Translocation of dynorphin neuropeptides across the plasma membrane. A putative mechanism of signal transmission.

Several peptides, including penetratin and Tat, are known to translocate across the plasma membrane. Dynorphin opioid peptides are similar to cell-penetrating peptides in a high content of basic and hydrophobic amino acid residues. We demonstrate that dynorphin A and big dynorphin, consisting of dynorphins A and B, can penetrate into neurons and non-neuronal cells using confocal fluorescence microscopy/immunolabeling. The peptide distribution was characterized by cytoplasmic labeling with minimal signal in the cell nucleus and on the plasma membrane. Translocated peptides were associated with the endoplasmic reticulum but not with the Golgi apparatus or clathrin-coated endocytotic vesicles. Rapid entry of dynorphin A into the cytoplasm of live cells was revealed by fluorescence correlation spectroscopy. The translocation potential of dynorphin A was comparable with that of transportan-10, a prototypical cell-penetrating peptide. A central big dynorphin fragment, which retains all basic amino acids, and dynorphin B did not enter the cells. The latter two peptides interacted with negatively charged phospholipid vesicles similarly to big dynorphin and dynorphin A, suggesting that interactions of these peptides with phospholipids in the plasma membrane are not impaired. Translocation was not mediated via opioid receptors. The potential of dynorphins to penetrate into cells correlates with their ability to induce non-opioid effects in animals. Translocation across the plasma membrane may represent a previously unknown mechanism by which dynorphins can signal information to the cell interior.

In cell signaling, information is conveyed across the plasma membrane to the cell interior. Specific receptors at the outer surface of the plasma membrane that recognize extracellular chemical stimuli transfer a signal across the plasma membrane to specific effector molecules at the inner surface of the membrane or within the cytoplasm and trigger the cellular response. However, not all information is transmitted through cell-surface receptors. Nitric oxide, for example, passes through the plasma membrane and induces cellular responses by interacting with intracellular proteins.
Short basic peptides referred to as cell-penetrating peptides (CPPs) 1 are able to translocate across the plasma membrane into cells (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16). Several prototypical CPPs include the human immunodeficiency virus transactivator of transcription (Tat) protein and the Antennapedia homeodomain or represent oligoarginine polymers or chimeric peptides (1)(2)(3)(4)(5)(6). In functional studies translocation of Antennapedia into neuronal cells induces augmented morphological differentiation (2), whereas penetration of Tat into HeLa cells trans-activates the stably transfected reporter gene through cis-Tat-responsive elements in the human immunodeficiency virus promoter (1). Translocation of Tat into T-cells leads to degradation of IB␣, induction of apoptosis, and secretion of interleukin (7). The mechanisms of peptide penetration are not clear. Uptake mediated through macropinocytosis or clathrin-mediated or caveolar endocytosis has been proposed (7)(8)(9)(10). Other models, suggesting translocation of the CPPs across the lipid bilayer through stable or transiently formed pores or inverted micelles, have also been hypothesized (11). The basic nature due to the high content of arginine or lysine, the amphipathic character, and/or the presence of a hydrophobic core sequences have been suggested as important structural features of CPPs (12)(13)(14). Little is known about the cellular functions of CPPs; they have generally been studied as vectors for intracellular delivery of cargo molecules with pharmacologic potential (11,15). A role for Tat and homeodomain proteins has been suggested in intercellular communication and cell signaling (16).
Neuropeptides are a group of short, 3-35 amino acid peptides that act as neurotransmitters in the central and peripheral nervous system. They are released from axon terminals or dendrites of neurons, diffuse to pre-or postsynaptic neuronal structures, and bind to and activate membrane G proteincoupled receptors. Signals induced upon peptide interaction with receptors trigger a cascade of secondary messengers. Several actions of neuropeptides such as substance P and dynorphin A (Dyn A) are not mediated through their cognate receptors (17)(18)(19)(20). The proinflammatory actions of substance P, including stimulation of release of vasoactive and inflammatory compounds from peritoneal mast cells, involve a receptorindependent mechanism (21). This mechanism is apparently based on direct interaction of substance P with the ␣ subunit of heterotrimeric G proteins, following the peptide translocation across the plasma membrane (22).
Dyn A and dynorphin B (Dyn B), prodynorphin-derived opioid peptides, are endogenous ligands for -opioid receptors that mediate many activities of dynorphins (23)(24)(25)(26)(27). Dynorphins critically regulate pain processing, motor behavior, and memory acquisition and also modulate reward induced by addictive substances (24 -27). Dyn A and big dynorphin (Big Dyn), a 32-amino acid prodynorphin-derived peptide consisting of Dyn A and Dyn B (28,29) (see Table I), are rich in arginine and lysine residues and are probably the most basic neuropeptides (30). Because hydrophobicity and/or basic nature are general properties of CPPs (Table I), we hypothesized that the basic dynorphins may also translocate across the plasma membrane, which may be relevant for interneuronal communication in the central nervous system. To test this hypothesis, we studied the ability of dynorphins to penetrate into cells using immunocytochemistry, confocal fluorescence microscopy, and fluorescence correlation spectroscopy on live and/or fixed cells. We also investigated the kinetics of translocation of the dynorphin peptides, their intracellular distribution, and interactions with phospholipid membranes relevant for the translocation into cells.

EXPERIMENTAL PROCEDURES
Peptides-Big Dyn, Dyn A, Dyn B, Big Dyn 6-26 (Table I), and biotinylated forms of these peptides with biotin attached to the Nterminal tyrosine via 6-aminohexanoic acid were synthesized at the Department of Medical Biochemistry and Microbiology, Uppsala University. Peptides were purified by reversed-phase chromatography on Vydac C18 218 TP 1022 and Sephasil C8 columns in a 0.1% trifluoroacetic acid acetonitrile/water solvent system and finally by gel filtration on a Superdex column in 1 M acetic acid. They were analyzed by analytical reversed-phase chromatography and matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS). Purity was determined by integration of the UV peak at ϭ 215 nm. The purity of all peptides was Ͼ98%.
