Direct Cytosolic Delivery of Polar Cargo to Cells by Spontaneous Membrane-translocating Peptides*

Background: Spontaneous membrane-translocating peptides were discovered by screening in synthetic lipid vesicles. Results: The translocating peptides carry membrane-impermeant cargos directly across cell membranes and drive systemic biodistribution in small animals. Conclusion: These peptides constitute a new class of delivery vehicle for membrane-impermeant cargos. Significance: Spontaneous membrane-translocating peptides could expand the universe of useful drugs. Direct cellular entry of potentially useful polar compounds into cells is prevented by the hydrophobic barrier of the membrane. Toward circumventing this barrier, we used high throughput screening to identify a family of peptides that carry membrane-impermeant cargos across synthetic membranes. Here we characterize the plasma membrane translocation of these peptides with polar cargos under a variety of conditions. The spontaneous membrane-translocating peptides (SMTPs) delivered the zwitterionic, membrane-impermeant dye tetramethylrhodamine (TAMRA) into cells even when the conditions were not permissive for endocytosis. They also delivered the larger, anionic membrane-impermeant dye Alexa Fluor 546 but did not deliver a quantum dot nanoparticle. Under all conditions, the SMTP-cargo filled the cytoplasm with a diffuse, non-punctate fluorescence that was partially excluded from the nucleus. d-Amino acid peptides behaved identically in vitro, ruling out proteolysis as an important factor in the diffuse cellular distribution. Thus, cytosolic delivery of SMTP-cargo conjugates is dominated by direct membrane translocation. This is in sharp contrast to Arg9-TAMRA, a representative highly cationic, cell-penetrating peptide, which entered cells only when endocytosis was permitted. Arg9-TAMRA triggered large scale endocytosis and did not appreciably escape the endosomal compartments in the 1-h timescales we studied. When injected into mice, SMTP-TAMRA conjugates were found in many tissues even after 2 h. Unconjugated TAMRA was rapidly cleared and did not become systemically distributed. SMTPs are a platform that could improve delivery of many polar compounds to cells, in the laboratory or in the clinic, including those that would otherwise be rejected as drugs because they are membrane-impermeant.


Direct cellular entry of potentially useful polar compounds into cells is prevented by the hydrophobic barrier of the membrane. Toward circumventing this barrier, we used high throughput screening to identify a family of peptides that carry membrane-impermeant cargos across synthetic membranes.
Here we characterize the plasma membrane translocation of these peptides with polar cargos under a variety of conditions. The spontaneous membrane-translocating peptides (SMTPs) delivered the zwitterionic, membrane-impermeant dye tetramethylrhodamine (TAMRA) into cells even when the conditions were not permissive for endocytosis. They also delivered the larger, anionic membrane-impermeant dye Alexa Fluor 546 but did not deliver a quantum dot nanoparticle. Under all conditions, the SMTP-cargo filled the cytoplasm with a diffuse, non-punctate fluorescence that was partially excluded from the nucleus. D-Amino acid peptides behaved identically in vitro, ruling out proteolysis as an important factor in the diffuse cellular distribution. Thus, cytosolic delivery of SMTP-cargo conjugates is dominated by direct membrane translocation. This is in sharp contrast to Arg 9 -TAMRA, a representative highly cationic, cellpenetrating peptide, which entered cells only when endocytosis was permitted. Arg 9 -TAMRA triggered large scale endocytosis and did not appreciably escape the endosomal compartments in the 1-h timescales we studied. When injected into mice, SMTP-TAMRA conjugates were found in many tissues even after 2 h. Unconjugated TAMRA was rapidly cleared and did not become systemically distributed. SMTPs are a platform that could improve delivery of many polar compounds to cells, in the laboratory or in the clinic, including those that would otherwise be rejected as drugs because they are membrane-impermeant.
There are many exogenous polar compounds that could be useful in vitro or in vivo if they could be effectively delivered to the interior of cells. Such potentially useful compounds include biochemically active druglike molecules, metabolites and metabolite analogs, imaging agents, and more. In the laboratory and in the clinic, there is a need for an efficient, nonperturbing method of delivering such membrane-impermeant compounds into cells. Currently, exogenous polar compounds can be delivered to cells by detergent-like membrane disruption (1) (e.g. lipidic transfection reagents), electroporation (2), lipid vesicle fusion (3), pore-forming peptides (4), and membranepermeabilizing proteins (5). However, these methods are inefficient and are potentially toxic to cells. Furthermore, most cannot be used in vivo. Polar compounds can also be delivered to cells via endocytosis, pinocytosis, or phagocytosis (6 -8). This is accomplished by targeting the cargos to cell surfaces through conjugation to antibodies or other affinity reagents that are endocytosed into cells after cell surface binding and clustering (9). These methods can potentially be used in vivo but require that the cell or tissue of interest have active processes that can be specifically targeted and utilized.
For the past few decades, cell-penetrating peptides (CPPs) 3 have also been pursued as a generic cargo delivery platform. The field of cell-penetrating peptides began with the discovery that the HIV transactivating transcription factor protein Tat could enter uninfected cells directly from solution. This led to the characterization of the Tat "protein transduction domain," a ϳ9-residue, highly cationic N-terminal peptide sequence, RRKRRQRRR. The Tat transduction domain, now called a CPP, was shown to have generic cell penetrating activity; it can deliver large polar biomolecules into cells, including peptides, proteins, DNA, and RNA (10 -13). Another well studied CPP, penetratin, is likewise highly cationic and can deliver large polar molecules to cells (12)(13)(14)(15). The exact lengths and sequences of these CPPs are not critical for their biological activity. Simple physical-chemical mimics such as polyarginine (n Ͼ 6) have very similar cargo delivery capabilities (13, 16 -19).
