J. Temsamani, Synt:em, Parc Scientifique Georges Besse, Nimes, France. Fax: + 33 4 66 04 86 67, Tel.: + 33 4 66 04 86 66, E-mail: firstname.lastname@example.org
The pAntp peptide, corresponding to the third helix of the Antennapedia homeodomain, is internalized by a receptor-independent process into eucaryotic cells. The precise mechanism of entry remains unclear but the interaction between the phospholipids of plasma membrane and pAntp is probably involved in the translocation process. In order to define the role of peptide–lipid interaction in this mechanism and the physico-chemical properties that are necessary for an efficient cellular uptake, we have carried out an Ala-Scan mapping. The peptides were labeled with a fluorescent group (7-nitrobenz-2-oxo-1,3-diazol-4-yl-; NBD) and their cell association was measured by flow cytometry. Furthermore, we determined the fraction of internalized peptide by using a dithionite treatment. Comparison between cell association and cell uptake suggests that the affinity of pAntp for the plasma membrane is required for the import process. To further investigate which are the physico-chemical requirements for phospholipid-binding of pAntp, we have determined the surface partition coefficient of peptides by titrating them with phospholipid vesicles having different compositions. In addition, we estimated by circular dichroism the conformation adopted by these peptides in a membrane-mimetic environment. We show that the phospholipid binding of pAntp depends on its helical amphipathicity, especially when the negative surface charge density of phospholipid vesicles is low. The cell uptake of pAntp, related to lipid-binding affinity, requires a minimal hydrophobicity and net charge. As pAntp does not seem to translocate through an artificial phospholipid bilayer, this might indicate that it could interact with other cell surface components or enters into cells by a nonelucidated biological mechanism.
The pAntp peptide, a 16 amino-acid segment corresponding to the third helix of the homeodomain of Antennapedia protein is known to be internalized by different cell lines [1–4]. Following conjugation, pAntp has been used as a vector for intracellular delivery of molecules such as peptides [5–8] and oligonucleotides [9,10]. The pAntp peptide is taken up into the cytoplasm and the nucleus of cells both at 37 °C and 4 °C indicating that this internalization does not occur by endocytosis . In addition, the reverso, enantio, and retro-inverso forms of pAntp are also imported into cells demonstrating that the internalization is a receptor-independent process [11,12]. Phosphorus NMR studies show that the interaction between pAntp peptide and a dispersion of brain phospholipid induces the lamellar to inverted hexagonal phase transition . It was therefore proposed that the pAntp peptide interacts with phospholipids and translocates through the bilayer by transient formation of inverted micelles in the membrane. This model implies that the peptide remains in an aqueous environment and may explain why hydrophilic compounds linked to the peptide are transported into cells . It has been demonstrated that different amphipathic peptides can translocate through a model membrane either by transient formation of pores in the membrane [15–17] or in potential-dependent manner [18,19]. However, the mechanism by which the peptide translocates across the bilayer remains to be elucidated.
The pAntp peptide adopts an α helical structure in SDS micelles but this structure does not seem important for (pAntp) cellular uptake as analogues in which the helical structure is lost can still translocate into cells . It was previously reported that tryptophanes or the presence of a net positive charge are important for peptide import into cells [1,20] but no systematic studies have been carried out to define the key properties for pAntp translocation activity.
To build up a sequence-activity relationship, we have carried out an Ala-Scan mapping in which each amino-acid of pAntp was replaced by an alanine residue. These substitutions modify the parameters associated with the amphipathic α helix (hydrophobicity and hydrophobic moment). The analogues were fluorescently labeled with an NBD group at the N-terminus in order to assess the relationships between physico-chemical properties and cell association, cellular uptake, lipid-membrane binding and conformation of pAntp peptide.
Flow cytometry was used to measure cellular uptake of pAntp peptide and its analogues in nonadherent human leukemia K562 cells. To accurately measure the fraction of internalized peptide, we used a method based on the selective chemical extinction of the NBD label by dithionite, a membrane-impermeable compound. This reagent was used previously to measure transbilayer distribution of NBD-labeled phospholipids in membrane cells [21,22], in erythrocyte membranes , and to measure the translocation of an amphipathic NBD-labeled peptide through phospholipid vesicles . We have determined the binding affinity of pAntp and its analogues to phospholipid vesicles by generating binding isotherms based on the NBD fluorescence  and studied the conformation of these peptides in a membrane-mimicking environment by circular dichroism spectroscopy.
