d-Maurocalcine, a Pharmacologically Inert Efficient Cell-penetrating Peptide Analogue*

Maurocalcine has been the first demonstrated animal toxin acting as a cell-penetrating peptide. Although it possesses competitive advantages, its use as a cell-penetrating peptide (CPP) requires that analogues be developed that lack its characteristic pharmacological activity on ryanodine-sensitive calcium channels without affecting its cell-penetrating and vector efficiencies. Here, we present the synthesis, three-dimensional 1H NMR structure, and activity of d-maurocalcine. We demonstrate that it possesses all of the desired features for an excellent CPP: preserved structure, lack of pharmacological action, conserved vector properties, and absence of cell toxicity. This is the first report of a folded/oxidized animal toxin in its d-diastereomer conformation for use as a CPP. The protease resistance of this new peptide analogue, combined with its efficient cell penetration at concentrations devoid of cell toxicity, suggests that d-maurocalcine should be an excellent vector for in vivo applications.

Over the last 15 years, several peptides have been described as possessing the property of accumulating inside cells. These peptides have collectively been denominated cell-penetrating peptides (CPP) 4 (1). The common characteristic of these peptides is that they are unusually enriched in positively charged amino acids (2). The HIV-1 TAT peptide, penetratin, and the chimeric transportan are well known and frequently used CPP (3,4). Reports on the use of CPP as vectors for the cell delivery of non-permeable membrane compounds are now on an exponential rise. As a consequence, many novel applications have been developed that are based on the cell delivery of drugs, peptides, proteins, antibodies, oligonucleotides, peptide nucleic acids, siRNA, cDNA, imaging agents, or nanoparticles (5)(6)(7). The mechanism(s) by which CPP enter cells remains largely debated, but the consensual view is that two non-concurrent pathways coexist: direct membrane translocation and one of many forms of endocytosis (8). This discussion is not only semantic because the intracellular distribution and fate of CPP differs with its mode of entry. If direct translocation is used, the peptide accumulates in the cytoplasm and reaches the nucleus, whereas, if endocytosis is the main pathway of cell entry, peptides end up in late endosomes. Translocation is the favored pathway for the cell delivery of cargoes by CPP because most drugs are active when delivered in the cytoplasm or in the nucleus, but not in late endosomes (4,9).
While searching for pharmacological agents regulating the activity of ryanodine receptors (RyR), our group came across a cationic toxin of 33-mer, maurocalcine (L-MCa), which was originally isolated and purified from a Tunisian scorpion, Scorpio maurus palmatus (10). RyR are intracellular calcium channels that are inserted into the membrane of endoplasmic reticulum and that control Ca 2ϩ release from intracellular stores. Several lines of evidence suggested that L-MCa represents the first member of a novel class of CPP. First, it was demonstrated that an extracellular application of L-MCa triggers an almost immediate Ca 2ϩ release from internal stores in myotubes, suggesting rapid access of L-MCa to its binding site (11). Second, it was discovered that the binding site of this toxin is located on the cytoplasmic side of RyR (12). Kinetic and RyR topological considerations therefore suggest that cell entry of L-MCa more likely relies on membrane translocation to achieve rapid cytoplasmic accumulation than on endocytosis. Third, L-MCa acts as an efficient vector for the delivery of proteins, peptides, nanoparticles, and drugs inside cells. At the structural level, the peptide folds along an inhibitor cystine knot motif with three disulfide bridges connected according to the pattern Cys 3 -Cys 17 , Cys 10 -Cys 21 , and Cys 16 -Cys 32 . Landmark properties of CPP are also found in L-MCa because (i) 12 of the 33 residues of L-MCa are positively charged, (ii) many of the charged residues are critical for cell penetration, and (iii) L-MCa interacts with critical membrane components required for cell penetration, such as proteoglycans and negatively charged lipids (13,14). Altogether, these considerations made L-MCa the first example of a folded and oxidized peptide toxin acting as CPP. Compared with other CPP, it possesses some competitive features, such as greater stability, cell penetration at lower concentrations, very low toxicity, and membrane translocation as the expected mode of cell penetration. Besides it is one of the few CPP whose cell penetration can be studied independently of cargo attachment because it possesses a direct physiological readout under the form of Ca 2ϩ release from internal stores. Two negative aspects related to the use of L-MCa instead of other CPP are (i) longer length of the peptide and (ii) pharmacological activity associated with cell penetration. Peptide length has little economical impact now because the cost of peptide production has sharply decreased. In contrast, pharmacological activity may represent a burden if one wants to use L-MCa as a vector for its application in vivo. To circumvent this problem, several strategies have been developed based on the assumption that the structural requirements for binding onto RyR were more stringent than those for cell penetration. First, single point mutagenesis of L-MCa was employed in order to alter peptide pharmacology without affecting cell penetration. Using this strategy, new L-MCa analogues have been defined with reduced or complete loss of pharmacological effects (13). Nevertheless, none of the analogues totally preserved the cell penetration efficiency of wild-type L-MCa. Second, an L-MCa analogue was produced in which all cysteine residues were substituted with isosteric 2-aminobutyric acid residues (15). The resulting peptide was unfolded due to the importance of disulfide bridges for folding/oxidation and lacked pharmacological properties. Interestingly, this analogue preserved correct cell penetration efficacy, albeit lower than L-MCa. In addition, lack of secondary structures and absence of disulfide bridges resulted in the loss of the competitive advantage of MCa over other CPP. Herein, we pursued another strategy with the aim to produce a cellpenetrating competent MCa analogue, similar in properties to L-MCa but without any pharmacological activity. We describe the chemical synthesis, successful folding, and oxidation of D-MCa; provide the three-dimensional structure of the resulting analogue; and demonstrate that it has cell penetration properties similar to those of L-MCa. In addition, D-MCa loses the protease sensitivity of L-MCa and is a non-toxic vector for cells. This is the first description of a folded/oxidized CPP build with D-amino acids designed on the basis of an animal toxin sequence.

