-bottomed flask and dried under a stream of N2 gas to produce a thin, homogeneous lipid film. For giant vesicle preparation, the dried lipid film was gradually hydrated with 10 mM Tris-HCl buffer for more than 24 h at 4 °C to be stripped off the glass surface.32 The obtained giant vesicle suspensions were centrifuged at 13 000 rpm for 10 min to remove contaminating multilamellar vesicles.33 For LUV preparation, the lipid film was vortex mixed in Tris buffer to obtain a multilamellar vesicle suspension. The resultant suspension was subjected to five cycles of freezingthawing and was then passed through a miniextruder equipped with two stacked 0.1 μm polycarbonate filters (Avanti, Alabaster, AL). In calcein leakage experiments, the film was dispersed in a 10 mM Tris-HCl buffer containing 60 mM calcein, and then the calcein-containing vesicles were separated from the free calcein by gel chromatography using a Sephadex G-50 column. The phospholipid concentration was determined by the Bartlett method.34 An average particle size of 110120 nm and a ζ potential of 25 mV for LUV were confirmed using a NICOMP 380ZLS potential/particle sizer (NICOMP, Santa Barbara, CA). Circular Dichroism (CD) Spectroscopy. Far-UV CD spectra were recorded from 190 to 250 nm at 25 °C using a J-600 CD spectropolarimeter with a 2 mm quartz cuvette. A polyarginine sample was diluted to 0.3 residual mM in 10 mM Tris-HCl buffer to obtain the CD spectrum. For the polyarginine-LUV mixture, polyarginine was incubated with LUV for 1 h prior to measurements. The spectrum was corrected by subtracting the buffer baseline or a blank sample containing an identical concentration of LUV. The R-helix content (%) of polyarginine was determined from mean residue ellipticity [θ] at 222 nm as described by Scholtz et al.35 ½θ222 ½θcoil 100 ½θhelix ½θcoil ! 2:5 þ 100t ¼ 40 000 1 n
R helix content ð%Þ ¼ ½θhelix
½θcoil ¼ 640 45t
ð1Þ
where [θ]222 is the measured mean residue ellipticity at 222 nm expressed in degrees cm2 dmol1, [θ]helix and [θ]coil are the mean residue ellipticities of the completely helical and coiled forms of the peptide (at 222 nm, expressed in degrees cm2 dmol1), respectively, n is the number of amino acid residues, and t is the temperature in °C. Fluorescence Studies. All fluorescence measurements were carried out using a Hitachi F-4500 fluorescence spectrophotometer. For steady-state fluorescence measurements, LUV was labeled with DPH or TMA-DPH by adding small aliquots of stock solution of probes in DMF to yield a phospholipid/probe molar ratio of 200:1 or 100:1, respectively. For the sample labeled with dansyl-PE, SBPL and dansyl-PE were mixed in chloroform at a phospholipid/probe molar ratio of 200:1 before the preparation of LUV. For the fluorescence anisotropy experiments, we measured fluorescence intensities (I) of I00, I090, I900, and I9090, where the numbers in the subscript of I indicate the direction of the plane of polarization of the polarizer and the analyzer. For example, I00 and I090 are the fluorescence intensities detected through a polarizer oriented parallel and perpendicular to the direction of polarization of the excitation beam. The fluorescence anisotropy, r, is given by r ¼
I0 0 GI0 90 I0 0 þ 2GI0 90
ð2Þ
where G = I900/I9090. DPH and TMA-DPH were excited at 360 nm, and the fluorescence was detected at 430 nm. Dansyl-PE was excited at 336 nm, and the fluorescence was detected at 513 nm. To evaluate water penetration into the membrane interface, the deuterium isotope exchange measurements were performed by monitoring the emission spectra of dansyl-PE in D2O buffer and comparing it to that in H2O 7100
dx.doi.org/10.1021/la200917y |Langmuir 2011, 27, 7099–7107
Langmuir
ARTICLE
Figure 2. Far-UV CD spectra of polyarginines bound to SBPL LUV. The phospholipid concentration was 0.5 mM. The concentrations of PLA and R8 were 0.1 and 0.4 residual mM, respectively.