Materials and methods
Fmoc-PAL-PEG-PS Support Resin and 9-fluorenylmethyloxicarbonyl (Fmoc) amino acids were purchased from Perseptive Biosystems (Hamburg, Germany). Other reagents used for peptide synthesis or NBD labeling included N,N′-diisopropylcarbodiimide (DIPCDI, Fluka), 1-hydroxybenzotriazole (HOBT, Perkin Elmer), N,N-diisopropylethylamine (DIEA, Fluka), dimethylformamide (Perseptive Biosystems) and 4-chloro-7-nitrobenz-2-oxa-1,3-diazol-4-yl (NBD-Cl, Fluka). The lipid POPC (1-palmitoyl-2-oleyl-sn-glycero-3-phosphocholine) and POPG (1-palmitoyl-2-oleyl-sn-glycero-3-phosphoglycerol) were obtained from Avanti Polar Lipids (Alabama, USA). Sodium dithionite, SDS and poly l-lysine were supplied by Sigma (St Louis, MO, USA). Reagents for cell biology include Opti-MEM, RPMI and fetal bovine serum, obtained from Gibco (Cergy-Pontoise, France), and bovine serum albumine (BSA,Pierce).
Peptide synthesis and labeling
All peptides were synthesized according to the Fmoc-tBu strategy using an AMS 422 (ABIMED, Germany). The peptide sequences are reported in Table 1 and helical-wheel projection of pAntp peptide is shown in Fig. 1. Labeling of the N-terminus of the Ala-Scan peptides with the NBD probe was achieved as described elsewhere . Briefly, resin-bound peptide was treated with piperidin [20% (v/v) in dimethylformamide] in order to remove the N-terminal Fmoc protecting group. NBD-Cl was added in dry dimethylformamide (fivefold molar excess) in the presence of DIEA (twofold molar excess) for 6 h under agitation in the dark to selectively label the N-terminal amino group.
Table 1. Sequences of the pAntp peptide and its Ala-Scan analogues.
The resin was washed with dimethylformamide and treated with deprotecting mixture to cleave the peptides from the resin and deprotect the side-chains. Peptide purification was accomplished by RP-HPLC (Water-prep LC 40, Water) under trifluoroacetic acid 0,01%/acetonitrile gradient condition. Purity as assessed by reverse phase analytic chromatography (Beckman Gold equipped with a Diode Array detector), was over 95% for all peptides by the criterion of relative UV absorbencies at 220 nm and 460 nm. The molecular masses were validated by MALDI-TOF MS (Voyager Elite, Perseptive Biosystems, UK).
K562 cells were cultured in RPMI supplemented with 10% fetal bovine serum. Cells were diluted to 3 × 105 cells per mL one day before the experiment. About 2 × 105 cells per well were plated on 10-well glass coverslip (7 mm diameter) coated with poly l-lysine at a concentration of 0.1% (w/v in water). Cells were preincubated in Opti-MEM at 37 °C for 30 min before incubation with the peptides. Each peptide solution was prepared at a concentration of 20 µm in Opti-MEM by dilution of a stock solution of peptide at 100 µm in phosphate buffer 25 mm (pH 7.5). The cell medium was removed and the peptide solution (40 µL) was added to the cells. After 15 min at 37 °C, the cells were rinsed three times with NaCl/Pi and fixed with a formaldehyde solution (3.7%) containing 1.5% methanol for 10 min at room temperature. The cells were washed three times with NaCl/Pi before being dried and mounted in Mowiol.
The confocal microscope was a Biorad 1024 CLSM system using a Nikon Optiphot II upright microscope and a Argon-Krypton ion laser with excitation line at 488 nm (emission filter: 522/532 nm). A 60 × planachromatic objectif Nikon lens was used. Series of optical sections were collected and projected onto a single image plane in the large sharp 1024 software and processing system at 512 × 512 pixels resolution.
