123
Chonlstry and Physics of Lipids, 62 (1992) 123"138 Elsevier ScientificPublishers Ireland Ltd.
Interaction of antioxidants with depth-dependent fluorescence quenchers and energy transfer probes in lipid bilayers Jessica S. Hinzmann, Rose L. McKenna, Tonya S. Pierson, Fusan Han, Ferenc J. K6zdy and Dennis E. Epps The Upjohn Company, 7000 Portage, Kalamazoo, It,ll 49001 (USA)
(Received December 16th, 1991; revision received March 26th, 1992; accepted April 24th, 1992) Three fluorescent, lipophilic, beterocyclicantioxidants were incorporated into lipid bilayers and exposed to depth-dependent nitroxyl fatty acid quenchers. The Stern-Volmerplots curved upward at low quencher concentrations.Quantitative analysis of the retorts showedthat this behavioris consistentwith complexformationbetweenquencherand fluorescentantioxidant,wherethe complex is 2-3 timesmore f l u ~ t than the parent fluorophore.At higher quencherconcentrations,both free antioxiclantand 'bright complex'are quencheddynamically,albeit quenchingof the latter is lessefficient.The complexprobably results from ionic, hydrogen bond and ~r-lrinteractiom. Formation of such a 'bright complex' is also observablein a homogeneoussolution of the reactants in cyclohexane.Additionalevidencefor the complexationof these antioxidants with fatty acids in fipid bilayersis provided by the fact that energytransfer from the antioxidantsto anthroyloxyfatty acids occursat surfaceconcentrationswhere radiativeenergytransfer betweenfree moleculesshould not be efficient.For directlyprobing the relativedepths of these fluorophoresin lipid bilayerswe used the aqueous quenchersacrylamideand iodide. They showedthat, in terms of increasingdepth in the bilayer,the order was U-78,517f < U-78,518e < U-75,412e.Our results, in toto, demonstrate that the Lazaroid antioxidants are incorporated into the lipid bilayer wherethey occupystrictlydefinedpositionsand orientations.Complexationwith fatty acylchains should be mechanisticallyrelevant, since it may enhance antioxidant activity by hindering free radical chain propagation. Key words: antioxidants; fluorescence;bilayers; quenching; energy transfer
Introduction Free radical-induced tissue damage occurs in a variety of diseases, such as myocardial reperfusion injury [1], traumatic head and spinal cord injury [2] and in carbon tetrachloride-induced liver toxicity [3]. The possibility that lipid peroxidation is the common cause o f these irreversible processes has led to the development o f antioxidants as potential therapeutic agents [4,5]. Since the agent ultimately responsible for cellular damage may be any one of a broad spectrum o f free radicals located in a variety o f compartments, the design o f a given antioxidant drug involves not only the optimization o f the intrinsic scavenging ability of the drug toward a Correspondsnce to: Dennlt E. Epps, The Upjohn Company, 7(~0 Portage, Kahummoo, MI 49001, USA.
given free radical, but also its partitioning into the target compartment. Reactive free radicals diffuse only short distances; propagation o f the free radical reaction can therefore be inhibited only in situ. For optimal action, t h e drug must be strategically located, either extracellularly to intercept radicals prior to diffusion into the cell, or within the cell or cell membrane in order to protect locally the target proteins, iipids and nucleic acids. For example, Vitamin E, a prominent biological free radical scavenger, is anchored in biological membranes by its phytol chain and is the major inhibitor of the peroxidation of membrane lipids [6]. In addition to its chemical action, vitamin E also modifies the host lipid bilayer's physical properties, especially viscosity and permeability. The strict localization of vitamin E in the lipid bilayer prevents it from reacting with free radicals
0009-3084/92/$05.00 1992 ElsevierScientificPublishers Ireland Ltd. Printed and Published in Ireland
124
essential for the functioning of the cell which are located in other compartments. Thus, a method of locating the antioxidants in membranes should help in optimizing drug design and in understanding the mechanism of action of antioxidants. Fluorescence spectroscopy is an ideal tool for determining the location of lipophilic fluorophores within biological membranes, either by measuring energy transfer to fatty acids tagged with anthroyloxy groups at different carbon atoms on the acyl chain, or by depth-dependent quenching by nitroxyl fatty acids [12]. Many of our experimental antioxidant drugs happen to contain fluorescent moieties and were readily amenable for similar studies. We chose to work with a model system of artificial bilayer lipid membranes composed of egg lecithin and cholesterol in a molar ratio of 4:1, which mimics the essential physical properties and permeability of cell membranes. The fluorescence properties of U-78,517f, U-78,518e and U-75,412e and their locations within the bilayer determined by fluorescence spectroscopy, are described herein.
U-78517F CH3
CH3
N~
CH3 • 2HCI • H20
U-78518E
CH3
~H2 ~=~
CH$
CH3 'ox HCI
U-75412E
~
H2.
NH
Materials a m l ~ All antioxidants were synthesized and purified at The Upjohn Company and obtained as solids from the Biological Screening Office. Their chemical structures are shown in Fig. 1. Vitamin E, naphthalene and p-terphenyl were purchased from Aldrich Chemical Company. Fluorescent and spin-labelled fatty acids were from Molecular Probes, Inc. 'Ultra Pure' acrylamide was from Schwartz Mann., egg lecithin from Avanti Polar Lipids and cholesterol from Sigma Chemical Company.