εr, with a gradient of ∼70 at the membrane surface, 3040 at the membrane interface,36 and ∼5 at the hydrocarbon center.37,38 To examine the effect of the dielectric constant on the secondary structure of polyarginine, we measured the CD spectra of polyarginine in aqueous buffer/ethanol solutions with different polarities. εr of mixed solutions was calculated from the following equation Figure 1. (A) Far-UV CD spectra of PLA554 in aqueous buffer/ethanol mixtures. The volume fractions of ethanol were (a) 0, (b) 72, (c) 76, (d) 80, and (e) 99%, corresponding to dielectric constants of 78, 39, 37, 35, and 25, respectively. The concentration of polyarginine was 0.3 residual mM. (B) R-Helix contents of polyarginine calculated from [θ]222 were plotted as a function of the dielectric constant of the solvent. buffer from 450 to 600 nm at an excitation wavelength of 336 nm. The D2O/H2O fluorescence intensity ratio is calculated from the integrated intensity of emission spectra from 500 to 550 nm in D2O and H2O buffer. For calcein leakage measurements, the polyarginine solution was added to the calcein-entrapped LUV at 25 °C. The fluorescence emission intensity was measured at 513 nm using an excitation wavelength of 490 nm. The percent leakage of calcein was determined using the fluorescence intensity corresponding to 100% leakage obtained in the presence of 5% Triton X-100. Isothermal Titration Calorimetry (ITC) Experiments. ITC measurements were carried out on a Microcal MCS ITC calorimeter. SBPL LUV was placed in the 1.3507 mL reaction cell, and a solution of polyarginine in a 250 μL titration syringe was injected into LUV in the cell. Prior to the measurements, the peptide solution and vesicle suspension were degassed under vacuum for 10 min. The injections were performed automatically at 25 °C under 400 rpm stirring. Enthalpies of binding of polyarginine to LUV were corrected for heats of polyarginine dilution and dissociation; these values were determined by titrating polyarginine into buffer alone. Confocal Laser Scanning Microscopy. The laser scanning confocal imaging system (Zeiss, LSM-410) equipped with an argon laser was used for confocal laser scanning microscopy. The fluorescence of FAM-labeled polyarginine was excited at 488 nm, and the emission was observed through a band filter (515565 nm). Samples were prepared by mixing giant vesicles and FAM-labeled polyarginine in a glassbottomed dish (Matsunami Glass Ind., Osaka, Japan) at a ratio of PLA or R8 to lipid (arginine residue/mol of phospholipid) of 0.6 or 0.03, respectively. To prevent photobleaching, the confocal microscope was operated under conservative laser intensity and time exposure conditions.
’ RESULTS Secondary Structure of Polyarginine in Solution or Bound to LUV. It is known that lipid membranes have a dielectric constant,
εr ¼ 78:3Vwater þ 25:37Vethanol
ð3Þ
where 78.3 and 25.37 are εr values of water and ethanol at 25 °C, respectively, and Vwater and Vethanol are volume fraction of aqueous buffer and ethanol, respectively. Figure 1A shows the CD spectra of PLA554 in Tris buffer/ethanol mixtures with different polarities. In Tris buffer (εr = 78), PLA554 exhibited a typical random coil structure (negative peak at ∼190 nm and positive peak at ∼215 nm). With a decrease in the dielectric constant, the secondary structure of PLA554 changed from random coil to R-helix (negative peaks at ∼208 and ∼222 nm, positive peak at ∼190 nm). Figure 1B compares the change in the secondary structure of polyarginines in different dielectric constant environments. This indicates that the degree of transition from a random coil to an R-helix in the low-εr environment is much greater for the longer polyarginine, especially PLA554. We next examined the structural change in polyarginine upon binding to LUV. Figure 2 shows the CD spectra of polyarginine in the presence of SBPL LUV, demonstrating that the lipid binding induced a slight but significant change in secondary structure from random coil to R-helix for PLA293 and PLA554, whereas R8 and PLA69 still exhibited random coil or random coil-like structure. The calculated R-helix content (4.3 and 10.8% for PLA293 and PLA554, respectively) corresponds to 13 and 60 arginines in the R-helical structure for PLA293 and PLA554, respectively. Interestingly, from these R-helix contents of PLA293 and PLA554, the dielectric constant environment for PLA bound to LUV is estimated to be about 3537 on the basis of the dielectric analysis shown in Figure 1B. This suggests that PLA is bound at the membrane interface region. Effects of Binding of Polyarginine on the Structure and Stability of LUV. To investigate the effects of binding polyarginine on the structure of the hydrocarbon region of SBPL membranes, we measured the fluorescence anisotropy of membrane probes DPH and TMA-DPH embedded in LUV. Because DPH and TMA-DPH are located about 7.8 and 10.9 Å from the hydrophobic center of the bilayer membranes, respectively,39,40 the fluorescence anisotropy of these probes reflects the fluidity of the hydrocarbon region in membranes.39 As shown in Figure 3A, 7101
dx.doi.org/10.1021/la200917y |Langmuir 2011, 27, 7099–7107
Langmuir
ARTICLE
Figure 4. Release of calcein from SBPL LUV 5 min after the addition of polyarginine.