Flow cytometry of cell association of peptide
K562 cells were cultured in RPMI medium supplemented with 10% fetal bovine serum. Cells were diluted at 3 × 105 cells·mL−1 1 day before the experiment. Cell association was measured by flow cytometry using a FACScan (Becton Dickinson). NBD-labeled peptides (1 µm) were incubated with K562 cells (5 × 105 cells·mL−1) in Opti-MEM medium at 37 °C for various periods of time (final volume was 0.5 mL). Thereafter, the cells were washed twice and then resuspended in 0.5 mL of ice-cold NaCl/Pi for flow cytometric analysis. Cell-associated fluorophores were excited at 488 nm and fluorescence was measured at 525 nm. A histogram of fluorescence intensity per cell (1 × 104) was obtained and the calculated mean of this distribution was considered as representative of the amount of cell-associated peptide. A stock solution of dithionite (1 m) was freshly prepared in 1 m Tris solution (pH 10). Following flow cytometric analysis, 5 µL of dithionite stock solution was added to cells maintained at 4 °C and the fluorescence of internalized peptides was measured after 5 min.
For preparation of small unilamellar vesicles (SUVs), a lipid film of desired composition was dried for 3 h and then suspended in buffer [20 mm Tris, 150 mm NaCl and 1 mm EDTA (pH 7.4)] to give a final concentration between 25–30 mm. The suspension was sonicated in ice-cold water for 25 min using a titanium tip ultrasonicator. Titanium and lipid debris were removed by centrifugation at 100 000 g. Dynamic light-scattering measurements (Brookaven Instruments Ltd, USA) confirmed the existence of a single population of vesicles (more than 95% of mass content) with a mean diameter of 31 ± 1 nm. The lipid concentration was determined in duplicate by phosphorus analysis .
Fluorescence measurements were performed in 3 mL quartz cuvettes at 25° ± 0.1 °C under constant magnetic stirring using a SLM AB-2 fluorometer (SLM Instruments, Inc., Urbana, USA). For measuring the binding of NBD-pAntp and its 16 analogues to SUVs, excitation and emission wavelengths were set at 464 nm and 530 nm, respectively. Excitation and emission band-pass were both set at 4 nm. Aliquots of a concentrated stock solution of liposomes (7.5 mm) were added to the peptide solution (0.5 µm). Each measurement was performed after equilibrium was reached. The contribution of scattered light to the observed signal was negligible. The binding isotherms were analyzed as partition equilibrium employing the following formula:
where Xb is defined as the molar ratio of bound peptide per total lipid, Kp corresponds to the partition coefficient, and Cf represents the equilibrium concentration of free peptide in the solution. Knowing the fluorescence intensity as function of the total lipid concentration, the fraction of bound peptide fb can be calculated and it is then possible to calculate Cf, as well as the extent of peptide binding, Xb. In practice, it was assumed that the peptides were initially partitioned only over the outer leaflet of the SUV (60% the total lipid). Therefore, values of Xb were corrected as follows:
The curve resulting from plotting Xb* vs free peptide, Cf, is referred to as the conventional binding peptide isotherm. Surface partition coefficients Kp were calculated from the initial slopes of the conventional binding isotherm (for a detailed description, see ).
Circular dichroism measurement
CD spectra of peptides were obtained on a Jobin Yvon CD6 spectropolarimeter. The spectra were recorded from 185 to 260 nm at room temperature in a quartz optical cell of 0.02 cm path-length. Baseline spectra for SDS were obtained prior to the peptide spectra. The peptides were scanned at a concentration of 75 µm in buffer (150 mm NaF, 20 mm Tris, 0.1 mm EDTA) in the presence of SDS (12.6 mm) to obtain a SDS-to-peptide molar ratio equal to 170. At this ratio, no change in the CD spectra was observed. The helical content was determined from the mean residue ellipticity [Θ] at 222 nm according to the relation [Θ]222 = −30 300[α] − 2340 where [α] is the amount of helix .
Calculation of peptide hydrophobicity and hydrophobic moment
We have calculated for each peptide the helical hydrophobic moment <µH> for the entire sequence on the basis of a perfect helical structure following the formula of Eisenberg . We have calculated the mean residue hydrophobicity <H> for the entire sequence. For calculation of these parameters, we have used the hydrophobic scale based on the octanol/water partition of individual N-acetyl amino-acid residues . This scale correlates well with the hydrophobic scale concerning uncharged amino acid, based on the partition of a set of analogues of an amphipathic peptide from water to a POPC/POPG bilayer .
