Label-Free Confocal Raman Mapping of ... - ACS Publications

Feb 5, 2018 - that CPPs may have on cell function, still remain to be fully clarified. In this work, we employed confocal Raman microscopy (CRM) and a...
0 downloads 3 Views 3MB Size
Download PDF
Article Cite This: Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Label-Free Confocal Raman Mapping of Transportan in Melanoma Cells Patrick. J. Cosme,†,§ Jing Ye,†,§ Shalondria Sears,‡ Ewa P. Wojcikiewicz,‡ and Andrew C. Terentis*,† †

Department of Chemistry and Biochemistry and ‡Department of Biomedical Science, Florida Atlantic University, Boca Raton, Florida 33431, United States S Supporting Information *

ABSTRACT: Cell-penetrating peptides (CPPs) are promising vectors for the intracellular delivery of a variety of membrane-impermeable bioactive compounds. The mechanisms by which CPPs cross the cell membrane, and the effects that CPPs may have on cell function, still remain to be fully clarified. In this work, we employed confocal Raman microscopy (CRM) and atomic force microscopy (AFM) to study the infiltration and physiological effects of the amphipathic CPP transportan (Tp) on the metastatic melanoma cell line SK-Mel-2. CRM enabled the detection of label-free Tp within the cells. Raman maps of live cells revealed rapid entry (within 5 min) and widespread distribution of the peptide throughout the cytoplasm and the presence of the peptide within the nucleus after ∼20 min. Principal component analysis of the CRM data collected from Tp-treated and untreated cells showed that Tp Raman bands were not positively correlated with lipid Raman bands, indicating that Tp entered the cells via a nonendocytic mechanism. Analysis of intracellularly recovered Tp by mass spectrometry showed that Tp remained intact in SKMel-2 cells for up to 24 h. The Raman spectroscopic data also showed that, although Tp was predominantly unstructured (random coil) in aqueous solution, it accumulated to high densities within the cells with mostly β-sheet and α-helical structures. AFM was employed to measure the effect of Tp treatment on cell stiffness. These data showed that Tp induced a significant increase in cell stiffness within the first hour of treatment, which was partially abated after 2 h. It is hypothesized that the increase in cell stiffness was the result of cytoskeletal changes triggered by Tp. KEYWORDS: transportan, cell-penetrating peptide, confocal Raman microscopy, atomic force microscopy

■

Transportan (Tp) is a chimeric CPP first synthesized and studied in the mid 1990s.7 Tp is formed by the conjugation of two smaller peptides, a fragment of the neuropeptide galanin forming the N-terminal side and the wasp venom peptide mastoparan forming the C-terminal side, with both fragments joined by a central lysine residue. The full 27 amino acid sequence, GWTLNSAGYLLGKINLKALAALAKKIL, classifies it as a primary amphipathic peptide with a net positive charge.3 Biotinylated Tp (biotin-Tp) was imaged within fixed epithelial cells by fluorescence microscopy employing a fluorescent avidin probe, showing that biotin-Tp was localized in the nucleus and nuclear membrane and also diffusely throughout the cytoplasm after 1 h incubation with 10 μM of peptide at 37 °C.8 In earlier work, biotin-Tp was detectable in the plasma membrane after 1 min, ER and Golgi complex membranes in the next 2−3 min, and in the nuclear membrane after 15 min incubation.9 The speed of cell entry and impotency of endocytosis inhibitors suggested a nonendocytic uptake mechanism.8 On the other

INTRODUCTION Cell-penetrating peptides (CPPs) are peptides that are able to cross the plasma membrane of most cell types with high efficiency and low cytotoxicity. Interest in CPPs stems from their potential therapeutic value, which includes attaching CPPs to a wide variety of bioactive compounds to improve cellular delivery.1−3 Although they have been identified since the 1980s, the structural features that determine CPP function and the mechanisms by which CPPs cross the cell membrane still remain unclear. We now know that a net positive charge is a defining characteristic of many CPPs, as is the ability to cross the cell membrane via both active (endocytic) and passive (nonendocytic) transport mechanisms. Critical parameters that appear to influence the mechanism of uptake include the extracellular peptide concentration, peptide sequence, the cell line and its membrane components, whether cargo is conjugated to the CPP, the nature of the cargo, and the nature of the peptide-cargo link.1,3−6 For some CPPs, an apparent threshold for the extracellular concentration has been observed, generally in the low micromolar range, where endocytic pathways appear to predominate below the concentration threshold, whereas nonendocytic pathways are more active above the threshold.3,4 © XXXX American Chemical Society

Received: Revised: Accepted: Published: A

July 12, 2017 November 26, 2017 February 4, 2018 February 5, 2018 DOI: 10.1021/acs.molpharmaceut.7b00601 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics

