Interaction of Poly (l-arginine) with Negatively Charged DPPG

2152

Biomacromolecules 2009, 10, 2152–2161

Interaction of Poly(L-arginine) with Negatively Charged DPPG Membranes: Calorimetric and Monolayer Studies Christian Schwieger† and Alfred Blume* Martin-Luther-Universita¨t Halle-Wittenberg, Institute of Chemistry-Physical Chemistry, Mu¨hlpforte 1, 06108 Halle, Germany Received March 20, 2009; Revised Manuscript Received June 26, 2009

The interaction of poly(L-arginine) (PLA) with dipalmitoyl-phosphatidylglycerol (DPPG) bilayer membranes and monolayers was studied by differential scanning calorimetry (DSC), isothermal titration calorimetry (ITC), and monolayer experiments. The binding of PLA affected the main transition temperature of lipid bilayers (Tm) only marginally. Depending on the PLA chain length, Tm was slightly increased or decreased. This finding was attributed to the superposition of two counteracting effects on the transition after PLA binding. The main transition enthalpy (∆Hm) was decreased upon PLA binding and the formation of a ripple phase (Pβ′) was suppressed. ITC experiments showed that two distinct processes are involved in binding of PLA to gel phase (Lβ′) membranes. At low peptide content the binding reaction is endothermic, and at high peptide concentration the binding becomes exothermic. However, the enthalpy of binding to fluid (LR) membranes was exothermic for all peptide-to-lipid ratios. The temperature dependence of PLA binding to fluid palmitoyl-oleoyl-phosphatidylglycerol (POPG) membranes showed a decrease in binding enthalpy with increasing temperature (∆RCp < 0), indicating hydrophobic contributions to the free energy of binding. For longer PLA chains, the binding enthalpy for LR membranes was more exothermic than for shorter chains. Monolayer adsorption experiments showed two consecutive binding processes. At low initial surface pressures (π0) a condensation of the lipid film (∆π < 0) is first observed after PLA injection into the subphase, followed by an increase in film pressure (∆π > 0) due to insertion of peptide side chains into the monolayer. At higher π0 only an increase in film pressure can be observed due to the insertion of the side chains. ∆π increases with increasing π0. The insertion of the peptide into the monolayer is corroborated by the observed shift of π-A isotherms to higher molecular areas. All presented experiments show that the binding of PLA to DPPG membranes has not only electrostatic but also nonelectrostatic contributions.

Introduction Electrostatic interactions at lipid membrane surfaces play an important role in protein binding to membranes and have been intensively studied using a variety of techniques. In many protein agglomerations of positively charged amino acids (mainly lysines and arginines) are found that provide an electrostatic binding domain for binding to the negatively charged plasma membrane.1-3 Moreover, it was found, that many cell penetrating peptides (CPPs) or protein transduction domains (PTDs) have sequences rich in arginine and lysine.4,5 During the last years it became clear that the accumulation of arginines plays a key role in membrane translocation. Thus many studies were undertaken to determine the parameters that ensure and enhance cellular uptake of natural and synthetic arginine rich peptides.6-8 This work was initialized by the finding that HIV-TAT (the 13 amino acid transduction domain of the HI virus), which comprises six Arg and two Lys residues, enters cells with ease.9 Substituting the lysines with arginines enhanced the cellular uptake.10 Simple oligoarginines were shown to cross the cell membrane even more readily than HIV-TAT.11-13 By contrast, homopolymers of lysine, ornithine, and histidine were not internalized.11 This shows that not only the charge but some specific properties of the guanidyl group are responsible for this behavior. Moreover, it was shown that the only prerequisite for * To whom correspondence should be addressed. Tel.: 49-345-5525850. Fax: 49-345-5527157. E-mail: [email protected]. † Present address: Institut Nationale de la Recherche Agronomique (INRA), Unite´ Biopolymers, Interaction, Assemblage (BIA), BP 71627, 44316 Nantes Cedex 3, France.

cellular uptake is the guanidyl function (the charged end group of the arginine side chain). This was deduced from the fact that the variation of side chain length,12,13 backbone spacings,14 backbone chemistry,12,15 and stereochemistry11 did not inhibit (sometimes even enhanced) the cellular uptake. Despite the clear evidence for cellular uptake, the mechanism is not clearly understood. Especially, it is not clear whether the translocation involves endocytosis or whether the peptides are directly transported through the hydrophobic membrane barrier. Although the former pathway seems to be more probable for a highly charged peptide, there is some indication that also the latter is used.16-18 A key factor is probably the pronounced ability of the guanidyl group to form strong bidentate hydrogen bonds with H-bond accepting counteranions.19,16,20 It was shown that such conjugates of poly- and oligoarginines with amphiphilic anions, such as aliphatic acids, sulphates, or phosphates, can be transferred into and across hydrophobic solvents such as octanol16 or chloroform.17,21 Also, the partitioning of polyarginine (PLA) into22 and the translocation through17 negatively charged lipid model membranes was shown. Due to its ability to adapt to different environments by anion binding, PLA was titled “molecular chameleon”.23 Despite the big interest in the so-called “arginine magic”, only a few articles were published that deal with well-defined model systems to elucidate the detailed binding mechanism of arginine homopolymers to lipid membranes. The group of Tsogas examined the binding of arginine monomers, PLA, and guanidinylated dendrimers to dihexadecylphosphate (DHP) containing vesicles. They reported that arginine monomers

10.1021/bm9003207 CCC: $40.75  2009 American Chemical Society Published on Web 07/15/2009