Fluorescently labeled Dyn A (5-carboxytetramethylrhodaminedynorphin A, Rh-Dyn A, C 124 H 175 N 33 O 27 , M r ϭ 3242) and Dyn B (5carboxytetramethylrhodamine-dynorphin B, Rh-Dyn B, C 99 H 135 N 23 O 21 , M r ϭ 2552) were synthesized in the form of 5-carboxytetramethylrhodamine attached to the N terminus and purified by preparative reversed-phase HPLC on a Vydac C18 column (10 m, 300 Å, 21 ϫ 250 mm). Two solvent systems, acetonitrile/water containing 0.1% trifluoroacetic acid and methanol/water containing 0.1% trifluoroacetic acid, were used to assess the purity. The final purity of the peptide, determined in both analytical systems by integration of the UV peak at ϭ 214 nm, was Ͼ98%. The molecular weight of the fluorescent peptide was determined by electrospray ionization-mass spectrometry (ESI MS).
Transportan-10 (TP10, C 104 H 185 N 27 O 23 , M r ϭ 2182) was synthesized in a stepwise manner in a 0.1-mmol scale on a peptide synthesizer (model 431A; Applied Biosystems, Framingham, MA) using a t-butoxycarbonyl strategy of solid phase peptide synthesis. t-Butoxycarbonyl amino acids were coupled as hydroxybenzotriazole esters to a p-methylbenzylhydrylamine resin (Bachem, Bubendorf, Switzerland) to obtain C-terminally amidated peptides. The rhodamine moiety was coupled manually to the N-terminal amino group. The fluorescently labeled peptide (5-(and-6)-carboxytetramethylrhodamine-TP10, Rh-TP10, C 129 H 205 N 29 O 27 , M r ϭ 2594) was finally cleaved from the resin with liquid HF at 0°C for 60 min in the presence of p-cresol and p-thiocresol. The purity of the peptide was Ͼ98%, as demonstrated by HPLC on an analytical Discovery C-18 reversed-phase HPLC column (15 ϫ 0.46 cm, 5 m, Supelco), and the correct molecular mass was obtained by using MALDI-TOF MS (Voyager-DE STR; Applied Biosystems).
Dyn B and Big Dyn were labeled with 125 I using the direct chloramine-T method. The labeled peptides were separated from free iodine by HPLC on a Vydac C18 column using a linear gradient 15-40% of acetronitrile/water with 0.04% trifluoroacetic acid (31).
Proteins and Antibodies-Rabbit polyclonal antisera were generated against rat Dyn A and Dyn B, which had been conjugated to keyhole limpet hemocyanin via their N-terminal cysteine residues. IgG fractions were purified with protein A-Sepharose. Alexa Fluor 488-conjugated goat anti-rabbit IgG F(abЈ) 2 fragments, transferrin from human serum-conjugated with Alexa Fluor 546, streptavidin-conjugated with Alexa Fluor 633, and anti-mouse IgG antibodies conjugated with Alexa Fluor 546 were purchased from Molecular Probes, Leiden, The Netherlands. Monoclonal antibodies against the marker of the Golgi complex GM 130 from the Organelle sampler kit were purchased from BD Transduction Laboratories. The monoclonal antibody against ERp29, the marker of the endoplasmic reticulum, was a kind gift from Dr. S. Mkrtchian, Karolinska Institutet (32).
Peptide Translocation Studies-In the assay with biotinylated peptides, HeLa cells were incubated with 1.0 M of the biotinylated peptides for 1 h at 37°C. Thereafter, the cells were washed three times with PBS, fixed in 3.7% paraformaldehyde solution in PBS for 1 h at 4°C, again washed three times with PBS, incubated in 100 mM glycine for 5 min, permeabilized with 0.1% Triton X-100 for 5 min, and after washing were incubated in blocking buffer for 10 min. For peptide visualization, cells were incubated with streptavidin-Alexa Fluor 633 conjugate for 1 h.
In the antibody labeling assay, HeLa, PC12, or COS-1 cells were incubated with 1.0 M Big Dyn, 10 M Dyn A, or 10 M Dyn B for 1 h at 37°C in the presence of 10% fetal calf serum or in its absence. The medium was removed, and cells were washed three times with PBS prior to fixation in 3.7% paraformaldehyde solution in PBS for 1 h at 4°C. Cerebellar granule cells were incubated with peptide for 2 h and fixed in 4% paraformaldehyde in PBS. In some experiments the fixation was performed in 2% paraformaldehyde solution in PBS for 5 min at room temperature. Fixed cells were washed three times with PBS, incubated in 100 mM glycine for 5 min, and permeabilized with 0.1% Triton X-100 for 5 min. After washing with PBS, the cells were incubated in a blocking buffer (0.2% bovine serum albumin in PBS) for 10 min and then overnight at 4°C with rabbit anti-dynorphin B-antiserum, IgG fraction, or rabbit anti-dynorphin A-antiserum, IgG fraction. After washing, the cells were incubated with goat anti-rabbit IgG F(abЈ) 2 fragments conjugated with Alexa Fluor 488, washed, and mounted in Dako fluorescence media.
For confocal microscopy, live HeLa cells were incubated with 10 M fluorescently labeled Rh-Dyn A or Rh-Dyn B for 15 min. After three additional washings with the medium, cells were viewed under a confocal microscope.
Colocalization Studies-For colocalization studies with transferrin, HeLa cells were incubated with 1.0 M Big Dyn and 25 g/ml transferrin conjugated with Alexa 546 for 1 h, fixed, and labeled with rabbit anti-dynorphin B antibody and goat anti-rabbit IgG F(abЈ) 2 fragments conjugated with Alexa Fluor 488. For colocalization with the markers of the endoplasmic reticulum ERp29 and the Golgi complex GM 130, HeLa cells were incubated with 1.0 M Big Dyn for 1 h, fixed, and stained with rabbit anti-dynorphin B antibody and monoclonal antibodies against ERp29 or GM 130 as primary antibodies and goat anti-rabbit IgG F (abЈ) 2 fragments conjugated with Alexa Fluor 488 and anti-mouse IgG antibodies conjugated with Alexa Fluor 546 as secondary antibodies.