EXPERIMENTAL PROCEDURES
Products-N-␣-Fmoc-L-amino acid, N-␣-Fmoc-D-amino acid, Wang-Tentagel resin, and reagents used for peptide synthesis were obtained from Iris Biotech. Solvents were analytical grade products from Acros Organics. Enzymes (trypsin, endoproteinase, and Asp-N) were obtained from Roche Applied Science.
Solid-phase Synthesis-The chemical synthesis of L-MCa was performed as previously described (10). D-MCa was chemically synthesized by the solid-phase method (16) using an automated peptide synthesizer (CEM Liberty). The peptide chain was assembled stepwise on 0.24 mEq of Fmoc-D-Arg-Pbf-Wang-Tentagel resin using 0.24 mmol of Fmoc-D-amino acid derivatives. The side chain protecting groups were as follows: trityl for Cys and Asn; tert-butyl for Ser, Thr, Glu, and Asp; Pbf for Arg; and tert-butylcarbonyl for Lys. Reagents were at the following concentrations: Fmoc-amino acids (0.2 M Fmoc-amino acids-OH in dimethylformamide), activator (0.5 M (1H-benzotriazole-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate in dimethylformamide), activator base (2 M N,N-diisopropylethylamine in N-methylpyrrolidone) and deprotection agent (5% piperazine, 0.1 M 1-hydroxybenzotriazole in dimethylformamide), as advised by PepDriver (CEM). After peptide chain assembly, the resin was treated for 4 h at room temperature with a mixture of trifluoroacetic acid/water/triisopropylsilan/ DTT (92.5:2.5:2.5:2.5). The peptide mixture was then filtered, and the filtrate was precipitated by adding cold t-butylmethyl ether. The crude peptide was pelleted by centrifugation (10,000 ϫ g, 15 min), and the supernatant was discarded. The reduced peptide was dissolved in 200 mM Tris-HCl buffer, pH 8.3, at a final concentration of 1 mM and stirred for 72 h at room temperature under air for oxidation and folding. Oxidized D-MCa was purified by high performance liquid chromatography (HPLC) using a Vydac C18 column (218TP1010, 25 ϫ 10 cm). Elution of D-MCa was performed with a 10 -60% acetonitrile linear gradient containing 0.1% trifluoroacetic acid. The purified fraction was analyzed by analytical reversed phase HPLC (Vydac C18 column 218TP104, 25 ϫ 4.6 cm). The pure fraction was lyophilized and quantified. 25 mg of D-MCa was purified, representing a theoretical yield of 25%. D-MCa was characterized by MALDI-TOF mass spectrometry. The reaction mixture was stirred for 2.5 h in the dark and then centrifuged. The resin was washed with N-methylpyrrolidone five times, followed by three washes with methanol. The final resin was dried and subjected for 4 h at room temperature to peptide cleavage with a mixture of trifluoroacetic acid/water/ triisopropylsilan/DTT (92.5:2.5:2.5:2.5). Next, crude FAM-D-MCa was oxidized as described for D-MCa.
