Research Article Received: 5 June 2009 Accepted: 9 August 2009 Published online in Wiley Interscience: 29 September 2009 (www.interscience.wiley.com) DOI 10.1002/jrs.2483 Confocal Raman microscopy for simultaneous monitoring of partitioning and disordering of tricyclic antidepressants in phospholipid vesicle membranes Christopher B. Foxa and Joel M. Harrisa,b∗ Characterization of drug–membrane interactions is important in order to understand the mechanisms of action of drugs and to design more effective drugs and delivery vehicles. Raman spectra provide compositional and conformational information of drugs and lipid membranes, respectively, allowing membrane disordering effects and drug partitioning to be assessed. Traditional Raman spectroscopy and other widely used bioanalytical techniques such as differential scanning calorimetry (DSC) and nuclear magnetic resonance (NMR) typically require high sample concentrations. Here, we describe how temperature-controlled, optical-trapping confocal Raman microscopy facilitates the analysis of drug–membrane interactions using micromolar concentrations of drug, while avoiding drug depletion from solution by working at even lower lipid concentrations. The potential for confocal Raman microscopy as an effective bioanalytical tool is illustrated using tricyclic antidepressants (TCAs), which are cationic amphiphilic molecules that bind to phospholipid membranes and influence lipid phase transitions. The interaction of these drugs with vesicle membranes of differing head-group charge is investigated while varying the ring and side-chain structure of the drug. Changes in membrane structure are observed in Raman bands that report intra- and intermolecular order versus temperature. The partitioning of drugs into the membrane can also be determined from the Raman scattering intensities. These results demonstrate the usefulness of confocal Raman microscopy for the analysis of drug–membrane systems at biologically relevant drug concentrations. Effective tools for monitoring drug–membrane c 2009 John Wiley & Sons, Ltd. interactions are crucial for rational design of new drugs. Copyright Keywords: confocal Raman microscopy; drug–membrane interactions; tricyclic antidepressants; phospholipid vesicles Introduction 498 Drug–membrane interactions constitute an essential factor in understanding the effects of drugs and improving their design. Drugs interact with membrane assemblies in multiple ways, whether they are being transported through a membrane, binding to proteins in the membrane, acting directly on the phospholipid molecules themselves, or being carried in liposomes. Furthermore, the cooperative nature of protein–membrane interactions means that protein function is strongly influenced by the membrane bilayer phase or structure. Thus, numerous techniques have been employed to elucidate drug–membrane activity and the influence of drugs on lipid phase transition behavior,[1,2] including differential scanning calorimetry (DSC),[3 – 8] vibrational spectroscopy,[5,7 – 9] X-ray diffraction (XRD),[10] nuclear magnetic resonance (NMR),[10,11] fluorescence spectroscopy,[3] and molecular dynamics simulations.[12 – 15] Despite the variety of methods employed to study lipid phase transitions and drug–membrane interactions, there are major drawbacks with many of these methods: for high sensitivity methods such as fluorescence, there is a need for labeling; the influence of the fluorescent probe on the structure and behavior of the system must be considered.[16,17] With less sensitive methods such as DSC, NMR, XRD, or traditional infrared and Raman spectroscopies, high drug and lipid concentrations are required, typically in the range of 10–100 mM.[5,18,19] While Raman scattering is a weak effect, this technique is well suited to study phospholipid structure; the J. Raman Spectrosc. 2010, 41, 498–507 method requires no labeling and is compatible with aqueous buffers, requiring a minimum of sample preparation. When integrated with a confocal microscope, the efficiency of Raman spectroscopy can be greatly improved for investigating small colloidal structures[20 – 22] including lipid vesicles or liposomes.[23] With a focused excitation laser beam in a confocal Raman microscope, it is possible to optically trap individual vesicles so that Raman scattering is collected from a small (<1.5 fl) volume dominated by the trapped vesicle,[23] such that contributions from the surrounding solution are minimized.[24] With modest incident lasers powers of a few milliwatts, the confocal Raman microscope allows lipid vesicles to be trapped indefinitely in the laser focus and their spectra to be acquired over long observation times while conditions are varied. The partitioning of drugs in a lipid membrane can be quantified in this experiment, from the Raman scattering intensity of the drugs that accumulate in the lipid bilayer of the trapped vesicle.[25] Changes in lipid membrane structure due to its interactions with drugs or other small molecules can be characterized by relative Raman peak intensities that respond to ∗ Correspondence to: Joel M. Harris, Department of Chemistry, University of Utah, 315 South 1400 East, Salt Lake City, UT 84112-0850, USA. E-mail: [email protected] a Department of Bioengineering, University of Utah, Salt Lake City, UT, USA b Department of Chemistry, University of Utah, Salt Lake City, UT, USA c 2009 John Wiley & Sons, Ltd. Copyright Confocal Raman microscopy of tricyclic antidepressants in vesicle membranes J. Raman Spectrosc. 2010, 41, 498–507 Materials and Methods Reagents and materials DPPC and DPPG of >99% purity were purchased from Avanti Polar Lipids (Alabaster, AL). Amitriptyline hydrochloride, nortriptyline hydrochloride, and protriptyline hydrochloride were obtained from Sigma (St. Louis, MO). An extruder was purchased from Avanti Polar Lipids. Track-Etch polycarbonate membranes were made by Nucleopore (Pleasanton, CA). Water was quartz-distilled and then filtered with a Barnstead (Boston, MA) NANOpure II and had a minimum resistivity of 18.0 M cm. Buffer components were supplied by Mallinckrodt (Paris, KY). Sample preparation One milligram of phospholipid (Tm 41 ◦ C) in chloroform was pipetted into a 15-ml glass vial and dried in vacuum for at least 3 h to remove the solvent. The film was hydrated with 1 ml of buffer above Tm and stored at ∼3 ◦ C. Lipid solutions were used within 1 week of initial hydration. When removed from the refrigerator for use, lipid solutions were incubated for at least 1 h at a temperature above the Tm . In the case of pure DPPC, vesicles were prepared by extruding the hydrated solution, at temperatures 10–20◦ above the Tm of the phospholipid, 15 times through a polycarbonate filter having pore openings of 1 µm diameter. DPPG forms vesicles spontaneously, so that extrusion was not necessary. The vesicles dispersions were diluted 500-fold into solutions containing 50 mM phosphate buffer (pH 7.2). A 40-µl aliquot of this solution was pipetted into a small-volume copper sample cell having a UV-ozone-cleaned coverslip window adhered to its base.