Polyarginine-Lipid Membrane Interaction

influence the lipid phase behavior because the main transition temperature Tm was decreased and the transition enthalpy ∆H was increased. ITC experiments showed that arginine binds stronger to the gel phase than to fluid DHP membranes. They further argued that some monomers penetrate the membrane and are transported into the vesicle interior.24 From ζ-potential and fluorescence experiments they concluded that PLA is located at the vesicle interface as long as the membrane is in the gel state but that it penetrates the membrane in the liquid crystalline state.25 Guanidylated dendrimers induced fusion of the vesicles and, similar to the monomers, a reduction of Tm. Under certain conditions the dendrimers might also be translocated into the vesicle interior.26 In contrast to these findings, Goncalves et al.27 reported that nonarginine (R9) does neither insert nor translocate through a palmitoyl-oleoyl-phosphatidylglycerol/palmitoyl-oleoyl-phosphatidylcholine (POPG/POPC; 3:7) bilayer but binds with some distance to the headgroup region. However, they determined high binding constants (8.2 × 104 M-1) and binding enthalpies (-2.5 kcal/mol) from isothermal titration calorimetry (ITC) experiments and evaluated the electrostatic contribution of binding to be 33% of the total free energy. Similar calculations yielding similar values were performed by Hitz et al. on the basis of fluorescence experiments.28 They reported that PLA binding to POPG/POPC (7:3) membranes leads to an increase in bilayer rigidity followed by the release of the aqueous content of the vesicle. Furthermore, they showed that PLA binding to the vesicles follows a biphasic kinetic, with a first step being probably electrostatic adsorption and the second step due to nonelectrostatic effects. With this work we want to contribute to the understanding of the binding mechanism of PLA to negatively charged membranes. It is not the objective of our work to examine the translocation through the lipid bilayer. Rather we want to reveal the mechanism of binding, which always has to precede a possible translocation. The focus of this work is directed to the influence the peptide binding has on the membrane properties. Our model system comprises PLA of different chain lengths and dipalmitoyl-phosphatidylglycerol (DPPG) vesicles or monolayers. The use of the DPPG, which has a transition temperature of 41 °C, allowed us to study the influence of PLA binding on the membranes phase behavior which was determined in a series of DSC experiments. ITC studies revealed the binding characteristics in both gel and fluid membrane state. Finally we deduced mechanistic information from monolayer experiments. Throughout this study we were guided by the attempt to deconvolute two different binding processes, one of which being electrostatic and the other nonelectrostatic in nature. The existence of these two contributions will be evident from data presented herein.

Experimental Section Materials. Poly(L-arginines). The poly(L-arginines) (PLA) with mean degrees of polymerization (DP; determined by viscosity measurements) of 19, 184, 906, and 1184 with chloride as counterion were purchased from Sigma-Aldrich (Steinheim, Germany) and used without further purification. Even though throughout this paper the different PLA samples are abbreviated by the mean DP, the samples are polydisperse. PLA concentrations are given in mol of arginine monomer/L. Lipids. 1,2-Dipalmitoyl-sn-glycero-3-phosphoglycerol (DPPG) and 1-palmitoyl-2-oleoyl- sn-glycero-3-phosphoglycerol (POPG) were obtained from Lipoid GmbH (Ludwigshafen, Germany) and used without further purification. Vesicle Preparation. For ITC experiments, lipids were dispersed in aqueous solution containing 100 mM NaCl by cyclic heating over the

Biomacromolecules, Vol. 10, No. 8, 2009

2153

phase transition temperature and repeated vortexing. Vesicles were then prepared by repeated extrusion (15 times) through a 100 nm polycarbonate membrane using a LiposoFast-Extruder (Avestin) at 10 °C above Tm (ca. 50 °C for DPPG membranes). Vesicle size was determined by dynamic light scattering (DLS) using an ALV-NIBS/HPPS DLS instrument (ALV-Laser Vertriebsgesellschaft m.b.H., Langen, Germany). For differential scanning calorimetry (DSC) and IR spectroscopic measurements lipids were dispersed in 100 mM NaCl solution and sonicated at temperatures above phase transition temperatures. For negatively charged vesicles, this leads to vesicles with a diameter of about 100 nm or less, as determined by DLS or visualized by electron microscopy. In this size range, the majority of the vesicles are by experience mostly unilamellar even though a fraction of multilamellar vesicles might be present. Vesicle preparations were stored in the refrigerator at about 4 °C. Methods. DSC. DSC experiments were performed with a Microcal VP-DSC (MicroCal Inc., Northhampton, USA). In all experiments we used a heating rate of 1 °C/min and a time resolution of 4 s. Lipid and peptide samples were prepared separately and mixed directly before the measurements. Lipid concentrations ranged from 1 to 4 mmol/L. The peptide was prepared in a concentration to give the desired lipid to peptide ratio Rc after adding it in a 1/1 (v/v) mixing ratio to the lipid suspension. The mixing order was always peptide to lipid and never vice versa. The reference cell was always filled with a 100 mM NaCl solution. At least three up- and down scans were performed for each sample to check for reproducibility. If not stated otherwise the presented curves originate from the second heating scan. The temperature cycling through the phase transition assured that the peptide was distributed homogeneously because only one peak was observed for the second and all further temperature scans. All presented curves are baseline corrected subtracting the buffer-buffer baseline. ITC. ITC measurements were performed with a MicroCal VP-ITC (Microcal, Inc., Southhampton, MA). The sample cell (1.447 mL) was loaded with the 2 mM vesicle suspension. The injection syringe (300 µL) was filled with 20 mM PLA solution. Both samples were degassed 10 min before the experiment. PLA was injected in steps of 10 µL into the sample cell that was stirred by the rotating injection syringe with 320 rpm. The equilibration time after each injection was set to 900 s to allow the cell feedback system to return to the baseline. The reference power offset was 20 µcal/s. Heats of dilution were evaluated by injecting PLA into NaCl solution. Data were evaluated with the ITC module for ORIGIN software, which is supplied by MicroCal, Inc. Monolayer Adsorption Experiments. The adsorption experiments of PLA on lipid monolayers were performed in a homemade PTFE trough, which holds a volume of 11.15 mL and has a surface area of 7.1 cm2 The trough and the plastic hood that covered the whole experimental setup were thermostatted at 20 °C with a circulating water bath. The surface pressure was recorded with a Wilhelmy plate (Riegler and Kirstein GmbH, Wiesbaden, Germany). As subphase a 100 mM NaCl solution in ultrapure water was used (pH ca. 6). The subphase was stirred during the experiment using a small magnetic stirring bar to accelerate diffusion of added solutes. DPPG was dissolved in a mixture of CHCl3 and methanol. Appropriate amounts of this solution to reach the desired initial surface pressure were spread on the water surface. The solvent was allowed to evaporate and the lipid film to equilibrate. After a constant surface pressure was reached for a period of at least 30 min, 10 µL of a 15 mM PLA solution was injected into the subphase, yielding a bulk concentration of 13.45 µM. For injection, a feed-through from the edge of the trough to the subphase interior was used. Monolayer Pressure-Area Isotherms. Surface pressure vs area isotherms were recorded using a film balance equipped with moveable barriers and Wilhelmy plate (Riegler and Kirstein GmbH, Wiesbaden, Germany). A circulating water bath was used to maintain the trough temperature at 20 °C. As subphase either a 100 mN NaCl solution or

2154


Figure 1. DSC plots of the gel-to-liquid-crystalline phase transition of DPPG/PLL complexes with an equimolar charge ratio (Rc ) 1) and different PLA chain length. The vertical dotted line is added to indicate the phase transition temperature of pure DPPG vesicles.