Proteolytic Digestion of L-MCa and D-MCa-The stability of D-MCa to proteolytic digestion was investigated under the same conditions used to determine the disulfide bridge of L-MCa toxin. D-MCa (400 g) was dissolved in 50 mM sodium phosphate buffer and endoproteinase Asp-N (enzyme/substrate, 1:200 (w/w), 37°C, 36 h) or trypsin (enzyme/substrate, 1:32 (w/w), 37°C, 36 h). The digestion was terminated by acidification with 10% aqueous trifluoroacetic acid (TFA). The product was analyzed directly by reversed phase HPLC with a linear gradient of 1-60% acetonitrile into water containing 0.1% trifluoroacetic acid.
Circular Dichroism-The folding states of L-and D-MCa were checked by far-UV CD. The CD spectra were recorded on a Jasco 810 dichrograph using 1-mm-thick quartz cells in 300 l of water. CD measurements were performed at 298 K, using a wavelength ranging from 260 to 190 nm. Peptide concentrations were 0.1 mM for these measurements. D-MCa thermal stability was assessed using CD by following changes in the spectrum with increasing temperature at a fixed wavelength (200 nm). Measurements were performed in the temperature range of 20 -100°C with data pitch of 10°C step and temperature rise of 5°C/min. NMR Experiments-Purified D-MCa was solubilized in 500 l of a mixture of H 2 O and D 2 O (9:1, v/v) at a final concentration of 1.9 mM. Amide proton exchange rates were determined after freeze-drying of the sample and solubilization in 100% D 2 O. All 1 H NMR spectra were recorded on a Bruker DRX500 Avance III spectrometer equipped with a QX1 probe with z axis gradients. The temperature was set to 300 K, and the spectra were recorded with 2048 complex points in the directly acquired dimension and 400 points in the indirectly detected dimension (4,096 ϫ 512 points for the DQF-COSY). Solvent suppression was achieved using excitations sculpting with gradients (17). Two-dimensional spectra were acquired using the states-time-proportional phase increment method (18) to achieve F1 quadrature detection (19). NOESY spectra were acquired using a mixing time of 100 ms. TOCSY was performed with a spin locking field strength of 8 kHz for 80 ms. The individual amide proton exchange rates were determined by recording series of six NOESY spectra (10-h duration for each experiment) at 300 K using the D 2 O sample. Amide proton exchange still giving rise to nuclear Overhauser effect (nOe) correlations after 60 h of exchange were considered as slowly exchanging and therefore engaged in a hydrogen bond, the partner of which was identified on the sight of preliminary calculated structures. All spectra were processed with NMRPipe (20). Spectrum Analysis and Experimental Restraints-The identification of amino acid spin systems and the sequential assignment were done using the standard strategy described by Wüthrich (21), applied with NMRview 5.2 graphic software (22). The comparative analysis of COSY and TOCSY spectra recorded in water gave the spin system signatures of the peptide. The spin systems were then sequentially connected using NOESY spectra. The integration of nOe data were performed by measuring peak volumes. On the basis of known distances in regular secondary structures, these volumes were translated into upper limit distances with CYANA 2.1 (23). Remaining nOe were assigned using the CANDID/ NOEASSIGN automatic procedure of CYANA 2.1. Chemical shifts have been deposited in the BMRB Data Bank with accession number 16605. The calculation, after modification of libraries to integrate D-amino acid configuration, consisted of seven cycles of iterative automated nOe assignment and structure calculation of 250 conformers in each cycle. At the end of each CYANA run, unambiguously assigned peaks were converted in distance restraints and used as inputs for the next calculation steps. To keep the similar assignment condition, the same nOe calibration parameter, calculated by CYANA during the first run of nOe assignment, was used in all other runs of calculation. The final structure calculations with CYANA were started from 600 conformers, and a simulated annealing with 20,000 time steps per conformer was done using the CYANA torsion angle dynamics algorithm (24,25). The 100 best solutions were refined using a short restrained molecular dynamics simulation in explicit solvent (26,27) in the program XPLOR-NIH (28). At the end of the refinement, the 20 lowest energy solutions were selected to form the final ensemble. The quality of the structure was analyzed with the PROCHECK-NMR (29) and WHATIF (30) programs. Superposition of the structures was performed using the McLachlan algorithm as implemented in the program ProFit (A. C. R. Martin; available on the World Wide Web). All structure representations were made with the program PyMOL (DeLano Scientific, Palo Alto, CA). The atomic coordinates and experimentally derived restraints have been deposited in the Protein Data Bank with accession number 2KQL.