Microscopy-Confocal images were recorded using a Leica TCS SP confocal laser-scanning microscope (Leica Microsystem, Heidelberg, Germany). Images were prepared using Photoshop 6.0 (Adobe; San Jose, CA).
FCS-In this method, statistical analysis of the time course and amplitudes of spontaneous fluctuations in fluorescence intensity in a small volume element was performed to evaluate dynamic processes (34,35). FCS was performed on a ConfoCor instrument (Carl Zeiss) built according to the principles described previously (34,35) with confocal illumination of a laser volume element of 0.2 fl. For focusing optics, a Zeiss Neofluar 40 ϫ NA 1.2 objective for water immersion was used in an epi-illumination set up. Dichroic (Omega 540 DRL PO 2 ) and band pass (Omega 565 DR 50) filters were used to separate the excited from the emitted radiation. The rhodamine fluorophore was excited with the 514 nm wavelength of an argon laser. The fluorescence intensity was detected by an avalanche photo diode (SPCM-200, EG & G).
Prior to FCS measurements, HeLa cells were washed and incubated with fluorescently labeled peptides Rh-Dyn A, Rh-Dyn B, Rh-TP10, or free rhodamine (Rh, in the form of tetramethylrhodamine-5-(6)-isothiocyanate, Molecular Probes) in phenol red-free Iscove's medium with 10% calf serum at room temperature. Fluorescence intensity was measured at different intervals after incubation in the medium and at different subcellular locations. The average fluorescence intensity was estimated from 10-s measurements.
Binding of 125 I-Big Dyn and 125 I-Dyn B to HeLa Cells-HeLa cells were dissociated with trypsin and suspended in the RPMI 1640 medium. The cells were incubated with 125 I-Big Dyn (60,000 cpm/1 ml) or 125 I-Dyn B (65,000 cpm/1 ml) for 10, 30, or 60 min at 37°C. In some of the experiments, the cells were preincubated with 100 M poly-L-lysine or 100 M unlabeled Big Dyn for 30 min. The cell-bound radioactivity was separated from the free label by centrifugation at 5000 rpm for 5 min. Washing did not reduce the cell-bound radioactivity.
Mass Spectrometry-Electrospray (ES) ionization tandem mass spectrometry (ESI MS/MS) was performed on a Q-TOF (Micromass), mass spectrometer equipped with an orthogonal sampling ES-interface (Zspray, Micromass). Samples were introduced via gold-coated nano-ES needles (Protana). A capillary voltage of 800 -1000 V was applied together with a cone voltage of 40 -45 V and collision energy of 4.2 eV. The sample aerosol was desolvated in a stream of nitrogen. During collision-induced dissociation, the collision energy was in the range of 15-30 eV, and argon was used as the collision gas.
For MS measurements, HeLa cells were incubated with 20 M Big Dyn in the presence of fetal calf serum for 1 h at 37°C, washed, detached with Versene (EDTA) solution (Invitrogen), and centrifuged at 1000 rpm for 10 min at 4°C. Cells were extracted with 1 M acetic acid at 95°C for 5 min. The suspension was ultrasonicated and centrifuged at 14,000 rpm for 20 min. The supernatant was diluted with trifluoroacetic acid (0.1% final), loaded onto Sep-Pak columns (C18; Waters Inc., UK), equilibrated with 0.1% trifluoroacetic acid, and after washing the columns with 0.1% trifluoroacetic acid and 1% trifluoroacetic acid in 30% acetonitrile, the peptides were eluted with 1% trifluoroacetic acid in 60% acetonitrile. The dried peptide fraction was dissolved in 0.1% trifluoroacetic acid and loaded onto C18 Zip Tips (Millipore) activated with 0.1% trifluoroacetic acid in 70% acetonitrile, 0.1% trifluoroacetic acid in 50% acetonitrile, and finally equilibrated in 0.1% aqueous trifluoroacetic acid. After washing the Zip Tips with 0.1% aqueous trifluoroacetic acid, the peptides were eluted with 1% acetic acid in 60% acetonitrile.
Vesicle Preparation-Small, unilamellar 70% 1-palmitoyl-2-oleoylphosphatidylcholine, 30% palmitoyl-2-oleoyl-phosphoglycerol (POPC/ POPG) vesicles that are less than 100 nm in diameter and with a 30% negative surface charge density were prepared from the highest quality POPC and POPG (Avanti Polar Lipids, Alabaster, AL) and used without further purification. Weighed amounts of POPC and POPG at the desired concentrations (with the chosen POPC/POPG molar ratio) were dissolved in chloroform to ensure complete mixing of the components. The solvent was removed under high vacuum for 3 h, and the dried lipids were dispersed in 10 mM potassium phosphate buffer, pH 7.0, and cooled. The ice-cooled dispersion was sonicated under nitrogen using a Heat System model 350 A sonifier, with the microtip at a low output control (setting 4) and 50% duty cycle, until maximal transparency was reached (ϳ30 min). Titanium particles and lipid debris were removed by centrifugation at 25,000 ϫ g for 1 h.
Determination of Dynorphin Concentration-The concentration of dynorphins in aqueous stock solutions was determined spectrophotometrically at ϭ 280 nm on a CARY 4 spectrophotometer. A molar absorptivity at 280 nm of 5690 M Ϫ1 cm Ϫ1 for one Trp and 1280 M Ϫ1 cm Ϫ1 for one Tyr residue was applied in the calculations (36).
Fluorescence Spectroscopy-Fluorescence was measured on a 50B luminescence spectrometer (PerkinElmer Life Sciences) using the FL-WINLAB operating software at room temperature and 4 mm optical path length. Tryptophan fluorescence was excited at 280 nm, and the emission wavelength scanned from 300 to 500 nm. Tyrosine fluorescence was excited at 275 nm, and the emission wavelength was scanned from 285 to 400 nm. Scans were recorded with 4-nm excitation and emission bandwidths at a scan speed of 250 nm min Ϫ1 . Three scans were recorded and averaged for each sample.