100 mM NaCl containing 50 µM PLA of different chain length was used. The solutions were prepared in ultrapure water with no pH correction (pH ca. 6). DPPG was spread from a CHCl3/methanol (2:1, vol/vol) solution using a microsyringe with a precision of 0.33 µL. After evaporation of the solvent the film was compressed with a rate of 2 Å2 molecule-1 min-1.

Results and Discussion DSC. In the course of this study we performed a series of DSC experiments to study the influence of PLA binding on the phase behavior of a DPPG membrane. Thereby we varied (i) the peptide chain length, (ii) the lipid-to peptide charge ratio Rc (always based on the concentration of arginine monomers), and (iii) the membrane composition. DSC curves of pure DPPG and its complexes with PLA of different chain lengths at a mixing ratio of Rc ) 1 are presented in Figure 1. The two apparent pKaapp values are 2.9 and 12.5 for DPPG and PLA, respectively, that is, DPPG is fully deprotonated and PLA is fully protonated at neutral pH. Therefore, at a ratio of Rc ) 1, the lipid and peptide charges are formally compensated. The average PLA chain length was varied in four steps from 69 to 1183 monomer units. The main endothermic phase transition of the pure DPPG membrane (lowest curve) is the gel-to-fluid phase transition with the transition temperature Tm. The phase transition temperature of about 40.8 °C and enthalpy of 10.7 kcal/mol agrees well with the values reported in literature.29-34 By addition of PLA, this transition seems to be only marginally affected. The two shorter peptides, PLA 69 and PLA 184 decrease Tm slightly by ∆Tm ) -0.8 and -0.2 °C, respectively. The two longer PLA, namely PLA 649 and PLA 1183 increase Tm by ∆Tm ) 0.2 and 0.1 °C, respectively. One would expect a much larger shift to higher temperature upon electrostatic binding of a positively charged polyelectrolyte to the negatively charged DPPG bilayer. Shielding the electrostatic repulsion between neighboring lipid molecules is reducing their lateral distance and results in a gain in van der Waals energy, stabilizing the gel phase and shifting the transition to higher temperature. The shift attributed to electrostatic shielding (∆Tel) amounts to 5.5 °C.35 This behavior could also be shown for poly(L-lysine) (PLL) binding to DPPG membranes.36 If ∆T exceeds 5.5 °C,

Schwieger and Blum

additional effects due to specific binding (including hydrogen bonding), structural changes, and changes in hydration can be assumed. If ∆T is less than 5.5 °C or becomes negative, then a perturbation of the chain packing due to additional hydrophobic interactions of the ligand with the hydrophobic membrane interior can be deduced. This is definitely the case for PLA binding to DPPG bilayers. PLA binding produces electrostatic screening of the headgroup charges, but this effect is counteracted by opposing effects that shift the phase transition to lower temperatures. These different effects are balanced in a way that the overall shift of Tm is close to zero. This notion is supported by the fact that positive as well as negative shifts are observed for PLA of different chain lengths. It can be assumed that the principle effects of binding are the same for all PLA molecules but that these effects are balanced slightly differently for PLA of different chain length, resulting in a slightly positive ∆Tm in the one case and in a slightly negative ∆Tm in the other. In general, such chain length dependencies are expected for polyelectrolytes binding to charged surfaces. This arises from the fact that the main driving force is the entropy gain achieved by counterion release.37,38 This effect is larger the more counterions are released upon the binding of one polymer chain. Moreover, hydration water is released during the binding process which is amplifying the described effect.39,40 Consequently, the binding constant, which has an entropic and an enthalpic contribution, is increasing for longer polyelectrolyte chains. This increase in binding constant is counteracted by steric and kinetic effects.41,42 The so-called pretransition from the gel phase Lβ′ to the ripple phase Pβ′ at 33.8 °C is abolished upon addition of PLA. The ripple phase is an undulated bilayer phase that is existent between the two lamellar phases Lβ′ and LR and is attributed to partial melting of the acyl chains.43 It is very sensitive to changes in the bilayer environment and structure. A suppression of the pretransition was reported for different peptides and anesthetics43 as well as for high salt concentrations44 or small unilamellar vesicles with high curvature (SUV). In the present case, screening of the surface charges, changes in hydration structure and a pronounced stabilization of the lamellar bilayer geometry might be responsible for the disappearance of this transition. The enthalpy of the main transitions (∆Hmain) is affected as well and shifted to slightly lower values. For all PLA/DPPG complexes ∆Hmain is about -1.5 kcal/mol lower than the main transition enthalpy of the pure DPPG membrane. With respect to the overall gel to fluid transitional enthalpy of a free membrane, which includes the pretransition enthalpy (∆Hgelffluid ) ∆Hpre + ∆Hmain), the decrease is even higher and results in values of about -2.5 kcal/mol. The width of the transition peaks is related to the cooperativity of the transition into the liquid-crystalline phase. The cooperative unit (c.u.) is essentially the number of molecules that change their phase state at the same time. It can be calculated on the basis of a simple two state model,45-47 yielding

c.u. )

∆HvH 4RT2Cmax ) 2 ∆Hcalor. ∆Hcalor.