Determination of NBD-pAntp cellular uptake
We have developed a protocol to determine the fraction of NBD-labeled peptide taken up into K562 cells by flow cytometry. It is based on a method previously introduced by McIntyre and Sleight  to measure the transbilayer distribution of NBD-phospholipid analogues. The distribution is determined by comparing the fluorescence intensity before and after addition of sodium dithionite, an essentially membrane-impermeant ion, which suppresses irreversibly the fluorescence of the accessible NBD-moiety localized on the external cell surface.
We incubated NBD-pAntp with K562 cells at either 37 °C or 4 °C. After 1 h incubation, we measured the kinetics of quenching of NBD-pAntp by dithionite treatment using a spectrofluorometer. The dithionite reaction was performed at low temperature (4 °C) because it has been shown that dithionite can cross biological membranes very rapidly at room temperature . Upon stabilization of the initial fluorescence, dithionite was added to the cuvette to initiate quenching. After 5 min, the fluorescence reaches a minimum and the remaining fluorescence corresponds to 20.5 ± 1.88% of the initial fluorescence intensity at 37 °C and 5 ± 0.28% of the initial fluorescence intensity at 4 °C (Fig. 2A). The experimental curve obtained could be fitted by the double-exponential equation:
This allows a more precise estimation of the percentage of internalized peptide (A) by distinguishing between a first phase, corresponding to the rapid quenching of the peptide localized in the outer leaflet of the plasma membrane (with constant rate k1), and a second phase with a slow reaction rate (k2), resulting from passage of dithionite through the plasma membrane . The fit gives A equal to 29% (k1 = 3.86 min−1; k2 = 0.077 min−1; correlation coefficient R = 0.99) for the upper trace (incubation at 37 °C) and A equal to 13.6% (k1 = 4.41 min−1; k2 = 0.265 min−1; R = 0.985) when cells were incubated with peptide at 4 °C. This result indicates that cellular import of pAntp peptide occurs at both temperatures.
Increasing dithionite concentration did not result in a modification of the extent of quenching, indicating that all the accessible NBD label was quenched by 3 mm dithionite (data not shown). Fluorescence spectra analysis showed a blue-shift in maximum fluorescence between free (maximum fluorescence at 534 nm) and the cell-associated NBD-pAntp (maximum fluorescence at 529 nm). This reflects a relocalization of NBD-moiety in a more hydrophobic environment, suggesting an interaction of the peptide with cell membrane phospholipids. The emission maximum of cell-associated NBD-pAntp was similar prior to and after addition of dithionite as indicated by fluorescence spectra (Fig. 2B). This indicates that the decrease in fluorescence is due only to dithionite reduction and not to a diminution of the fluorescence maximum caused by a red-shift in fluorescence emission. Thus, the measured fluorescence after dithionite treatment corresponds actually to the fraction of peptide localized in the intracellular compartment.
To measure the kinetics of import of NBD-pAntp into cells, we incubated the peptide with cells at 37 °C for various times and then determined the remaining fluorescence following treatment with dithionite. The dithionite quenching trace (Fig. 3A) showed that the extent of quenching decreased as a function of time at 37 °C, indicating a protection of the NBD-label from reduction by dithionite. This protection presumably reflects pAntp internalization into cells. Figure 3B shows the fraction of peptide taken up by cells. The uptake is very rapid and seems to reach a plateau after 1 h of incubation. These results show that the use of dithionite permits efficient measurement of the fraction of internalized peptide inside the cells. The kinetics of cell association of NBD-pAntp in K562 cells, measured by flow cytometry, was similar to the kinetics of translocation determined with dithionite (data not shown).
We also examined the cellular localization of NBD-pAntp peptide in K562 cells after 15 min incubation at 37 °C, using confocal microscopy. The NBD-pAntp was localized in the nucleus and to a larger extent in the cytoplasm (Fig. 4). This confirms that the NBD-pAntp peptide is localized in cells after a short incubation time, thus corroborating with internalization kinetic determined above. During these experiments, we observed a rapid photobleaching of NBD fluorescence during the confocal analysis of the fixed cells. The rapid diminution of the initial fluorescence of the NBD-pAntp peptide localized inside the cells after incubation at 4 °C, lower than that observed at 37 °C, did not allow us to acquire adequate images.