CD Spectroscopy-CD spectra were recorded from 190 to 250 nm, with a 0.2 nm step resolution and 100 nm min Ϫ1 scanning speed on a Jasco J-720 CD spectropolarimeter equipped with a PTC-343 temperature controller. The spectra were collected and averaged over 40 scans. All experiments were performed at 25°C, using quartz cells of 1.0-mm optical path length. Background signals were subtracted from the CD spectra of the peptides. Contributions of spectral components from different secondary structures were estimated by computer fittings of the CD spectra with the VARSELEC program (37). For short, flexible peptides, these estimations should not be taken as exact evaluations of secondary structure contribution, but rather as reflecting relative changes in secondary structure under varying conditions.

RESULTS
Streptavidin and Immunolabeling of Dynorphins Translocated into Cells-To determine whether dynorphins translocate across the plasma membrane into cells, we initially applied confocal laser-scanning microscopy on fixed cells using streptavidin and anti-dynorphin antibodies for labeling of biotinylated ( Fig. 1) and non-biotinylated ( Fig. 2) peptides, respectively. In the assay with biotinylated peptides, HeLa cells were incubated with 1.0 M biotinylated Big Dyn, Dyn A, Dyn B, or Big Dyn 6-26 for 1 h at 37°C in the presence of fetal calf serum, fixed, labeled with streptavidin Alexa Fluor 633 conjugate, and viewed under a confocal microscope. Punctate cytoplasmic labeling, with negligible signal in the cell nucleus and on the plasma membrane, was observed in cells that were incubated with Big Dyn. Minimal labeling was evident with Dyn A, Dyn B, or Big Dyn 6-26. Because this assay detects biotinylated peptides with the same sensitivity, the results imply that Big Dyn translocates into cells more efficiently than the other dynorphins.
Translocation was also observed when HeLa cells were incubated with 1.0 M non-biotinylated Big Dyn, fixed, and labeled with anti-Dyn B antibodies ( Fig. 2A). These antibodies raised against the C terminus of Dyn B demonstrated 100% crossreactivity with Big Dyn, but do not react with Dyn A or the N-terminal dynorphin fragment [Leu,Arg 6 ]enkephalin. The Big Dyn distribution pattern was identical to that observed for the biotinylated peptide. This pattern and the intensity of labeling were nearly the same regardless of whether the cells were incubated in the presence of 10% fetal calf serum or in its absence. Thus, serum peptides and proteins at much higher concentrations than that of Big Dyn did not interfere with the uptake of this peptide. Fixation of cells under mild conditions, 2% paraformaldehyde solution for 5 min at room temperature instead of 3.7% paraformaldehyde solution for 1 h at 4°C, did not change the pattern of the Big Dyn distribution in cells.
Translocation into cerebellar granule (Fig. 2B), PC12, and COS-1 cells (data not shown) was assessed upon incubation with 1.0 M Big Dyn for 1 or 2 h (for cerebellar granule cells) at 37°C. Cells were then fixed and immunolabeled with anti-Dyn B-antibodies. Big Dyn was taken up and displayed similar distribution patterns in all cell types tested. To test whether translocation into cells can be observed at higher concentrations of Dyn A and Dyn B, HeLa cells were incubated with 10 M Dyn A or Dyn B for 1 h at 37°C, fixed, and immunolabeled with anti-Dyn A or anti-Dyn B antibodies, respectively. The anti-Dyn A antibodies were raised against the C terminus of Dyn A. At this concentration intense intracellular immunola-beling of Dyn A (Fig. 2C) but not Dyn B was observed with a distribution pattern similar to that of Big Dyn. Translocation of Dyn A was also evident when a biotinylated form of this peptide at 10 M concentration was incubated with HeLa cells for 1 h at 37°C in the presence of fetal calf serum and labeled with streptavidin Alexa Fluor 633 conjugate after cell fixation. Biotinylated Big Dyn 6-26 studied at the same concentration and under the same conditions did not enter the cells.
HeLa cells do not express opioid receptors at detectable levels, as evident from binding experiments with opioid ligands (data not shown, see Refs. 38 and 39). Nonetheless, in order to block opioid receptors that might be expressed at low levels, HeLa cells were preincubated with the general opioid antagonist naloxone (100 M) or the selective -opioid receptor antagonist nor-binaltorphimine (10 M), and penetration of Big Dyn and Dyn A was studied. Differences between treated and untreated cells were not observed, demonstrating that opioid receptors are not involved in the translocation process.
Confocal Microscopy Study on Live Cells-Translocation of fluorescently labeled Dyn A and Dyn B into live HeLa cells was studied by confocal microscopy. Cells were incubated with 10 M Rh-Dyn A or Rh-Dyn B for 15 min or longer in phenol red-free medium containing 10% fetal calf serum, washed three times with medium, and viewed under a confocal microscope. Intracellular fluorescence of Rh-Dyn A was observed in 30 -50% of the cells that did not apparently differ in morphology from the nonlabeled cells. Fluorescence was most intense in the cytoplasm, although some bright fluorescent domains were visible in the cell nucleus. The signal was generally much weaker at the plasma membrane and in the cell nucleus ( Fig. 3 A and  C). Rh-Dyn B did not penetrate into cells under the same conditions (Fig. 3, B and D). Thus, Dyn A exhibits a higher potential to translocate into live cells compared with Dyn B.
FCS Study-FCS, a confocal technique for examining molecular events in real time and with single molecule detection sensitivity (40), was applied to investigate peptide translocation into live cells, including dependence of translocation on the peptide concentrations and kinetics of translocation. Interactions of Rh-Dyn A with live cells were compared with those of Rh-Dyn B, Rh-TP10, a prototypical CPP, and the fluorescent label alone.