Phospholipid vesicles Small unilamellar phospholipid/cholesterol vesicles (4:1, tool:tool) were prepared by the ethanol injection procedure of Batzri and Korn [7]. Briefly, for a typical vesicle pre~ration, 100 rag of egg lecithin and 12.5 m g dmtesterol were dissolved in 2 ml ethanol and injected with rapid stirring into 50 ml 0.01 M Tris-HCl, 150 mM NaCI, pH 7.4. The solution wasooacentrated to a small volume by ultrafil~tion through an
C'-O ~ ' J
O
~
.
'CH3
Fig, 1. Chemicalstructuresof antioxidant~
A m i s h YM-100 ~ t e r and applied to a 2.5 x 95 cm Selthatose CL-48 c o n m . Fractions were eluted with the same buffer at a flow rate of 0.46 mi/min and the eluate was monitored at 210 nm. Peak fractions were pooled and concentrated to - 1 0 ml by ultra~tration, as described above. IAI~i ~ s eontmt of the coneen. tra..auivo0kle ~ ,wan detetmtn~ rain8 the premdme of Bins!eft[8].
Spectral measurements
For UV/VlS 8 ¢ ~ r 0 1 ~ ~ t ~ 2200 ~ t n t m e n t i n ~
we ~ 0 a C . ~
with a ~
~m-
125
purer and controlled by Spectra Calc software. Steady-state fluorescence measurements were made in the ratio mode, using either a Hitachi F2000 or a Perkin Elmer MPF66 instrument and the spectral data were processed using the software packages supplied by the manufacturers. Fluorescence quantum yields of our compounds, Ounk, were calibrated with naphthalene in ethanol (~ref = 0.21 ) or with quinine sulfate in 0.1 M sulfuric acid (Oref = 0.55 ), at the excitation wavelength of 295 nm for all compounds. Quantum yields were calculated from the relationship ~unk = ~ref ×
lunk/Iref X Aref/Aun k
(1)
where lunk and lref are the integrated fluorescence intensities and Aunk and Aref the absorbances at the exciting wavelength.
Dynamic fluorescence measurements Multifrequency fluorescence measurements of the intrinsic fluorescence of the antioxidants given in Fig. 1 were made using a cross correlation phase and modulation fluorometer. The harmonic content of a high repetition rate mode-locked NdYAG laser was used to generate frequencies from 20 to 300 MHz. The Nd/YAG laser (Spectra Physics) was used to synchronously pump a dye laser (Rhodamine 6G) whose pulse train was frequency-doubled with an angle-tuned frequency doubler utilizing a KDP crystal [9]. The autocorrelated pulse width was 10-12 ps and the cross correlation frequency was 25 Hz. The dye laser was tuned to the emission maximum of the dye, 590 nm, which, upon doubling, yielded an ultraviolet light of 295 nm (2-5 mW) for excitation. The latter was passed through a half-wave plate prior to entering the sample compartment to eliminate any horizontally polarized light. Data were collected with the emission polarizer set at the magic angle of 55° and emission was observed through interference filters with 10 nm bandpasses. Usually 9-10 different modulation frequencies were used and data were collected until phase and modulation deviations reached less than 0.2 and 0.004, respectively. Temperature was maintained at 25° using a Haake circulating bath; lifetime least squares analysis was performed using
the software provided by ISS, Inc., Champaign, IL. Absorbances at 295 nm were kept below 0.09 optical density (OD) to eliminate absorptive screening. A solution of p-terphenyl in cyclohexane 0" = 1 ns) was used as the reference lifetime standard.
Fluorescence quenching The fluorescent antioxidants were added in small aliquots of ethanol or DMSO to the vesicle~ and allowed to equilibrate for 5-10 rain before use. All operations were performed at 25°C and absorbances of the antioxidant/vesicle complexes were kept at < 0.1 to preclude absorptive screening. Fluorescence intensities were measured at 330 nm for U-78,517fand at 380 nm for U-78,518e and U-75,412e. Measurements were signal-averaged for 10-16 s and corrected for dilution. The data for quenching by acrylamide were fitted to the linear form of the Stern-Volmer equation: Fo = 1 + Ksv [Q] F
(2)
where Ksv is the Stern-Volmer quenching constant and [Q] is the quencher concentration. Alternatively, F as a function of [Q] was analyzed by non-linear least squares fitting. For upwardcurving Stern-Volmer plots, the equation should either include a static quenching component, or take into account the quenching within a 'sphere of action': Fo = (1 + Ksv [Q]) e vQ F
(3)
where V is the volume of the sphere within which the probability of quenching is unity.
Energy transfer to anthroyloxy fatty acids In one set of experiments, anthroyloxy fatty acids were added in ethanol to 300-400 /~M phospholipid vesicles in buffer to yield a molar ratio of 1:250 and allowed to equilibrate for 1-4 h before use. Individual Lazaroids were then added and after l rain, the fluorescence emission of the acceptor was measured at 450 nm with excitation at 295 nm. All fluorescence intensities were corrected for dilution. In a second set of experiments,
126
energy transfer was measured by determining the loss of donor emission in the presence of increasing amounts of acoeptor. Because of a slight overlap between the emission spectra of the donors and the acceptor, the fluorescence emission was measured not only at a wavelength where the emission of the donor dominates, but also at one where the emission of the acceptor is preponderant. Using the known quantum yields of emission at these wavelengths, the data were then corrected for the slight contribution of the acceptor.