Cell association and cell uptake of pAntp and analogues
We then measured cellular association (no addition of dithionite) and cellular uptake (after addition of dithionite) of the 16 Ala-Scan analogues (Table 1) using flow cytometry. Figure 5 shows that in general, substitution of amino-acid residues by alanine had the same influence on both cell association and cellular uptake. This may imply that the quenching extent was similar for each peptide. This observation was confirmed by the direct measurement of dithionite quenching of cell-associated analogues with a fluorimeter (data not shown).
Replacement of several hydrophobic residues (Ile3, Ile5, Trp6, Phe7, Trp14) by alanine resulted in a 3- to 10-fold decrease in uptake. The decrease in cellular uptake resulting from mutation of Met12 was less dramatic. The substitution of non charged polar residues (Gln, Asn) does not influence substantially the internalization of parent compound, except for the mutation of Gln2 which resulted in an increase of pAntp cellular uptake. The mutation of charged residues (Arg,Lys) reduces significantly the internalization ability of pAntp, implying that a relationship exists between the peptide net charge and translocation activity.
Characterization of binding isotherms
The NBD fluorescence is sensitive to the polarity of its environment, an effect whom has been used for binding studies . A blue-shift in the emission maximum from 537 toward 530 nm, and an increase in the fluorescence intensity were observed when NBD-pAntp was in a more hydrophobic environment at 25 °C (data not shown). The increase in NBD fluorescence upon binding to membrane phospholipids was used for the generation of binding isotherms for pAntp and analogues, from which partition coefficients could be calculated. It should be noted that the coefficient partition depends on the concentration of lipid accessible to peptide. We have therefore examined the topology of NBD-pAntp upon binding to SUVs using dithionite treatment. Figure 6 shows that, after incubation of the peptide (5–15 min) with lipid vesicles, the NBD fluorescence is entirely extinguished by dithionite treatment indicating that the totality of peptide remains at the outer leaflet. Using calcein efflux assay, we have checked that pAntp peptide does not allow dithionite to cross membrane by permeabilizing membrane (data not shown). In addition, identical results were obtained with the analogues of peptide indicating that mutation does not modify the topology of the parent compound (data not shown).
The NBD-labeled peptide, at a concentration of 0.5 µm, was titrated with the vesicles of defined phospholipid composition. Conventional binding curves were obtained by plotting the resulting increase in fluorescence intensity of NBD-peptide as a function of lipid-to-peptide molar ratio (Fig. 7A). The curves obtained by plotting Xb* (the molar ratio of bound peptide per 60% of the total lipid) vs Cf (the equilibrium concentration of free peptide in the solution) are referred to conventional binding isotherms (Fig. 7B). The surface partition coefficient Kp was estimated by extrapolating the initial slopes of the curves to Cf values of zero. The shape of the binding isotherm of NBD-Antp, Ap6 given as example (Fig. 7B) and analogues (data not shown) was linear indicating that peptide accumulation at the surface is a simple phenomenon without cooperative association .
Determination of partition constants
The Kp values of pAntp peptide and its analogues are determined by a titration experiment with POPC/POPG (mol/mol 95/5) SUVs. The results shown in Fig. 8A indicate that the Kp of NBD-pAntp is about 21 000 m−1. Substitution of the hydrophobic residues (Ile3, Trp6, Phe7, Trp14) by alanine decreases the binding affinity of the pAntp peptide. The corresponding surface partition coefficients of these analogues were 3- to 11-fold lower than that of pAntp. This suggests that these residues allow hydrophobic interaction between the peptide and the phospholipids and are important for anchoring the peptide into the membrane. Mutation of the charged residues (except Arg1, Arg10) did not modify drastically the lipid-binding affinity of pAntp peptide suggesting that an increase of hydrophobicity compensates the decrease of net charge. However, there is a lack of correlation between sequence-independent properties (hydrophobicity, net charge of peptide) and the measured surface partition coefficient. For example, substitution of charged residues Arg1 and Arg10 which diminishes the net charge of pAntp has opposite effect on its lipid-binding affinity. This suggests that a sequence-dependent parameter (the helical hydrophobic moment) influences the interaction between pAntp and the lipid membrane.