[40 – 46,48,49,51,52] Vesicles collected in the optical trap were either individual or small aggregates of ∼2–8 vesicles; both types of samples were observed, and no structural differences between membranes were detected. Since vesicles containing DPPG were not extruded, a range of vesicle sizes was present. An effort was made to analyze vesicles near 1 µm diameter; however, larger vesicles up to 4 µm in diameter were also trapped when smaller vesicles could not be found. Temperature-dependent Raman microscopy The temperature-dependent Raman microscope setup has been described previously.[37] Briefly, the beam from a 647.1-nm krypton ion laser was passed through a band-pass filter and a 4× beam expander mounted on a Nikon TE 300 inverted fluorescence microscope and then through another band-pass filter. The expanded laser beam was then reflected by a dichroic beam splitter into a 100× oil immersion 1.4 NA microscope objective where it was focused to a 0.6-µm-diameter spot inside the sample cell. The same objective was used to collect the Raman scattering from the sample, which was passed through the beam splitter, then through two high-pass filters, and then focused on to a monochromator with 50 µm slit width. A charge-coupled device (CCD) camera was employed to collect the spectra over three pixels or 78 µm in the vertical dimension. The spectral resolution was 6 cm−1 . Laser power at the sample was ∼23 mW and integration time was 1 min. In general, three samples at each data point were collected; however, some spectra were deleted because of low signal (ambiguous results) or because vesicles were dislodged from the optical trap as a result of air bubbles formed during the temperature ramp. White light correction and baseline fitting of c 2009 John Wiley & Sons, Ltd. Copyright www.interscience.wiley.com/journal/jrs 499 changes in intra- and intermolecular order.[5,7,9,25 – 35] In a confocal Raman microscopy experiment, these structural changes in the membrane can be combined with quantitative information on the degree of drug partitioning,[25] yielding a detailed picture of the drug–membrane interactions. An important advantage of opticaltrapping confocal Raman microscopy for such studies is that membrane effects of very low, clinically relevant concentrations of drug can be investigated. Samples are typically prepared with a few micromolar lipid concentrations, which yield a sufficient number density of vesicles that is relatively easy to locate and trap a vesicle for study. This low lipid concentration and corresponding small membrane volume avoids depleting clinically relevant (µM) drug concentrations through membrane partitioning, which is a significant limitation of the less sensitive methods that require 10 mM or higher lipid concentrations. Lipid membrane assemblies undergo characteristic thermal phase transitions,[30,32,35,36] which are characterized by changes in lipid structural order. Small-molecule drugs most often decrease and widen the main lipid phase transition (Tm ) depending on drug concentration,[37] but may also increase Tm if they enhance lipid packing.[27,38,39] Lipid phase transitions can be detected in confocal Raman microscopy by the use of a temperature-controlled stage and metal sample cell, allowing sample temperatures to be varied from ∼5 to 55 ◦ C while requiring only small sample volumes (<40 µl).[37] While temperaturecontrolled confocal Raman microscopy has been used to study membrane structural changes in phase transitions of individual vesicle membranes prepared from pure synthetic lipids,[35] the technique has not been applied to the study of the effects of drugs on the structure and phase transitions of lipid membranes, where both the drug partitioning as a function of temperature and membrane phase transition can simultaneously be investigated. In this study, the partitioning of tricyclic antidepressants (TCAs) into both zwitterionic and negatively charged phospholipid membranes and their influence on the membrane thermal phase transitions are investigated using optical-trapping confocal Raman microscopy. Tricyclic antidepressants are cationic, amphiphilic, small-molecule drugs; the mechanism of action of TCAs is not fully understood, but it is clear that they bind to phospholipid membranes.[2] TCA–membrane interactions are critical because TCAs are known to induce lipidosis, or intracellular accumulation of lipids.[40 – 49] Literature reports on the activity of TCAs indicate strong but varying membrane interactions depending on the particular drug and lipid structure, including effects on lipid phase transitions.[49,50] The TCAs chosen for this study have structural variations in both the ring and side-chain structure. With similar pKa values ranging from ∼9.4 to 10.0, the drugs have essentially the same charge state (+1) at the experimental pH of 7.2. Variations in their membrane activity are explored in terms of the structural differences of the drugs or variations in lipid headgroup structure. Vesicles composed of 1,2-dipalmitoyl-sn-glycero-3phosphatidylcholine (DPPC) and 1,2-dipalmitoyl-sn-glycero-3[phospho-rac-(1-glycerol)] (DPPG) in solutions containing varying concentrations of TCAs are monitored by confocal Raman microscopy as a function of temperature in order to determine the drug partitioning and disordering behavior. The utility of confocal Raman microscopy for investigating low concentrations of drug in drug–membrane studies is discussed. C. B. Fox and J. M. Harris the Raman spectra with a fourth-order polynomial function were done in Matlab (MathWorks), as described previously.