imated from the following equation: 02(/) = a2(/)countins statistics+ (0.0218 /)2, where the coefficient of [ was calculated from the variation in intensities of the monitored reflections. Cell parameters were determined by least-squares fit of CuKoq 2q values (X(CuKa0 = 1.5402 A) for 25 high-20 reflections. Lp corrections appropriate for a monochromator with 50% perfect character were applied. The calculated overall absorption coefficient was 2.03/ram. Remits mal l)i~anlea
Crystal structure The bulk drug lot (A)0309FCWI40A of U-83,836e, the negative enantiomer of U-78,517f, was recrystallized using the methanol-heptane 'Binary Vapor Diffusion System' method. Although crystals could also be obtained by the methanoi-acetoue system, only crystals from the first system were stable during the mounting procedure. A clear, chunky, prism-shaped crystal of 0.2 x 0.2 x 0.5 mm size, obtained from methanol-heptane, was selected and mounted on a glass fiber. Diffraction data were collected at -150°C, using a Nicolet P21 X-ray diffractometer controlled by a Harris computer, with graphite-monochromatized CuK~ radiation (~(CuKa)= 1.5405 A). All 3146 unique reflections were measured to a 20m~ of 138° for Laue group mmm; 2597 intensities were >3a. The 0/20 step-scan technique was used with a scan speed of 4°/min and a scan width >3.4 °. Ten reflections were periodically monitored and showed no deterioration. Standard deviations in the intensities were approx-
The spectroscopic properties of our antioxidants, relevant to this study, are shown in Table I. The excitation maxlma of U-78,517f, U-78,518e and U-75,412e were essentially invariant as a function of solvent. The emission maxima were somewhat solvent dependent, but not sufficiently Stokes-shifted to classify the Lazaroids as 'environment-sensitive' probes. It is interesting to note that there is only one fluorescence maximum for U-78,518e, even thoegh, taken individually, both the chromanol moiety (as in Vitamin E) and the piperazinylpyridinyl ring system are fluorescent. With excitation at 295 nm, a very small peak is observed at approximately 330 nm, corresponding to that of the chromanol moiety. However, it is evident that the greatest proportion of the energy emitting from the chromanoi moiety is transferred to that of the piperazinylpyridinyt ring via a F6rster energy transfer mechanism, since the emission maxima of this compound is > 386 ran in all solvents. Thus only the fluorescence of the
TABLE I Fluorescence excitation and emission maxima of antioxidants in various solvents Compound
Ethanol
Methanol
Isopropanol
Cyclohexane
Vesides a
Xm U78,517f U78,518e U75,412e
285 310 310
326 405 391
296 312 311
aEgg l e c i t h i n / ¢ h o ~ (4:1, mol/mot) v e s k J ~ ' ~
332 399 385
296 309 310
326 386 385
as ~descri~in ~ .
298 310
336 401 388
292 310
328 387 391
127 TABLE II Quantum .yields (4D)of Lazaroids in variom solvents Compound
Ethanol
Methanol
Isopropanol
Cyclohexane
Vesicles
U78,517f U78,518e U75,412e
0.083 0.08 0.024
0.029 0.044 0.022
0.044 0.123 0.04
0.03 0.143 0.055
0.07 0.23 0.104
piperazinylpyridinyl ring system is observed; it is this ring system which is also the origin of the fluorescence of U-75,412e. In addition, the excitation and emission maxima of the fluorescent Lazaroids are in a range where one should expect significant energy transfer to membrane localized probes, such as depth-dependent anthroyloxy fatty acids. The Lazaroids could thus act as intrinsic probes for the determination of their membrane localization, without significant perturbation of the bilayer by additional bulky extrinsic probes. Fluorescence quantum yields (~) as a function of solvent are shown in Table II. In general, they are dependent upon the hydrophobicity of the solvent. On the other hand, the relatively high quantum yield observed for each compound inserted into unilamellar vesicles is due not only to the solvent environment, but also to the rigidity of the fluorophore within the bilayer. The solvent dependence of the fluorescence lifetime of U-78,517f is shown in Table III. The data were analyzed allowing for a finite Raman TABLE IlI Fluorescence lifetimes 0") of U-78,517f (ns) c
contribution to the total lifetime. As expected, the Raman contribution to the total lifetime is greatest for vesicles, 5.4% of the total lifetime. The lifetime of U-78,517f in vesicles is intermediate between that of the compound in ethanol and in methanol. The fluorescent chromanol moiety of U-78,517f is identical to that of Vitamin E and it would be expected that the fluorescence lifetimes would be similar. Indeed, this was found to be the case: the lifetime of U-78,517f in ethanol, 1.8 ns for TI is identical to that for Vitamin E, 1.8 ns [12]. In addition, the lifetime of U-78,517f in small egg lecithin/cholesterol unilamellar vesicles is 1.49 ns, comparable to that determined for Vitamin E in egg lecithin vesicles, 1.31 ns. Thus, substitution of the phytol chain of Vitamin E by the dipyrrolidinyl-pyrimidinyl-piperazinyl ring system does not appreciably alter the fluorescence lifetime of the chromanol moiety. The lifetime measurements also indicate that the fluorescence of U-78,517f is not very environment-sensitive. The lifetimes of the piperazinylpyridinyl moieties of U-78,518e and U-75,412e were too short to be measured with our instrumentation.