Preparation of Heavy SR Vesicles-Heavy SR vesicles were prepared following the method of Kim et al. (31). Protein concentration was measured by the Biuret method. Ca 2ϩ Release Measurements-Ca 2ϩ release from heavy SR vesicles was measured using the Ca 2ϩ -sensitive dye antipyrylazo III. The absorbance was monitored at 710 nm by a diode array spectrophotometer (MOS-200 optical system, Biologic, Claix, France). Heavy SR vesicles (50 g) were actively loaded with Ca 2ϩ at 37°C in 2 ml of a buffer containing 100 mM KCl, 7.5 mM sodium pyrophosphate, 20 mM MOPS, pH 7.0, supplemented with 250 M antipyrylazo III, 1 mM ATP/MgCl 2 , 5 mM phosphocreatine, and 12 g/ml creatine phosphokinase (25). Ca 2ϩ loading was started by sequential additions of 50 and 20 M CaCl 2 . In these loading conditions, no calcium-induced calcium release interfered with the observations. At the end of each experiment, Ca 2ϩ remaining in the vesicles was determined by the addition of 4 M Ca 2ϩ ionophore A23187 (Sigma), and the absorbance signal was calibrated by two consecutive additions of 20 M CaCl 2 .
3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) Assay-Cells were seeded into 96-well microplates at a density of ϳ8 ϫ 10 4 cells/well. After 2 days of culture, the cells were incubated for 24 h at 37°C with L-MCa or D-MCa at a concentration of 10 M. Control wells containing cell culture medium alone or with cells, both without peptide addition, were included in each experiment. 0.1% saponin was used as a toxic agent for comparison. The cells were then incubated with MTT for 30 min. Conversion of MTT into purple-colored MTT formazan by the living cells indicates the extent of cell viability. The crystals were dissolved with DMSO, and the opti-cal density was measured at 540 nm using a microplate reader (Biotek ELx-800, Mandel Scientific Inc.) for quantification of cell viability. All assays were run in triplicates.
Confocal Microscopy-For analysis of the subcellular localization of FAM-L-MCa or FAM-D-MCa in living cells, cells were incubated with the fluorescent peptides for 2 h and then washed with phosphate-buffered saline (PBS) alone. The plasma membrane was stained with 5 g/ml rhodamine-conjugated concanavalin A (Molecular Probes) for 5 min. Cells were washed once more. Live cells were then immediately analyzed by confocal laserscanning microscopy using a Leica TCS-SPE operating system. Rhodamine (580 nm) and FAM (517 nm) were sequentially excited, and emission fluorescence was collected in z-confocal planes of 10 -15-nm steps.
In some experiments, FAM-D-MCa incubation with CHO cells coincided with a 20-min incubation with 50 nM LysoTracker red DND-99 before confocal acquisition.
FACS-FAM-L-MCa or FAM-D-MCa was incubated for 2 h with CHO cells to allow cell penetration. The cells were then washed twice with PBS to remove excess extracellular peptide. Next, the cells were treated with 1 mg/ml trypsin (Invitrogen) for 10 min at 37°C to detach cells from the surface and centrifuged at 500 ϫ g before suspension in PBS. For experiments concerning endocytosis inhibitors, CHO cells were initially washed with F12K and preincubated for 30 min at 37°C with different inhibitors of endocytosis: (i) 1 mM amiloride, (ii) 5 M cytochalasin D, (iii) 5 mM nocodazole, or (iv) 5 mM methyl-␤-cyclodextrin (all from Sigma). The cells were then incubated for 2 h at 37°C with 1 M FAM-D-MCa. For all of these experimental conditions, flow cytometry analyses were performed with live cells using a Becton Dickinson FACS LSR II flow cytometer (BD Biosciences). Data were obtained and analyzed using FCS express software (De Novo). Live cells were gated by forward/ side scattering from a total of 50,000 events.
Statistical Analyses-All data are given as mean Ϯ S.D. for n number of observations, and statistical significance (p) was calculated using Student's t test.