[37] A water-cooled Peltier stage (Technical Video, Ltd.) was integrated with a copper block to surround the sample cell and allow efficient heat transfer.[23] Temperature measurements were collected from a well in the copper block using a thermocouple. Temperature was changed at a rate of approximately 1 ◦ C/min, and reported values were averaged from recorded temperatures after each 1-min signal integration. All peak intensities were obtained from the maximum peak intensity value in the region near the indicated wavenumbers. The peak ratio from the C–H stretching region, where scattering from the terminal methyl C–H symmetric stretching mode (2934 cm−1 ) is ratioed to the scattering of the C–H asymmetric stretching mode (2883 cm−1 ), is plotted versus temperature to monitor the membrane phase transition. A smoothed first derivative of the temperature-dependent ratios is computed with a five-point second-order Savitzky–Golay filter,[37] where the peak of the numerical derivative reports the inflection point of the melting transition. Drug partitioning data were plotted using the intensity of the C C ring stretch of the drug (1600 cm−1 ) ratioed to the C O stretching mode (∼1730 cm−1 ) of the lipid, since the latter was not observed to change significantly with temperature. The scattering intensities of the C C ring stretch of all three drugs, as 200 mM standard solutions prepared in methanol, were measured and compared. The intensities of amitriptyline and nortriptyline were equivalent, while the scattering from protriptyline was threefold greater, likely due to the greater polarizability of the conjugated double bond in the central ring of this compound. The observed partitioning ratio of this drug was divided by 3 to account for its greater Raman scattering cross section. The temperature-dependent drug partitioning ratios were smoothed using a three-point moving average. In general, the ‘below Tm ’ drug partitioning values were taken 6 ◦ C below the measured Tm of the lipid sample and the ‘above Tm ’ drug partitioning ratios were taken 6 ◦ C above the measured Tm , although some samples did not span this entire temperature range. For these cases, the closest data point was used as long as it was below or above the Tm . The peak ratio I1600 /I1730 from control lipid samples was subtracted from the drug-containing samples. Self-modeling curve resolution (SMCR) analysis of temperaturedependent Raman spectra from lipid samples has been described in a previous publication.[53] Briefly, this method projects the Raman spectra into eigenvector space, eliminating eigenvectors that represent random noise, and rotating the minimal basis set of eigenvectors, representing correlation in the data, back into real space by assuming non-negativity in pure component spectra and their temperature profiles. SMCR is used here to compare the pure component spectra representing the lipid fluid phase of control lipid samples and lipid samples containing drug. Results 500 The structures of the tricyclic antidepressant drugs amitriptyline, nortriptyline, and protriptyline, and their octanol/water and liposome/water partition coefficients[35] are shown in Fig. 1. The liposome formulation used to determine the liposome/water partition coefficients consisted of DPPC (46%) : DPPG (24%) : cholesterol (30%).[47,54] The drug structures differ in the number of methyl groups at the end of the aliphatic chain and the location of the C C bond that www.interscience.wiley.com/journal/jrs is found in the center ring or on the aliphatic chain. The structures and phase transition temperatures of the 16-carbon acyl chain lipids DPPC (zwitterionic) and DPPG (anionic) are also presented in Fig. 1. The Raman spectrum of an optically trapped DPPC vesicle with peak assignments is shown in Fig. 2, where it is compared to Raman scattering from pure DPPG and DPPC : DPPG 1 : 1 molar ratio vesicles at room temperature. The Raman scattering from samples of pure DPPC vesicles was typically more intense than that of vesicles of DPPG; there is a greater likelihood of DPPC vesicles being multilamellar (more than one lipid bilayer), while DPPG vesicles are typically unilamellar due to electrostatic repulsion between the charged head groups.[47] Spectra were collected continuously for 1 min integration times as the temperature was slowly changed (∼1 ◦ C/min), and contained information indicative of changes in the lipid structural order. By monitoring changes in peak intensities and wavenumbers of Raman scattering from phospholipid vesicles, differences in acyl chain conformation and the corresponding intra- and intermolecular membrane order can be determined.[5,7,9,26 – 31] Phase changes in phospholipid assemblies can be detected in either the C–C stretching (∼1060–1120 cm−1 ) or the C–H stretching (∼2800–3100 cm−1 ) regions of the spectrum. The methylene C–C stretching region reflects the evolution from trans (∼1065 and 1114 cm−1 ) to gauche (∼1086 cm−1 ) conformations in the lipid acyl chains as a membrane becomes more disordered.[7,31,55] The methylene C–H stretching region also reflects membrane order and consists of three strong bands that represent C–H symmetric stretching (∼2845 cm−1 ), C–H asymmetric stretching (∼2883 cm−1 ), and terminal methyl C–H symmetric stretching (∼2934 cm−1 ). Despite overlap between these bands, the peak intensity ratio for the 2934 to 2883 cm−1 bands correlates with changes in interchain interactions as well as intramolecular conformations as the membrane becomes disordered.