(1)

with ∆HvH being the van t’Hoff enthalpy of melting, evaluated from the heat capacity maximum of the calorimetric curve, ∆Hcalor. being the calorimetrically determined transition enthalpy, and Cmax being the heat capacity at the peak maximum. From Figure 1 it is obvious that the cooperativity is decreased upon binding of the shortest PLA and is increased upon binding



2155

Figure 2. DSC plots of the gel-to-liquid crystalline phase transition of DPPG vesicles complexed with various amounts of (a) PLA 69 and (b) PLA 1181. The vertical dotted line is added to indicate the phase transition temperature of the pure DPPG vesicles.

of the longer ones. The c.u. changes from 108 to about 180 molecules. An increased cooperativity could be due to a better lateral coupling of the lipids in the bilayer, which might be facilitated when the headgroup charges are screened by the bound peptide. In Figure 2 we present the DSC curves of DPPG/PLA complexes at different lipid to peptide ratios Rc based on charges. Obviously two situations can be distinguished, which are the lipid excess (Rc > 1) and peptide excess situation (Rc < 1). The transition temperature Tm of the peptide excess complexes is always decreased with respect to Tm of the pure DPPG membrane and the DSC curves do not change any more below an Rc value of 0.5. This indicates that excess peptide in the solution remains unbound and does not influence the membrane organization any more. Slight changes in the DSC curves are seen between value of Rc ) 1 (which is charge compensating ratio) and Rc ) 0.5. Complexes formed under peptide excess are obviously overcharged, leading to a surface charge reversal. This is a common effect for polyelectrolyte binding.48 The extent of overcharging is dependent on PLA chain length, as can be deduced from the fact that the curves recorded at Rc ) 1 resemble more the lipid excess behavior for longer PLA chains and more the peptide excess behavior for shorter PLA chains. This means that the saturation limit is shifted to lower Rc values the longer the PLA chain is, that is, overcharging is more pronounced for longer PLA chains. This can be easily understood because unbound loops and ends of the polymer that extend into the solution are responsible for the overcharging effect. These loops are getting more frequent with longer polymer chain length.49 In the lipid excess regime (Rc > 1) Tm is unaffected (longer PLA) or decreases only slightly (shorter PLA). It is obvious, that here the calorimetric peaks change continuously in position, shape, peak heights and integral area dependent on the peptide content. Peaks get smaller, i.e. transitions get more cooperative with increasing peptide content, with the highest cooperativity being achieved at the saturation ratio. This can be explained by assuming that all of the lipid molecules are in the bound state. Furthermore ∆H increases with the peptide content. According to Paphahadjopoulos,50 this situation is typical for binding of charged peptides to membrane surfaces. However, for this simple case, the transition enthalpy should be higher than that of the pure membrane, which is not the case here. Even the final ∆H that is reached is lower than ∆H0, suggesting a perturbation of the bilayer by an insertion of the peptide. PLA binding can obviously not be explained with a simple model, but rather a combination of electrostatic and

hydrophobic interactions takes place. The peak symmetry is different in lipid and peptide excess complexes. In the case of lipid excess, the peak shows a shoulder on the high temperature side of the peak; in the case of peptide excess, the peak is shifted to lower temperature and is asymmetric on the low temperature side. Apparently, the hydrophobic interaction leading to a perturbation of lipid chain packing is more pronounced in the case of peptide excess. ITC. ITC was used to measure the binding enthalpy of PLA when titrated stepwise to lipid vesicles. Figure 3 shows the titration curves and the integral heats of reaction of PLA of different length to extruded DPPG vesicles at 10 °C below Tm. The injection peak profiles clearly show that the binding of PLA to DPPG vesicles involves two processes, one being endothermic and the other being exothermic. Up to a binding of approximately 0.4 equiv of PLA (Rc ) 2.5) the binding enthalpy is endothermic. Further titration to higher peptide content leads to exothermic binding of PLA, which is indicating a change in binding mechanism. The highest amount of heat is released at a peptide content of 0.5 (Rc ) 2). Further titration results in a decay of the negative reaction enthalpy to zero. The same experiments performed at a temperature 10 °C above Tm (see Figure 4) shows that the binding of PLA to fluid membranes is exothermic for all titration steps. Initially, the exothermic effects become larger after each titration step. This might be caused by an endothermic process that is more and more overcompensated by the exothermic process. But also a cooperative binding would explain this initial decrease. Cooperativity could be due to structural changes either of the membrane or the peptide during the binding process that facilitates subsequent binding. Possibly a change of the peptides secondary structure upon binding produces such an effect. The maximal heat is released at a binding ratio of about one. Above this ratio a sharp decrease of the reaction enthalpy is observed, indicating a high binding constant (105 to 107 M-1) and a 1 to 1 (charge ratio) binding stoichiometry. The binding of PLA to fluid state vesicles is clearly chain length dependent as shown in Figure 5. With increasing PLA chain length the reaction enthalpy increases and the binding constant becomes higher as judged from the shape of the titration curves. Comparison of the binding of PLA of the same chain length to gel and fluid state vesicles shows that binding to fluid state vesicles is more exothermic than to gel state vesicles. The differences in reaction enthalpies range between -0.8 kcal/mol (PLA 69) and -1.5 kcal/mol (PLA 649). As binding to the two

2156


Schwieger and Blum

Figure 3. ITC titration curves of gel phase DPPG vesicles (LUV, r ) 50 nm, 2 mM total lipid) with PLA of different chain length (c ) 20 mM) at 30 °C. Top, heating power vs time; bottom, integral heats of reaction vs inverse mixing ratio.

Figure 4. ITC titration curves of fluid phase DPPG vesicles (LUV, r ) 50 nm, 2 mM total lipid) with PLA of different chain length (c ) 20 mM) at 50 °C. Top, heating power vs time; bottom, integral heats of reaction per mol of PLA vs inverse mixing ratio.

states of the DPPG membrane and the phase transition of pure DPPG and DPPG with bound PLA can be regarded as a cyclic process these reaction enthalpies are connected according to gffl ∆tHL/P ) ∆tHLgffl + ∆BHfl - ∆BHg