RESULTS
Chemical Synthesis of D-MCa and Enzymatic Stability-The amino acid sequence and the expected disulfide bridge organi-  OCTOBER 29, 2010 • VOLUME 285 • NUMBER 44 zation of D-MCa are illustrated in Fig. 1A. Solid-phase chemical synthesis of D-MCa was achieved stepwise on 0.1 mmol of Fmoc-D-Arg-(2,2,4,6,7-pentamethyldihydrobenzofuran-5-sulfonyl)-Wang-Tentagel resin by means of an Fmoc strategy. The amount of target peptide linked to the resin was 0.085 mmol, indicating an 85% yield of peptide assembly. Indeed, a relative homogeneity of crude D-MCa was obtained after TFA treatment as witnessed by analytical C 18 reversed phase HPLC (Fig. 1B). Crude D-MCa was then folded/oxidized for 72 h in alkaline conditions, and the main peptide was purified to Ͼ99% homogeneity by semipreparative C18 reversed phase HPLC. The purity of D-MCa is illustrated by analytical C 18 reversed phase HPLC, and the folding/oxidation process is witnessed by the shift in elution time (Fig. 1B). MS analyses (MALDI-TOF technique) of crude and folded/oxidized D-MCa provide experimental molecular masses (M ϩ H) ϩ of 3866.5 and 3860.4 Da, respectively. The shift in experimental molecular mass of 6.1 Da upon folding/oxidation is in agreement with the engagement of all six cysteine residues in the formation of three disulfide bridges. The yield of D-MCa synthesis, following peptide assembly, TFA treatment, oxidative folding, and purification was 25%. The enzymatic susceptibility of D-MCa was determined by in vitro incubation of D-MCa with either trypsin or endoproteinase Asp-N (Fig. 1C). As shown by analytical C 18 reversed phase HPLC, D-MCa was totally resistant to the action of these enzymes, whereas L-MCa was significantly degraded. This is a significant advantage of D-MCa over L-MCa if in vivo applications are pursued with these CPP.

Cell Penetrating D-Maurocalcine Peptide
CD Analysis and Thermal Stability of D-MCa-The CD spectrum of D-MCa was determined to assess its secondary structure and compared with the CD spectrum of L-MCa ( Fig. 2A).
Measurements were performed at wavelengths ranging from 260 to 190 nm. The far-UV spectrum of D-MCa shows essentially -* transitions of the amide chromophore of the peptide backbone and a large positive contribution between 200 and 210 nm, indicating the presence of ␤-strand structures, which is coherent with the structure of L-MCa (32). Interestingly, the CD spectrum of D-MCa is a mirror image of the one recorded for L-MCa, revealing the persistence of the conformation and the presence of ␣-carbons in D-configuration. The presence of disulfide bridges in D-MCa should represent a competitive advantage over other CPP in terms of peptide stability. To assess this point, the CD spectral of D-MCa were recorded at several different temperatures (20, 40, 60, 80, and 100°C). All CD spectra coincided well, indicating that the structure of D-MCa was remarkably stable with increasing temperatures (Fig. 2B). A thermal denaturation/renaturation cycle has been performed on D-MCa from 20 to 100°C by following the ellipticity value at 200 nm. This value remains constant within the temperature range, demonstrating that D-MCa structure is resistant to thermal denaturation (Fig. 2C).
Determination of the Three-dimensional Solution Structure of D-MCa-NMR resonance assignment and structure calculation were performed for D-MCa. The spin systems were identified on the basis of both COSY and TOCSY spectra. Once the assignment procedure was carried out, almost all protons were identified, and their resonance frequencies were determined ( Table 1). The three-dimensional structure was determined by using 390 nOe-based distance restraints (including 158 intraresidue restraints, 116 sequential restraints, 36 medium range restraints, and 80 long range restraints). The distribution of these nOe-based distance restraints along the sequence is shown in Fig. 3A. In addition, 12 hydrogen bond restraints derived from hydrogen-deuterium exchange data have been included as well as nine distance restraints derived from the three disulfide bridges (Cys 3 -Cys 17 , Cys 10 -Cys 21 , and Cys 16 -Cys 32 ) as previously determined for L-MCa (32). Altogether, the final experimental set corresponds to 11.81 constraints/residue on the average. The final step of calculation, including the whole set of restraints (see Table 2), led to 900 solutions, from which the 100 best were energy-minimized in explicit solvent. The 20 best solutions (no residual violation greater than 0.1 Å) were kept for analysis (Fig. 3B). The root mean square (r.m.s.) deviation calculated on the ensemble, excluding C-and N-terminal residues, is 1.1 Ϯ 0.2 Å for backbone atoms and 2.32 Ϯ 0.24 Å if all non-hydrogen atoms are included ( Fig. 3A and Table 2). If only the ordered regions (amino acids 8 -10, 20 -23, and 30 -32) are considered, these r.m.s. deviation values drop to 0.43 Ϯ 0.08 and 1.37 Ϯ 0.19 Å, respectively. These values indicate a lower resolution for the unordered regions (Fig. 3B). This is confirmed by the individual r.m.s. deviation values and by the scarcity of constraints in these regions (Fig. 3A). The correlation with the experimental data shows no nOe-derived distance violation greater than 0.1, and the Ramachandran plot, after modification of the library to include D-amino acid configurations, shows (in PROCHECK software nomenclature) 95.5% of the residues in the allowed regions and 0.5% in the disallowed regions.