[5,9,31] The effect of increasing amounts of nortriptyline on the lipid main phase transition profile of DPPC, DPPG, and DPPC : DPPG 1 : 1 is shown in Fig. 3 by monitoring the change in the C–H stretching peak ratio I2934 /I2883 with temperature. The drug decreases Tm and widens the transition profile width. Similar effects were noticed for the other TCAs, protriptyline and amitriptyline. The Tm value for each sample is taken as the inflection point on the slope of the I2934 /I2883 curve and is calculated using a 5-point Savitzky–Golay filter as described in the Experimental section. In Figs 4 and 5, the effects of the TCAs on lipid Tm values are plotted according to drug concentration. The influence of the drugs on different lipids is apparent. However, when acting on the same lipid, no significant difference is seen between the disordering effects of the three drugs. Table 1 presents the resulting Tm values for comparison. Interestingly, the disordering effects of TCAs, while obviously influencing Tm , do not appear to increase overall lipid disorder once the fluid phase is reached. This is observed by comparing the pure component spectra obtained by SMCR representing the fluid phase DPPC membrane both with and without drug in Fig. 6. The temperature-dependent partitioning of the drugs is apparent in Fig. 7(a), where a Raman spectrum from a vesicle sample in a 500 µM amitriptyline solution at low temperature is compared with one at high temperature. In Fig. 7(b), the Raman scattering peak representing the C–C ring stretch of the drug (1600 cm−1 )[25,56] is ratioed to the C O stretch of the lipid (∼1730 cm−1 ). In DPPC, the drug concentration in the membrane significantly increases near the Tm , revealing that the drug preferentially partitions into the fluid phase membrane. This c 2009 John Wiley & Sons, Ltd. Copyright J. Raman Spectrosc. 2010, 41, 498–507 Confocal Raman microscopy of tricyclic antidepressants in vesicle membranes Figure 1. Structures of the tricyclic antidepressant drugs and phospholipid molecules. Reported drug partition coefficients[47] and lipid phase transition temperatures[47,54] are also shown. 500 C-C stretch (gauche) C-H antisymmetric stretch C-H symmetric stretch CH2 twist CH2 bend C-C stretch (trans) 400 Raman Intensity / 10 PE terminal CH3 symmetric stretch C-C stretch (trans) C-N symmetric stretch DPPC 300 1 min integration C=O stretch 200 DPPG 100 DPPC:DPPG 1:1 0 600 900 1200 1500 1800 2100 2400 2700 3000 Wavenumber / cm-1 Figure 2. Raman spectra and peak assignments of optically trapped phospholipid vesicles. DPPC is compared to DPPG and DPPC : DPPG 1 : 1 molar ratio vesicles. temperature-dependent partitioning is seen in the other lipid samples as well, but it is usually less dramatic and sometimes not apparent at all (Fig. 8). Moreover, the lipid samples containing DPPG generally contain much more drug at all temperatures compared to DPPC, presumably due to electrostatic interactions (see the section on Discussion). Discussion Membrane disordering effects of TCAs J. Raman Spectrosc. 2010, 41, 498–507 c 2009 John Wiley & Sons, Ltd. Copyright www.interscience.wiley.com/journal/jrs 501 The application of optical-trapping confocal Raman microscopy for the elucidation of several aspects of drug–membrane interactions has been developed here using TCAs. TCAs provide an intriguing model for the study of the influence of cationic amphiphiles of varying structure in different lipid environments. Small-molecule drugs with membrane affinity can have varying effects on phospholipid vesicles. Small organic molecules have been shown to increase lipid order (higher Tm ) or disrupt lipid order (lower Tm ), depending on how the structure of the molecule intercalates into the lipid bilayer.[57] The first result apparent from Figs 3–5 is that TCAs have disordering effects on lipid bilayers, both neutral and anionic, as manifested by the decreased Tm values. This indicates the manner of drug packing in the lipid bilayer. Small molecules that increase lipid order tend to fit neatly into the gaps between lipid chains,[58] but TCAs are sufficiently bulky to create steric disruption. Indeed, an NMR study has shown that the TCAs are probably located with their cyclic rings near the acyl chains and their polar carbon tail ‘snorkeling’ up near the lipid headgroup; in other words, TCAs are mostly in the lipid interfacial/headgroup region, disrupting lipid packing and providing the acyl chains with more room to move.[59] In general, the literature suggests that TCAs are located with the rings near the lipid acyl chains and the charged tail near the lipid headgroup,[60] although one report concludes that the drug conformation in DPPC is quite mobile but preferentially assumes a state where the charged tail is folded back on the rings.[49,51,61] Indeed, ring flexibility in TCA differs depending on specific drug structure.[42] The results in Figs 3–5 align well with the general picture from the literature; the fact that the drugs lower the lipid Tm means that they must affect the acyl chain packing in the lipid. Also, evidence for electrostatic interactions is clear: the drug has greater disordering effects in anionic membranes. Moreover, the partitioning of the TCAs shown in Fig. 8 reveals that more drug is present in anionic membranes, as discussed in more detail below. Figures 3–5 and Table 1 clearly show that TCAs have a greater effect on lipid order in DPPG compared to DPPC as manifested by the lipid Tm values at different drug concentrations. This finding is in accordance with another study that examined TCA effects on DPPC and DPPG at higher drug concentrations,[62] but contradicts an earlier report that examined the effects of several different cationic amphiphilic drugs with varying phospholipid molecules.[45] In this earlier report, it was found that at very C. B. Fox and J. M. Harris 1.0 0 µM drug (pure lipid) 20 µM drug 100 µM drug 500 µM drug 0.9 DPPC DPPG DPPC:DPPG 1:1 46 44 42 0.8 40 0.7 38 0.6 36 DPPC 34 0.5 32 0.4 30 1.0 I2934/I2883 28 0.9 46 0.8 44 42 0.7 40 0.6 Tm/ °C DPPG 0.5 38 36 34 0.4 32 1.0 30 0.9 28 0.8 46 0.7 44 42 0.6 40 DPPC:DPPG 1:1 0.5 38 36 0.4 20 25 30 35 40 45 50 55 34 32 Temperature /°C Figure 3. Effects of increasing concentrations of nortriptyline on the phase transition profiles of DPPC (top), DPPG (middle), and DPPC : DPPG 1 : 1 (bottom) as revealed by the I2934 /I2883 lipid peak ratio. 30 28 0 502 high (mM) drug concentrations, the drug overcomes the influence of the phospholipid headgroup, resulting in similar Tm values for DPPA (1,2-dipalmitoyl-sn-glycero-3-phosphatidic acid), DPPG, and DPPC, even though the pure lipids have different Tm values.[48] The depressions in Tm values for DPPC and DPPG shown in Figs 4 and 5 are in good agreement to values obtained from fluorescence studies using higher drug concentrations,[48] suggesting that a plateau in lipid disordering is reached. This plateau effect has been reported for other cationic amphiphilic drugs in DPPG.