(2)

gffl being the gel to fluid transition enthalpy of the with ∆tHL/P lipid/peptide complex, ∆tHLgffl being the transition enthalpy of the free lipid, and ∆BHfl and ∆BHg being the binding enthalpies of the peptide to the fluid and the gel phase, respectively. Thus, the binding enthalpies contribute to the phase transition enthalpy measured in the DSC experiments (see Figures 1 and 2). In our case, the difference between the last two terms in eq 2 is negative, because ∆BHfl is more exothermic than ∆BHg. There-

fore, the transition enthalpy of the membrane with bound gffl peptides ∆tHL/P should be lower compared to the transition enthalpy of the free membrane. This confirms partially the observed unexpected reduction of transition enthalpies reported above. The ITC experiments indicate that the binding stoichiometry of PLA to gel phase or fluid phase membranes is different. Whereas binding to fluid vesicles shows a 1:1 stoichiometry, gel state vesicles seem to be saturated at a peptide content of 0.5 (Rc ) 2) or below. This stoichiometry of two lipids per peptide monomer is expected if one assumes that binding is electrostatic, leads to charge compensation, and that only the lipids of the outer monolayer of the vesicle are accessible. A stoichiometry of 1, as found for PLA binding to fluid vesicles, would mean that all the lipids of a vesicle are



2157

Figure 5. Accumulated total heat of reaction per mol of lipid released upon PLA binding to DPPG vesicles at 30 °C (circles) and 50 °C (squares). The differences between both curves at a molar charge ratio of 1 are given in the graphs.

Figure 6. Left: Heats of reaction resulting from stepwise titration of PLA 649 (vinj ) 10 µL, c ) 20 mM) to POPG vesicles (LUVs, r ) 50 nm, c ) 2 mM) at different temperatures. Right: temperature dependence of reaction enthalpy.

accessible, that is, the vesicle is ruptured or the peptide can be translocated into the interior of the vesicle. Arginine-rich macromolecules have been found in the lumen of vesicles and cells after application to the outside.11,13,16,23 The pathway of translocation is an issue of an ongoing discussion and not yet clear. We could show by cryotransmission-electron-microscopy that fluid POPG vesicles are not completely ruptured upon PLA binding. Moreover, we performed dye release experiments that show that the vesicle content is not completely mixed with the external volume (Reuter et al., unpublished). Thus, it might be that transient pores are induced18 that allow the peptide to translocate into the vesicle and bind to the inner monolayer. The notion that more PLA interacts with lipid membranes in the fluid state than in the gel state was also reported by Tsogas et al.25 who showed by zeta potential measurements that more PLA was necessary to neutralize anionic DHP containing vesicles in the fluid phase than in the gel phase. Moreover, they showed by fluorescence quenching that the PLA content in the bulk phase was reduced to a higher extent in the presence of fluid vesicles than in the presence of gel vesicles. They argued that PLA was partially incorporated into the lipid bilayers at temperatures above Tm. Translocation pathways through the hydrophobic core of the membrane have been discussed before.51 It was shown that PLA

can be transferred through hydrophobic environments (e.g., chloroform, octane) after complexation with amphiphilic anions.16,23 Also, anionic phospholipids have been shown to alter the hydrophilicity of PLA, drastically facilitating a transport to hydrophobic media.18,21 Hitz et al. state that only 25-30% of the free energy of interaction between PLA and POPG containing vesicles is of electrostatic origin. The rest is attributed to hydrogen bonding and/or hydrophobic interactions. That hydrophobic interactions indeed play a role in PLA/PG interaction could be shown by a temperature-dependent ITC experiment (see Figure 6). The reaction enthalpies observed upon titration of fluid POPG vesicles with PLA 649 decrease with increasing temperature. The temperature dependence of ∆RH is given by

∂∆RH ) ∆RCp ∂T

(3)

where ∆RCp is the heat capacity change during the reaction. The observed negative ∆RCp is indicative for a release of water from hydrophobic surfaces.40,52 This implies that hydrophobic interactions between the reactants take place or that the hydrophobic contacts between the membrane lipids are increased. ∆RCp is also influenced by ion binding to the membrane surface with subsequent release of hydration water from the

2158


Schwieger and Blum

Figure 7. Left: Change in surface pressure π as a function of time after injection of PLA 184 under a DPPG monolayer with different starting surface pressures π0 (subphase ) 100 mM NaCl solution in H2O). PLA was injected underneath the monolayer at t ) 0. The top scale refers to the surface pressure/area isotherm of DPPG at 20 °C shown as dotted curve. Right: changes in surface pressure 1 h after injection of PLA (9) and at the point of maximal change (O) vs initial surface pressure π0.

interaction zone. However, these effects would produce a positive ∆RCp and, thus, counterbalance the effect that arises from hydrophobic dehydration. It was shown that the electrostatic binding of oligoarginine (R9) to heparinsulfate (a polyanion) produces a positive ∆RCp27 as expected. Thus, it is unlikely that the negative value of ∆RCp measured here can be attributed to headgroup interactions. ∆RCp of PLA binding to POPG vesicles was estimated to be -17.7 cal mol-1 K-1, which would formally correspond to the removal of two hydrogen atoms from contact with water,53 but since the hydrophilic contribution to ∆RCp is not known, this value remains uncertain. A similar temperature dependence of ∆RH was also observed using saturated DPPG instead of POPG (data not shown). Monolayer Experiments. To further elucidate the mechanism of PLA binding to PG membranes, we performed experiments with lipid monolayers at the air/water interface. A monolayer is a simplified membrane model, mimicking only one leaflet of a membrane. Thus, not all processes that take place in volume phases can be observed. Possible processes such as aggregation, fusion, pore formation, or translocation that might account for energetic effects cannot be examined in monolayer experiments. Only the first steps of binding are detected and can be separated from subsequent structural changes that often make volume experiments difficult to interpret. Monolayers of DPPG were spread on an aqueous subphase containing 100 mM NaCl at different surface pressures π0. After injection of PLA into the subphase, we recorded the changes of surface pressure as a function of time (Figure 7). The shape of the recorded surface pressure curves is strongly dependent on the initial pressure π0. Two general situations can be distinguished: when PLA is injected underneath a monolayer at low π0, the surface pressure decreases first and increases in a second step, reaching an equilibrium value higher than π0. If PLA is injected underneath a monolayer at high π0, the surface pressure increases, reaching a plateau about 20 min after injection (see Figure 7). At low surface pressures (π), that is, larger area per molecule (Am), the monolayer is in the liquid expanded phase (LE), which is comparable to the liquid crystalline phase (LR) of a bilayer. At higher surface pressure and smaller area per molecule the