The three-dimensional conformation of D-MCa (Protein Data Bank accession code 2KQL) consists of a compact disulfide-bonded core, from which several loops and the N terminus emerge. The main element of regular secondary structure is a double-stranded anti-parallel ␤-sheet comprising residues 20 -23 and 30 -32. A third peripheral extended strand composed by residues 8 -10 is almost perpendicular to the doublestranded anti-parallel ␤-sheet (Fig. 3C). MCa is the first scorpion toxin known to adopt an ICK motif, whereas this fold has been observed for several other animal toxins from spider and shell venoms (33). To compare the native structure with the D-configuration herein described, we recalculated the L-configuration by using the previously published NMR data set (32) with the same protocol as for the D-configuration (Fig. 3C). This newly calculated structure of L-MCa is pretty close to the previously published one (r.m.s. deviation: 0.51 Å on the ordered regions) (32). To objectively compare the L-and D-MCa threedimensional structures, we modified the Protein Data Bank file of D-MCa in order to draw the peptide in the L-configuration. The superimposition of both structures is shown in Fig. 4. The r.m.s. deviation values between both structures are 1.18 Å for backbone atoms (this value drops to 0.44 Å on ordered structures) and 2.07 Å for all non-hydrogen atoms. Therefore, the molecule, synthesized in D-configuration for ␣-carbons adopts a fold that is a pure mirror image of the L-configuration (Fig.  3C). This inversion is also seen on the organization of the molecular surface, leading to (i) the conservation of the basic patch and of the overall electrostatic anisotropy (the resulting dipole moment is conserved both in direction and strength) and (ii) the geometry inversion of the toxin surface involved in interaction with RyR (not shown). The conservation of the basic patch implies that membrane translocation properties should be well conserved, whereas geometry inversion of toxin surface should affect the ability of D-MCa to recognize RyR.
D-MCa Is a Pharmacologically Inert Analogue of L-MCa-D-Isomer ligands are supposed to lose their ability to recognize their pharmacological targets. L-MCa is known to bind onto a discrete binding site of RyR (12), thereby triggering an increase in channel opening probability and the occurrence of a long lasting subconductance state (34). Also, channel opening by L-MCa can indirectly be monitored by the conversion of a low affinity binding site for ryanodine to a high affinity one (11) and by Ca 2ϩ release from purified sarcoplasmic reticulum (SR) vesicles (35). We therefore investigated the effects of D-MCa on these two paradigms (Fig. 5). As shown, L-MCa potently increases [ 3 H]ryanodine binding on heavy SR vesicles known to contain RyR. Maximal binding stimulation is 6.2-fold and occurs with an EC 50 of 17.8 nM. In contrast to L-MCa, D-MCa had no effect on [ 3 H]ryanodine binding even for concentrations   OCTOBER 29, 2010 • VOLUME 285 • NUMBER 44 up to 1 M (Fig. 5A). We have repeatedly illustrated that L-MCa triggers Ca 2ϩ release from purified heavy SR vesicles (11,35,36). Under similar experimental conditions, heavy SR vesicles, initially loaded with Ca 2ϩ , did not respond to an extravesicular application of 1 M D-MCa. In contrast, when this application was followed by an application of a much lower concentration of L-MCa (60 nM), a massive Ca 2ϩ release was observed (Fig.  5B). Altogether, these data argue that D-MCa, although a mirror structure of L-MCa, is totally pharmacologically inert with regard to RyR channel activation.

Labeling of D-MCa by FAM and Cell Penetration Properties-
To investigate the cell penetration properties of D-MCa, it was first covalently coupled to FAM using a peptide bond at the N terminus of the peptide (Fig. 6A). FAM is represented here by two isomers that may differ slightly in hydrophobicity during HPLC purification. After synthesis, FAM-D-MCa was purified by reversed phase C18 HPLC. As shown, the purified folded/ oxidized FAM-D-MCa differs significantly from D-MCa in elution time, suggesting that the resulting cargo-vector chimera was more hydrophobic than the vector itself (Fig. 6B). MS analysis of pure folded/oxidized FAM-D-MCa provides an experimental molecular mass (M ϩ H) ϩ of 4215.8 Da, confirming the correct assembly of the cargo-vector chimera molecule (Fig. 6B,  inset). Next, we investigated whether FAM-D-MCa could penetrate into live CHO cells. 1 M FAM-D-MCa was incubated for 2 h with live CHO cells, labeled for plasma membrane with concanavalin A-rhodamine, and examined immediately by confocal microscopy. As shown, FAM-D-MCa was present in all cells, but the distribution was variable from cell to cell (Fig. 6C). This distribution was fully comparable with the distribution observed for FAM-L-MCa, synthesized according to the same principles as FAM-D-MCa (synthesis not described here). These data indicate that D-MCa preserves the cell penetration properties of L-MCa. Confocal images further suggest that the labeling of D-MCa-FAM is intracellular rather than membranous. To reinforce this observation, we analyzed how much of the D-MCa-FAM colocalized with the plasma membrane staining (Fig. 6D). As shown, a very small fraction of the D-MCa-FAM was colocalized with concanavalin A-rhodamin staining (6.2% of total FAM-colored pixels).