[45] Indeed, in Figs 4 and 5, the Tm of DPPG values show a plateau effect, where 20–100 µM drug shows similar disordering behavior as 500 µM drug. However, the decrease in the Tm of DPPC follows a continuous decrease with increasing drug concentration and no apparent plateau, perhaps because the effective ‘plateau’ concentration is higher than the concentrations studied here (≤500 µM). This difference in TCA plateau behavior effects between DPPC and DPPG could arise because the attraction of the drug to zwitterionic DPPC derives from relatively weak hydrophobic interactions, whereas the interaction of the drug with anionic DPPG www.interscience.wiley.com/journal/jrs 100 200 300 400 500 Drug concentration / µM Figure 4. Effects of increasing concentrations of protriptyline (top), amitriptyline (middle), and nortriptyline (bottom) on lipid Tm values. includes a strong electrostatic attraction that quickly saturates the lipid headgroups with the accumulation of drug.[63] It has been found that electrostatic interactions with cationic drugs dominate in mixtures of neutral and anionic lipids.[49] In addition, another TCA drug (chlorpromazine) was shown to have a greater disordering effect on DPPG than DPPC monolayers; this effect was shown by IR spectroscopy to be in the headgroup region instead of the acyl chains.[44,49] Finally, while chlorpromazine was found to lower the Tm of neutral as well as anionic membranes, greater disrupting effects were observed using NMR in the anionic membrane.[64] In short, while there is disagreement about the extent of the disordering effects of TCAs in lipids, it is clear that anionic lipids are more strongly affected than zwitterionic lipids. c 2009 John Wiley & Sons, Ltd. Copyright J. Raman Spectrosc. 2010, 41, 498–507 Confocal Raman microscopy of tricyclic antidepressants in vesicle membranes 46 Protriptyline Amitriptyline Nortriptyline 45 44 43 42 41 40 39 DPPC 38 42 40 Tm/ °C 38 36 34 32 30 28 DPPG 26 46 44 activity), but there was no apparent difference between amitriptyline and nortriptyline.[43] In an analogous system to amitriptyline and nortriptyline, an NMR study concluded that the extra terminal methyl group in imipramine has more of a lipid headgroup disordering effect than desipramine in a 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC)-1palmitoyl-2-oleoyl-sn-glycero-3-phospho-(1’-sn-glycerol) (POPG) POPG mixture, probably due to steric effects; however, it is also noted that these differences may not be directly biologically interpretable since desipramine is, ironically, more prone to induce lipidosis than imipramine.[61] Finally, one study concluded that the aliphatic tail present on TCAs is essential for their effects on neutral lipid membranes since iminodibenzyl, which lacks the hydrocarbon tail, showed minimal effects.[51] Therefore, while some reports indicate that an extra terminal methyl group leads to more greater disordering, this finding is not conclusive and is not apparent from the results in Figs 4 and 5 and Table 1. It is also interesting to note that the level of disorder in control lipid samples due to temperature is not significantly increased when drug is present. In other words, while the presence of drug decreases the main phase transition temperature (destabilizing the gel phase), the extent of lipid disorder reached in the final fluid phase does not significantly increase with the presence of drug. This is apparent in Fig. 3, where the peak intensity ratio representing lipid disorder in drug solutions is not significantly greater than control lipid samples once the fluid phase is reached. Other Raman scattering peak intensities also appear similar to control lipid samples, as revealed in comparing the pure component spectra obtained from SMCR representing the fluid-phase DPPC membrane with or without drug (Fig. 6). 42 40 Drug partitioning between membrane and solution 38 36 34 32 30 DPPC:DPPG 1:1 28 0 100 200 300 400 500 Drug concentration / µM Figure 5. Comparison of the effects of each of the TCA drugs on DPPC (top), DPPG (middle), and DPPC : DPPG 1 : 1 (bottom) Tm values. J. Raman Spectrosc. 2010, 41, 498–507 c 2009 John Wiley & Sons, Ltd. Copyright www.interscience.wiley.com/journal/jrs 503 Studies correlating the structure of different TCAs to lipid membrane disordering effects have come to different conclusions. For instance, using a fluorescence depolarization membrane probe, nortriptyline and protriptyline were observed to have similar disordering effects on lipid Tm in mixed liposomes composed of DPPC and 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC), but no effect could be seen in pure DPPC liposomes at 100 µM drug.[65] The TCAs amitriptyline, nortriptyline, imipramine, and desipramine were shown by calorimetry to have similarly lowered dipalmitoyllecithin lipid Tm at 1 : 1 drug : lipid molar ratios, but the extra methyl group on imipramine and amitriptyline caused a multiphase thermal profile.[66] In contrast, another fluorescence study found that amitriptyline and nortriptyline had a greater lipid disordering effect than imipramine and desipramine in 1,2-dimyristoylsn-glycero-3-phosphocholine (DMPC), DMPC/cholesterol, and synaptosomal membranes (including effects on ATPase Drug partitioning between membrane and solution is an important parameter in its pharmacological function. For instance, it has been found that membrane/plasma partitioning values for TCAs are correlated to drug concentrations in brain homogenate.[43] Several interesting conclusions can be drawn from Figs 7 and 8. In Fig. 8, it apparent that each of the drugs partitions preferentially in membranes containing DPPG, confirming that electrostatic interactions play a significant role in the membrane affinity of the drug molecules. This is an important conclusion since biological lipid assemblies consist of a multiplicity of lipid molecules including charged, neutral, saturated acyl chains, unsaturated acyl chains, and headgroup modifications.[67,68] In the case of brain tissue, where antidepressants are designed to have significant effects, anionic lipids (including phosphatidylglycerols) form a major component of the lipid content.[39] In fact, lipid composition has been shown to be important in mental disorders, and antidepressants can alter this membrane lipid content.[69] Therefore, the effect of TCAs on different lipid compositions is integral to understanding their biological functions, and it is apparent from Fig. 8 that anionic lipids are important for increased TCA affinity for lipid membranes. An additional conclusion that can be drawn from Figs 7 and 8 is that TCA partitioning into the membrane increases significantly above the lipid Tm compared to below Tm . In other words, TCAs are able to bind much more readily to fluid-phase lipids, as suggested by previous studies.