Cells characterized by slightly elongated polygonal morphology were analyzed, whereas those possessing a rounded shape and blebbing of the plasma membrane were presumed apoptotic and were excluded from the study (41). For measurements on the plasma membrane, the laser beam was focused on the upper surface of a cell attached to the floor of the Nunc chamber (Fig. 4A, solid circle). For measurements in the cytoplasm, the laser focus was positioned 6 m down from the cell-surface level and then 3 m above this level and 3 m to the right and to the left of this point (Fig. 4A, open circles). Local concentration of the peptides in the medium was measured 30 m above the plasma membrane surface (Fig. 4A,  cross).
FCS measurements demonstrated that Rh-Dyn A translocates efficiently into live HeLa cells at all concentrations tested as follows: 10, 25, and 200 nM (Fig. 4, B and C, and Fig. 5, A and C). The Rh-Dyn A accumulation in the cytoplasm exhibited a hyperbolic type kinetic profile with a rapid initial uptake phase lasting 1-5 min and with a half-time of less than 50 s. This phase was followed by a steady-state phase. Steady-state levels in the cytoplasm were generally proportional to the concentration of Rh-Dyn A in the medium (Fig. 4B). Fluorescence intensity associated with the plasma membrane was also proportional to the peptide concentration in the medium and was similar to that in the cellular interior (Fig. 4B). At 10 and 25 nM peptide concentrations in the medium, the steady-state fluorescence levels were similar in the medium, at the membrane, and in the cytoplasm. At 200 nM, however, the steady-state levels in the medium were greater than those in the cells and at the membrane.
Autocorrelation analysis of the FCS data showed that Rh-Dyn A was generally present in large complexes in the medium, on the membrane, and in the cytoplasm, formed with serum or cellular components. Evaluations obtained by fitting the experimental autocorrelation curves in the cytoplasm with a threecomponent model (40,42) suggest that about 50% of Rh-Dyn A formed complexes with intracellular components characterized by diffusion time diff Ϸ 1 s; about 15% of Rh-Dyn A was associated with smaller complexes characterized by diff Ϸ 6 ms, whereas about 25% of Rh-Dyn A was detected in the free form ( diff ϭ 0.13 ms). The determination of diffusion times was complicated because of the large size of the complexes; the values given should be considered as a first approximation. For the same reason, calculations of concentrations of Rh-Dyn A based on the measurement of diffusion times in the cells were not possible. By using Rh-Dyn A concentrations and fluorescence intensity values in the medium as a standard, and by assuming that spectral properties are the same in the studied compartments, the local steady-state concentrations of the peptide in the cytoplasm and on the membrane were estimated (Table II). The local Dyn A concentrations in the cytoplasm were in the nanomolar range (from 10 to 50 nM) at 10 -200 nM concentrations in the medium.
In the dissociation experiments, HeLa cells preincubated with Rh-Dyn A for 30 min were washed three times, and the incubation was continued in the fluorescent peptide-free medium (Fig. 5A). Washing decreased the concentration of Rh-Dyn A in the medium more than 10 times, whereas the peptide levels at the membrane and in the cytoplasm decreased by a factor of 2 and 5, respectively. The new steady-state levels were reached rapidly, within several minutes after washing.
Rh-Dyn B interactions with HeLa cells were investigated at 10 and 100 nM and at 1 M peptide concentrations in the medium. At 10 and 100 nM concentrations, translocation of Rh-Dyn B was virtually not observed; the fluorescence intensities in the cytoplasm did not exceed the background levels in control cells. At 100 nM, significant peptide accumulation was observed on the plasma membrane (Fig. 5C). At 1 M concentrations, limited translocation of Rh-Dyn B was seen with fluorescence levels in the cytoplasm higher than the background levels but considerably lower than those associated with the plasma membrane.
Rh-TP10, a prototypical CPP (Table I), showed a membrane/ cytoplasm distribution and kinetic profile of translocation into HeLa cells similar to Rh-Dyn A (Fig. 5, B and C). The penetration pattern of these two peptides markedly contrasted with that of free rhodamine used for comparison; rhodamine showed an initial accumulation predominantly on the plasma membrane with slow onset of penetration into the cytoplasm (Fig. 5C).
Uptake of Radioactive Dynorphins by HeLa Cells-Interactions of Big Dyn and Dyn B with live HeLa cells were also evaluated in binding assays with radioactive peptides (Fig. 6). Live HeLa cells were incubated with 125 I-Big Dyn or 125 I-Dyn B for 10, 30, and 60 min at 37°C, and thereafter bound and free peptides were separated by centrifugation. Binding of labeled Big Dyn to cells was much higher than that of Dyn B and reached its maximum level after 10 min of incubation. Preincubation of HeLa cells with either 100 M unlabeled Big Dyn or 100 M poly-L-lysine (Fig. 6) for 30 min did not change the cellular uptake of 125 I-Big Dyn. This implies that Big Dyn translocation occurs without saturation, and apart from the electrostatic interactions, other types of interactions contribute to Big Dyn binding to and translocation into the cells.
Mass Spectrometry Analysis of Big Dyn Translocated into HeLa Cells-The forms of Big Dyn translocated into the cells were analyzed by mass spectrometry (Fig. 7). HeLa cells were incubated with 20 M Big Dyn for 1 h at 37°C, washed, suspended in EDTA/PBS solution, and centrifuged, and the peptides were extracted from the cell pellet and purified on a Sep-Pak column. A peptide peak with an m/z ratio corresponding to Big Dyn was detected in the mass spectra. The peptide was confirmed to be Big Dyn by ESI MS/MS using collisioninduced dissociation with argon. Degradation products were not detected by the MS analysis.
