Biochemistry 1988, 27, 4639-4645
`
`4639
`
`Fluorescence Indicates a Calcium-Dependent Interaction between the Lipopeptide
`Antibiotic LY 146032 and Phospholipid Membranes
`
`Jeremy H. Lakey* and Marius Ptak
`Centre de Biophysique MolZculaire, CNRS, and UniversitZ dOrlZans, I A, Avenue de la Recherche Scientifique, 45071 Orleans
`Cgdex 2, France
`Received June 15, 1987; Revised Manuscript Received January 21, 1988
`
`ABSTRACT: LY 146032 is one of the A21978C family of calcium-dependent antibiotics. This paper reports
`on its interactions with membranes as studied by its intrinsic fluorescence. The Trp residue was found to
`have a low fluorescence yield because of Forster-type energy transfer to the kynurenine residue (Kyn) ( E
`= 465 nm in HzO) was a sensitive probe of the
`= 5000 a t 364 nm). However, the Kyn fluorescence (A,,,
`membrane interactions, and it was used in steady-state fluorescence measurements including fluorescence
`polarization anisotropy. Initial binding of the peptide to phospholipid vesicles occurs in calcium-free solutions.
`When calcium is added, the resulting 10-fold fluorescent enhancement and 15-nm blue shift show that it
`causes the antibiotic to penetrate further into the lipid bilayer. Calcium is bound with an association constant
`of 151 M-l, while a phospholipid titration in the presence of calcium gave an association constant of 5 X
`1 O3 M-' for egg phosphatidylcholine. Magnesium and cadmium cause very slight fluorescence enhancements,
`but a more significant effect is caused by the trivalent lanthanide ions. Analysis of these data indicates
`that the calcium-selective site is on the peptide and that ion binding to the phospholipid headgroups has
`a secondary role. Comparison with the divalent cation dependent antibiotics bacitracin and amphomycin
`shows that LY 146032 has a quite different activity and that a calcium-dependent membrane interaction
`could account for results obtained in vivo.
`
`L Y 146032 (Figure l), an acidic lipopeptide antibiotic
`produced by Streptomyces roseosporus, is effective against
`a range of Gram-positive pathogenic bacteria (Eliopolous et
`al., 1985). However, its in vitro antibiotic activity is unusual
`because it requires the presence of calcium ions. For example,
`it has been shown that in the case of Streptococcus faecalis
`the addition of 1 mM calcium chloride to the culture medium
`causes a 30-fold decrease in the minimum inhibitory concen-
`tration (Eliopolous et al., 1985). This effect was found to be
`unique to calcium, and media supplements of Ba2+, Zn2+ and
`Mg2+ had no comparable ability to lower the minimum in-
`hibitory concentration. LY146032 has been shown to inhibit
`test infections in mice, but clearly it was not possible to observe
`the effect of a varying Ca2+ concentration.
`The possibility that its action was that of a calcium iono-
`phore was excluded by experiments on black lipid films and
`phospholipid vesicles. These showed that although it did in-
`crease the bilayer conductivity, this increase was neither
`calcium dependent nor sufficiently large to account for the
`observed antibacterial activity (Lakey & Lea, 1986). These
`experiments did show, however, that in common with other
`lipopeptides, e.g., polymixins, octapeptins (Storm et al., 1977),
`and iturins (Maget-Dana et al., 1985), LY146032 interacts
`strongly with bilayer membranes (Lakey & Lea, 1986). This
`common property of lipopeptides is thought to be due to in-
`sertion of the fatty acyl moiety into the membrane, and as a
`group their modes of action are varied but always membrane
`based. If the action of LY146032 involves a membrane-based
`step, it serves a purpose other than directly increasing mem-
`brane permeability, and so another measure of this interaction
`
`is required. In this paper we use the intrinsic fluorescence of
`LY 146032 to study its interactions with phospholipid vesicles
`and calcium.
`The peptide sequence of LY 146032 is identical with that
`of the A21978C family of peptides produced by S. roseosporus
`(Debono et al., 1987). They differ solely in the length of their
`fatty acyl tails, LY 146032 having a straight-chain decanoyl
`side chain (Figure 1). As the chemical structure of the cal-
`cium-dependent antibiotic (CDA) from Streptomyces coeli-
`color (Lakey et al., 1983) is unknown, its relationship with
`LY146032 is unclear.
`
`EXPERIMENTAL PROCEDURES
`LY146032 was a gift from Lawrence Day of Eli Lilly &
`Co. Egg yolk phosphatidylcholine was purified according to
`the method of Singleton et al. (1965). DMPC' and DPPC
`were purchased from Sigma, verified pure by thin-layer
`chromatography, and used without further purification. GdC13
`was from UCB, Brussels, Belgium, and EuC1, was from Fluka.
`L-Kyn was purchased from Sigma, and all other chemicals
`were of the best available grade.