There are now many so-called cell-penetrating peptides described in the literature (10,12,13,15,20). The mechanism of cell entry of these peptides has been widely studied (13,(17)(18)(19)(21)(22)(23). Although there is still some disagreement on the contribution of spontaneous membrane translocation, in recent years, a consensus has begun to emerge that the highly cationic CPPs (such as Tat and polyarginine) are actively taken up into cells mostly by one or more type of endocytosis (10, 12, 13, 15, 19, 21, 24 -26). Here we give experimental evidence that strongly supports this idea. The fact that cationic CPPs can deliver nanoparticle cargos as well as cargos to which they are not covalently attached (16,27,28) also supports the idea that endocytosis is involved. Once endocytosed, there is the complicating question of cargo release from endosomes that has sometimes been attributed to spontaneous membrane translocation (16). As we show below, endosomal release can be very inefficient (19,25,29,30) and can effectively prevent the delivery of some endocytosed cargos to the cytosol. Given this roadblock to delivery, various approaches based on physical or osmotic lysis of endosomes have been tested to increase the efficiency of endosomal release postuptake (22,29,(31)(32)(33)(34). Still, despite significant efforts over the last two decades, the highly cationic cell-penetrating peptides have yet to achieve their potential as delivery vehicles in the laboratory or in the clinic.
We hypothesized that peptides that can carry polar compounds across synthetic bilayer membranes by spontaneous translocation without permeabilization would also be capable of direct delivery of small, membrane-impermeant compounds into cells. Spontaneous membrane-translocating peptides (SMTPs) that are not dependent on endocytosis may have significant advantages over cationic CPPs for the delivery of nonmacromolecular classes of membrane-impermeant cargos. Such peptides will 1) function in vitro and in vivo; 2) be applicable to all cell types; 3) not be dependent on endocytosis, transporters, or other cellular processes; and 4) be non-toxic to cells. Guided by this hypothesis, we used high throughput screening of a peptide library to discover a remarkable family of SMTPs that may have these critical properties (28). Most importantly, these SMTPs translocate rapidly across synthetic lipid bilayer membranes and across living cell membranes without membrane disruption, probably as monomers (28,35), even when the cargo is a relatively large polar dye that is membrane-impermeant in its free form (28).
Here we characterize the utility of this novel peptide family to carry polar cargo moieties into living cells without cell lysis or toxicity. Additionally, we examine the mechanism, rate, and extent of translocation. Our results show that the delivery of polar, membrane-impermeant cargos up to 1000-Da molecular mass is efficient and occurs when endocytosis of cationic cellpenetrating peptides and other markers are blocked. Finally, we show that membrane-impermeant cargos attached to SMTPs become systemically distributed and are found in many tissues in small animals, strongly suggesting that the SMTPs described here will have genuine utility as a drug delivery platform in vivo.

EXPERIMENTAL PROCEDURES
Peptide Synthesis and Labeling-All peptides were synthesized and purified by Bio-synthesis, Inc. Reverse phase HPLC and MALDI mass spectrometry were used to verify purity and identity. Peptides were labeled with TAMRA-5-maleimide or AF546-maleimide by incubating the peptide with a 2-fold molar excess of dye in dimethylformamide. After 1 h at room temperature, the reaction was quenched by addition of 5% water and 5% diisopropylethylamine. Purification of the labeled peptide was done by bulk ion exchange using PolyCatA resin and by reverse phase HPLC. Purified peptides were dissolved at ϳ200 M in dimethyl sulfoxide, and the stock solution was diluted to 0.5-2 M just prior to use.
Cellular Translocation-Chinese hamster ovary (CHO) cells were seeded on cover glass slides with four chambers. Seeding density was 1 ϫ 10 4 cells/chamber with a 1.7-cm 2 seeding area. Cells were seeded 24 -48 h prior to experimentation with ϳ500 l of fresh, filter-sterilized Dulbecco's modified Eagle's medium (DMEM) to which 1.5 g/liter sodium bicarbonate and 0.8 g/liter glucose were added along with non-essential amino acids and fetal bovine serum. The cells were incubated at 37°C in 5% CO 2 . To perform translocation experiments at 37°C, fresh medium was added to adherent cells, which were conditioned for 30 min at 37°C prior to addition of markers. To perform translocation experiments at 22°C, slides with adherent cells were removed from the incubator and washed with PBS at 37°C and then with PBS at room temperature. Cells were conditioned at room temperature for at least 15 min. In a translocation experiment, either Alexa Fluor 488-dextran 3000 (FD3) was added to achieve a concentration of 10 g/ml or the plasma membrane stain NBD-lysophosphatidylethanolamine was added to a final concentration of 50 nM. Dye-labeled peptides were added from stock dimethyl sulfoxide (DMSO) solution to a final concentration between 0.5 and 5 M peptide. Most experiments were performed at 1-2 M peptide. The final DMSO concentration in the cell culture was usually 1% (v/v) or less, which we showed had no effect on cells.