Fluorescence quenching Solvent
~'1
f] a
Tavgb
Methanol Ethanol Vesicles
1.26 1.8 1.49
0.986 0.976 0.946
1.21 1.76 1.41
"Weighted average of the two lifetimes. bFraction of the total lifetime in the slow component. c2 mole% U-78,517f in 300 ~M small unilamellar vesicles prepared as described in Methods. The data were analyzed by fLxinga Ranmn component to the lifetime of.001 ns and allowing the fraction of the Raman component to float. This was necessary because the Raman fine of water with excitation at 295 nm occurs at 328 nm.
Fluorescence quenchers probe the microenvironments of fluorescent moieties, be they in proteins or biological membranes. Acrylamide is a neutral quencher which penetrates somewhat into the lipophilic interior of membranes and proteins, whereas iodide is an ionic quencher restricted to the aqueous phase and it can provide insight into the electrostatic environment around a fluorophore. For more precise location within the lipid bilayer of membranes, depth-dependent quenchers, such as spin-labelled fatty acids, had also been used.
128
Fluorescence quenching experiments with membrane-soluble agents were initially performed by adding small aliquots of doxyl fatty acids in ethanol or methanol, or acrylamide and iodide in buffer, to a fixed amount of antioxidant incorporated into various amounts of phospholipid/ cholesterol vesicles at a concentration of 200-500 ttM total phospholipid. Increasing the amount of vesicles resulted in a decrease in quenching for any given quencher concentration. However, the quenching was independent of phospholipid concentration when data were compared for identical quencher concentrations expressed in terms of mole quencher/mole phospholipid. This then shows that all the fluorescence phenomena occur exclusively in the membrane, i.e. within the experimental error of our techniques, the Lazaroids partition exclusively into the lipid phase and the contribution of any of the membrane-soluble agents in the aqueous phase is negligible. Subsequent quenching experiments were performed at only one concentration of vesicles. Acrylamide quenches all of the-membraneincorporated fluorescent Lazaroids, as shown in Fig. 2. Stern-Volmer plots for U-78,517f and
3.00
2.50
IJ_ ~2.00 0 It_ 1.50
1.00
O. r
...., !.I¸ i .....................................
Fig. 2. ~
fits to ~
U.78,517f.
oto
o~ o~ o~ ACRYLAMIDE, M
q ~
of ~ t m d
~
,,
""~:~o
eatioxi.
~ ; ~ A - ~ U.75Jl12e; O.~O, U-711,$10e;,~0~-O,
U-75,412e are linear and fully consistent with Eqn. (2) (Table IV), i.e. static quenching is not observable for this system. In contrast, the plot for U-78,518e is curving upward at higher concentrations, indicating either the presence of a static quenching component o r quenching within a sphere of action. We have analyzed the data according to both of these models and we found that the fit was significantly better for Eqn. (3) than for the equation corresponding to the simultaneous dynamic and static quenching. The agreement between the experimental points and the theoretical curve calculated from Eqn. (3) is shown in Fig. 2. The deviation from linearity allowed us to calculate the radius of the quenching sphere of action around U-78,518e to be 5.8 A. This value is in reasonable agreement with that determined for quenching of the indole ring by acrylamide [111.
The quenching constants in Table IV show that U-78,517f is the Lazaroid most readily quenched by acrylamide and that the other two compounds are about equally quenched. The concentration of acrylamide should be highest at the lipid/water interface and diminish progressively with increasing depth into the lipid layer. Our results thus indicate that, among our Lazaroids, the fluorescent moiety of membrane-bound U-78,517f is the closest to the aqueous phase and those of U-78,518e and U-75,412e are located somewhat deeper in the lipid phase. This conclusion is also supported by the fact that, as compared to the acrylamide quenching of many water-soluble fluorophores, the Ksv'S of the Lazaroids are low. Thus, the fluorescent moiety of all Lazaroids tested is imbedded in the lipid portion of the bilayer, as was previously demonstrated for vitamin E. In other words, the Lazaroids are located, in all likelihood, in the same general area as Vitamin E. It is worth noting that none of the Stern-Volmer plots in Fig. 2 curve downward. Downward curvature would be indicative of the presence of multiple fluorophore populations with different acc~sibilities, such as fluorophores in the inner and outer leaflets of the bilayer, or fluorophores in different states of protonation. One would thus conclude that none of the Lazaroids used is able to translocate across the
129 TABLE IV
Fluorescence parmneters governing the quenching and energy transfer for the Lazaroids in phmphofipid bilayers Experiment
Parameter
Acrylamide
Ksv (M -])
KI
K,v2 (M -I) j~
Nitroxyl fatty acids
5-DS
U-75AI2e
KL (l/mole%) Kc (l/mole%) Kd (mole%)
fc/fL 7-DS
KL (l/mole%) K¢ (l/mole%) Kd (mole%)
f¢lft. 10-DS
KL (l/mole%)
K¢ (I/mole%) K~l(mole%)
fo/fL Anthroyloxy-fattyacids
6-AS 9-AS 12-AS
Kd (Eqn. Kd (Eqn. /CO(Eqn. /~ (Eqn. Kd (Eqn. Kd (Eqn.
U78,518e
U-78,517f
0.58 4. 0.002
0.56 4. 0.12
1.44 4. 0101
0.46 ± 0.06 0.43 ± 0.05
0.70 4- 0.06 0.33 4- 0.03
0.25 ± 0.01
Chonlstry and Physics of Lipids, 62 (1992) 123"138 Elsevier ScientificPublishers Ireland Ltd.