Figure 8. Surface pressure vs molecular area isotherms of DPPG spread on different subphases: (a) 100 mM NaCl, (b) 100 mM NaCl + 0.5 mM PLA 69, (c) 100 mM NaCl + 0.5 mM PLA 184, (d) 100 mM NaCl + 0.5 mM PLA 1183. The dotted lines indicate the transition pressure of the respective monolayers. The upward pointing vertical arrows indicate the expected pressure increase in the adsorption experiments performed at constant area. The downward pointing arrows indicate the decrease of πtr upon polypeptide adsorption.

monolayer is in the so-called liquid condensed phase (LC), which is comparable to the gel phase (Lβ′) bilayers. The transition pressure (πtr) observed under the conditions chosen here is 10 mN/m (see Figure 8 or underlying π-A isotherm in Figure 7). The initial decrease in surface pressure upon PLA binding to LE monolayers indicates that the lipids get condensed and the molecular area is lowered. This is due to electrostatic adsorption to the interface and shielding of the headgroup charges. The subsequent increase in surface pressure is interpreted as an insertion of the arginine side chains into the headgroup region of the monolayer, concomitantly decreasing the available molecular area per lipid molecule. This binding mechanism is also found in the LE/LC phase transition region. When all lipids are in the condensed state (π0 > 15 mN/m), no decrease of π is detected upon PLA interaction, but after injection of PLA, an increase of π is seen. This increase is higher


the higher the initial surface pressure of the monolayer. The absence of the initial decrease of π implies that lipids, which are organized in a liquid condensed monolayer are not further condensed by PLA adsorption. The peptide side chains, though, still insert between the lipid headgroups. At the monolayer-bilayer equivalence pressure, which was found to be about 30 mN/ m,54 PLA addition increases the surface pressure, that is, the peptide inserts into the monolayer. Because at this pressure the lipid organization in the monolayer should be the same as in a bilayer at the same temperature, it can be assumed that also for gel state DPPG vesicles the side chains of PLA can penetrate into the lipid monolayer. The difference of π reached after 1 h of interaction and the initial values π0 are given in Figure 7b. Interestingly, these ∆π values are almost constant in the range of π0 < 20 mN/m, which is the range of LE monolayers and the coexistence range LE/ LC. When all lipids are in the condensed state ∆π increases with π0. This relationship is quite unexpected because commonly reported relationships for peptide or protein binding are inverse, that is, ∆π is inversely proportional to π0, which is explainable with the peptide being more readily inserted in a more loosely packed monolayer.55-59 An explanation for the increasing ∆π values with increasing π0 can be given after inspection of the π-A isotherms, which are presented in Figure 8. The isotherm of DPPG on a pure NaCl subphase shows a transition pressure πtr of 10 mN/m, with a corresponding molecular area of 80 Å2 and a collapse area of 40 Å2. These values correspond well with DPPG isotherms, reported in the literature.60,61 If DPPG is spread on PLA containing subphases, the molecular area per lipid is increased in the LE phase as well as in the LC phase, indicating an insertion of PLA into the monolayer. This confirms the interpretation of the pressure increase detected in the adsorption experiments. The ∆π values given above correspond to the pressure difference of the pure DPPG isotherm and the isotherm of DPPG on a PLA containing subphase at a constant molecular area (see up pointing arrows in Figure 8). This pressure difference increases when the compressibility of both monolayers decreases under compression (i.e., the isotherms become steeper). The effect of an increasing pressure difference is even more pronounced when the compressibility of the DPPG monolayer with inserted PLA decreases more than that of the pure DPPG monolayer. This can be indeed observed for the isotherms of LC phase monolayers (see Figure 8). This increasing difference in surface pressure between both isotherms with decreasing Am is equivalent to the increase in ∆π with increasing initial monolayer pressure π0, which was observed in the adsorption experiments. This correlation exists as long as the inserted polypeptide is not squeezed out from the monolayer at the exclusion pressure πex. In the case of PLA inserted into DPPG monolayers πex is higher than 40 mN/m. The high exclusion pressures indicate a strong interaction between the lipids and inserted PLA. The high exclusion pressures are also responsible for the unusual relation between ∆π and π0. The lower compressibility of monolayers with bound PLA as compared to pure DPPG monolayers indicates that the insertion of PLA leads to a better ordering of the lipids. Also the reduction of the transition pressure πtr shows that the interaction with PLA favors the formation of the LC phase. Figure 8 shows that the extent of πtr decrease depends on the PLA chain length. The longer the PLA chain, the lower is πtr. This correlates well with the increase in Tm with increasing PLA chain length that was detected by DSC.


2159

The π-A isotherms further show that also the magnitude of increase in average molecular area per lipid molecule dependents on the PLA chain length. In general, the increase is higher the longer the PLA chain, which implies that longer PLA chains insert to a higher extent into the monolayer than shorter ones. The monolayer experiments support the interpretation of the DSC data as well as of the ITC data given before. The increase of π after injection of PLA and the increased molecular areas in the π-A isotherms show that the peptide interacts not only superficially but inserts to a certain extent into the lipid membrane or monolayer, respectively. This finding presented here differs from the interpretations of Goncalves et al.27 These authors stated on basis of 2H NMR experiments that the interaction of R9 with POPG/POPC bilayers is only at the surface of the membrane. We presume that these differences could be due to the different degrees of polymerization of the peptides used in the experiments.