Laser-scanning confocal fluorescence microscopy was performed on living, unfixed cells using a Nikon inverted confocal microscope with lasers of 488, 543, and 633 nm used to excite NBD/Alexa Fluor 488, TAMRA/Alexa Fluor 546, and Texas Red-casein/Qdot-streptavidin, respectively. Quantitation of dye intensities was done using the program ImageJ. For each measurement of relative intensity (inside/outside), multiple fields of cells were imaged under conditions where the intensity is a linear function of dye concentration (see text). In each field, the ratio for multiple cells was measured by integrating across a rectangular area for each cell that included unambiguous cell cytosol and a volume of external solution. Results from multiple cells in multiple fields (n ϭ 6 -12) were averaged together. Uncertainties are standard errors.
The translocation rate was also measured with confocal microscopy. After addition of FD3, a field of appropriate cells was identified followed by addition of 0.5-2 M SMTP-TAMRA. The exclusion of FD3 from the cells is used as an aid to setting the focal plane through the center of the cells. After addition of peptide-TAMRA, the same cells are imaged every 2 min without changing instrument settings to assess the rate of translocation from the ratio of the cytoplasmic intensity to the external intensity.
Cytolysis and Cytotoxicity-Cytolysis was measured by incubating cells with 2 M peptide-TAMRA in the presence of 50 nM SYTOX Green, a membrane-impermeant DNA-binding dye. We counted the proportion of cells that were SYTOX Green-positive after 30 -45 min. For a positive control, we used the lytic peptide melittin from bee venom, which causes almost 100% of cells to be SYTOX Green-positive after 5 min. Long term cytotoxicity was measured using the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay as described elsewhere (36). In this case, the peptides were added to cells upon seeding, and the cell viability was measured after 48 h. Here also the lytic toxin melittin was used as a positive control.
Cargo Biodistribution in Mice-Female BALB/c mice from Charles River Laboratories were injected with 400 l of solutions of free TAMRA, peptide-conjugated TAMRA, or PBS buffer. TAMRA concentrations were 5-15 M for peptides and 10-fold higher for free TAMRA. Intravenous injections were done via the tail vein. Intraperitoneal injections were made in the lower quadrant of the abdomen just under the subcutaneous fat. After injection, mice were returned to their cages and observed for 30 min for signs of distress. None were observed in any treatment. At 2 h, mice were anesthetized with ketamine and exsanguinated by cardiac puncture. Following blood collection, animals were sacrificed by cervical dislocation, and tissues were harvested and frozen without preservatives. Roughly 100 mg of each tissue or the whole organ if it was smaller was repeatedly frozen in liquid nitrogen and manually homogenized followed by dissolution of the homogenate in 0.5 ml of PBS buffer with 0.5% Tween 20 and 0.4 mg/ml each of proteinase K and trypsin. After 3 h of digestion at 37°C, Tween 20 was increased to 3.5% (v/v), and the solution was boiled for 15 min. After cooling, samples were centrifuged at 14,000 rpm in an Eppendorf centrifuge to pellet undigested debris. Similar extraction procedures have been reported by other groups (37)(38)(39)(40). The supernatants were added to the wells of a 96-well plate at dilutions of 1, 1:2, 1:4, 1:8, and 1:16. Measurements were made in a BioTek Synergy plate reader with an excitation filter of 540/20 nm (peak/width) and an emission filter of 590/40 nm. Each plate also contained a TAMRA concentration standard from 2 M to 20 pM in 10-fold dilutions as well as a 10-fold serial dilution of the solution that was injected into the mice. Calculated concentrations are corrected for the tissue weight actually processed and for the actual injected concentration. We assume that nmol of TAMRA/g of tissue is equivalent to M TAMRA in the tissue.

RESULTS
Peptides and Cargos-For this work, we studied three representative SMTPs from the family that was identified in a high throughput screen (28) and a D-amino acid variant of one of them ( Fig. 1). For controls, we studied ONEG, an observed negative peptide directly generated in the screen, and DNEG, a designed negative peptide that was engineered from TP3 by replacing its three critical arginines with aspartate (28). For a highly cationic cell-penetrating peptide control, we studied Arg 9 , a mimic of Tat (12). All peptides have a free N-terminal amino group and a C-terminal cysteine-amide. All cargos were coupled to the sulfhydryl side chain of cysteine using maleimide chemistry.
The dye TAMRA-maleimide was used as the cargo in most experiments. TAMRA-maleimide is a zwitterionic 430-Da visible dye that has one negative charge and one positive charge ( Fig. 1). TAMRA makes an ideal cargo to test the properties of translocating peptides because it is membrane-impermeant in free form (28,41) (see also supplemental Figs. S24 and S29) and because its visible fluorescence provides a robust intensity at wavelengths with very little cellular autofluorescence or other forms of interference. TAMRA is quantifiable by fluorescence microscopy to concentrations as low as a 1 nM and is not especially prone to photobleaching. Furthermore, TAMRA gives a stable intensity inside cells and is not readily subject to cellular redox chemistry or other paths to degradation. In a few experiments, we also studied peptides conjugated to Alexa Fluor 546, an anionic, 1030-Da dye with two sulfate moieties (Fig. 1). Finally, we studied biotinylated peptides bound to a streptavidin-quantum dot conjugate.