In many cells, D-MCa-FAM penetration results in a mixed distribution (cytoplasm in addition to a punctuate distribution). These coincident distributions have been interpreted as evidence that two mechanisms of cell penetration of FAM-D- MCa coexist, translocation for cytoplasm distribution and a form of endocytosis for endosomal punctuate distribution, as evidenced in earlier studies (14,15,37). The use of either one of these two entry pathways is possibly under the influence of the nature and size of the cargo. Therefore, we reinvestigated the nature of FAM-D-MCa cell entry by using various inhibitors of the endocytosis route (Fig. 7). LysoTracker red stains endosomal structures. As shown, 73.5% of endosomal structures were also positive for FAM-D-MCa (Fig. 7, A and B). In addition, a significant fraction of FAM-D-MCa staining (55.4%) was endosome-negative, suggesting the coexistence of two types of cell distributions (more than half not related to endosomes and less than half endosomal). These ratios between both distributions differ from the one observed using fluorescent streptavidine as cargo. With streptavidine, we have shown that the macropinocytosis inhibitor, amiloride, blocks 80% of cell entry (14). Here, we quantified by fluorescence-activated cell sorting (FACS) the effects of nocodazole (microtubule formation inhibitor),   1 ϩ x b /c b )), where y 0 ϭ 0.06 at MCa ϭ 0 nM, a ϭ 6.2 is the maximal binding stimulation factor over basal value, b ϭ Ϫ1 is the slope coefficient, and c ϭ 17.8 nM is the EC 50 for L-MCa effect. D-MCa is without effect on [ 3 H]ryanodine binding. B, effect of D-MCa and L-MCa on Ca 2ϩ release from heavy SR vesicles. Heavy SR vesicles were actively loaded with Ca 2ϩ by five sequential additions of 2 M Ca 2ϩ in the monitoring chamber. The addition of 1 M D-MCa has no effect on Ca 2ϩ release, whereas a subsequent application of 60 nM L-MCa produces a massive sustained Ca 2ϩ release. Application of 0.5 mM EGTA chelates the released Ca 2ϩ and lowers the absorbance.  which was not observed for streptavidine as cargo (14). This result indicates that macropinocytosis remains the major mode of FAM-D-MCa cell penetration by endocytosis. These data also show that the nature of cargo influences to what extent endocytosis is used over cell translocation for cell penetration. In particular, with a small cargo, such as FAM, a significant fraction of the peptide is susceptible to enter through membrane translocation even if endocytosis is a major route of cell entry.  trations of FAM-D-MCa for 2 h, and cells were treated with trypsin for plastic support detachment and immediately evaluated for fluorescence intensity by FACS. As shown, the cell penetration of FAM-D-MCa is dose-dependent, with a positive signal starting at 100 nM and signals still increasing at a concentration of 3.3 M (Fig. 8A). The dose-dependent cell penetration of FAM-L-MCa was identical to that of FAM-D-MCa at all concentrations tested (Fig. 8B). The average fluorescence of each CPP-cargo complex was plotted as a function of concentration (Fig. 8C). As shown, FAM-D-MCa penetrates with the same dose dependence as FAM-L-MCa. Concentrations required for half-maximal cell penetration were 1.38 and 1.78 M for FAM-D-MCa and FAM-L-MCa, respectively. Finally, a good CPP is also one that presents no cell toxicity. Both D-MCa and L-MCa were evaluated for their toxic effects by incubation with CHO cells for 24 h. As shown, both peptides had no toxic effects at concentrations of 10 M, contrary to 0.1% saponin (Fig. 8D).