VA) and cultured according to previously published procedures.12 For live or fixed cell Raman experiments, the cells were seeded in 35 mm collagen-coated, glass-bottom dishes (MatTek Corp., Ashland, MA) and left to incubate overnight in full medium under 5% CO2 atmosphere at 37 °C. Adherent cells were typically at 50−70% confluency (∼10,000 cells/cm2) following overnight incubation. For fixed cell experiments, the medium was replaced with full medium spiked with Tp at the desired final concentration (10−25 μM) and incubated for 1 h at 37 °C and 5% CO2 atmosphere. Tp-spiked medium was then removed, and the cells were gently washed three times with phosphate-buffered saline (PBS). The cells were then fixed with 2% paraformaldehyde solution in PBS (10 min at RT) before being rinsed twice with PBS, once with Milli-Q water, and dried at RT. For live cell experiments, the complete medium was replaced with medium without serum and phenol red and spiked with Tp at the desired final concentration (10−25 μM). Raman measurements were then performed immediately on adherent cells in the glass-bottom dish at RT. For some live cell experiments (e.g., Figure 2 and Figure S2), Raman measurements were first performed on cells in the glass-bottom dish before adding Tp, and then Tp was added to the medium in the dish carefully (without disturbing the dish) so that Raman measurements could be performed on the same cells following the addition of Tp. Results presented are representative of at least three independent experiments. Raman Spectroscopy. All Raman spectroscopy was performed with a LabRam HR-800-INV Raman microscope system (Horiba Scientific, Edison, NJ) and the 647 nm line from a Beamlok 2060 mixed argon/krypton ion laser (Spectra Physics, Santa Clara, CA). Drop deposition Raman spectra of Tp, and Tp:DPPC samples were obtained by depositing 2 μL of each stock solution onto a #1.5 glass coverslip and drying at RT (∼30 min). Spectra were collected from the deposit under the microscope with a 40× microscope objective, 5 mW laser power, and 2 min acquisition time per point. A line scan of the sample consisting of 12 points was performed, and the raw spectra were averaged to produce the final spectrum. Raman spectra of water and trifluoroethanol (TFE)/water solutions of Tp were acquired using a macro-sampling port attached to the spectrometer as described previously.12 For the present study, solutions of Tp (5 μL, 3−10 mM) were placed in 0.5 mm square i.d. glass capillary tubes (VitroCom, Mountain Lakes, NJ) and sealed at both ends. A 5 cm focal length convex lens was used to focus 200 mW of 647 nm laser light into the sample. Spectral acquisition times were 10 min. Solvent spectra were run under the same conditions for subsequent background correction. Solvent-subtracted Tp Raman spectra were then manually baseline corrected. The results presented are representative of at least two independent sets of measurements. Confocal Raman Mapping. Live and fixed cell Raman map acquisitions were performed with the LabRam HR-800INV Raman microscope system (Horiba Scientific, Edison, NJ) with an Olympus IX-71 inverted microscope and Märzhäuser XY-scanning stage. Data were collected by raster scanning the 647 nm, 3.5 mW laser over a rectangular area of the cell with spectra obtained at 1−3 μm step intervals while employing 200 μm confocal hole diameter and 100 μm spectrometer entrance slit width. The laser was focused with an Olympus 100× UPLFLN phase-contrast oil immersion objective with 1.3 NA. The X-Y and Z spatial resolutions were determined to be ∼1

hand, biotin-Tp complexed with streptavidin-gold nanoparticles or Texas red-labeled avidin was observed to enter HeLa, human Bowes melanoma, and Cos-7 cells via different forms of endocytosis.10,11 Previously, we employed confocal Raman microscopy (CRM) with 13C-labeling to detect penetratin, a cationic peptide, in live melanoma cells.12 The observed rapid entry and widespread distribution of the peptide within cells, as well as the lack of correlation between peptide and lipid Raman signals, indicated that the mechanism of cell entry under the conditions of study was likely nonendocytic. One advantage of the CRM approach over fluorescence microscopy approaches is that CRM does not require the attachment of a fluorophore, which can potentially alter the physicochemical and hence uptake properties of the peptide. Another advantage is that it can provide structural information on the peptide directly within the cell. For example, on the basis of the position of the amide I vibrational band in the Raman spectra, we determined that the secondary structure of penetratin was a mixture of random coil and β-strand in the cytoplasm with the peptide accumulating as β-sheets in the nucleus. This information provides a new perspective and possibly vital clues on the uptake mechanism and intracellular fate of the peptide. In the present study, we employed CRM to follow the uptake of Tp by metastatic melanoma cells (SK-Mel-2s). We achieved this without any labeling of the peptide by monitoring the strong Raman signals of the Trp residue in the peptide. Previous Raman studies of the same cell line revealed that the endogenous Trp levels are intrinsically low in these cells.12,13 Therefore, the presence of Tp inside the cells could be readily detected by way of the Raman bands associated with the Trp residue of the peptide. We used difference spectroscopy and principal component analysis (PCA) independently to resolve the Raman spectrum of Tp from the background cellular Raman spectrum. We then interpreted the Tp Raman spectra to determine the distribution and structure of the peptide inside cells. In addition to CRM, we employed atomic force microscopy (AFM) for the first time to determine the effect of Tp on the stiffness of cells, gaining unique insights on how the uptake of Tp may induce cytoskeletal changes.