Interaction of antioxidants with depth-dependent fluorescence quenchers and energy transfer probes in lipid bilayers Jessica S. Hinzmann, Rose L. McKenna, Tonya S. Pierson, Fusan Han, Ferenc J. K6zdy and Dennis E. Epps The Upjohn Company, 7000 Portage, Kalamazoo, It,ll 49001 (USA)
(Received December 16th, 1991; revision received March 26th, 1992; accepted April 24th, 1992) Three fluorescent, lipophilic, beterocyclicantioxidants were incorporated into lipid bilayers and exposed to depth-dependent nitroxyl fatty acid quenchers. The Stern-Volmerplots curved upward at low quencher concentrations.Quantitative analysis of the retorts showedthat this behavioris consistentwith complexformationbetweenquencherand fluorescentantioxidant,wherethe complex is 2-3 timesmore f l u ~ t than the parent fluorophore.At higher quencherconcentrations,both free antioxiclantand 'bright complex'are quencheddynamically,albeit quenchingof the latter is lessefficient.The complexprobably results from ionic, hydrogen bond and ~r-lrinteractiom. Formation of such a 'bright complex' is also observablein a homogeneoussolution of the reactants in cyclohexane.Additionalevidencefor the complexationof these antioxidants with fatty acids in fipid bilayersis provided by the fact that energytransfer from the antioxidantsto anthroyloxyfatty acids occursat surfaceconcentrationswhere radiativeenergytransfer betweenfree moleculesshould not be efficient.For directlyprobing the relativedepths of these fluorophoresin lipid bilayerswe used the aqueous quenchersacrylamideand iodide. They showedthat, in terms of increasingdepth in the bilayer,the order was U-78,517f < U-78,518e < U-75,412e.Our results, in toto, demonstrate that the Lazaroid antioxidants are incorporated into the lipid bilayer wherethey occupystrictlydefinedpositionsand orientations.Complexationwith fatty acylchains should be mechanisticallyrelevant, since it may enhance antioxidant activity by hindering free radical chain propagation. Key words: antioxidants; fluorescence;bilayers; quenching; energy transfer
Introduction Free radical-induced tissue damage occurs in a variety of diseases, such as myocardial reperfusion injury [1], traumatic head and spinal cord injury [2] and in carbon tetrachloride-induced liver toxicity [3]. The possibility that lipid peroxidation is the common cause o f these irreversible processes has led to the development o f antioxidants as potential therapeutic agents [4,5]. Since the agent ultimately responsible for cellular damage may be any one of a broad spectrum o f free radicals located in a variety o f compartments, the design o f a given antioxidant drug involves not only the optimization o f the intrinsic scavenging ability of the drug toward a Correspondsnce to: Dennlt E. Epps, The Upjohn Company, 7(~0 Portage, Kahummoo, MI 49001, USA.
given free radical, but also its partitioning into the target compartment. Reactive free radicals diffuse only short distances; propagation o f the free radical reaction can therefore be inhibited only in situ. For optimal action, t h e drug must be strategically located, either extracellularly to intercept radicals prior to diffusion into the cell, or within the cell or cell membrane in order to protect locally the target proteins, iipids and nucleic acids. For example, Vitamin E, a prominent biological free radical scavenger, is anchored in biological membranes by its phytol chain and is the major inhibitor of the peroxidation of membrane lipids [6]. In addition to its chemical action, vitamin E also modifies the host lipid bilayer's physical properties, especially viscosity and permeability. The strict localization of vitamin E in the lipid bilayer prevents it from reacting with free radicals
0009-3084/92/$05.00 1992 ElsevierScientificPublishers Ireland Ltd. Printed and Published in Ireland
124
essential for the functioning of the cell which are located in other compartments. Thus, a method of locating the antioxidants in membranes should help in optimizing drug design and in understanding the mechanism of action of antioxidants. Fluorescence spectroscopy is an ideal tool for determining the location of lipophilic fluorophores within biological membranes, either by measuring energy transfer to fatty acids tagged with anthroyloxy groups at different carbon atoms on the acyl chain, or by depth-dependent quenching by nitroxyl fatty acids [12]. Many of our experimental antioxidant drugs happen to contain fluorescent moieties and were readily amenable for similar studies. We chose to work with a model system of artificial bilayer lipid membranes composed of egg lecithin and cholesterol in a molar ratio of 4:1, which mimics the essential physical properties and permeability of cell membranes. The fluorescence properties of U-78,517f, U-78,518e and U-75,412e and their locations within the bilayer determined by fluorescence spectroscopy, are described herein.
U-78517F CH3
CH3
N~
CH3 • 2HCI • H20
U-78518E
CH3
~H2 ~=~
CH$
CH3 'ox HCI
U-75412E
~
H2.
NH
Materials a m l ~ All antioxidants were synthesized and purified at The Upjohn Company and obtained as solids from the Biological Screening Office. Their chemical structures are shown in Fig. 1. Vitamin E, naphthalene and p-terphenyl were purchased from Aldrich Chemical Company. Fluorescent and spin-labelled fatty acids were from Molecular Probes, Inc. 'Ultra Pure' acrylamide was from Schwartz Mann., egg lecithin from Avanti Polar Lipids and cholesterol from Sigma Chemical Company.