Dynorphin Colocalization with Markers of Subcellular Organelles-The pattern of distribution of dynorphins in the cells (Figs. 1-3) suggests that these peptides are associated with subcellular membrane structures. This was investigated in colocalization experiments with markers of the endoplasmic reticulum (ERp29) (32), Golgi complex (GM 130), and clathrinmediated endocytosis (transferrin). HeLa cells incubated with 1.0 M Big Dyn for 1 h were fixed and probed with polyclonal anti-Dyn B antibody and monoclonal anti-ERp29 or GM 130 antibodies. Colocalization was evident for Big Dyn and ERp29 (Fig. 8A). Colocalization of Big Dyn with GM 130 (Fig. 8B) and transferrin (Fig. 8C) was not observed. The latter observation suggests that clathrin-mediated endocytosis was not involved.
Dynorphin Interactions with Lipid Membranes-The first event in the CPPs translocation into cells is their interaction with the negatively charged plasma membrane. Peptide-membrane interactions are generally modeled with phospholipid vesicles. To characterize the dynorphin-membrane interactions, the binding of dynorphins to small, unilamellar, and partially negatively charged POPC/POPG vesicles was studied by fluorescence and CD spectroscopy.
Phosphate buffer (10 mM, pH 7.0) was used for the spectroscopic measurements because no vesicle aggregation by peptides was observed at a lipid to peptide ratio (L/P) of Յ25 in this buffer. At higher ionic strength, in 50 mM phosphate buffer, pH 7.0, rapid and on-the-minute time scale, aggregation of vesicles was induced by Big Dyn at an L/P of Յ45, as is evident from the increase in light scattering. Fluorescence emission spectra of tryptophan (in Big Dyn, Dyn A, and Big Dyn 6-26) and tyrosine (in Dyn B) in the presence of vesicles or in their absence are shown in Fig. 9. In the buffer solution, the tryptophan and tyrosine residues demonstrated fluorescence emission maxima at emission ϭ 353 and 303 nm, respectively, which are close to the values measured for free tryptophan and tyrosine in aqueous medium (43). In the presence of the 30% negatively charged vesicles, the tryptophan emission maxima were shifted toward shorter wavelengths ( Big Dyn emission ϭ 341 nm, Dyn A emission ϭ 344 nm, and Big Dyn 6-26 emission ϭ 342 nm), and increases in fluorescence intensity were observed. Practically the same fluorescence spectra were observed at all investigated lipid/peptide ratios, ranging from L/P ϭ 25 for Dyn A and Big Dyn 6-26 to L/P ϭ 325 for Dyn A and L/P ϭ 395 for Big Dyn 6-26, or from L/P ϭ 50 to L/P ϭ 172 for Big Dyn, indicating that the binding process was already completed at the lower L/P ratios. In hydrophobic solvents, the emission maximum of tryptophan shifts toward shorter wavelengths, and in a non-polar solvent like methylcyclohexane, emission ϭ 300 nm (44). The data indicate that the tryptophan residue may be partially buried among the head groups of the POPG. A similar localization in the head group region of the lipid bilayer has been observed for the CPPs penetratin and transportan (45)(46)(47).
Upon interaction with vesicles, the tyrosine fluorescence spectrum of Dyn B showed an increase in fluorescence intensity (Fig. 9). This may be due to hydrogen bonding of the hydroxyl group of tyrosine with the glycerol, phosphate, or carboxyl groups of the polar lipids (48) but may also occur because of acquisition of secondary structure. Thus, all four peptides bind strongly to lipid vesicles, suggesting their efficient binding to the lipid phase of the plasma membrane. Induction of secondary structure in the peptides upon interactions with membranes may be relevant for the translocation process. In the aqueous buffer solution, the CD spectra demonstrated a random coiled structure of the studied dynorphins with the residual value of ␣-helical contribution estimated to be about 10% for all peptides (Fig. 10). The ␣-helical structure was revealed by the shift of the zero-crossing toward longer wavelengths and an increased negative amplitude around 222 nm. For Big Dyn, Dyn A, and Dyn B, the ␣-helix value increased to ϳ30% upon interaction with the lipid vesicles at an L/P ϭ 40 or higher. There was virtually no induction of secondary structure in Big Dyn 6-26. Similar CD results were obtained with large unilamellar vesicles with the same lipid composition, which were prepared by an extrusion procedure and had ϳ100 nm diameter (data not shown). Thus, both Big Dyn and Dyn A with high translocation potential are able to form secondary structures upon interaction with lipid membranes, whereas Big Dyn 6-26, which does not penetrate into cells, does not acquire secondary structure. On the other hand, Dyn B with low translocation potential formed an ␣-helical structure. Together, these results imply that the formation of secondary structure may be relevant but is not sufficient for the translocation process.

DISCUSSION
The present study demonstrates that the opioid peptides Big Dyn and Dyn A, which are similar to several groups of CPPs in the high content of basic and hydrophobic amino acids (Table  I), are able to translocate across the plasma membrane of neuronal and non-neuronal cells. Dyn B and Big Dyn 6-26 do not penetrate into cells (Fig. 11). Big Dyn showed higher translocation potential compared with that of Dyn A. Dyn A and transportan-10, a prototypical CPP penetrated into cells with similar efficacy. The translocated dynorphins were predominantly located in the cytoplasm where they were associated with the endoplasmic reticulum. Cells were incubated with 20 M Big Dyn for 1 h at 37°C in the complete medium containing 10% fetal calf serum. After extraction and purification by reversed-phase chromatography, the peptide fraction was analyzed by ESI tandem mass spectrometry. Big Dyn was detected as a full size peptide. Degradation products were not observed. The specificity of penetration was remarkable considering that all peptides belong to the dynorphin family (Table I; Fig.  11). Big Dyn 6-26, the most basic dynorphin peptide studied, was not able to penetrate into cells, suggesting that the translocation is not solely based on electrostatic interactions with the plasma membrane. This peptide may be considered as a Dyn A derivative in which the N-and C-terminal domains are swapped (Fig. 11). The lower translocation potential of Big Dyn 6-26 compared with Dyn A demonstrates that the primary sequence or the ability to form secondary structure, in which these peptides differ, may be relevant for efficient penetration into the cells.