■

EXPERIMENTAL SECTION Peptide Samples. Tp with amino acid sequence GWTLNSAGYLLGKINLKALAALAKKIL, amidated at the C-terminus (MW = 2840.5), was either purchased from 21st Century Biochemicals (Marlborough, MA) or synthesized in-house on a protein technology PS3 peptide synthesizer using rink amide MBHA resin as previously described.12 Tp without the Cterminal amide modification (MW = 2841.5) was purchased from AnaSpec (Fremont, CA) and used for some live cell mapping studies (Figure 4 only) and all fixed cell mapping studies. Peptide stock concentrations were determined using an Evolution 60S UV−vis spectrophotometer (Thermo Scientific, Madison, WI) and a 1 cm quartz cuvette. Tp concentration was calculated from the absorbance at 280 nm, assuming an extinction coefficient of 5,500 M−1 cm−1. Sample mixtures of Tp and 1,2-dipalmitoyl-sn-glycero-3-phosphocholine (DPPC, Avanti Polar Lipids, Alabaster, Al) were prepared by mixing 1 mM Tp and 5 or 20 mM DPPC solutions in methanol in a 1:1 volume ratio. Cell Culture and Treatments. Human metastatic melanoma cells (Sk-Mel-2, catalogue number HTB-68) were purchased from American Type Culture Collection (Manassas, B

DOI: 10.1021/acs.molpharmaceut.7b00601 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics

the cell, Κ, and the angle formed by the plane of the surface and the indenter (AFM cantilever), θ, as K 4 F= α2 2(1 − v 2) πtan θ

and 5 μm, respectively, based on scans performed across standardized, micrometer-sized polystyrene beads.13 Cell Viability Test. At the end of each live cell Raman mapping experiment, medium was removed; cells were washed gently with PBS, and then Trypan Blue solution was added to the dish to test for the viability of cells. Uptake of Trypan Blue by the cells was visually assessed with bright-field microscopy. Overall, cell viability was typically 90% for cells treated with 25 μM of Tp after 2 h at RT on the microscope stage. MALDI-TOF MS of Intracellular Transportan. Intracellular Tp was recovered and analyzed by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) as described in detail previously.14 Briefly, adherent Sk-Mel-2 cells were incubated with 25 μM Tp for 3, 6, or 12 h in full medium at 37 °C and 5% CO2 atmosphere. Control cells were incubated under identical conditions without the addition of Tp. Cells were then washed, detached with trypsin-EDTA solution, lysed with 0.1% HCl solution, and centrifuged, and the resulting supernatant (“post-lysis supernatant”) was desalted using a C18 ZipTip pipet tip (EMD Millipore, Darmstadt, Germany). The desalted supernatant was immediately prepared for MALDI-TOF MS by mixing 1 μL of supernatant with 1 μL of α-cyano-4-hydroxycinnamic acid matrix solution in a microcentrifuge tube. One microliter of the mixture was then spotted on the MALDI plate and allowed to dry. In addition to the treatment and control cell samples, 1 μL of pure α-cyano-4-hydroxycinnamic acid matrix solution and 1 μL of 50:50 authentic Tp solution mixed with matrix were also spotted on the MALDI plate and analyzed as controls. MALDITOF MS was performed in positive reflector mode using a Voyager MALDI-TOF mass spectrometer (Applied Biosystems, Foster City, CA). Raman Data Processing. Raman spectrum background and baseline corrections were performed manually within OriginPro 2016 (OriginLab Corp., Northampton, MA). The amide I band fitting procedure, based on the method of Maiti et al.15 and detailed in our previous publication,14 involved nonlinear least-squares fitting of multiple Voigt functions within OriginPro 2016. Raman spectra from Raman maps were batchprocessed using COBRA,16 a background removal program written for MATLAB that utilizes a wavelet transform. Background-corrected spectra were mean centered and normalized prior to the principal component analysis (PCA) using MATLAB R2013a (Mathworks, Natick, MA).13 Atomic Force Microscopy. For AFM measurements, adherent SK-Mel-2 cells were cultured and prepared in 35 mm glass-bottom dishes as described above for the Raman mapping studies. All measurements were performed on live cells in full medium spiked with 10 μM Tp or medium alone (control). AFM measurements were performed immediately after treatment (0 h) and 1 and 2 h after treatment. Cell stiffness was tested by lowering an untreated MLCT AFM cantilever (Bruker, Billerica, MA) onto the surface of a cell at a constant rate of 5 μm/s and exerting a force against the cell. Measurements were performed on a minimum of five different cells at each time point, and 100 force−indentation curves were acquired for each cell. Experimental cantilever spring constants were calculated using the method described by Hutter and Bechhoefer and ranged from 25 to 35 pN/nm.17 Young’s modulus was calculated by fitting a modification of the Hertz model to the force−indentation curves.18 In this model, the force−indentation relation is a function of Young’s modulus of

The indenter angle, θ, and the Poisson ratio, ν, are assumed to be 55° and 0.5, respectively, and α is the indentation and F is the force. Least squares analysis of the force−indentation curves was performed using Igor Pro (WaveMetrics Inc.) to obtain the Young’s modulus. Statistical analyses of the data were performed using OriginPro 2016 statistical tools.