Figure 3. (A) Effects of PLA binding on the fluorescence anisotropy of DPH and TMA-DPH in SBPL LUV. (B) Effects of polyarginine binding on the D2O/H2O fluorescence intensity ratio and wavelength of the maximum fluorescence of dansyl-PE in SBPL LUV. The phospholipid concentration was 0.1 mM, and the concentrations of PLA and R8 were 0.1 and 0.003 residual mM, respectively.
the finding that the fluorescence anisotropy value, r, of DPH and TMA-DPH in LUV was not affected by the binding of polyarginine indicates that there is no detectable interaction between polyarginine and the hydrocarbon region of SBPL membranes in the equilibrium state. This is in contrast to the binding of PLA to pure phosphatidylglycerol vesicles in which PLA strongly affects the membrane structure with significant decreases in the gel to liquid-crystalline phase-transition enthalpy.41 We next used dansyl-PE as a membrane probe for the interface region because its dansyl group is located about 19 Å from the bilayer center.42 The wavelength of maximum fluorescence (WMF) and the deuterium isotope exchange of dansyl-PE were used to assess the degree of hydration and packing in the interface region of LUV. The dansyl fluorophor with an exchangeable hydrogen such as dansyl-PE is known to have a greater quantum yield in D2O relative to that in H2O because of the reduced rate of proton transfer. An increase in the fluorescence intensity of dansyl-PE in D2O compared to that in H2O, therefore, indicates the exposure of the probe to water, and the D2O/H2O intensity ratio reflects the hydration or lipid packing in the interface region.4345 As shown in Figure 3B, the binding of polyarginine to LUV induced the decreases in the D2O/ H2O fluorescence intensity ratio and the WMF of dansyl-PE, and this trend was more significant for longer polyarginine: the order of decrease in the D2O/H2O fluorescence intensity ratio and WMF was R8 ≈ PLA69 < PLA293 ≈ PLA554. These results suggest that polyarginine binds to the membrane interface region, with the degree of membrane insertion being greater for the longer polyarginine. We further evaluated the membrane interaction of polyarginine by measuring the release of entrapped fluorescence dye, calcein, from LUV. Figure 4 shows that the calcein release from SBPL LUV with an increasing amount of polyarginine. R8 caused
Figure 5. Isothermal titration calorimetry for PLA554 (5 residual mM) injected into SBPL LUV (10 mM). Each peak in the heat flowchart corresponds to the injection of 5 μL aliquots of PLA554 at 25 °C.
no leakage of calcein up to a 0.6 arginine residue/lipid molar ratio, whereas all PLAs induced a significant leakage of calcein from LUV. This indicates that the binding of PLA causes a structural perturbation in lipid vesicles, in sharp contrast to no such perturbation by the binding of R8.46,47 ITC Measurements for the Binding of Polyarginine to LUV. To compare the interaction of polyarginine with lipid membranes further, ITC measurements were employed. First, we measured the enthalpy of binding by injecting polyarginine into SBPL LUV at a low arginine/lipid molar ratio (