Measurement of Spontaneous Translocation-We used laserscanning confocal fluorescence microscopy on living cells to monitor cellular entry of the fluorescent dye TAMRA attached to peptides. Imaging was either done in the presence of a fluorescent dextran, FD3, to mark the aqueous phase or in the presence of NBD-lysophosphatidylethanolamine to label the plasma membrane.
To study spontaneous translocation, we needed to inhibit the ability of cells to undergo endocytosis with the smallest possible disruption to the physical state of the cell membrane. However, FIGURE 1. Peptides and cargos. In this work, we studied three spontaneous membrane-translocating peptides, TP1,TP2, and TP3, discovered in high throughput screen (28) and a D-amino acid version of TP2. For negatives, we studied ONEG, an observed negative from the screen; DNEG, a designed negative made by replacing three critical arginines in TP3 with aspartate; and Arg 9 , a canonical cationic cell-penetrating peptide. All peptides have N-terminal amines and C-terminal amides. A C-terminal cysteine was used for cargo attachment. For "cargos," we used mostly TAMRA-maleimide, a polar, membrane-impermeant dye. In a few experiments, we used Alexa Fluor 546maleimide, a large, anionic membrane-impermeant dye.
there are many classes of endocytosis, each with different characteristics (9), and there is no validated, agreed upon method of blocking them all (42). For this work, we used the endocytosis of Arg 9 -TAMRA, a classical cell-penetrating peptide, as the most relevant probe of endocytosis. In the literature, endocytosis is sometimes inhibited by incubation at 4°C (43); however, the changes in the membrane that occur near freezing, especially formation of gel-like states of the membrane lipids (44), can obscure spontaneous translocation. We have shown preliminary evidence that endocytosis in CHO cells is mostly inhibited when the cells are brought to room temperature and medium is replaced by PBS (28). The behavior of Arg 9 -TAMRA in Fig. 2 confirms the effect of temperature on endocytosis in living CHO cells. See also supplemental Figs. S1-S7. At 37°C in growth medium, Arg 9 -TAMRA was rapidly taken up and concentrated into small intracellular organelles shown by highly punctate fluorescence. The fact that these organelles also contained fluorescent dextran (FD3) taken up from the external medium indicates that they must be a class of endocytic vesicle. In addition to dextran, we have also shown that proteins in the external solution were taken up in Arg 9 -induced endosomes (supplemental Fig. S1) and that some of the Arg 9 -containing organelles became acidified (supplemental Fig. S5). Uptake of Arg 9 -TAMRA was rapid and began only a few minutes after peptide addition. The images in Fig. 2 were taken after 30 -45 min with 2 M Arg 9 -TAMRA. A higher Arg 9 concentration or longer incubation led to many more Arg 9 /FD3-positive endosomes per cell (supplemental Fig. S4). CHO cells incubated at 37°C without Arg 9 did not take up any FD3 or soluble protein into endosomes (supplemental Fig. S2). Thus, Arg 9 and other cationic CPPs trigger the onset of large scale endocytosis that does not occur in their absence. This may explain why cationic CPPs can "deliver" nanoparticles and cargo to which they are not covalently attached (27,45). It may also help explain their potential for toxicity (see below).
The behavior of Arg 9 -TAMRA changed dramatically at 22°C (room temperature) at which it accumulated on the plasma membranes of CHO cells but did not enter cells detectably and did not trigger any detectable endocytosis ( Fig. 2 and supplemental Figs. S6 and S7). We conclude that CHO cells in PBS at FIGURE 2. Mechanism of Arg 9 -TAMRA entry into living cells. Upper panels, living CHO cells were incubated with 10 g/ml FD3 and 2 M Arg 9 -TAMRA either at 37°C in full growth medium or at 22°C in PBS. After 30 -45 min, live cells were imaged by confocal laser-scanning fluorescence microscopy without washing away free peptide. Middle panels, higher magnification images after incubation with Arg 9 -TAMRA at 37°C in medium. Lower panels, representative TAMRA and FD3 intensity scans across the cells indicated above by yellow boxes. Left, 37°C. Right, 22°C. The external intensity was defined as 100%, and the cell compartments are marked as follows: PM, plasma membrane; C, cytosol; N, nucleus. OCTOBER 11, 2013 • VOLUME 288 • NUMBER 41 22°C cannot undergo large scale peptide-triggered endocytosis at least in the timescales of Ͻ1 h that we examined in this work. Although these results are for cells in PBS, we observed exactly the same inhibition of endocytosis of Arg 9 -TAMRA when the cells were simply brought to room temperature in full growth medium (not shown).

Spontaneous Membrane-translocating Peptides
To verify that decreasing the temperature to 22°C is enough to essentially eliminate endocytosis on the relevant timescale, we also studied the endocytosis of wheat germ agglutinin and the receptor-mediated endocytosis of EGF triggered by its binding to EGF receptor. In both cases, large scale endocytosis occurred at 37°C but essentially ceased at 22°C even in full growth medium. See supplemental Figs. S8 -S15. Although we cannot prove with absolute certainty that all forms of endocytosis were completely blocked at 22°C in PBS, we have shown that several common classes were severely inhibited in the 1-h timescales of these translocation experiments.