■

RESULTS Raman Spectrum of Transportan. The Raman spectrum of Tp is shown in Figure 1 with the top spectrum collected

Figure 1. Raman spectrum of Tp recorded from a dry deposited sample (top spectrum) and as a 10 mM solution in water (bottom spectrum). Key: CH = CH2 and CH3 bending modes; A1 and A3 = amide I and amide III modes; νCC = aliphatic carbon bond stretches; W and Y = tryptophan and tyrosine ring modes; Sv = solvent peak.

from a dry deposit of the peptide and the bottom spectrum from a 10 mM aqueous solution of the peptide. All of the bands in the spectrum were readily assigned to specific peptide vibrational modes or functional groups from the results of previous peptide and protein Raman studies. The strongest band in the spectrum, with partially resolved peaks at 1437 and 1451 cm−1, is due to −CH2 and −CH3 group deformations.19,20 Several strong bands visible throughout the spectrum are associated with the aromatic ring vibrations of Trp (e.g., 758, 1012, 1339, 1552 cm−1) and Tyr (e.g., 836/856, 1339, 1618 cm−1).14,19,21−23 Other strong bands at 903, 957, 1106, 1129, and 1153 cm−1 are assigned to aliphatic carbon−carbon stretching (νCC) modes.15,22,24−26 Although νCC bands are generally weak in Raman spectra, they appear prominently in the Raman spectrum of Tp due to the high abundance of aliphatic side chain-containing Ala, Lys, and Leu residues in the sequence. Also featured prominently in the spectrum are bands associated with the amide I and amide III vibrational modes at 1660 and ∼1245 cm−1, respectively. The positions of the amide I and III bands can be used to assign the secondary structures of proteins and peptides.14,15,27,28 The amide I peaks at 1660 cm−1 in the drop deposition spectrum and 1656 cm−1 C

DOI: 10.1021/acs.molpharmaceut.7b00601 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics

absence (or slightly negative intensity) of the 1003 cm−1 band in the difference spectrum. The difference spectrum in Figure 2 matches almost perfectly the spectra of pure Tp in Figure 1 with no missing or additional bands evident. The difference spectrum does, however, display some differences in relative band intensities and amide I and III band shapes compared to the spectra in Figure 1, which are attributed to contrasting Tp secondary structures and local environments within living cells. PCA was employed as an alternative method of extracting the Tp spectrum from the background cell Raman signals (Figure 3). PCA, which is a multidimensional data analysis technique,13

in the aqueous solution spectrum are indicative of significant βsheet and α-helical structural components, respectively. A detailed analysis of the amide I band profiles for these and other Raman spectra will be presented later. Live Cell Raman Mapping. CRM was employed to detect the presence of unlabeled Tp within single, living melanoma cells. For a typical experiment, a small region (∼5 × 5 μm) inside of either the cytoplasm or the nucleus of an adherent cell was chosen to perform a laser scan. A scan was first performed before the addition of the peptide and repeated across the same region of the cell a certain time after the addition of the peptide. Figure 2 shows the results of a cytoplasm scan,

Figure 2. Tp in the cytoplasm of a live SK-Mel-2 cell. (A) Phase contrast image of the cell captured with a 100× phase contrast objective. (B) Averaged Raman spectra collected from the scan area indicated by the small box shown in (A). Spectra were recorded either ∼60 min after treatment with 25 μM Tp (center spectrum) or before treatment (bottom spectrum). The difference spectrum, treatment− control, is shown as the top spectrum. Scan parameters were 5 × 6 μm scan area, 1.0 μm step-size, 2 × 10 s dwell-time per map point.

Figure 3. PCA of the Raman spectra collected in the experiment shown in Figure 2. The scores for the first two PCs are shown in (A) as a scatter plot with mean scores (solid symbols) and SEMs of the scores (crossed bars) for the control and treatment groups also shown. The means of the PC1 (but not PC2) scores for the two groups are significantly different at the 0.05 level using the 2- sample t test. Loads for the first two PCs are presented in (B).

detecting the presence of Tp 60 min after the addition of 25 μM of Tp. Panel A in Figure 2 shows a phase contrast image of the cell and an outline of the region scanned by the laser. Panel B shows the average of 30 Raman spectra collected within the 5 × 6 μm scan region before Tp addition (bottom spectrum) and the averaged spectrum collected after Tp addition (middle spectrum). The appearance of new bands that can be attributed to Tp is clearly evident in the middle spectrum. The difference of the two Raman spectra, treatment minus control, is shown in Panel B of Figure 2, top spectrum. The difference spectrum was calculated by first scaling the intensities of the treatment and control spectra so that they had matching intensities for the phenylalanine ring breathing mode appearing at 1003 cm−1. This intense, narrow band appears ubiquitously in the Raman spectra collected from cells but does not appear in the Raman spectrum of Tp because it does not have a Phe residue (Figure 1). Thus, a good spectral subtraction is confirmed by the

enables other Raman signal changes that may accompany the appearance of Tp in cells to be determined. Panel A in Figure 3 shows a scatter plot of the scores for the first two principal components (PCs). The first principal component (PC1) separates the Raman spectra collected from the cell before Tp addition (control) and after Tp addition (treatment). PC1 encapsulates 30.5% of the total variance in the data. A plot of the PC1 loads (Figure 3, Panel B) shows that the main Raman spectral features separating the control from the treatment spectra are the Raman bands associated with Tp (c.f., Figure 1). The second principal component (PC2) encapsulates 6.7% of the data variance and displays Raman bands associated mainly with lipids (i.e., 715, 1083, 1270, 1304, 1437, and 1655 cm−1).13 In the cytoplasm of cells, point-to-point differences in the local D