`Small unilamellar vesicles (SUV) were prepared by soni-
`cation. Phospholipid in ethanol was dried as a thin film in a
`glass cup, a standard buffer solution (standard buffer, 200 mM
`KCl/10 mM cacodylate, pH 6.0) was added, and sonication
`by an MSE sonicator was carried out under nitrogen. Sub-
`sequently, the vesicles were centrifuged for 1 h at 40 000 rpm
`(1 30000g) to remove multilamellar liposomes and titanium
`fragments. All stages of the preparation were performed above
`the phase transition temperature for the lipid concerned. PC
`
`*Address correspondence to J.H.L. at the European Molecular Bi-
`ology Laboratory, Meyerhofstrasse, Postfach 10.2209, D-6900 Heidel-
`berg, FRG. J.H.L. Thanks the Science and Engineering Research
`Council (Swindon, U.K.) for the award of a NATO postdoctoral fel-
`lowship.
`
`Abbreviations: SUV, small unilamellar vesicles; PC, phosphatidyl-
`choline; DMPC, L-a-dimyristoyl-PC; DPPC, L-a-dipalmitoyl-PC; L-Kyn,
`free kynurenine; P-Kyn, kynurenine residue in LY 146032; FPA,
`fluorescence polarization anisotropy; T,, phase transition temperature.
`
`0006-2960/88/0427-4639$01.50/0
`
`0 1988 American Chemical Society
`
`
`1 of 7
`
`FRESENIUS-KABI, Exh. 1033
`
`

`
`4640 B I O C H E M I S T R Y
`
`L A K E Y A N D P T A K
`
`t
`L - O r n
`f
`
`"'y,
`
`L - A S P
`N
`*
`D - A l a
`f
`L -Asp
`
`
`G ~ Y
`\
`D - Ser
`I
`(L - threo) 3 - MeGlu
`1
`L - Kyn
`.oAo
`
`L - Thr
`t
`t
`L - A S P
`L - A m
`t
`L -Trp
`I
`NH
`
`FIGURE 1: Structure of LY146032 (decanoyLA21978C): Kyn, ky-
`nurenine; Orn, ornithine; Me-Glu, methyl glutamate.
`
`concentrations were determined by measuring the lipid
`phosphorus according to the method of Chen et al. (1956).
`Chemical Modification of LY146032. Acetylation of the
`aromatic amine on the kynurenine residue of LY146032 was
`carried out as follows: 0.5 mg of LY146032 was dissolved in
`a solution of 330 pL of distilled water, 100 pL of 1 M HCl,
`and 60 p L of acetic anhydride by vortexing. After the addition
`of 20 mg of solid sodium acetate, the mixture was vortexed
`again and then purified by the method of Bohlen et al. (1980)
`with 2-propanol/water (7/1) as the eluant.
`Fluorescence Techniques. Fluorescence measurements were
`carried out on a Kontron SFM25 with slits set to 5 nm (ex-
`citation) and 10 nm (emission). The spectrofluorometer was
`interfaced with an Apple IIe microcomputer for data collection
`and manipulation. Temperature control was by circulating
`water through the four-cell holder in which the temperature
`was measured by a thermocouple placed in a buffer-filled
`cuvette.
`The fluorescence results are expressed as either the relative
`fluorescence at a given wavelength (F) or, where possible, the
`relative quantum yield (@) calculated as the integrated
`fluorescence emission between 400 and 520 nm.
`Polarization experiments used the SFM25 polarization
`accessory under remote control via the Apple IIe. The
`fluorescence intensity components (Iw, Ivh, Ihv, Ihh), in which
`the subscripts refer to the horizontal (h) or vertical (v) pos-
`itioning of the excitation and emission polarizers, respectively,
`were used to calculate the steady-state fluorescence polari-
`zation anisotropy (FPA):
`FPA = ( I , - IvhG)/(zw + 21,hG)
`(1)
`where G is the grating factor that corrects for wavelength-
`dependent distortions of the polarizing system.
`(2)
`= I h v / I h h
`All experiments were carried out in 1 cm path length cu-
`vettes containing at least 2.5 mL of sample with an absorption
`at 364 nm of <0.1. Absorption measurements were performed
`on a Beckman DU-8 or an LKB Biochrom spectrometer.
`The association of LY 146032 with phosphatidylcholine,
`Ca2+, and Gd3+ (as ligands) was analyzed by the method of
`Lehrer and Fasman (1966). The association constant K, is
`related to the sample fluorescence at a particular ligand
`concentration [SI by
`Ka = [ ( F - F o ) / ( F ~ - F)I(l/[sI)
`(3)
`where the fluorescence intensities of the peptide alone, the
`peptide at [SI, and the peptide at an infinite concentration of
`S are denoted by Fo, F, and F,, respectively. F, was calculated
`
`FIGURE 2: Absorption spectra of native and modified LY146032 in
`water: solid line, unmodified LY146032; broken line, LY146032 after
`acetylation of its kynurenine residue.