The binding affinity of peptides increases with the negative charge density of vesicles. The surface partition coefficients of peptides, determined with POPC/POPG (75 : 25) SUVs, were about 10-fold higher than those determined with POPC/POPG (95 : 5) SUVs (Fig. 8B). This suggests that the electrostatic interactions between positively charged peptides and negatively charged phospholipids increase the binding affinity of the peptide to the bilayer. The analogues Ap3, Ap6 and Ap7 have the lowest binding affinity to POPC/POPG (75 : 25) SUVs but the difference between this group of peptides and those with higher binding affinity are less pronounced than with bilayers of low negative surface charge. This indicates that the loss of hydrophobic interaction could be compensated by electrostatic interactions between peptides and phospholipid bilayers.
Secondary structure of analogues
The peptides CD spectra in the presence of SDS micelles are typical for an α helical conformation as reflected by a negative ellipticity at 208 and 222 nm and a maximum ellipticity at 193 nm (data not show). The extent of α helical secondary structure in pAntp and its analogues was estimated from their mean residue ellipticity at 222 nm (Fig. 9). The percentage of helical content of analogues is similar to the helical content of NBD-pAntp (34%) ranging from 28% to 38% except for several peptides (Ap10, Ap13, Ap16) which exhibit a helical content higher than 40%. In contrast, the Ap6 analogue had the lowest helical content (20%).
Relationship between physicochemical parameter, lipid-binding and cell-association
Helical hydrophobic moment and hydrophobicity are two parameters used to describe helical amphipathic peptides . Our first approach to find a relation between these descriptors and lipid binding-affinity was the clustering of the peptides into several classes (4–5 peptides per class) as a function of their respective affinities for POPC/POPG (95 : 5) SUVs. Then, we plotted for each peptide the helical hydrophobic moment <µH> against the mean hydrophobicity <H> (Fig. 10). The hydrophobic plot reveals a relationship between the Kp values of peptides and the calculated parameters. We have then carried out a multiple linear regression analysis on the complete set of peptide with Kp value as a dependent variable (17 values), and with parameters <µH> and <H> as independent variables. We obtained an equation Kp = −29.76 + 97.6 <H> + 105.36 <µH> (R2 = 0.72; Fischer Test F = 17.73) demonstrating the correlation between Kp and these both parameters. The same statistical analysis, but with Kp values obtained with POPC/POPG (75 : 25) SUVs, does not reveal first a significant correlation (R2 < 0.7) between lipid-binding affinity and the descriptors. However, when the Kp values associated with analogues Ap14, Ap15 and Ap16 were excluded from analysis, a significant correlation was found with Kp = −51.35 + 276 <H> + 573 <µH> (R2 = 0.81; F = 24.88).
Hydrophobic plot indicated that diminution of hydrophobicity and amphipathicity of pAntp peptide decreases its cellular uptake. This suggests that a threshold value of these parameters is necessary for cellular uptake of pAntp peptide. However, no correlation between the amount of internalized peptide and the physicochemical parameters <H> and <µH> was determined. In order to improve our analysis, we have clustered peptides as function of theirs net charge (6+ or 7+). Figure 11 shows internalization as function of <H> for these both clusters and indicates that a correlation exists between <H> and the internalization of peptide having the higher net charge (I = −23.5 + 571 <H> R2 = 0.78). Additionally, we found a relevant correlation existing between lipid binding affinity of these peptides for POPC/POPG (95 : 5) SUVs and theirs uptake into cell (R2 = 0.78). At opposite, no pertinent correlation was found for mutants with lower net charge. Statistical analysis including hydrophobicity and hydrophobic moment calculated using the Wimley and White scale  gave similar conclusion about the relationship between these parameters and experimental results. However, such a scale does not allow to describe efficiently the relationship between these parameters and the Kp values determined with POPC/POPG (75 : 25) SUVs.
Our approach to studying the sequence elements important for the pAntp peptide internalization was based on the use of Ala-scan mapping in which each amino acid in the sequence was successively substituted by an alanine. The mutation of the primary sequence of an α helical amphipathic peptide changes its global physico-chemical properties . Labeling peptides with the NBD group allowed the comparison of cellular uptake and phospholipid-binding of pAntp analogues and analysis of the physico-chemical properties required for translocation.