DOI: 10.1021/acs.molpharmaceut.7b00601 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics concentration of lipids constitutes the major biochemical variance in the Raman maps (other than the presence/absence of Tp) due to the heterogeneous distribution of lipid vesicles. Note that the treatment and control spectral groups are not significantly separated from each other with respect to PC2 (Figure 3, Panel A), indicating that the distribution of lipid vesicles was the same for both the Tp-treated and untreated cells. Tp was detectable in the cytoplasm of cells within minutes of peptide addition, which is consistent with previous fluorescence microscopy studies of Tp.9 Tp was not detectable in the nucleus of cells in the present study until ∼20 min after peptide addition. Figure S1 shows a scan that was performed across an area of a cell spanning both the cytoplasm and the nucleus, which began immediately after peptide treatment and was completed ∼20 min after initial treatment. The image shows the accumulation of Tp in the cytoplasm up to the nuclear membrane and the almost complete absence of Tp from the nucleus. After ∼20 min, significant quantities of Tp were detected in the nuclei of cells. Figure S2 shows the results of a scan performed within the nucleus of a cell 50 min after the addition of 25 μM Tp. The Raman spectrum of Tp in the nucleus was calculated from the difference between intensitynormalized treatment and control spectra (Figure S2, panel B). The resultant spectrum is very similar to the one derived for Tp in the cytoplasm (Figure 2), though there are some differences in the amide I and amide III band regions. PCA for these nuclear data confirms that the separation of control and treatment spectra is due to the Tp Raman spectral features, shown as PC1 in Figure S3. PC2 in this case consists mainly of Raman bands associated with nucleic acids, i.e., DNA and RNA (e.g., 783, 810, 1249, 1319, 1481, 1570 cm−1).13 Thus, as expected, the main point-to-point variance of the Raman spectral data within the nucleus of cells for both the control and treatment groups is the intensity of DNA/RNA Raman signals. Raman maps were also collected from live cells over larger areas to better visualize the overall distribution of the peptide throughout the cells and how this evolves with time (Figure 4). The size of Raman maps of live cells is limited by the length of time that it takes to collect a map and the ability of live cells to remain viable during that time. In the present case, we collected maps with areas 180−405 μm2 and laser dwell times of 15 s with low laser power (3.5 mW), limiting map acquisition times to less than 15 min to ensure cell viability. The Raman maps were generated by calculating the ratio of intensities (i.e., integrated peak areas) for the 1012 cm−1 tryptophan and 1003 cm−1 phenylalanine bands. A baseline value for this ratio was determined from multiple scans performed on untreated cells (Table S1). For treated cells, any spectrum in a Raman map that had a peak ratio exceeding the baseline value was ascribed a nonzero intensity in the map, representing the presence of Tp. The intensity of red color in the map is proportional to the magnitude of the peak area ratio (above the baseline value). The first scan (Figure 4, panel A) was initiated immediately after peptide addition and detected almost no peptide in the cell. The second scan that was initiated 45 min after peptide addition (panel B) shows a large amount of Tp in the cell, which was detected at almost every point in the Raman map with somewhat higher concentrations in the outer reaches of the cytoplasm. Subsequent scans of different cells at later time points (panels C−I) show more or less the same pattern of widely distributed peptide across the cytoplasm and nucleus of the cells, though with generally higher concentrations within

Figure 4. Raman maps of Tp in live SK-Mel-2 cells. The rectangular false-color Raman maps of Tp (red) are superposed on the phase contrast images of the cells (100× objective). The intensity of red, proportional to the concentration of Tp, is based on the ratio of the areas of the 1012 and 1003 cm−1 peaks in the Raman spectrum recorded at each point in the map. Each map was recorded sequentially, starting from (A) and then every ∼45 min thereafter, finishing with (I). The peptide concentration was 20 μM, and the mapping parameters were 3.0 μm (A−C), 2.5 μm (E, F), or 2.0 μm (all others) step size and 15 s dwell time. Scale bar, bottom left in each micrograph, indicates 10 μm.

the cytoplasm. PCA was performed on all of the Raman spectra collected for the maps shown in Figure 4, forming a data set consisting of maps from 9 treated cells and 4 untreated cells (∼500 Raman spectra in total). A scatter plot of the scores for PC1 and PC2 is presented in panel A of Figure S4 showing significant separation between treatment and control clusters. The means of the PC1 and PC2 scores for the two groups were significantly different at the 0.05 level using the 2-sample t test. A plot of the loads (Figure S4, panel B) shows that PC1 is comprised of mainly Tp Raman bands (negative peaks in the loads plot) and lipid bands (positive peaks). PC2 displays mainly Tp bands (negative peaks) and some protein and lipid bands (positive peaks).13 Recovery of Transportan from Live Cells. CPPs, like most peptides and proteins, are susceptible to proteolytic degradation, which can have a significant influence on the cellular processing and therapeutic potential of the CPP. We therefore studied whether Tp remains intact in live cells or whether it is degraded intracellularly after uptake. Tp was incubated with live SK-Mel-2 cells under the same conditions as were employed for the CRM experiments. Tp was incubated with cells for 3, 6, and 24 h and then extracted from the cells using established procedures.29 Tp stock solutions and postlysis supernatants of Tp-treated and untreated cells were analyzed with MALDI-TOF MS (Figure S5). Tp stock sample used for the cell experiments displayed an observed mass of 2844.7 Da (theoretical 2840.5 Da; Figure S5, panel A). Postlysis supernatants of cells taken at 3 h (not shown), 6 h (Figure S5, panel B), and 24 h (not shown), all gave a similar mass spectrum to that of the Tp stock solution with a major peak at 2845.4 Da. Significantly, the mass spectrum of postlysis supernatants taken from untreated cells did not display the same peak (Figure S5, panel C). Together, the data show that SK-Mel-2 cells under the conditions of study did not degrade Tp intracellularly within the first 24 h of incubation. Fixed Cell Mapping. SK-Mel-2 cells were treated with 20 μM of Tp and incubated for 1 h at 37 °C before washing and fixing. Large area Raman maps were then collected from 11 E