DISCUSSION
MCa is one additional member of the exponentially growing list of reported CPP. It is, however, one of the rare CPP that is fully natural and that presents such a well defined three-dimen-sional structure. However, because of its intrinsic pharmacological properties, it requires the design of new analogues that take advantage of its peculiar cell penetrating efficacy without the drawback of its activity. One design strategy, based on the replacement of cysteine residues by 2-aminobutyric acid residues and the consequent loss of secondary structure, led to the successful production of a pharmacologically inert but potent cell-penetrating MCa analogue. This strategy, however, eliminated some of the MCa features that distinguish it from other classical CPP. In particular, this analogue is no longer folded, and it loses some of its cell penetrating efficacy. The present strategy, consisting of successfully producing D-MCa, was aimed to circumvent both of these drawbacks. In addition, the use of D-amino acids renders the peptide resistant to protease action, making it a particularly useful vector for in vivo applications. Although several natural toxin/peptides are known to contain D-amino acids, such as several conopeptides (38 -40), none have a primary structure based on the sole use of D-amino acids. The presence of such amino acids in the sequence has been linked to toxin activity. In r11a, a 46-amino acid I 1 -conotoxin (41), has a phenylalanine at position 44, which undergoes epimerization from an L-Phe to a D-Phe, a transition linked to a dramatic gain in pharmacological activity (39). Similarly, isomerization of L-Phe to D-Phe also enhanced biological activity of conomap-Vt, a linear peptide from Conus vitulinus (40). The complete chemical synthesis of a toxin with only D-amino acids has been seldom reported. In one pioneering work, Di Luccio et al. (42) reported the chemical synthesis of D-maurotoxin, a four-disulfide-bridged toxin active on potassium channels. Interestingly, the kinetic features of the in vitro oxidation/ folding of D-maurotoxin are indistinguishable from those of L-maurotoxin. In contrast, the effects of PDI (which catalyzes breakage and reformation of disulfide bridges) and PPIase (which controls isomerization between the cis and trans configurations of prolyl imidic peptide bonds) on in vitro oxidation/folding of maurotoxin were stereo-selective. In another example, the 35-amino acid residue sea anemone ShK toxin that blocks potassium channels was also synthesized with amino acids that have a D-configuration at ␣-carbon (43). This peptide also folded as a mirror image of L-ShK. Curiously, pharmacological experiments and docking simulation analyses indicate that D-ShK has the ability to recognize and block K v 1.3 channels, albeit with a reduced affinity. This may occur because of the 4-fold symmetry of K v 1.3 channels and because of the compact set of interactions taking place (Lys 22 filling in the ion selectivity filter and Arg 24 and Arg 29 interacting with His 404 of the channel). In another example of a violation of the traditional lock-and-key model of ligand-target interaction, it was found that the D-enantiomer of GsMTx4, a peptide of the venom of tarantula Grammostola spatulata, remained active on stretchactivated cation channels (44). This effect is proposed to occur through local bilayer thinning without requiring a physical contact with the channel. In the case of MCa, we found that producing this CPP with D-amino acids also does not interfere with its folding and with the correct formation of disulfide bridges. The disulfide bridge pattern could not be assessed as usual (i.e. by limited trypsin digestion of the folded/oxidized peptide followed by MS analyses). In contrast, we confirmed the correct folding and disulfide bridge organization through 1 H NMR. This approach demonstrated that D-MCa is a mirror structure of L-MCa still displaying a significant basic face. D-MCa is a diastereomer rather than an enantiomer because only C ␣ are in the D-configuration. It is only for Thr 26 and Ile 28 that the side chain configurations differ from the exact mirror images. However, the different stereochemistry of the Thr and Ile side chains in D-MCa relative to the backbone did not cause any significant perturbations in the structure. Importantly enough, we observed that D-MCa was completely pharmacologically inert, indicating that the lock-and-key concept for the interaction of L-MCa with RyR is well preserved. Also, despite the affinity of MCa for negatively charged lipids of the plasma membrane (13), this finding also argues for a direct protein interaction of L-MCa with RyR in agreement with the cytoplasmic localization of the MCa binding site identified on RyR (12). Amazingly enough, D-MCa would therefore be the first example of a diastereomer toxin inactive on its pharmacological target. We also found that D-MCa keeps its ability to penetrate into cells, making this toxin the first known example of a folded/oxidized D-CPP. Because D-MCa appears to penetrate as well as L-MCa but also presents cytoplasmic localization, a sign of membrane translocation, these data indicate that MCa/lipid interactions are conformation-insensitive. This was expected already from the finding that unfolded MCa, in which cysteine residues were replaced by 2-aminobutyric acid residues, also penetrated efficiently in cells and also targeted associated cargoes to the cytoplasm (15). Besides cytoplasmic localization, both FAM-L-MCa and FAM-D-MCa also enter through endocytosis and end up in endosomal structures, albeit to a lesser extent than with streptavidine as cargo instead of FAM. These findings argue also that endocytosis remains a major contributor of the cell entry of MCa. We conclude that D-MCa is a competitive pharmacologically inert CPP that has the added advantage of being protease-resistant. It should also not be recognized by the immune system because it is probably resistant to proteolytic processing by T cells for presentation on major histocompatibility complexes. The maintained cell penetrating efficacy of this peptide should make it an ideal vector for in vivo applications.