Phospholipid vesicles Small unilamellar phospholipid/cholesterol vesicles (4:1, tool:tool) were prepared by the ethanol injection procedure of Batzri and Korn [7]. Briefly, for a typical vesicle pre~ration, 100 rag of egg lecithin and 12.5 m g dmtesterol were dissolved in 2 ml ethanol and injected with rapid stirring into 50 ml 0.01 M Tris-HCl, 150 mM NaCI, pH 7.4. The solution wasooacentrated to a small volume by ultrafil~tion through an
C'-O ~ ' J
O
~
.
'CH3
Fig, 1. Chemicalstructuresof antioxidant~
A m i s h YM-100 ~ t e r and applied to a 2.5 x 95 cm Selthatose CL-48 c o n m . Fractions were eluted with the same buffer at a flow rate of 0.46 mi/min and the eluate was monitored at 210 nm. Peak fractions were pooled and concentrated to - 1 0 ml by ultra~tration, as described above. IAI~i ~ s eontmt of the coneen. tra..auivo0kle ~ ,wan detetmtn~ rain8 the premdme of Bins!eft[8].
Spectral measurements
For UV/VlS 8 ¢ ~ r 0 1 ~ ~ t ~ 2200 ~ t n t m e n t i n ~
we ~ 0 a C . ~
with a ~
~m-
125
purer and controlled by Spectra Calc software. Steady-state fluorescence measurements were made in the ratio mode, using either a Hitachi F2000 or a Perkin Elmer MPF66 instrument and the spectral data were processed using the software packages supplied by the manufacturers. Fluorescence quantum yields of our compounds, Ounk, were calibrated with naphthalene in ethanol (~ref = 0.21 ) or with quinine sulfate in 0.1 M sulfuric acid (Oref = 0.55 ), at the excitation wavelength of 295 nm for all compounds. Quantum yields were calculated from the relationship ~unk = ~ref ×
lunk/Iref X Aref/Aun k
(1)
where lunk and lref are the integrated fluorescence intensities and Aunk and Aref the absorbances at the exciting wavelength.
Dynamic fluorescence measurements Multifrequency fluorescence measurements of the intrinsic fluorescence of the antioxidants given in Fig. 1 were made using a cross correlation phase and modulation fluorometer. The harmonic content of a high repetition rate mode-locked NdYAG laser was used to generate frequencies from 20 to 300 MHz. The Nd/YAG laser (Spectra Physics) was used to synchronously pump a dye laser (Rhodamine 6G) whose pulse train was frequency-doubled with an angle-tuned frequency doubler utilizing a KDP crystal [9]. The autocorrelated pulse width was 10-12 ps and the cross correlation frequency was 25 Hz. The dye laser was tuned to the emission maximum of the dye, 590 nm, which, upon doubling, yielded an ultraviolet light of 295 nm (2-5 mW) for excitation. The latter was passed through a half-wave plate prior to entering the sample compartment to eliminate any horizontally polarized light. Data were collected with the emission polarizer set at the magic angle of 55° and emission was observed through interference filters with 10 nm bandpasses. Usually 9-10 different modulation frequencies were used and data were collected until phase and modulation deviations reached less than 0.2 and 0.004, respectively. Temperature was maintained at 25° using a Haake circulating bath; lifetime least squares analysis was performed using
the software provided by ISS, Inc., Champaign, IL. Absorbances at 295 nm were kept below 0.09 optical density (OD) to eliminate absorptive screening. A solution of p-terphenyl in cyclohexane 0" = 1 ns) was used as the reference lifetime standard.
Fluorescence quenching The fluorescent antioxidants were added in small aliquots of ethanol or DMSO to the vesicle~ and allowed to equilibrate for 5-10 rain before use. All operations were performed at 25°C and absorbances of the antioxidant/vesicle complexes were kept at < 0.1 to preclude absorptive screening. Fluorescence intensities were measured at 330 nm for U-78,517fand at 380 nm for U-78,518e and U-75,412e. Measurements were signal-averaged for 10-16 s and corrected for dilution. The data for quenching by acrylamide were fitted to the linear form of the Stern-Volmer equation: Fo = 1 + Ksv [Q] F
(2)
where Ksv is the Stern-Volmer quenching constant and [Q] is the quencher concentration. Alternatively, F as a function of [Q] was analyzed by non-linear least squares fitting. For upwardcurving Stern-Volmer plots, the equation should either include a static quenching component, or take into account the quenching within a 'sphere of action': Fo = (1 + Ksv [Q]) e vQ F
(3)
where V is the volume of the sphere within which the probability of quenching is unity.
Energy transfer to anthroyloxy fatty acids In one set of experiments, anthroyloxy fatty acids were added in ethanol to 300-400 /~M phospholipid vesicles in buffer to yield a molar ratio of 1:250 and allowed to equilibrate for 1-4 h before use. Individual Lazaroids were then added and after l rain, the fluorescence emission of the acceptor was measured at 450 nm with excitation at 295 nm. All fluorescence intensities were corrected for dilution. In a second set of experiments,
126
energy transfer was measured by determining the loss of donor emission in the presence of increasing amounts of acoeptor. Because of a slight overlap between the emission spectra of the donors and the acceptor, the fluorescence emission was measured not only at a wavelength where the emission of the donor dominates, but also at one where the emission of the acceptor is preponderant. Using the known quantum yields of emission at these wavelengths, the data were then corrected for the slight contribution of the acceptor.