DOI: 10.1021/acs.molpharmaceut.7b00601 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics

constituting the amide I band at the end of the fit provides a quantitative (though not highly precise) estimate of the relative contribution of each of the secondary structure components. Although β-sheet and random coil conformations occur in the same wavenumber range, they may be differentiated by their typically narrower (β-sheet) and broader (random coil) band widths. β-Strand and PPII conformations also overlap in wavenumber range but cannot be distinguished on the basis of bandwidth. In the case of Tp, β-strand is more likely than PPII due to the absence of proline residues in its sequence and propensity to form β-sheets (vide infra). β-Turns are not considered in the analysis because they typically constitute only a small proportion of the overall secondary structure in proteins and peptides. Fitting results for solvent-corrected Tp Raman spectra in water and water:TFE (1:1) are shown in Figure 6 (panels A

different cells (Figure S6). These maps show that the rate of Tp uptake was variable from cell to cell. Some maps show a large amount of Tp distributed throughout the cell with the highest points of concentration being in the cytoplasm (e.g., cells E, F, and H), whereas other cells contained relatively little Tp, which is distributed in a punctate fashion mainly in the cytoplasm (e.g., cells B, I, and K). AFM Cell Stiffness Measurements. AFM was employed to test the effect of Tp on the overall stiffness of cells. This was achieved by lowering a cantilever onto the surface of cells at a constant rate to acquire force indentation curves, which were then fitted to obtain the Young’s modulus of the cell. Multiple force indentation curves were acquired for multiple cells either untreated (control) or treated with 10 μM Tp. The data are tabulated in Table S2 and graphed in Figure 5. The data show a

Figure 5. Bar graph of AFM cell stiffness measurement results that compare the average Young’s modulus for Tp-treated and untreated (control) SK-Mel-2 cells. Error bars represent the standard error of the mean. Figure 6. Fitting of the amide I band profile and adjacent Raman bands in the Raman spectrum of Tp in (A) water, (B) 1:1 water:TFE, (C) in the cytoplasm of a live cell, and (D) within the nucleus of a live cell. In each panel, the red curve is the experimental data; the dashed black curves are the individual Voigt functions representing each peak or amide band component, and the solid black curve represents the cumulative resultant of the fit.

significant increase in cell stiffness 1 h after treatment with Tp, which then begins to decrease at 2 h. One-way ANOVA of the four data groups determined that the mean modulus of elasticity was significantly different at the 0.05 level for all group comparisons except for the control versus + Tp (0 h) pairing. Nonparametric Kruskal−Wallis ANOVA also determined that the group populations were significantly different at the 0.05 level. Secondary Structure of Transportan. It is possible to estimate the secondary structure composition of proteins and peptides by modeling the amide I band profiles in IR and Raman spectra.30,31 For example, Maiti et al. modeled the amide I band profiles in Raman spectra obtained for αsynuclein in different environments.15 We previously employed a similar fitting procedure to characterize the secondary structures of penetratin,14 and we employ the same procedure here to characterize the structure of Tp under a variety of conditions. The modeling procedure assumes that the amide I band has three major secondary structure contributions: α-helix (constrained to 1650−1656 cm−1), β-sheet and random coil (1664−1670 cm−1), and β-strand and PPII (1674−1685 cm−1). A detailed justification for why these wavenumber ranges were chosen to represent each structure type can be found in previous publications.12,14,15,19,27,28,31,32 Voigt profiles are assumed for each of the three amide I band components as well as adjacent Raman bands. Nonlinear least-squares fitting of the spectrum is performed by allowing each of the Voigt functions to vary in peak position within their respective wavenumber range constraints as well as peak width and peak area. The relative area of each of the three Voigt profiles

and B). In pure water, the secondary structure of Tp was 76% random coil and 24% α-helix. In water:TFE, the proportion of α-helix increased to 38% and random coil decreased to 62% (Table 1). The large band widths for the second Voigt component in each fit (49 and 44 cm−1) indicate that this component is random coil rather than β-sheet (Table 1). A fit of the amide I region for the drop deposition sample resulted in α-helix, β-sheet, and β-strand components being implicated in relative proportions of 46, 29, and 25%, respectively (Figure S7 and Table 1). The amide I regions for Tp in the cytoplasm and nucleus of cells were also fitted. Using the spectra obtained from mathematical subtraction of the cell background (i.e., spectra from Figure 2 and Figure S2), the fits indicated that Tp contained significant proportions of both α-helix and β-sheet with β-sheet being more strongly implicated in the nucleus, whereas Tp in the cytoplasm possessed a small β-strand component (Figure 6, panels C and D, and Table 1). When the spectra for Tp in the cytoplasm and nucleus derived from PCA were fitted (i.e., PC1 in Figure 3 and Figure S3), the results were similar though the β-sheet proportions were higher and the widths of the Voigt components markedly lower when F