Mechanism of SMTP Cargo Delivery-Having shown that 22°C is sufficient to block CPP endocytosis on short timescales, we proceeded to study the mechanism of SMTP entry into cells. To test our hypothesis that the SMTPs will translocate into cells without triggering or requiring endocytosis, we performed parallel experiments at 37°C in medium and at 22°C in PBS using TAMRA-labeled SMTPs ( Fig. 3 and supplemental Figs. S20 -S29). Under both conditions, the SMTP-TAMRA conjugates filled the cell cytosol with similarly intense, diffuse (non-punctate) fluorescence. No endosomal encapsulation of peptide or of FD3 was ever observed even at 37°C in growth medium, which are endocytosis-permissive conditions. The D-amino acid version of TP2 behaved identically (see below), indicating that peptide degradation was not responsible for the diffuse cytosolic distribution of the TAMRA cargo. See also supplemental Figs. S22 and S26. The fact that SMTP-TAMRA conjugates did not induce the encapsulation of FD3 at 37°C (Figs. 3-5 and supplemental Figs. S25-S27) verifies that, unlike Arg 9 , the SMTPs do not trigger endocytosis even when the conditions are permissive. The observation that SMTPs readily entered the cell cytosol at 22°C in PBS, conditions under which Arg 9 , wheat germ agglutinin, and EGF were excluded, verifies that SMTPs do not require endocytosis. The SMTP-TAMRA peptides can enter the cell cytosol predominantly, if not entirely, by directly translocating across the plasma membrane in the same way that that they spontaneously translocate across synthetic lipid bilayers (28,41). As we discuss below, the negative peptides in Fig. 1, ONEG and DNEG, did not significantly enter cells under any condition (see also supplemental Figs. S28 and S32). Experiments conducted in human epithelial A431 cells with TP2-TAMRA, Arg 9 -TAMRA, and the negative peptide ONEG-TAMRA (supplemental Figs. S29 -S31) showed the same behavior, indicating that the performance of these peptides is not cell type-specific. In support of these conclusions, we note that SMTP translocation into cells at 22°C was measurable just a few minutes after addition of peptide and was complete within about 30 min, whereas endocytosis of Arg 9 , wheat germ agglutinin, and EGF receptor/EGF was essentially undetectable for an hour or longer under the same conditions.
To further test our conclusion that SMTPs do not significantly interact with endocytic pathways to enter cells, we stud-ied the effect of a very high concentration of TP2D (50 M) on the endocytosis of Arg 9 -TAMRA and wheat germ agglutinin-TAMRA and on the receptor-mediated endocytosis of EGF-TAMRA. Despite the presence of a very high concentration of non-degradable TP2D, there were no detectable effects on the number, morphology, or distribution of endosomes over the 1-h timescale of our translocation experiments. See supplemental Figs. S16 -S19.
Spontaneous Translocation into Cells-In Fig. 4, we show additional translocation images collected under various conditions but always at room temperature in PBS under which endocytosis of Arg 9 and other markers are inhibited. In A and B, we show laser-scanning confocal fluorescence microscopy images of CHO cells incubated for 30 -45 min at room temperature with 1 M TP1-TAMRA. A and C show individual channels and a merged image of an experiment with an NBD-lysolipid plasma membrane stain added after removal of the excess peptide from the external space. In B and D, the cells were incubated with FD3 added prior to the peptide. In these experiments, neither peptide nor FD3 were washed away prior to  OCTOBER 11, 2013 • VOLUME 288 • NUMBER 41

Spontaneous Membrane-translocating Peptides
imaging. In all of these room temperature experiments, all of the SMTP-TAMRA conjugates in Fig. 1 accumulated in the cell cytoplasm in 30 -45 min, giving a mostly uniform (non-punctate) fluorescence. See supplemental Figs. S20 -S28 for additional images and data.
When external conjugates were not washed away, the TAMRA accumulated to a cytosolic intensity that was always similar to or slightly higher than the external intensity. The hydrophilic aqueous phase marker FD3 was completely excluded from the cells and was not taken up in endosomes at all. In Fig. 4C, we show translocation experiments performed with cells at low density and high density (upper left and upper right) and in the presence (lower left) and absence (lower right) of 500 nM rotenone, an ATP synthase inhibitor used to deplete ATP in the cell. These manipulations had little or no effect on peptide translocation after 30 -45 min of incubation at room temperature. In D, we show an experiment with FD3 and free unconjugated TAMRA demonstrating that the TAMRA cargo did not enter cells measurably over the course of at least several hours.
Quantitation of Translocation Rate-Next we measured the rate and extent of translocation for all of the peptides shown in Fig. 1

under identical conditions: CHO cells at 22°C in PBS.
Before we measured translocation, we tested the fluorophore concentration range over which confocal measured intensities were linear with concentration using solutions of peptide-TAMRA conjugate. Intensities were always linear with concentration up to about 75% of the maximum measurable value of the detector. Linearity was independent of the actual concentration range from 1 nM to 10 M. Thus, as long as the gain factor of the instrument was set low enough to ensure that the maximum intensity of a sample was not close to the maximum numerical value of the detector, the intensities were a linear function of concentration. These conditions were satisfied in all translocation experiments.
To perform quantitative translocation rate measurements, we used CHO cells to which we added FD3 as a marker of the external aqueous space followed by either peptide-TAMRA conjugates or other markers. The value we report is the ratio of intensity in the cell cytosol compared with the intensity outside the cell in solution minus appropriate background intensity. These measurements were made without washing away external peptide. For quantitation, we ignored the nucleus where the fluorescence was lower than that in the cytosol (but higher than background). In Fig. 5, we show an example of a quantitation image. The yellow rectangles indicate the areas over which intensities were integrated in this example. The plot in the lower right shows the intensities for FD3 and SMTP-TAMRA across the plasma membrane after 30 -45 min of incubation at 22°C. We used the FD3 signal to verify cell integrity and to delineate the boundary of the cell. The average TAMRA intensity on either side of the plasma membrane was used to calculate the inside/outside concentration ratio. In the ratios reported below, we give the average intensity ratio across all cells/preparations measured ϮS.E.