Conclusions In the work presented here we examined the binding behavior of the positively charged peptide PLA to negatively charged DPPG membranes as a function of PLA chain length, the lipid to peptide charge ratio Rc and the phase state of the membrane. DSC experiments revealed that the influence of PLA binding on the main phase transition temperature Tm of DPPG membranes is unexpectedly small. Tm is shifted to slightly lower or higher values dependent on the chain length of the binding peptide. This is attributed to the presence of different binding processes that are partially compensating each other for the overall effect. The unspecific electrostatic adsorption that stabilizes the gel phase (Lβ′) is overlaid by at least one binding process that favors the liquid crystalline LR phase. The cooperativity of the main phase transition is decreased in lipid excess complexes but increased in peptide excess complexes. This indicates that well-defined PLA/DPPG complexes are formed. To prove the existence and reveal the nature of different binding processes we performed ITC and monolayer experiments. ITC revealed indeed at least two different steps of PLA binding to gel state membranes, one of them being endothermic and the other being exothermic. Moreover, we found that the binding to fluid phase membranes is exothermic and therefore energetically favored. The binding enthalpies ∆RH and the binding constants K for PLA binding to fluid membranes are larger in comparison to the values found for binding to gel state membranes. Longer PLAs seem to bind with a slightly higher binding constant and definitely with a more negative ∆RH than the shorter ones. Because we could not determine reliable values for the binding constants, it remains unclear whether also an increase in the entropic contribution to the free energy of binding, which is normally observed for polyelectrolyte binding as a function of chain length, is present. Besides electrostatic effects also hydrophobic interactions contribute to PLA binding as shown by temperature dependent ITC measurements on fluid POPG membranes. The determined negative ∆RCp is indicative for the release of water molecules from hydrophobic surfaces. Thus, it can be assumed that hydrophobic parts of the arginine side chains get buried in the membrane during the binding process. That indeed the binding is not only at the membrane surface, but moieties of the peptide insert into a DPPG membrane could clearly be shown by monolayer adsorption experiments. The monolayer experiments provide also convincing evidence for at least two consecutive binding steps. The peptide has a

2160


condensing effect on monolayers of low initial pressure π0 (LE phase). This condensation, which is reflected by a surface pressure decrease (∆π < 0) is followed by an insertion of the peptide or parts of the peptide into the monolayer, which results in a final surface pressure increase (∆π > 0). Monolayers of high π0 (LC phase) are not condensed any more but still penetrated by the peptide. The initially surprising result that ∆π increases with increasing π0, can be explained by the larger slope of the π-A isotherm. Very instructive is the comparison of the data on PLA binding presentedherewithdatafromPLLbindingtoDPPGmembranes.36,62 PLL and PLA are very similar in structure. Both are highly charged polypeptides with positively charged side chains. The chemical structure of the backbone is identical and the length of the hydrophobic spacer of the side chain between backbone and charged group is similar. However, the chemical structures of the charged groups are different: in PLL an ammonium group is present and in PLA the cationic moiety is a guanidinium group. This structural difference causes a remarkably different binding behavior. Whereas PLL binding can be explained mainly by electrostatic effects, PLA binding shows additional features. For instance, binding of PLL increases the transition temperature of DPPG membranes as expected for an electrostatic adsorption.50 Furthermore, in contrast to PLA, PLL interacts more strongly with DPPG gel phase membranes than with fluid phase membranes (unpublished ITC data). Monolayer adsorption experiments that were performed with PLL show in principle the same characteristics as described above for PLA adsorption. But PLL shows a higher propensity for monolayer condensation and a lower propensity for monolayer expansion than PLA (unpublished data). This shows that the interaction of PLL with DPPG is governed more by electrostatic effects than the binding of PLA. We could show that the binding of PLA to negatively charged DPPG membranes shows typical electrostatic features but has also noticeable nonelectrostatic contributions. This confirms the results given by Hitz et al.28 and Goncalves et al.27 DSC experiments revealed the existence of at least two different binding mechanisms. ITC and monolayer experiments point toward a hydrophobic contribution to the binding process by partial insertion of the side chains into the bilayer or monolayer, respectively. Also, hydrogen bonding might be involved in the binding process, which was already reported to play a key role in arginine interaction with membrane lipids.16,18,20 Furthermore, secondary structure changes of the binding peptide might contribute to the free energy of binding.63 To elucidate more details of the binding mechanism, we performed FT-IR and X-ray experiments on the same system, which will be presented in a forthcoming paper. Acknowledgment. This work was supported by the Fonds der Chemischen Industrie.

References and Notes (1) Heimburg, T.; Angerstein, B.; Marsh, D. Biophys. J. 1999, 76, 2575– 2586. (2) Montal, M. FEBS Lett. 1999, 447, 129–130. (3) Wang, J.; Gambhir, A.; McLaughlin, S.; Murray, D. Biophys. J. 2004, 86, 1969–1986. (4) Magzoub, M.; Graeslund, A. Q. ReV. Biophys. 2004, 37, 147–195. (5) Futaki, S. AdV. Drug DeliVery ReV. 2005, 57, 547–558. ¨ . AdV. Drug DeliVery ReV. 2005, 57, 529. (6) Zorko, M.; Langel, U (7) Chung, S.-K.; Maiti, K. K.; Lee, W. S. Int. J. Pharm. 2008, 354, 16. (8) Pantos, A.; Tsogas, I.; Paleos, C. M. Biochem. Biophys. Acta 2008, 1778, 811. (9) Vives, E.; Brodin, P.; Lebleu, B. J. Biol. Chem. 1997, 272, 16010– 16017.