The penetration pattern observed by immunolabeling fixed cells and by confocal microscopy and FCS on live cells was virtually the same. Therefore, the fixation-induced dynorphin relocation from the plasma membrane into the cytoplasm did not contribute substantially to the overall penetration pattern in contrast to what was suggested for several CPPs (49,50).
The rates of Rh-Dyn A translocation have been determined at several concentrations in the medium and compared with those of transportan-10. Both peptides demonstrated rapid penetration kinetics with the steady-state levels being reached within 1-5 min after initiation of the reaction. Substantial amounts of the peptides translocated across the plasma membrane already during the first 50 -60 s of incubation. The rapid kinetics suggest that dynorphins penetrate into cells and interact with their intracellular targets shortly after they are released into the extracellular space from cells producing these peptides. The kinetic profile observed does not discriminate between two possible penetration mechanisms, mediated by endocytosis or based on peptide-induced perturbation with the membrane lipid phase, because both of them may operate at similar time scales (51,52).
The local Dyn A concentrations measured in the cytoplasm and on the membrane were generally proportional to the concentrations of the peptide in the medium (Table II). These cytoplasmic concentrations may be biologically relevant because they are in the range of those that are required for specific peptide-protein interactions (1-100 nM).
Mechanisms of Penetration and Intracellular Trafficking-Although the phenomenon of CPP translocation across the plasma membrane into the cell is well established, there is still a substantial controversy regarding the mechanisms involved. The penetration may occur either through the lipid bilayer or by the endocytotic internalization. Interacting with the lipid phase peptides may translocate through membrane by forming pores or inverted micelles or diffuse through the bilayer (11). Peptide diffusion has been considered as improbable because it requires high desolvation energy (53,54). Another model proposes that hydrophilic CPPs translocate across the membrane by forming pores through oligomerization of several peptide molecules. CPP oligomers may adopt an amphipathic helical membrane-spanning conformation in which hydrophobic residues juxtapose the lipid phase whereas the hydrophilic faces form a channel within the complex (55). According to the micelle model, electrostatic interactions of peptides with the polar heads of membrane lipids induce formation of inverted micelles with peptide inclusions in their hydrophilic core. Micelle opening at the inner surface of the plasma membrane would release peptides into the cytoplasm (56). The diffusion, pore, and micelle models imply that translocation is a rapid, temperatureand energy-independent process. Consistently with these predictions substance P, R 7 W, Tat 47-57, and the signal sequence hydrophobic region peptides have been demonstrated to translocate into live cells at 4°C or in the presence of metabolic inhibitors (22,(57)(58)(59). CPP translocation across model lipid membranes also supports this notion (60,61). Models of the second type consider clathrin-mediated endocytosis, caveolar endocytosis, and lipid raft-dependent macropinocytosis as translocation mechanisms (8 -10, 57). Endocytosis is followed by passage of peptide from the vesicle into the cytoplasm. A characteristic feature of this mechanism is energy dependence. Thus, fluorescent penetratin, Tat 47-57 and (VRR) 4 have been determined in the clathrin-positive endocytic vesicles, and endocytic entry has been blocked by metabolic inhibitors (9). Full-length Tat and its 11-amino acid transduction domain have been found to translocate into live HeLa cells through the caveolar endocytotic pathway originating from cell membrane lipid rafts (8). Rapid internalization of Tat fusion proteins by lipid raft-dependent macropinocytosis after an initial ionic cellsurface interaction has also been demonstrated (10). Although two types of translocation mechanisms obviously differ, they both have received substantial experimental support. The overall conclusion is that they both are relevant for the translocation process; different peptides may use different strategies for translocation, although the same peptide could translocate through alternative pathways (11).
Dynorphins translocated into cells colocalize with FM 1-43, a general marker of endocytosis (preliminary data), implying that this pathway is involved. Colocalization with caveolin-1, a marker of caveolar endocytosis, was observed (preliminary data), although colocalization with transferrin, a marker of clathrin-mediated endocytosis, was not evident (see Fig. 8), suggesting that caveolar but not clathrin-mediated endocytosis plays a role. On the other hand, a fraction of Big Dyn pene- trated into HeLa cells through energy-independent mechanisms, as evident from experiments carried out under conditions when endocytosis was inhibited either by incubation at 4°C or in the presence of 0.45 M sucrose so that transferrin uptake was completely abolished (preliminary data). Thus, both energy-dependent (through caveolar endocytosis) and energy-independent (occurring at 4°C or the presence of 0.45 M sucrose) pathways might mediate dynorphin translocation across the plasma membrane. Identification of the potential role of each pathway in the translocation of dynorphins into the cells, including neurons, requires further investigation.
CPPs may be released from endosomes and macropinosomes into the cytosol (9,10,62). Tat and oligoarginine translocate from the plasma membrane through an endocytotic pathway to the Golgi apparatus and endoplasmic reticulum (62). This pathway may be common for the import of CPPs, viruses, and toxins, including ricin and Shiga toxin (63,64). Colocalization of Big Dyn with ERp29, an endoplasmic reticulum marker, suggests that dynorphins are transported to the endoplasmic reticulum through the same pathway.
An interesting question is how dynorphins capable of penetrating through membrane are retained within the secretory granules. If translocation potential is solely determined by the endocytotic mechanism, dynorphins would not penetrate through membranes of the secretory granules. However, if penetration is based on the interactions with lipid bilayer, the dependence of translocation potential on membrane characteristics should be studied to answer this question. These characteristics may include lipid composition, electrochemical transmembrane gradient, and pH in and outside of membrane structure that are obviously different for the plasma and secretory granule membranes. The possibility that dynorphins are complexed in the secretory granules with negatively charged molecules such as chromogranins/secretogranins and therefore are not able to diffuse out of the vesicles should be also investigated.