[70] The TCA chlorpromazine was found to have a ∼3.5 times higher affinity for egg phosphatidylcholine (eggPC) and egg phosphatidic acid (eggPA) monolayers compared to the pure lipids, presumably due to more disorder in the egg C. B. Fox and J. M. Harris Table 1. Lipid phase transition temperatures influenced by tricyclic antidepressant drugsa 0 µM prot 0 µM ami 0 µM nor 20 µM prot 20 µM ami 20 µM nor 100 µM prot 100 µM ami 100 µM nor 500 µM prot 500 µM ami 500 µM nor a DPPC (◦ C) σ No. samples DPPG (◦ C) 43.43 44.13 42.7 43.2 44.08 42.53 41.9 41.87 41.83 39.93 39.17 39.93 0.34 0.75 0.31 1.15 1.24 0.46 0.38 1.09 0.21 0.86 0.69 0.86 3 3 3 3 2 3 3 3 3 3 3 3 41.65 39.9 40.43 33.7 34.95 31.98 30.7 32.13 31.62 29.6 31.2 32.05 σ 1.67 1.16 1.12 2.91 1.05 0.69 2.02 1.46 0.83 4.83 No. samples DPPC : DPPG (◦ C) σ No. samples 1 3 3 3 3 3 3 3 3 3 1 3 41.33 43.53 39.55 40.48 39.78 36.55 35.8 40.28 38.35 33.45 34.53 30.78 1.72 2.16 1.56 0.39 0.39 0.99 0.07 0.75 2.26 4.12 1.1 0.11 3 3 2 2 2 2 2 3 2 3 2 2 prot, ami, and nor refer to protriptyline, nortriptyline, and amitriptyline, respectively. Control DPPC Raman Intensity DPPC in 500µM amitriptyline 1000 1100 1200 1300 1400 1500 2900 3000 3100 Wavenumber / cm-1 Figure 6. Fluid-phase spectra of DPPC vesicle samples in buffer without drug (solid line) or in buffer with 500 µM amitriptyline (dotted line) obtained by SMCR indicating no significant difference in lipid disorder. 504 lipids.[67,71 – 74] This affinity for fluid phase membranes most likely is because the ordered gel phase lipid membrane exhibits a high enthalpy penalty for drug insertion due to the disruption in the chain packing, whereas the disordered fluid phase membrane has a lower enthalpy barrier for drug insertion. This can be understood more fully by considering that the gel lipid membrane consists of ordered acyl chain trans conformers with a high degree of chain coupling, whereas the fluid lipid membrane has high gauche conformer content along with chain decoupling and rotation.[35,41] Indeed, an increase in bilayer volume and especially bilayer area occurs in fluid-phase membranes,[33,35] allowing greater room for the drug to intercalate within the lipids. Another observation from Fig. 8 is that the three TCAs show similar partitioning behavior. The octanol/water partition coefficients of cationic amphiphilic drugs and TCAs have been shown to have a positive correlation with their ability to disrupt both DPPG and DPPC lipid membranes.[75] However, amitriptyline, which has a much higher octanol/water partition coefficient than either protriptyline or nortriptyline, did not exhibit a significantly different www.interscience.wiley.com/journal/jrs effect on lipid Tm in DPPC or DPPG vesicles (Fig. 5). It should be remembered that octanol/water partition coefficients are not equivalent to liposome/water partition coefficients, especially for amphiphilic molecules.[45,63] For example, liposome/water partition coefficients for protriptyline, amitriptyline, and nortriptyline were determined using membranes composed of a mixture of DPPG, DPPC, and cholesterol and found to be significantly larger than the octanol/water partition coefficients (especially for nortriptyline and protriptyline, see Fig. 1).[2] The octanol/water partition coefficients were not able to predict the partitioning behavior of the TCAs (Fig. 8). Amitriptyline has significantly higher hydrophobicity (i.e. octanol/water partition coefficient) than either protriptyline or nortriptyline, so it might be expected to show greater partitioning, at least in the neutral DPPC vesicles. Instead, all three drugs appeared to have similar partitioning behavior in all three lipid environments studied. This is expected on the basis of the liposome/water partition coefficients of the three drugs listed in Fig. 1. Complicating the partitioning discussion is a small proportion of neutral, deprotonated TCA at physiological pH, which has been shown to have a higher DPPC liposome/water partition coefficient than the charged form, especially when the TCAs have an extra methyl group (such as with amitriptyline).[57] Protriptyline has a higher pKa value (10.0) than either amitriptyline (9.4) or nortriptyline (9.7),[41] so that if the unprotonated form was having a large effect on the partitioning, then protriptyline would show significantly less partitioning than the other two forms, which it does not. Confocal Raman microscopy and drug–membrane interactions There are several key advantages inherent in the characterization of drug–membrane interactions by optical-trapping confocal Raman microscopy. This technique overcomes the limited sensitivity of Raman scattering through the use of an efficient, high-numericalaperture objective for excitation and scattered light collection in a small sampling volume that is filled with the vesicle or vesicle aggregate being studied. While the local concentrations of lipid and partitioned drug are sufficiently high within this volume to be readily detected, the concentration of drug averaged over the entire sample can be quite small, close to in vivo levels. It is estimated that typical plasma concentrations of TCAs would be c 2009 John Wiley & Sons, Ltd. Copyright J. Raman Spectrosc. 