To measure cellular uptake of NBD-labeled pAntp and its analogues, we have used the dithionite method [21,22]. We report here that dithionite treatment of cell-associated NBD-labeled peptide can be used to estimate the peptide fraction localized inside the cell. First, we determined by flow cytometry that after 1 h of incubation with K562 cells at 37 °C and 4 °C, a fraction corresponding, respectively, to about 29% and 13.6% of cell-associated NBD-pAntp was localized intracellularly. The percentage of total internalized peptide was similar to the one reported in another cell line using a different method based on the derivatization of noninternalized peptide by diazotized 2-nitroalanine [3,4]. In addition, our results show that the pAntp peptide internalization in human leukemia K562 cells can occur at 37 °C and to a lesser extent at 4 °C, supporting the view that the internalization mechanism does not involve entirely endocytosis [1,12]. The kinetics of internalization of NBD-pAntp are rapid and reach a plateau after 1 h, corroborating with the results obtained in other cell lines such as lymphocyte cell  and embryonic nerve cells [1,12]. We confirmed by confocal microscopy that indeed the peptide is inside the cell.
In order to determine the key properties necessary for pAntp internalization, we measured cell association and cell uptake of each NBD-pAntp analogue by flow cytometry. The kinetics of cell association of NBD-pAntp was similar to the kinetic of internalization determined by dithionite assay. The amount of cell-associated peptide represents the sum of internalized peptide and of peptide remaining on the cell surface after cells wash. Interestingly, treatment with dithionite reveals that, at equilibrium, the fraction of cell-associated peptide protected from reduction is similar for each analogue. This could mean that the amount of internalized peptide depends on the quantity of peptide initially associated with cell surface, suggesting that the affinity of the peptide for plasma membrane is a determining factor for cell internalization.
Our Ala-scan internalization results show that hydrophobicity of pAntp plays an important role in cellular uptake. For example, substitution of aromatic residues by an alanine diminishes cell internalization. This corroborates previous results showing the importance of Trp residue for pAntp internalization . Mutation of a charged residue (Arg, Lys) within the sequence of pAntp resulted also in a significant decrease in cell uptake. This could explain why analogues of pAntp, in which the first residue (Arg1) or the last residue (Lys16) are not present within the sequence, enter poorly into cells . Our data, concerning the importance of the net charge of the pAntp peptide for its translocation ability, corroborate also those obtained recently by an Ala-Scan mapping of a biotinylated pAntp peptide . However, in these studies, it was reported that internalization of pAntp peptide does not depend significantly on its hydrophobicity . The discrepancy observed with our results could be due to a difference in experimental conditions used such as the cell type, peptide concentration and peptide labeling.
Fluorescence spectra studies suggested that the peptide interacts with the phospholipid matrix of the cellular membrane and supported the involvement of a lipid–peptide interaction during the translocation process . We therefore investigated the role of the sequence in the phospholipid binding of pAntp peptide. Liposomes are lipid bilayer models suitable for determining by fluorescence the influence of substitution on lipid-binding affinity [24,32,34].
We show that both the helical hydrophobic moment and the hydrophobicity modulate the binding of pAntp peptide to the lipid bilayer containing reduced negative surface charge. In addition, the slight diminution of hydrophobicity and helical hydrophobic moment, obtained by substitution of hydrophobic residue, results in a strong decrease of pAntp affinity towards phospholipids. This suggests that pAntp peptide has the minimal helical amphipathicity required to interact with bilayers of low negative surface charge. With an increase in vesicle negative charge, electrostatic interactions between the peptide and the membrane become predominant and compensate for the loss of hydrophobic interactions. Moreover, a statistical analysis indicates that a modification of hydrophobicity or helical hydrophobic moment in the N-terminal half of pAntp peptide could change its affinity for POPC/POPG (75 : 25) SUVs. In contrast, substitutions within the highly charged C-terminal half of the peptide do not influence lipid-binding of peptide probably because electrostatic interactions are predominant and are not influenced by the modification of the hydrophobicity or helical amphipathicity values in this segment. This corroborates a previous study demonstrating that hydrophobic interaction and global helical propensity are essential for the lipid-binding of an amphipathic peptide to a neutral surface but not to a negatively charged bilayer .