Structural Aspects of Translocation-Big Dyn, Dyn A, Dyn B, and Big Dyn 6-26 interacted strongly with charged phospholipid vesicles, suggesting that these peptides may tightly bind to the lipid phase of the plasma membrane. However, a binding event is apparently not sufficient for their translocation across the plasma membrane because dynorphins show different ability to penetrate into cells. This notion is supported by the FCS observations that both Dyn A and Dyn B adsorb on the membrane of HeLa cells, whereas only Dyn A penetrates into the cytoplasm.
Big Dyn, Dyn A, and Dyn B showed induction of ␣-helical structure at the levels of 30% upon interactions with phospholipid membranes. Among these peptides, Big Dyn and Dyn A, which penetrate into cells, have higher positive charge compared with Dyn B. Dyn B, which does not penetrate into cells, contains three basic residues, compared with 10 and 5 basic residues for Big Dyn and Dyn A, respectively (Table I). The high positive charge has been demonstrated to be important for the binding of CPPs to glycosaminoglycans on the plasma membrane. Big Dyn 6-26 demonstrated a lower ability to form secondary structure in the presence of phospholipids and was not able to translocate across the plasma membrane. Thus, two structural features, the ability to form ␣-helix and high positive charge, may in conjunction determine the membrane translocation potential of Big Dyn and Dyn A.
Our structural findings are in agreement with other conformational studies of Dyn A (65) in aqueous and hydrophobic solutions (66, 67) and in membrane model systems such as dodecylphosphocholine micelles (68, 69) and dimyristoylphosphatidylcholine bilayers (70). These studies reveal a largely unstructured backbone conformation of Dyn A in aqueous solutions and the formation of secondary ␣-helical structure in the presence of lipids. There are no structural data on Big Dyn published previously, but two heptapeptide repeats in this peptide are characterized by alternating hydrophobic and hydrophilic residues (see Fig. 2 in Ref. 71). A link of heptapeptide repeats is a distinctive feature of an ␣-helical coiled structure (72,73) in which the amphipathic helices interact through their hydrophobic surfaces. Formation of an amphipathic ␣-helix or ␣-helical coiled structure on the water-membrane interface may be a prerequisite for the tight binding of Big Dyn to the membrane followed by its translocation into the cell either through the endocytotic or pore/micelle mechanisms.
Potential Intracellular Targets for Dynorphins-Dyn A and Big Dyn differ from other opioid peptides in their ability to induce non-opioid responses. Although the precise mechanisms underlying non-opioid actions are not known, the NMDA receptors are apparently involved (20, 74 -80). Thus, Dyn A produces neurological dysfunctions, hindlimb paralysis, and also a long lasting state of allodynia that are blocked by NMDA receptor antagonists (74 -76, 78 -80). These non-opioid Big Dyn and Dyn A actions are critical for the development of chronic neuropathic pain. Prodynorphin knock-out mice do not maintain chronic pain state induced by spinal nerve ligation (81,82), whereas inhibition of degradation of endogenous Big Dyn produces nociception (80). Big Dyn was 100-fold more effective than Dyn A in inducing non-opioid nociceptive effects, whereas Dyn B was not active (potency order Big Dyn Ͼ Dyn A Ͼ Ͼ Dyn B) (79,80). The rank of potency correlates well with the ability of these peptides to translocate across the plasma membrane and suggests that peptide translocation is relevant for physiological processes such as the development of chronic pain. A conserved sequence of the C terminus of the NR1 subunit of the NMDA receptor has been identified as a target for Dyn A. 2 Synthetic peptides containing this epitope form stable complexes with Dyn A, functionally antagonizing the potentiation of NMDA receptor-activated responses produced by Dyn A. These peptides prevent Dyn A-induced cell death and reduce dynorphin-evoked allodynia in the spinal cord. The identified sequence of the NR1 subunit is localized intracellularly, and translocation of Dyn A across the plasma membrane may be a prerequisite for dynorphin-NMDA receptor interactions in the cells.
Substance P stimulates degranulation of mast cells even though the mast cells do not express receptors for this peptide (22). A rapid uptake of substance P into mast cells was demonstrated, and peptide applied intracellularly activated mast cell exocytosis. This effect was blocked by GDP␤S and pertussis toxin. Thus, substance P apparently translocates across the plasma membrane and induces mast cell degranulation through the G protein ␣ subunit-mediated pathway. G proteins can be activated by other cationic peptides besides substance P. By analogy, G proteins may be targeted also by basic dynorphins translocated into the cells. This assumption is supported by the observation that Dyn A stimulates histamine release from mast cells, an effect that is not mediated though opioid receptors (83)(84)(85)(86)(87).
Dyn A interacts with all three subtypes of opioid receptors with reasonably high affinity (88). A large proportion of ␦-opioid receptors are located in the neuronal cytoplasm where they are associated with the endoplasmic reticulum and Golgi apparatus (89,90). In dendrites and dendritic spines, ␦-opioid receptors are preferentially localized to membranes of the smooth endoplasmic reticulum and spine apparatus. Dynorphins translocated into cells may potentially interact with intracellular ␦-opioid receptors and also withand -opioid receptors undergoing transport and recycling. If intracellular receptors are functional, their interaction with dynorphins might elicit biological response through changes in receptor biogenesis.
The release of neurotransmitters at the synapse is mediated through fusion of vesicles with the plasma membrane (51,(91)(92)(93)(94). To maintain synaptic transmission, excess of membrane is retrieved by compensatory endocytosis (95) involving the rapid retrieval of small vesicles and bulk uptake of membrane similar to macropinocytosis (51, 96 -98). Whether retrieval endocytosis plays a critical role in dynorphin translocation across the plasma membrane at the synapse is important to evaluate in future studies. Statistical analysis demonstrated that opioid and other neuropeptides are characterized by a 2-fold higher average content of basic and hydrophobic residues compared with a random protein sample and non-regulatory peptides (30). These properties that are critical for translocation across the plasma membrane are common for neuropeptides and CPPs. Translocation of neuropeptides, including substance P and dynorphins, into the cell and their interaction with intracellular targets such as G proteins might represent an evolutionary ancient mechanism of intercellular communications and signal transduction that may have operated before specific receptors for these peptides evolved.