2010, 41, 498–507 Confocal Raman microscopy of tricyclic antidepressants in vesicle membranes (a) 400 DPPC with 500 µM amitriptyline at 27 °C DPPC with 500 µM amitriptyline at 46 °C Raman Intensity / 10 PE 350 300 250 200 150 ring C=C stretch 1600 cm-1 100 50 0 1000 1500 2000 2500 3000 Wavenumber / cm-1 (b) 3.0 2.5 amitriptyline 2.0 I1600/I1730 + N 1.5 CH3 CH3 1.0 0.5 0.0 25 30 35 40 45 50 with 5 nm bilayer thickness,[47] the total membrane volume is 2.2 µl. If the concentration of amitriptyline is 20 µM in solution, there would be 32 mM drug in the membrane volume due to partitioning. However, this would require 73 nmol of drug when only 20 nmol are available in solution! Therefore, the results obtained from such an experiment would not represent the solution concentration conditions which originally existed; i.e. there is not an excess of 20 µM drug to partition into the lipid, so partitioning lowers the actual concentration of drug in solution and in the lipid. In contrast, a typical Raman microscopy experiment requires 40 µl of 3 µM lipid dispersion, which gives a total membrane volume of 90 pl. If the concentration of amitriptyline is once again 20 µM, the membrane volume would require 3 pmol of drug to reach equilibrium with the solution. Since there is 800 pmol of drug available in solution, less than 0.4% depletion would take place. Therefore, results from Raman microscopy truly represent the effects of a 20 µM amitriptyline solution. Thus, confocal Raman microscopy is uniquely suited to investigate the influence of low concentrations of lipophilic drugs on the phase transitions of vesicle membranes. Despite the advantages of Raman microscopy, there are some disadvantages. Relatively large liposomes (diameter >500 nm) are required, in order for the vesicles to be visible through the eyepiece to guide them into the optical trap. While the electrostatic repulsion in charged lipids makes them more likely to be unilamellar, phosphatidylcholine vesicles, especially of large diameter, can be multilamellar,[83] and they appear to aggregate more as well. The question of lamellarity might introduce data interpretation problems if there is a very slow membrane permeability rate of the drug. However, in the case of TCAs, it was shown that 30 min of incubation at room temperature was more than enough time to equilibrate the drug in both zwitterionic (i.e. multilamellar) and anionic (unilamellar) lipid samples,[25,56] a condition which was easily satisfied in the experiments presented here. Temperature / °C Figure 7. Tricyclic antidepressants exhibiting temperature-dependent partitioning. (a) Comparison of the Raman spectrum of DPPC below and above Tm in 500 µM amitriptyline solution. (b) Plot of the amitriptyline : lipid ratio with increasing temperature in DPPC as revealed by the Raman scattering peak ratio I1600 /I2847 . J. Raman Spectrosc. 2010, 41, 498–507 Confocal Raman microscopy is shown to be an effective technique to elucidate drug–membrane interactions using micromolar concentrations of both lipid and drug. The results are more relevant to the biological situation because low concentrations of drug can be investigated while problems associated with depletion of the drug in solution are avoided by working at even lower lipid concentrations. Raman spectra provide conformational and compositional information for the lipid and drug, respectively, allowing membrane disordering and drug partitioning behavior to be assessed. Raman spectra are collected from a ∼1.3 fl volume of optically trapped lipid vesicles. In this study, it has been shown that the cationic and amphiphilic TCA drugs protriptyline, amitriptyline, and nortriptyline disorder and lower the phase transition temperature of both zwitterionic and anionic lipid membranes, while the effect on anionic membranes is much greater, consistent with interactions between the anionic lipid head groups and the cationic amphiphilic drugs. All three drugs preferentially partition into anionic membranes and fluid phase membranes, and overall partitioning results agree with previously reported liposome/water partition coefficients but not octanol/water partition coefficients. Acknowledgements This research was supported in part by the National Science Foundation under Grant CHE-0654229 and by Eli Lilly and Co. c 2009 John Wiley & Sons, Ltd. Copyright www.interscience.wiley.com/journal/jrs 505 on the order of micromolars, with some of that concentration bound to plasma proteins and thus unavailable for therapeutic activity.[76,77] Moreover, amitriptyline causes hemolysis of red blood cells and is toxic in a neurological animal model at ∼10 mM.[57,67,78] The critical micelle concentrations of TCAs are in the low millimolar range,[79] above which TCAs act as detergents that could dissolve lipid membranes.[44,79 – 82] Biologically relevant (µM) drug concentrations cannot be controlled in traditional methods to study lipid phase transitions such as DSC and NMR, where high (mM) concentrations of lipid are required to observe a response. Thus, if it is desired to examine the membrane effects of a highly lipophilic drug at micromolar concentrations in the presence of millimolar lipid, the drug in solution would be quickly depleted, where there is not sufficient drug available to partition into the membrane without depleting the drug concentration in solution. For example, amitriptyline has a liposome/water partition coefficient (logD) of 3.21.[44,63] This means that there will be 1600-fold greater concentration of amitriptyline in the lipid membrane than in solution. Thus, in a typical DSC experiment that requires 1 ml of a 3 mM lipid dispersion, with vesicles that are unilamellar and 1 µm in diameter Conclusions C. B. Fox and J. M. Harris DPPC aboveTm DPPC below Tm 2.0 DPPG above Tm DPPG below Tm DPPC:DPPG above Tm DPPC:DPPG below Tm 1.5 I 1600 / I 1730 1.0 0.5 0.0 protriptyline 0 100 amitriptyline 200 300 400 500 0 100 nortriptyline 200 300 400 500 0 100 200 300 400 500 Drug concentration / µM Figure 8. Partitioning of increasing concentrations of protriptyline (left), amitriptyline (middle), and nortriptyline (right) in DPPC, DPPG, and DPPC : DPPG 1 : 1 below and above Tm . Additional support to Chris Fox from a University of Utah Graduate Research Fellowship is gratefully acknowledged. References 506 [1] J. M. Sanderson, Org. Biomol. Chem. 2005, 3, 201. [2] J. K. Seydel, M. Wiese, Drug-Membrane Interactions: Analysis, Drug Distribution, Modeling, Wiley-VCH: Weinheim, Germany Verlag: 2002. [3] P. L. Bardonnet, V. Faivre, F. Pirot, P. Boullanger, F. Falson, Biochem. Biophys. Res. Commun. 2005, 329, 1186. [4] S. B. Hwang, T. Y. Shen, J. Med. Chem. 1981, 24, 1202. [5] I. Kyrikou, S. K. Hadjikakou, D. Kovala-Demertzi, K. Viras, T. Mavromoustakos, Chem. Phys. Lipids 2004, 132, 157. [6] H. Lygre, G. Moe, H. Holmsen, Acta. Odontol. Scand. 