imated from the following equation: 02(/) = a2(/)countins statistics+ (0.0218 /)2, where the coefficient of [ was calculated from the variation in intensities of the monitored reflections. Cell parameters were determined by least-squares fit of CuKoq 2q values (X(CuKa0 = 1.5402 A) for 25 high-20 reflections. Lp corrections appropriate for a monochromator with 50% perfect character were applied. The calculated overall absorption coefficient was 2.03/ram. Remits mal l)i~anlea
Crystal structure The bulk drug lot (A)0309FCWI40A of U-83,836e, the negative enantiomer of U-78,517f, was recrystallized using the methanol-heptane 'Binary Vapor Diffusion System' method. Although crystals could also be obtained by the methanoi-acetoue system, only crystals from the first system were stable during the mounting procedure. A clear, chunky, prism-shaped crystal of 0.2 x 0.2 x 0.5 mm size, obtained from methanol-heptane, was selected and mounted on a glass fiber. Diffraction data were collected at -150°C, using a Nicolet P21 X-ray diffractometer controlled by a Harris computer, with graphite-monochromatized CuK~ radiation (~(CuKa)= 1.5405 A). All 3146 unique reflections were measured to a 20m~ of 138° for Laue group mmm; 2597 intensities were >3a. The 0/20 step-scan technique was used with a scan speed of 4°/min and a scan width >3.4 °. Ten reflections were periodically monitored and showed no deterioration. Standard deviations in the intensities were approx-
The spectroscopic properties of our antioxidants, relevant to this study, are shown in Table I. The excitation maxlma of U-78,517f, U-78,518e and U-75,412e were essentially invariant as a function of solvent. The emission maxima were somewhat solvent dependent, but not sufficiently Stokes-shifted to classify the Lazaroids as 'environment-sensitive' probes. It is interesting to note that there is only one fluorescence maximum for U-78,518e, even thoegh, taken individually, both the chromanol moiety (as in Vitamin E) and the piperazinylpyridinyl ring system are fluorescent. With excitation at 295 nm, a very small peak is observed at approximately 330 nm, corresponding to that of the chromanol moiety. However, it is evident that the greatest proportion of the energy emitting from the chromanoi moiety is transferred to that of the piperazinylpyridinyt ring via a F6rster energy transfer mechanism, since the emission maxima of this compound is > 386 ran in all solvents. Thus only the fluorescence of the
TABLE I Fluorescence excitation and emission maxima of antioxidants in various solvents Compound
Ethanol
Methanol
Isopropanol
Cyclohexane
Vesides a
Xm U78,517f U78,518e U75,412e
285 310 310
326 405 391
296 312 311
aEgg l e c i t h i n / ¢ h o ~ (4:1, mol/mot) v e s k J ~ ' ~
332 399 385
296 309 310
326 386 385
as ~descri~in ~ .
298 310
336 401 388
292 310
328 387 391
127 TABLE II Quantum .yields (4D)of Lazaroids in variom solvents Compound
Ethanol
Methanol
Isopropanol
Cyclohexane
Vesicles
U78,517f U78,518e U75,412e
0.083 0.08 0.024
0.029 0.044 0.022
0.044 0.123 0.04
0.03 0.143 0.055
0.07 0.23 0.104
piperazinylpyridinyl ring system is observed; it is this ring system which is also the origin of the fluorescence of U-75,412e. In addition, the excitation and emission maxima of the fluorescent Lazaroids are in a range where one should expect significant energy transfer to membrane localized probes, such as depth-dependent anthroyloxy fatty acids. The Lazaroids could thus act as intrinsic probes for the determination of their membrane localization, without significant perturbation of the bilayer by additional bulky extrinsic probes. Fluorescence quantum yields (~) as a function of solvent are shown in Table II. In general, they are dependent upon the hydrophobicity of the solvent. On the other hand, the relatively high quantum yield observed for each compound inserted into unilamellar vesicles is due not only to the solvent environment, but also to the rigidity of the fluorophore within the bilayer. The solvent dependence of the fluorescence lifetime of U-78,517f is shown in Table III. The data were analyzed allowing for a finite Raman TABLE IlI Fluorescence lifetimes 0") of U-78,517f (ns) c
contribution to the total lifetime. As expected, the Raman contribution to the total lifetime is greatest for vesicles, 5.4% of the total lifetime. The lifetime of U-78,517f in vesicles is intermediate between that of the compound in ethanol and in methanol. The fluorescent chromanol moiety of U-78,517f is identical to that of Vitamin E and it would be expected that the fluorescence lifetimes would be similar. Indeed, this was found to be the case: the lifetime of U-78,517f in ethanol, 1.8 ns for TI is identical to that for Vitamin E, 1.8 ns [12]. In addition, the lifetime of U-78,517f in small egg lecithin/cholesterol unilamellar vesicles is 1.49 ns, comparable to that determined for Vitamin E in egg lecithin vesicles, 1.31 ns. Thus, substitution of the phytol chain of Vitamin E by the dipyrrolidinyl-pyrimidinyl-piperazinyl ring system does not appreciably alter the fluorescence lifetime of the chromanol moiety. The lifetime measurements also indicate that the fluorescence of U-78,517f is not very environment-sensitive. The lifetimes of the piperazinylpyridinyl moieties of U-78,518e and U-75,412e were too short to be measured with our instrumentation.