In Fig. 6, we show a set of example images from a time course experiment using CHO cells at room temperature and the peptide TP2-TAMRA. The amount of TP2-TAMRA inside the cells increased rapidly over the course of the first 20 min. Rate experiments like this were conducted for TP1, TP2, TP3, and the D-amino acid version of TP2 (Fig. 1). The SMTPs rapidly accumulated inside cells, approaching an inside/outside ratio of about 1 within 20 min (Fig. 7). As we show in Figs. 3-5 and supplemental Figs. S20 -S28, the inside/outside ratio reached an equilibrium value that is a little greater than 1 after 30 -45 min. It did not continue to increase after the 1st h. We currently do not know why the final concentration of SMTP-TAMRA inside the cell is higher than that outside, but we speculate that peptide accumulation could be affected by the residual transmembrane potential, which is negative inside, or by the interaction of SMTP-TAMRA peptides with the highly crowded protein environment of the cytosol.
We performed the same rate measurements for the dye controls FD3 and free unconjugated TAMRA and for the peptide controls DNEG-TAMRA, ONEG-TAMRA, and Arg 9 -TAMRA. None of the controls, including the classical cell-penetrating peptide Arg 9 , entered cells appreciably over the course of the experiment under these conditions (Fig. 7).
Other Cargos in Vitro-We tested the ability of TP2 to deliver two other cargos to cells. First, we coupled TP2 to Alexa Fluor 546, a membrane-impermeant dye that is much larger than TAMRA (1030 versus 430 Da) and much more polar because it has two anionic sulfate moieties (Fig. 1). As shown in Fig. 8, TP2-AF546 also translocated into the cytosol of CHO cells under conditions where endocytosis of Arg 9 did not occur. We also verified that TP2-AF546 translocates across synthetic lipid bilayers and that free AF546 is membrane-impermeant (not shown). Finally, we tested a quantum dot cargo that we expected to be too large to be deliverable by spontaneous translocation. We synthesized TP2-biotin and incubated it with a streptavidin-Qdot complex. After incubation of this complex with CHO cells at room temperature, we observed no evidence of Qdot intensity inside the cells (Fig. 8). Because it is known that cationic CPPs carry Qdots into cells by endocytosis (16), this experiment strengthens our conclusion that the translocating peptides carry cargos into cells by spontaneous translocation.
Cytolysis and Cytotoxicity-We also assessed the effect of the translocating peptides on the integrity and viability of living cells. First, we measured peptide-induced cytolysis using the membrane-impermeant, DNA-binding dye SYTOX Green in the presence of 2 M SMTP-TAMRA peptides, negative peptide controls, and Arg 9 . As shown in Fig. 9, we observed no measurable evidence of cytolytic activity at this concentration, which is the typical concentration of confocal microscopy experiments, indicating that the observed translocation is not due to membrane permeabilization. In a few experiments, we tested 10 M SMTP-TAMRA and observed no effect on CHO cells (not shown). As a positive control, we used the well known lytic peptide toxin melittin, which caused nearly 100% cell lysis (Fig. 9).
We also measured the effect of the peptides on the long term viability of CHO cells with the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay, which quantitates the activity of mitochondrial reductases (36,46) to assess the number and viability of cells. In this experiment, we measured the effect of peptide concentrations up to 30 M peptide. We observed little or no effect of peptides on cell viability up to 10 M peptide. At 30 M, TP3 and Arg 9 caused some measurable toxicity (Ͼ20%), whereas TP1 and TP2 did not. We performed a few experiments with TP2 up to 50 M and observed no effect on cells (not shown). The lytic control peptide melittin was essentially 100% lethal to cells at the same concentrations.
Cargo Distribution in Vivo-To obtain a preliminary assessment of the effect of SMTPs on the in vivo fate of a membraneimpermeant small molecule cargo, we assessed the biodistribution of TAMRA and peptide-TAMRA conjugates in mice. Female BALB/c mice were injected intravenously and intraperitoneally with solutions of ϳ100 M free TAMRA or ϳ10 M SMTP-TAMRA conjugates. This approach is similar to what others have done for CPPs (37)(38)(39)(40). In this single end point experiment, we sacrificed the animals at 2 h and harvested blood and tissues for analysis of TAMRA content. The results are shown in Fig. 10. In the intravenously injected animals, free TAMRA was essentially gone from all tissues by 2 h, consistent with the expected rapid clearance via the kidneys. TAMRA conjugated to L-amino acid SMTPs administered intravenously behaved very differently. The cargo was found at a concentration greater than 200 nM in many tissues even at 2 h but was essentially undetectable in the bloodstream. For these conjugates, TAMRA was found at high concentration in the large intestine but was also found in heart, muscle, and other tissues. Tissue distributions were not the same for TP1,TP2, and TP3. Importantly, TAMRA attached to D-amino acid SMTPs had an improved tissue distribution at 2 h with relatively high concentration (Ն500 nM) in most tissues, suggesting a systemic tissue distribution. The concentration of TP2D-TAMRA in the liver and large intestine was relatively much lower than that of the L-amino acid SMTP-TAMRA conjugates, and the concentration in other tissues was higher. This suggests that proteolytic degradation contributes to the long term tissue distribution and clearance of L-amino acid SMTP-cargo conjugates, likely shunting TAMRA into the same clearance pathway as free TAMRA. When free TAMRA was administered by intraperitoneal injection, it was detectable at 2 h but only in the  OCTOBER 11, 2013 • VOLUME 288 • NUMBER 41 large and small intestines and no other tissue, suggesting that it is being cleared and is not systemically distributed. Intraperitoneal administration of SMTP-TAMRA conjugates also gave rise to a high concentration in the large intestine at 2 h, but unlike the free TAMRA, it was also found at Ͼ200 nM in multiple other tissues. When administered by intraperitoneal injection, TP2D is unique in that its intestinal concentration was relatively low, and its concentration in other tissues was much higher.