DOI: 10.1021/acs.molpharmaceut.7b00601 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics

Table 1. Results of Multi-Voigt Peak Fitting of the Amide I Band in the Raman Spectrum of Tp Measured under Different Solvent Conditions and in Different Regions of Live Melanoma Cells peak positiona/widthb/percent contributionc environment water water:TFE (1:1) drop deposition cytoplasm nucleus

α-helix 1653/22/24 1652/20/38 1655/31/46 1657/21/58 1655/20/59 1658/21/52 1652/14/38

β-sheet or unstructured 1665/49/76 1664/44/62 1664/24/29 1672/20/39 1668/11/41 1669/27/48 1664/18/62

(diff)d (PCA)e (diff) (PCA)

(diff) (PCA) (diff) (PCA)

β-strand or PPII

1682/28/25 1692/19/4

Wavenumber (cm−1). bFwhm (cm−1) of either the Gaussian or Lorentzian component of the Voigt function, whichever was the larger of the two. Calculated from the area of the corresponding Voigt peak divided by the sum of the areas of all three Voigt peaks derived from the nonlinear leastsquares fit of the amide I band profile. dDiff = difference; the results of fitting the amide I band for Tp from the difference spectrum. eResults of fitting the amide I band for Tp from the PCA data. a c

such as Tat and oligoarginines, which are efficiently delivered to the cytosol via transduction.34,35 Endocytosed MAP was detected in the nucleus in less than 10 min, leveling off after ∼1 h. In contrast, YG(R)9 was internalized into the cytoplasm by transduction but only began to accumulate in the nucleus after 1 h. Furthermore, Chaumet et al. reported a novel endosomal route for the delivery of intact Pseudomonas exotoxin A (PE) from the cell surface to the nucleoplasm via nuclear envelope-associated endosomes.36 Although PE (∼69 kDa) is much larger than typical CPPs, it is conceivable that such a direct endosomal route to the nucleus could be available to CPPs as well. However, this route was not likely significant for Tp under the conditions studied because Tp was consistently observed in the cytoplasm before it appeared in the nucleus, and Tp levels in the nucleus always remained less than those in the cytoplasm. We speculate that the ∼20 min lag time for Tp nuclear entry could be the result of a reduced rate of diffusion of Tp within the cytoplasm, possibly due to its aggregation or binding to other cytosolic proteins. The reason that Tp levels in the nucleus did not reach as high as those in the cytoplasm may be attributable to residues 16−21 in the Tp sequence (LKALAA), which tested positively as a nuclear export sequence.37 Our study of live cells revealed that Tp was rapidly detected in most but not all cells scanned. This was the case even for cells in the same cell culture dish probed within minutes of each other, indicating that there is some condition of the cell (yet to be determined) that is required for rapid Tp entry to occur. Our fixed cell studies employing a pulse-chase approach provided further insights (Figure S6). These Raman maps were collected following 20 μM Tp treatment for 1 h, washing, and fixation. The maps show a high degree of variability in both peptide content and distribution from cell to cell. In contrast to the live cell maps, some of the fixed cell maps show a distinctly punctate peptide distribution (e.g., maps B, I, and K) more consistent with an endocytic mode of uptake or some other form of localized membrane barrier crossing. One possible reason for the different appearance of the live and fixed cell Raman maps is that whereas the fixed cells were incubated with Tp for only 1 h prior to washing and fixation, most of the live cells were incubated with Tp for much longer (up to 5 h) prior to Raman mapping. Consequently, the fixed cells contained much less Tp, whose local concentration was less than the detection threshold at many points within the maps, leading to a more punctate appearance for some of the fixed cell maps. Another factor is that active diffusion of Tp in live cells, either

compared to the corresponding spectral subtraction data (Figure S8 and Table 1). Raman Spectra of Transportan-DPPC Mixtures. To estimate the possible appearance of Raman spectra of Tp if it were entrapped in endosomes, we prepared 1:20 and 1:5 mol ratio mixtures of Tp with phospholipid 1,2-dipalmitoyl-snglycero-3-phosphocholine (DPPC). The Raman spectrum of pure DPPC displayed several strong Raman bands characteristic of those observed for lipid vesicles in live cells, such as the 715, 1061, 1128, 1295, and 1438 cm−1 bands.13 For the 1:20 Tp:DPPC mixture, the Raman spectrum was dominated by lipid Raman features. Only at the mole ratio of 1:5 were Tp Raman bands visible with lipid Raman bands in the Raman spectrum (Figure S9).

■

DISCUSSION Here, we studied the uptake of unlabeled Tp in a metastatic melanoma cell line (SK-Mel-2) by employing CRM with live and fixed cells. The use of unlabeled peptide avoids any possible artifacts associated with an attached fluorophore. CRM employing a scanning, focused laser with confocal optical detection ensures that the detected peptide occupies the volume within cells and is not membrane bound. Our experiments employed extracellular peptide concentrations in the range 10−25 μM, which are slightly higher than those employed for fluorescence microscopy studies.11 On the basis of previous CPP studies, these higher concentrations might be expected to favor nonendocytic modes of cell entry.2,33 Indeed, we observed cell entry within a few minutes of Tp treatment, detecting it first in the cytoplasm (Figure S1) and then within the nucleus after ∼20 min. From ∼60 min onward, the peptide was detectable at almost all points probed within the cells, though not uniform in concentration, and generally more concentrated in the cytoplasm than in the nucleus (Figure 4). The observed rapid entry and dispersed (i.e., nonpunctate) appearance of the peptide in the cytoplasm of cells are more indicative of a nonendocytic mode of cell entry. This is also supported by the MALDI-TOF MS results showing that Tp remained fully intact within the first 24 h of cell entry (Figure S5), indicating that Tp bypassed or evaded the cell’s endolysosomal system within this time frame. The delayed appearance of Tp in the nuclei of cells is not necessarily indicative of an endocytic mode of cell entry. For example, Zaro et al. reported that highly endocytosed CPPs such as the model amphipathic peptide (MAP) may be more efficiently transported to the nucleus than cationic peptides G