Fluorescence quenching Solvent
~'1
f] a
Tavgb
Methanol Ethanol Vesicles
1.26 1.8 1.49
0.986 0.976 0.946
1.21 1.76 1.41
"Weighted average of the two lifetimes. bFraction of the total lifetime in the slow component. c2 mole% U-78,517f in 300 ~M small unilamellar vesicles prepared as described in Methods. The data were analyzed by fLxinga Ranmn component to the lifetime of.001 ns and allowing the fraction of the Raman component to float. This was necessary because the Raman fine of water with excitation at 295 nm occurs at 328 nm.
Fluorescence quenchers probe the microenvironments of fluorescent moieties, be they in proteins or biological membranes. Acrylamide is a neutral quencher which penetrates somewhat into the lipophilic interior of membranes and proteins, whereas iodide is an ionic quencher restricted to the aqueous phase and it can provide insight into the electrostatic environment around a fluorophore. For more precise location within the lipid bilayer of membranes, depth-dependent quenchers, such as spin-labelled fatty acids, had also been used.
128
Fluorescence quenching experiments with membrane-soluble agents were initially performed by adding small aliquots of doxyl fatty acids in ethanol or methanol, or acrylamide and iodide in buffer, to a fixed amount of antioxidant incorporated into various amounts of phospholipid/ cholesterol vesicles at a concentration of 200-500 ttM total phospholipid. Increasing the amount of vesicles resulted in a decrease in quenching for any given quencher concentration. However, the quenching was independent of phospholipid concentration when data were compared for identical quencher concentrations expressed in terms of mole quencher/mole phospholipid. This then shows that all the fluorescence phenomena occur exclusively in the membrane, i.e. within the experimental error of our techniques, the Lazaroids partition exclusively into the lipid phase and the contribution of any of the membrane-soluble agents in the aqueous phase is negligible. Subsequent quenching experiments were performed at only one concentration of vesicles. Acrylamide quenches all of the-membraneincorporated fluorescent Lazaroids, as shown in Fig. 2. Stern-Volmer plots for U-78,517f and
3.00
2.50
IJ_ ~2.00 0 It_ 1.50
1.00
O. r
...., !.I¸ i .....................................
Fig. 2. ~
fits to ~
U.78,517f.
oto
o~ o~ o~ ACRYLAMIDE, M
q ~
of ~ t m d
~
,,
""~:~o
eatioxi.
~ ; ~ A - ~ U.75Jl12e; O.~O, U-711,$10e;,~0~-O,
U-75,412e are linear and fully consistent with Eqn. (2) (Table IV), i.e. static quenching is not observable for this system. In contrast, the plot for U-78,518e is curving upward at higher concentrations, indicating either the presence of a static quenching component o r quenching within a sphere of action. We have analyzed the data according to both of these models and we found that the fit was significantly better for Eqn. (3) than for the equation corresponding to the simultaneous dynamic and static quenching. The agreement between the experimental points and the theoretical curve calculated from Eqn. (3) is shown in Fig. 2. The deviation from linearity allowed us to calculate the radius of the quenching sphere of action around U-78,518e to be 5.8 A. This value is in reasonable agreement with that determined for quenching of the indole ring by acrylamide [111.
The quenching constants in Table IV show that U-78,517f is the Lazaroid most readily quenched by acrylamide and that the other two compounds are about equally quenched. The concentration of acrylamide should be highest at the lipid/water interface and diminish progressively with increasing depth into the lipid layer. Our results thus indicate that, among our Lazaroids, the fluorescent moiety of membrane-bound U-78,517f is the closest to the aqueous phase and those of U-78,518e and U-75,412e are located somewhat deeper in the lipid phase. This conclusion is also supported by the fact that, as compared to the acrylamide quenching of many water-soluble fluorophores, the Ksv'S of the Lazaroids are low. Thus, the fluorescent moiety of all Lazaroids tested is imbedded in the lipid portion of the bilayer, as was previously demonstrated for vitamin E. In other words, the Lazaroids are located, in all likelihood, in the same general area as Vitamin E. It is worth noting that none of the Stern-Volmer plots in Fig. 2 curve downward. Downward curvature would be indicative of the presence of multiple fluorophore populations with different acc~sibilities, such as fluorophores in the inner and outer leaflets of the bilayer, or fluorophores in different states of protonation. One would thus conclude that none of the Lazaroids used is able to translocate across the
129 TABLE IV
Fluorescence parmneters governing the quenching and energy transfer for the Lazaroids in phmphofipid bilayers Experiment
Parameter
Acrylamide
Ksv (M -])
KI
K,v2 (M -I) j~
Nitroxyl fatty acids
5-DS
U-75AI2e
KL (l/mole%) Kc (l/mole%) Kd (mole%)
fc/fL 7-DS
KL (l/mole%) K¢ (l/mole%) Kd (mole%)
f¢lft. 10-DS
KL (l/mole%)
K¢ (I/mole%) K~l(mole%)
fo/fL Anthroyloxy-fattyacids
6-AS 9-AS 12-AS
Kd (Eqn. Kd (Eqn. /CO(Eqn. /~ (Eqn. Kd (Eqn. Kd (Eqn.
U78,518e
U-78,517f
0.58 4. 0.002
0.56 4. 0.12
1.44 4. 0101
0.46 ± 0.06 0.43 ± 0.05
0.70 4- 0.06 0.33 4- 0.03
0.25 ± 0.01