Spontaneous Membrane-translocating Peptides
Intriguingly, in some cases (e.g. TP2D-TAMRA intravenously or intraperitoneally), we observed a high concentration of the cargo in whole brain tissue. However, until we perform much additional work, we cannot say with certainty whether the cargo had crossed the blood brain barrier or rather was being preferentially deposited in the brain vasculature.

DISCUSSION
Since the discovery and characterization of the cell penetrating capabilities of the cationic Tat sequence more than 20 years ago, there has been excitement about the possibilities for the use of CPPs as a generic delivery platform for polar molecules (10,12,13,15,20,26,40). Accordingly, there has been a significant amount of work done to engineer and modify the known CPPs to improve their utility and to explore the mechanistic principles of cell penetration. Because the highly cationic CPPs enter cells mostly by endocytosis, a significant effort has also been made to identify alternate classes of peptides that enter cells without requiring endocytosis or that enter cells by endocytosis followed by rapid lysis of endosomes. This is especially important for cargos, such as bioactive peptides, that are highly sensitive to lysosomal degradation. The state of the art for discovery or engineering of novel classes of CPPs is 1) to identify peptides that occur naturally (12,20,40); 2) to modify known sequences (40), including membrane-permeabilizing sequences (14,45); 3) to design new sequences from first principles (12); 4) to design sequences in silico using pattern recognition algorithms (47); or 5) to identify sequences that enter cells using high throughput approaches (48). A weakness inherent in most of these techniques is that the mechanism of entry of CPPs into cells cannot be specifically selected in advance because we do not know the molecular details well enough. As a result, many new CPPs, which are modeled in some way after known sequences, have similar mechanisms of cell entry.
Spontaneous Membrane-translocating Peptides-In our recent work, we have circumvented this weakness by identifying peptides specifically based on their ability to translocate spontaneously across synthetic lipid membranes without permeabilization. We hypothesized that such peptides will also deliver cargos spontaneously across cell membranes. To identify such peptides and test the hypothesis, we subjected a cationic/hydrophobic combinatorial peptide library to a high throughput screen in which we selected for soluble peptides that translocate rapidly across synthetic lipid bilayers without any measurable bilayer disruption (28).
In the work reported here, we characterized the entry of these SMTPs into living cells. Their behavior in synthetic bilayers was fully recapitulated in living cells. The SMTPs delivered several different membrane-impermeant cargos into cells without lysis or toxicity. Their mechanism of entry into cells was very different from that of the highly cationic CPPs. Although endocytosis was required for the entry of the cationic CPP Arg 9 into living cells, we observed little or no evidence that endocytosis was involved in the entry of the SMTPs and their polar cargos into cells even under conditions where endocytosis was not inhibited. Furthermore, SMTPs had no effect on endocytosis or endosome morphology, further indicating that they do not interact with the common endosomal pathways. Taken together, these results show that our high throughput approach to discovery of SMTPs has general utility in the design and discovery of novel peptide delivery platforms.
Mechanism of CPP Entry-In cells under endocytosis permissive conditions, highly cationic CPPs trigger endocytosis  Fig. 6 is shown. Images of the same field were collected approximately every 2 min after peptide addition under identical conditions. The inside/outside ratio was calculated for multiple cells as shown in Fig. 6 probably through a membrane damage response mechanism (49). We show this effect very clearly in Fig. 2 and in supplemental Figs. S1-S7. At low M concentration of Arg 9 -TAMRA, which is non-toxic and non-lytic, the peptide triggered large scale endocytosis in CHO cells and A431 cells. At timescales starting at 2 min and lasting up to at least 1 h, the TAMRA cargo was found concentrated in small intracellular organelles that also contained fluorescent dextran taken up from the medium. Arg 9 -TAMRA did not significantly escape the endosomal pathway on these timescales (see supplemental Figs. S1-S7). This is consistent with reports that show that endosomal release of CPP-cargo conjugates is sometimes very inefficient, one factor that currently limits their utility for macromolecular cargo delivery (19,25,29,30). Another factor that can limit the utility of cationic CPPs is that they do not enter cells that are not capable of endocytosis.
Despite the endocytosis-dependent entry of cationic CPPs into living cells, there are also reports of their spontaneous translocation across synthetic bilayers, suggesting that spontaneous translocation should also occur in cells. These reports have fueled a significant and longstanding controversy in the field. However, most of the cationic CPPs that "translocate" across synthetic bilayers only do so at a high peptide/lipid ratio (Ͼ1:100) or under conditions of strong membrane binding. Thus, at least some reports of "translocation" are probably artifacts that arise from cooperative, peptide-induced bilayer disruption, membrane fusion, permeabilization, or vesicle aggregation.