DOI: 10.1021/acs.molpharmaceut.7b00601 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics

also evident by the fact that high quality Raman spectra of Tp were attainable from cells when employing low laser power (5 min) to obtain similar spectral intensities. Previously reported experimental data show that peptide secondary structural traits, such as α-helicity, or secondary structural changes, may play a crucial role in mediating cell entry for some CPPs.1,41−45 For example, β-sheet conformation was critical for efficient entry into endothelial cells by an amphipathic 26-mer peptide,45 whereas a 22-mer CPP derived from the human milk protein lactoferrin exhibited a conformation-dependent uptake efficiency.44 Penetratin, however, displayed no reduction in uptake when helix formation was disrupted by proline insertion.46 It is not currently feasible to record the Raman spectrum of a peptide as it crosses the plasma membrane of a live cell in real time due to the spatial resolution and sensitivity limitations of the technique. We have shown here, however, that it is feasible to resolve the Raman spectrum of a peptide and determine its secondary structure within specific compartments of live cells after it has entered. Whether these spectra reflect the actual structure or condition of the peptide as it crosses the cell membrane is not clear. One scenario, conceivable for an amphipathic peptide like Tp, is that high local concentration of the peptide at the cell surface leads to the formation of transmembrane α-helical and/or β-sheet structures, which temporarily affect membrane organization to facilitate cell entry.47,48 If stable, these same peptide structures could be detected inside cells with CRM. It has been documented that the cationic peptides TAT and penetratin, as well as the cationic amphipathic peptide MPG, bind to negatively charged glycosaminoglycans (GAGs) at the surface of cells, causing them to cluster.49 This event triggers the activation of the Rac1 GTPase and remodeling of the actin network made visible by the formation of lamellipodia.50,51 Previously, it was proposed that actin network remodeling may favor CPP entry via macropinocytosis51 or increase membrane fluidity to promote nonendosomal cellular uptake.47,50 The present AFM results show that Tp markedly increased the stiffness of SK-Mel-2 cells within 1 h of treatment, which partially recovered after 2 h (Figure 5). The increase in cell stiffness could be the direct result of actin remodeling in these cells via the GAG-Rac 1 GTPase mechanism just described. Delaroche et al. similarly reported that arginine- and tryptophan-rich peptides could cross the cell membrane and remodel actin cytoskeleton in malignant cells, decreasing cell motility.52 The significance of these findings is that CPPinduced cell stiffness and the resulting reduction in cell motility could be an effective strategy for blocking cancer metastasis.

endosomally captured or otherwise, could have the effect of “blurring” or homogenizing the appearance of the live cell maps compared to the fixed cell maps where diffusion is not possible. A final factor to consider for the Raman maps is that the zresolution of the confocal Raman microscope employed (∼5 μm) is comparable to the depth of the adherent cells (∼10 μm). This could also contribute to a more homogeneous appearance of an actually more punctate distribution of the peptide within the cells, live or fixed. PCA of Raman spectra collected from Tp-treated and untreated cells showed that there was not a positive correlation between peptide and lipid Raman signals (Figure 3 and Figures S3 and S4). In Figure 3, PCA separated the Raman signals of Tp (shown in PC1) from those of lipids (shown in PC2), which are orthogonal PCs and therefore uncorrelated by definition. The mean PC2 scores for both treated and untreated cells were not significantly different, showing that although lipid Raman signals did vary from point to point in the Raman maps, they did so equally in both Tp-treated and untreated cells. Thus, the lipid variability was a natural feature of the cells themselves and not the result of Tp treatment. In Figure S4, Tp and lipid Raman bands appear together in the same PC (i.e., PC1), but lipid bands point positively whereas Tp bands point negatively. In this case, Tp and lipid Raman signals are inversely or negatively correlated because when Tp Raman band intensities are high, lipid Raman band intensities are low, and vice versa. We argue that this is the opposite trend of what would be expected if Tp were predominantly encapsulated in endosomes and that these experimental results provide evidence for a nonendocytic uptake mechanism. In further support, we showed that, when Tp is strongly associated with lipid, it is the lipid Raman signals that are most prominent in the Raman spectrum (Figure S9). In contrast, Tp Raman bands rarely coincided with lipid Raman bands in the Raman spectra collected from Tp-treated cells. To sum the data regarding what they indicate about the mechanism of Tp uptake, the rapid entry and predominantly cytosolic localization of Tp, the lack of positive correlation of Tp with lipid Raman signals (Raman/ PCA data), the full-length status of intracellular Tp after 24 h (MS data), and the rapid (