The N-terminal half of the pAntp peptide in the presence of SDS is highly ordered in an amphipathic α helix from residues Arg1 to Gln8 . Mutation of Trp6 decreases the helical content and lipid-binding, probably by perturbing the repartition of amino acids between the hydrophilic and hydrophobic sides. This suggests that the hydrophobic side of the helix containing Ile3, Trp6, and Phe7 is important for pAntp binding to the inner part of POPC/POPG membrane. In contrast, mutation of Arg10 or Lys13, which increases the helical amphipathicity, is associated with an increase of the helical content and lipid-binding affinity of pAntp. These mutations should allow the creation of a longer hydrophobic side interacting with the inner part of the phospholipid membrane. This means that the presence of several charged residues does not allow pAntp to adopt a complete amphipathic structure and disturbs its interaction with the phospholipid membrane especially when the charge-density is low.
Although the hydrophobic plot suggests that a minimal value for these both parameters could be necessary for cell uptake of pAntp peptide, no correlation exists between the helical amphipathicity and the amount of internalized peptides. It should be expected, however, that the amount of internalized peptide probably reflects the translocation capacity of the molecule as well as its affinity for cell surface. Interestingly, the modulation of hydrophobicity of pAntp peptide, without modifying its net charge, changes both its affinity for vesicles having low charge density and its uptake into cells. Thus, the interaction of pAntp peptide with cellular surface, if we assume a model in which the peptide interacts with neutral lipid matrix of cell membrane, is probably modulated by hydrophobicity of the peptide. In contrast, no correlation was found between the net charge of pAntp peptide, its lipid-binding affinity and its cellular uptake. In many cases, the substitution of one of the basic residues decreases the internalization of pAntp but increases its lipid-binding affinity. The net charge of the peptide could be important not only for the interaction of pAntp peptide with cell surface but also for others steps of the internalization process.
Taken together, these results indicate that certain physicochemical parameters modulate internalization of pAntp peptide and that a partial relationship exists between the lipid-binding affinity of this peptide and its uptake into cells. Absence of relevant relationship between the cellular uptake and the helical amphipathicity of peptide is in agreement with data obtained recently with a set of α-canonical amphipathic peptides . Indeed, the lack of correlation, between this structure-dependent property and cellular uptake corroborates previous conclusions, which indicate that the helical structure of pAntp is not required for peptide internalization [12,13].
The cellular uptake of the peptide is proportional to its cell association suggesting that the affinity of pAntp peptide for plasma membrane may be a determining factor for its translocation process. To explain its internalization, it was proposed earlier that pAntp peptide translocates into cells by inducing a lamellar to inverted micelle transition in plasma membrane [1,12]. Moreover, it was mentioned that the substitution of tryptophan within the sequence, which abolishes the translocation activity of pAntp, suppresses its membrane-destabilizing properties [13,14]. Our results suggest that the poor cellular uptake observed with several less hydrophobic analogues of pAntp could be explained also by the binding-affinity of peptide to lipid matrix of cellular membranes. The binding isotherm experiment indicates that pAntp association with a phospholipid membrane is a simple phenomenon of aggregation . In addition, we observed that NBD-pAntp peptide and analogues remain localized at the outer leaflet of the vesicle suggesting that no translocation of peptides occurs spontaneously through a lipid membrane. Calcein efflux assay indicates that no permeabilization occurs upon binding of peptide to lipid bilayer (data not shown). This suggests that pAntp, having a minimum helical amphipathicity, interacts with lipid membrane but cannot destabilize the bilayer organization as described for other positively charged amphipathic peptides . This agrees with a recent study suggesting that peptide import cannot be related to destabilization of phospholipid matrix of the plasma membrane but rather to lipid–peptide association which may play a key role in the translocation activity [3,4]. As the pAntp peptide does not seem to translocate through a lipid bilayer, this might indicate that it should interact with other cell surface components to cross cell membrane. It might also be possible that pAntp peptide enters into cells by a mechanism that involves both energy-dependent and independent process as reported for another cell-permeable peptide [3,4]. We are currently extending our studies to better understand the translocation mechanism.
This work was supported by a grant of ANRT (Association Nationale pour la Recherche et la Technologie) to GD and partially by the DGXII Research Program (contract no. BIO-CT98-0227). We acknowledge Drs J. M. Lhoste, R. Brasseur, and A. Rees for helpful discussions. We are indebted to Dr F. Roux and Mrs M. Paolini for peptide synthesis and purification, Dr J. Sainte-Marie for help with phosphorus analysis, Dr C. Braud for use of a circular dichroism spectropolarimeter, and Dr M. Jullien for use of a photon correlation spectrometer.