Schwieger and Blum (10) Futaki, S.; Suzuki, T.; Ohashi, W.; Yagami, T.; Tanaka, S.; Ueda, K.; Sugiura, Y. J. Biol. Chem. 2001, 276, 5836–5840. (11) Mitchell, D. J.; Kim, D. T.; Steinman, L.; Fathman, C. G.; Rothbard, J. B. J. Pept. Res. 2000, 56, 318–325. (12) Wender, P. A.; Rothbard, J. B.; Jessop, T. C.; Kreider, E. L.; Wylie, B. L. J. Am. Chem. Soc. 2002, 124, 13382–13383. (13) Wender, P. A.; Mitchell, D. J.; Pattabiraman, K.; Pelkey, E. T.; Steinman, L.; Rothbard, J. B. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 13003–13008. (14) Rothbard, J. B.; Kreider, E.; Pattabiraman, K.; Pelkey, E. T.; VanDeusen, C. L.; Wright, L.; Wylie, B. L.; Wender, P. A. CellPenetrating Pept. 2002, 141–160. (15) Umezawa, N.; Gelman, M. A.; Haigis, M. C.; Raines, R. T.; Gellman, S. H. J. Am. Chem. Soc. 2002, 124, 368–369. (16) Rothbard, J. B.; Jessop, T. C.; Lewis, R. S.; Murray, B. A.; Wender, P. A. J. Am. Chem. Soc. 2004, 126, 9506–9507. (17) Sakai, N.; Takeuchi, T.; Futaki, S.; Matile, S. ChemBioChem. 2005, 6, 114–122. (18) Tang, M.; Waring, A. J.; Hong, M. J. Am. Chem. Soc. 2007, 129, 11438–11446. (19) Onda, M.; Yoshihara, K.; Koyano, H.; Ariga, K.; Kunitake, T. J. Am. Chem. Soc. 1996, 118, 8524–8530. (20) Rothbard, J. B.; Jessop, T. C.; Wender, P. A. AdV. Drug DeliVery ReV. 2005, 57, 495–504. (21) Sakai, N.; Matile, S. J. Am. Chem. Soc. 2003, 125, 14348–14356. (22) Thoren, P. E. G.; Persson, D.; Esbjoerner, E. K.; Goksoer, M.; Lincoln, P.; Norden, B. Biochemistry 2004, 43, 3471–3489. (23) Sakai, N.; Futaki, S.; Matile, S. Soft Matter 2006, 2, 636–641. (24) Tsogas, I.; Tsiourvas, D.; Paleos, C. M.; Giatrellis, S.; Nounesis, G. Chem. Phys. Lipids 2005, 134, 59–68. (25) Tsogas, I.; Tsiourvas, D.; Nounesis, G.; Paleos, C. M. Langmuir 2005, 21, 5997–6001. (26) Tsogas, I.; Tsiourvas, D.; Nounesis, G.; Paleos, C. M. Langmuir 2006, 22, 11322–11328. (27) Goncalves, E.; Kitas, E.; Seelig, J. Biochemistry 2005, 44, 2692–2702. (28) Hitz, T.; Iten, R.; Gardiner, J.; Namoto, K.; Walde, P.; Seebach, D. Biochemistry 2006, 45, 5817–5829. (29) Zhang, Y.-P.; Lewis, R. N. A. H.; McElhaney, R. N. Biophys. J. 1997, 72, 779–793. (30) Huang, C. h.; Li, S. Biochim. Biophys. Acta 1999, 1422, 273–307. (31) Durvasula, R. V.; Huang, C.-H. Biochim. Biophys. Acta 1999, 1417, 111–121. (32) van Dijck, P. W.; de Kruijff, B.; Verkleij, A. J.; van Deenen, L. L.; de Gier, J. Biochem. Biophys. Acta 1978, 512, 84–96. (33) Jacobson, K.; Papahadjopoulos, D. Biochemistry 1975, 14, 152–161. (34) Findlay, E. J.; Barton, P. G. Biochemistry 1978, 17, 2400–2405. (35) Cevc, G.; Watts, A.; Marsh, D. FEBS Lett. 1980, 120, 267–270. (36) Schwieger, C.; Blume, A. Eur. Biophys. J., 2007, 36, 437–450. (37) Montich, G.; Scarlata, S.; McLaughlin, S.; Lehrmann, R.; Seelig, J. Biochim. Biophys. Acta 1993, 1146, 17–24. (38) Wagner, K.; Harries, D.; May, S.; Kahl, V.; Raedler, J. O.; Ben-Shaul, A. Langmuir 2000, 16, 303–306. (39) Lehrmann, R.; Seelig, J. Biochim. Biophys. Acta 1994, 1189, 89–95. (40) Garidel, P.; Blume, A. Langmuir 1999, 15, 5526–5534. (41) Novellani, M.; Santini, R.; Tadrist, L. Eur. Phys. J. B 2000, 13, 571– 578. (42) Ulrich, S.; Seijo, M.; Stoll, S. Curr. Opin. Colloid Interface Sci. 2006, 11, 268–272. (43) Heimburg, T. Biophys. J. 2000, 78, 1154–1165. (44) Blume, A.; Garidel, P. Handb. Therm. Anal. Calorim. 1999, 4, 109– 173. (45) Blume, A. Thermochim. Acta 1991, 193, 299–347. (46) Ackerman, T.; Neumann, E. Biopolymers 1967, 5, 649. (47) Ackermann, T.; Ruterjans, H., Z. Phys. Chem. (Frankfurt) 1964, 41, 116. (48) May, S.; Harries, D.; Ben-Shaul, A. Biophys. J. 2000, 79, 1747–1760. (49) Chodanowski, P.; Stoll, S. Macromolecules 2001, 34, 2320–2328. (50) Papahadjopoulos, D.; Moscarello, M.; Eylar, E. H.; Isac, T. Biochim. Biophys. Acta 1975, 401, 317–335. (51) Fuchs, S. M.; Raines, R. T. Cell. Mol. Life Sci. 2006, 63, 1819–1822. (52) Blume, A. Biochemistry 1983, 22, 5436–5442. (53) Gill, S. J.; Wadso, I. Proc. Natl. Acad. Sci. U.S.A. 1976, 73, 2955– 2958. (54) Blume, A. Biochim. Biophys. Acta 1979, 557, 32–44. (55) Maget-Dana, R. Biochim. Biophys. Acta 1999, 1462, 109–140. (56) Kimelberg, H. K.; Papahadjopoulos, D. Biochim. Biophys. Acta 1971, 233, 805–809. (57) Bringezu, F.; Majerowicz, M.; Maltseva, E.; Wen, S.; Brezesinski, G.; Waring, A. J. ChemBioChem 2007, 8, 1038–1047.

Polyarginine-Lipid Membrane Interaction (58) Demel, R. A.; London, Y.; Van Kessel, W. S. M. G.; Vossenberg, F. G. A.; Van Deenen, L. L. M. Biochim. Biophys. Acta 1973, 311, 507–519. (59) Dyck, M.; Loesche, M. J. Phys. Chem. B 2006, 110, 22143–22151. (60) Grigoriev, D.; Krustev, R.; Miller, R.; Pison, U. J. Phys. Chem. B 1999, 103, 1013–1018. (61) Sacre, M. M.; Tocanne, J. F. Chem. Phys. Lipids 1977, 18, 334–354.


2161

(62) Foerster, G.; Schwieger, C.; Faber, F.; Weber, T.; Blume, A. Eur. Biophys. J. 2007, 36, 425–435. (63) Wieprecht, T.; Beyermann, M.; Seelig, J. Biophys. Chem. 2002, 96, 191–201.

BM9003207

Interaction of Poly (l-arginine) with Negatively Charged DPPG

Recommend Documents