2003, 61, 303. [7] T. J. O’Leary, P. D. Ross, I. W. Levin, Biochemistry 1984, 23, 4636. [8] F. Severcan, I. Sahin, N. Kazanci, Biochim. Biophys. Acta 2005, 1668, 215. [9] M. Prochazka, J. Stepanek, P. Y. Turpin, Chem. Phys. Lipids 2004, 132, 145. [10] T. Mavromoustakos, I. Daliani, J. Matsoukas, in Bioactive Peptides in Drug Discovery and Design: Medical Aspects (Eds.: J. Matsoukas, T. Mavromoustakos), IOS Press: Fairfax, 1999, pp 13. [11] S. S. Fan, T. Y. Shen, J. Med. Chem. 1981, 24, 1197. [12] Y. Barenholz, Curr. Opin. Colloid Interface Sci. 2001, 6, 66. [13] D. Bemporad, C. Luttmann, J. W. Essex, Biochim. Biophys. Acta 2005, 1718, 1. [14] D. Bemporad, C. Luttmann, J. W. Essex, Biophys. J. 2004, 87, 1. [15] Y. Song, V. Guallar, N. A. Baker, Biochemistry 2005, 44, 13425. [16] A. Cruz, L. Vazquez, M. Velez, J. Perez-Gil, Langmuir 2005, 21, 5349. [17] H. Ferreira, M. Lúcio, J. L. F. C. Lima, A. Cordeiro-da-Silva, J. Tavares, S. Reis, Anal. Biochem. 2005, 339, 144. [18] H. N. Hunter, W. Jing, D. J. Schibli, T. Trinh, I. Y. Park, S. C. Kim, H. J. Vogel, Biochim. Biophys. Acta 2005, 1668, 175. [19] R. N. Lewis, Y. P. Zhang, R. N. McElhaney, Biochim. Biophys. Acta 2005, 1668, 203. [20] M. Lankers, J. Popp, W. Kiefer, Appl. Spectrosc. 1994, 48, 1166. [21] K. Ajito, Appl. Spectrosc. 1998, 52, 339. [22] M. P. Houlne, C. M. Sjostrom, R. H. Uibel, J. A. Kleimeyer, J. M. Harris, Anal. Chem. 2002, 74, 4311. [23] D. P. Cherney, J. C. Conboy, J. M. Harris, Anal. Chem. 2003, 75, 6621. [24] T. E. Bridges, M. P. Houlne, J. M. Harris, Anal. Chem. 2004, 76, 576. [25] C. B. Fox, R. A. Horton, J. M. Harris, Anal. Chem. 2006, 78, 4918. [26] W. Knoll, Biochim. Biophys. Acta 1986, 863, 329. [27] J. T. Mason, Methods Enzymol. 1998, 295, 468. [28] Y. Omura, S. Muraishi, Spectrochim. Acta A 1997, 53, 1783. www.interscience.wiley.com/journal/jrs [29] R. G. Snyder, J. R. Scherer, B. P. Gaber, Biochim. Biophys. Acta 1980, 601, 47. [30] N. Yellin, I. W. Levin, Biochim. Biophys. Acta 1977, 489, 177. [31] I. W. Levin, Adv. Infrared Raman Spectrsc. 1984, 11, 1. [32] A. Csiszar, E. Koglin, R. J. Meier, E. Klumpp, Chem. Phys. Lipids 2006, 139, 115. [33] C. J. Orendorff, M. W. Ducey, J. E. Pemberton, J. Phys. Chem. A 2002, 106, 6991. [34] R. J. Meier, A. Csiszar, E. Klumpp, J. Phys. Chem. B 2006, 110, 20727. [35] C. B. Fox, R. H. Uibel, J. M. Harris, J. Phys. Chem. B 2007, 111, 11428. [36] E. Mushayakarara, P. T. T. Wong, H. H. Mantsch, Biophys. J. 1986, 49, 1199. [37] C. B. Fox, G. A. Myers, J. M. Harris, Appl. Spectrosc. 2007, 61, 465. [38] R. Koynova, M. Caffrey, Biochim. Biophys. Acta 1998, 1376, 91. [39] O. G. Mouritsen, Life – As A Matter of Fat: The Emerging Science of Lipidomics, Springer: New York, 2005. [40] U. M. Joshi, P. R. Kodavanti, B. Coudert, T. M. Dwyer, H. M. Mehendale, J. Pharmacol. Exp. Ther. 1988, 246, 150. [41] M. Bennouna, J. Ferreira-Marques, S. Banerjee, J. Caspers, J. M. Ruysschaert, Langmuir 1997, 13, 6533. [42] M. G. Casarotto, D. J. Craik, J. Colloid Interface Sci. 1993, 158, 326. [43] B. R. Cater, D. Chapman, S. M. Hawes, J. Saville, Biochim. Biophys. Acta 1974, 363, 54. [44] N. Deo, T. Somasundaran, P. Somasundaran, Colloids Surf. B: Biointerfaces 2004, 34, 155. [45] A. Harder, H. Debuch, Chem. Phys. Lipids 1986, 39, 65. [46] C. Song, H. Holmsen, W. Nerdal, Biophys. Chem. 2006, 120, 178. [47] S. T. Burns, M. G. Khaledi, J. Pharm. Sci. 2002, 91, 1601. [48] R. Hanpft, K. Mohr, Biochim. Biophys. Acta 1985, 814, 156. [49] Z. Fisar, Gen. Physiol. Biophys. 2005, 24, 161. [50] M. J. Reasor, S. Kacew, Exp. Biol. Med. 2001, 226, 825. [51] J. S. Santos, D. K. Lee, A. Ramamoorthy, Magn. Reson. Chem. 2004, 42, 105. [52] R. Krulik, J. Sikora, P. Bures, K. Fuksova, Drug Metabol. Drug Interact. 1991, 9, 283. [53] A. Savitzky, M. J. E. Golay, Anal. Chem. 1964, 36, 1627. [54] C. Hansch, A. Leo, Substituent Constants for Correlation Analysis in Chemistry and Biology, John Wiley & Sons: New York, 1979. [55] C. Huang, J. T. Mason, I. W. Levin, Biochemistry 1983, 22, 2775. [56] H. Hauser, Biochim. Biophys. Acta 1984, 772, 37. [57] Z. Fisar, K. Fuksova, M. Velenovska, Gen. Physiol. Biophys. 2004, 23, 77. [58] P. Garidel, C. Johann, L. Mennicke, A. Blume, Eur. Biophys. J. 1997, 26, 447. [59] K. Lohner, Chem. Phys. Lipids 1991, 57, 341. [60] S. V. Dvinskikh, U. H. N. Durr, K. Yamamoto, A. Ramamoorthy, J. Am. Chem. Soc. 2007, 129, 794. [61] B. G. Sanganahalli, P. G. Joshi, N. B. Joshi, Life Sci. 2000, 68, 81. c 2009 John Wiley & Sons, Ltd. Copyright J. Raman Spectrosc. 2010, 41, 498–507 Confocal Raman microscopy of tricyclic antidepressants in vesicle membranes [62] M. G. Casarotto, D. J. Craik, J. Pharm. Sci. 2001, 90, 713. [63] B. Kursch, H. Lullmann, K. Mohr, Biochem.Pharmacol. 1983, 32, 2589. [64] A. A. Hidalgo, A. S. Pimentel, M. Tabak, O. N. Oliveira Jr, J. Phys. Chem. B 2006, 110, 19637. [65] W. Nerdal, S. A. Gundersen, V. Thorsen, H. Hoiland, H. Holmsen, Biochim. Biophys. Acta 2000, 1464, 165. [66] K. L. Audus, M. A. Gordon, Biochem. Pharmacol. 1985, 34, 705. [67] Z. Fisar, K. Fuksova, J. Sikora, L. Kalisova, M. Velenovska, M. Novotna, Neuroendocrinol. Lett. 2006, 27, 307. [68] Z. Fisar, R. Krulik, K. Fuksova, J. Sikora, Gen. Physiol. Biophys. 1996, 15, 51. [69] M. A. Yorek, in Phospholipids Handbook (Ed.: G. Cevc), Marcel Dekker: New York, 1993, p 745. [70] M. Baciu, S. C. Sebai, O. Ces, X. Mulet, J. A. Clarke, G. C. Shearman, R. V. Law, R. H. Templer, C. Plisson, C. A. Parker, A. Gee, Philos. Trans. A Math Phys. Eng. Sci. 2006, 364, 2597. [71] J. B. Custodio, L. M. Almeida, V. M. Madeira, Biochem. Biophys. Res. Commun. 1991, 176, 1079. [72] M. Luxnat, H.-J. Galla, Biochim. Biophys. Acta 1986, 856, 274. [73] D. A. Middleton, D. G. Reid, A. Watts, J. Pharm. Sci. 2004, 93, 507. [74] J. Zhang, T. Hadlock, A. Gent, G. R. Strichartz, Biophys. J. 2007, 92, 3988. [75] D. Marsh, Chem. Phys. Lipids 1991, 57, 109. [76] D. Pan, L. C. T. Shoute, D. L. Phillips, J. Phys. Chem. A 2000, 104, 4140. [77] D. Pan, L. C. T. Shoute, D. L. Phillips, J. Phys. Chem. A 1999, 103, 6851. [78] D. Razavi, J. Mendlewicz, Neuropsychobiology 1982, 8, 73. [79] N. Kitagawa, M. Oda, I. Nobutaka, H. Satoh, T. Totoki, M. Morimoto, Toxicol. Appl. Pharmacol. 2006, 217, 100. [80] M. Gutierrez-Pichel, S. Barbosa, P. Taboada, V. Mosquera, Colloid Polym. Sci. 2003, 281, 575. [81] P. Taboada, D. Attwood, J. M. Ruso, M. Garcia, V. Mosquera, Phys. Chem. Chem. Phys. 2000, 2, 5175. [82] R. H. Manzo, M. E. Olivera, G. L. Amidon, V. P. Shah, J. B. Dressman, D. M. Barends, J. Pharm. Sci. 2006, 95, 966. [83] J. Lemmich, K. Mortensen, J. H. Ipsen, T. Honger, R. Bauer, O. G. Mouritsen, Phys. Rev. E 1996, 53, 5169. 507 J. Raman Spectrosc. 2010, 41, 498–507 c 2009 John Wiley & Sons, Ltd. Copyright www.interscience.wiley.com/journal/jrs
© Copyright 2026 Paperzz