Such translocation-mimicking mechanisms are driven by strong partitioning of charged and/or amphipathic peptides into membrane interfaces. For this reason, anionic bilayers are usually required for the apparent translocation of cationic CPPs. Peptides with such interfacial activity (50) are unlikely to  be useful as drug delivery platforms because they will only function at high concentration, because of their potential for lysis and toxicity, and because of their poor pharmacokinetics (38). Furthermore, translocation across highly anionic bilayers at high cationic peptide concentration is not expected to predict translocation across the zwitterionic phosphatidylcholine, cholesterol-, and sphingomyelin-rich plasma membranes of mammalian cells at low peptide concentration (although it may predict the peptide-induced lysis of the anionic membranes of late endosomes).
Mechanism of SMTP Entry-The SMTPs we describe in this work are unique and behaved very differently than the highly cationic CPPs both in cells and in synthetic bilayers. Unlike most other reported translocating peptides, these peptides translocated across synthetic bilayers in a manner than is independent of the peptide concentration down to extremely low peptide/lipid ratios (1:6000) (28,41), indicating that they translocate as monomers and do not require strong binding or high bound peptide to lipid ratios. These peptides also translocate across zwitterionic phosphatidylcholine bilayers without bilayer disruption (35), and they bind only weakly to bilayers (28,35). The low steady state concentration of bound peptides coupled with an inherently fast translocation rate (28) ensures that they can translocate without causing a measurable disruption of the bilayer integrity (28,35,41). We recently showed using neutron diffraction that a characteristic feature of the SMTPs is that they insert into membranes without causing any change in bilayer structure (35). This is unusual for membrane-active peptides. The conserved spacing of the arginine residues within the LRLLR motif may allow for their interaction with lipid phosphates in the bilayer interface, whereas the hydrophobic residues promote insertion into the bilayer. These same factors are also critical for the function of the arginine-rich, membrane-embedded S4 sensor helices of voltagegated potassium channels (41), which have the same motif concatenated three times. In support of this idea, we note that the observed negative peptide, ONEG, which does not translocate, has a very similar amino acid composition as the SMTPs but has two adjacent arginines instead of the LRLLR motif. Furthermore, the designed negative peptide, DNEG, which has an identical hydrophobicity and polar/nonpolar pattern as TP3 (51) but without the cationic arginines, also does not translocate.
Effect on Cargo Distribution in Vivo-In a preliminary mouse model experiment of long term cargo fate in vivo, the unique membrane translocating capability of the SMTPs was also evident. Although the membrane-impermeant free TAMRA was efficiently cleared and was not found in most tissues, when it was attached to SMTPs, TAMRA was found in tissues, including muscle, heart, and even brain, suggesting a very significant improvement in cargo biodistribution. The use of D-amino acid SMTPs reduced the amount of TAMRA in clearance pathways and led to a further improvement in the systemic distribution of cargo. Although these results are preliminary, they are similar to published measurements of biodistribution of dye-labeled cell-penetrating peptides (37,38,40,52). Thus, these data strongly support the idea that the SMTPs we have discovered here can deliver membrane-impermeant cargos across cell barriers in living animals. More extensive characterization of the in vivo fate, tissue localization, and kinetics of SMTP-TAMRA biodistribution is currently underway.
Utility of Spontaneous Membrane-translocating Peptides-Peptide families with the properties of the SMTPs described here could have significant utility as delivery platforms for membrane-impermeant compounds. For example, in vitro, there is a vast array of small molecules that could be useful if delivered to cells. These include various inhibitors, activators, and other druglike molecules, bioactive peptides, metabolites or metabolite analogs, imaging agents, and more. SMTPs would also be very useful in vivo to deliver potential drugs to tissues where they are needed. Many biochemically active small molecules are rejected as drugs or become drugs with significant side effects due to poor pharmacokinetics often caused by the lack of spontaneous translocation across membranes. We speculate that the SMTPs described here will have a unique utility for increasing the efficiency of systemic in vivo delivery of such polar molecules to tissues. We envision a mechanism that is similar to that of almost all successful small molecule drugs: systemic drugs cross endothelial layers because they translocate across membranes and enter tissues and cells indiscriminately. By promoting the systemic delivery of small molecules that are membrane-impermeant in free form, the SMTPs described here have the potential to transform the field of drug discovery by increasing the efficiency of drug delivery and by expanding the universe of biochemically active small molecules that can be used as laboratory tools or systemic drugs. FIGURE 10. Tissue fate of TAMRA cargo in vivo. Solutions of free TAMRA or peptide-TAMRA conjugates were injected intravenously or intraperitoneally into mice. At 2 h, the mice were anesthetized, exsanguinated, and then sacrificed for tissue harvesting. The TAMRA content of unpreserved, unfixed tissues was measured by fluorescence. Roughly 100 mg of homogenized, protease-digested, and detergent-solubilized tissue was analyzed. Tissues from multiple PBS-injected animals served as background samples. Tissue concentrations of TAMRA are corrected for variations in the actual amount of TAMRA injected and represent a standard injection of 10 nmol of TAMRA. Lg. Int., large intestine; Sm. Int., small intestine.