`I
`I
`
`c l
`
`
`
`, n
`
`I
`I
`
`I
`
`Excitation wavelength
`FIGURE 3: Excitation spectra of LY 146032 kynurenine fluorescence
`in 1 mg/mL egg PC SW: F(450 nm), relative fluorescence at 450-nm
`emission wavelength without subtraction of the SUV base line; (-)
`standard buffer plus 50 mM CaCI,; (- - -) standard buffer plus 50
`mM CaCl, and 15 mM GdC13. The excitation spectrum of free t-Kyn
`in standard buffer is also shown (--) although not on the same scale.
`by extrapolation of the plot 1/(F - Fo) versus 1 /[SI to 1 /[SI
`= 0. In all cases, the plot of (F - Fo)/(Fm - F) versus [SI fitted
`to a straight line at nonsaturation concentrations of S , and Ka
`was calculated as the inverse of [SI at the intersection with
`(F - Fo)/(F.. - F) = 1 (Privat et al., 1974). The treatments
`according to Kolber and Haynes (1981) and Kauffman et al.
`(1983) were as described in the original references.
`
`RESULTS
`As shown in
`LYI 46032 Fluorescence Characteristics.
`Figure 1, the peptide contains four negatively charged groups
`[three aspartate and one methyl glutamate (Me-Glu)], one
`positively charged ornithine (Orn), and two fluorophores.
`These comprise a tryptophan residue situated between the lipid
`chain and the peptide headgroup and a kynurenine residue
`(Kyn) in the headgroup itself. The situation of the Trp should
`make it sensitive to insertion of the lipid moiety into a hy-
`drophobic bilayer, but unfortunately, this residue has a very
`poor fluorescence yield. When a solution of LY146032
`M peptide in standard buffer) was excited at 285 nm, the Trp
`emission was only 3.5% that of free L-Trp under the same
`conditions. The absorption spectrum clearly indicates (Figure
`2) a Trp (e = 5000 at 286 nm) concentration equal to that
`of the peptide, while the 'H NMR spectrum (D. Marion, A.
`Caille, J. Lakey, and M. Ptak, unpublished results) showed
`Trp and kynurenine to be present in equal proportions, thus
`excluding the possibility of Trp degradation having occurred
`(Creed, 1984). Furthermore, the excitation at 285 nm pro-
`duced a second emission at 465 nm, which was more intense
`when excited at 365 nm, indicating that Kyn was the second
`emission source. The excitation spectrum (Figure 3) for this
`Kyn emission clearly shows a strong Trp component around
`285 nm that is in between the Kyn absorbtion maxima at 256
`and 364 nm. It was thus assumed that the poor Trp emission
`is at least partly due to energy transfer. This Forster type of
`transfer results from the combination of the proximity of the
`two fluorophores and the overlap of the Trp emission spectrum
`
`
`2 of 7
`
`

`
`C A L C I U M - D E P E N D E N T A N T I B I O T I C L Y 1 4 6 0 3 2
`
`V O L . 2 7 , N O . 1 3 , 1 9 8 8 4641
`
`salt
`
`concn
`(mM)
`
`Table I: Effect of Multivalent Ions on P-Kyn Fluorescence
`re1
`quantum
`yield
`1
`0.80
`0.92
`0.98
`2.62
`1 .o
`10.7
`1.7
`1.2
`5.4
`
`LY 146032
`M in std buffer) CaC12
`CdC12
`MgCI2
`GdCIS
`LY 146032
`( M + 1 mg/mL CaCI2
`egg PC in std
`CdCl2
`buffer)
`MgC12
`GdClq
`
`50
`50
`50
`15
`
`50
`50
`50
`5
`
`max
`emission
`(nm)
`465
`465
`465
`46 5
`456
`462
`450
`462
`462
`452
`
`l b
`
`" T 0
`
`4b 50
`30
`20
`[Ca"]/mM
`FIGURE 5: LY 146032 kynurenine fluorescence enhancement in
`phosphatidylcholine SUVs by calcium ions. Samples contained 1
`mg/mL egg PC SUV and
`M LY146032 in standard buffer.
`Calcium was added from 1 or 3 M CaC1, stock solution and mixed,
`and the fluorescence yield was measured as the integral between 400
`and 520 nm after base line subtraction. The fluorescence increase
`was instantaneous on this time scale. A,,,, wavelength of maximum
`emission; 4, fluorescence yield; &, fluorescence yield in the absence
`of multivalent ions; A,,,
`365 nm.
`
`[Gd3+ I/ mM
`FIGURE 6: Fluorescence and light scattering increases in LY146032
`solution on GdCl, addition. ( 0 ) Fluorescence expressed as the fraction
`of maximum fluorescence attained (FIF,). The cuvette contained
`M LY146032 in standard buffer. Gd3+ was added
`2.5 mL of
`from 0.5 M GdC13 stock. A,,,
`365 nm; A,,,
`455 nm. (A) Same
`conditions as above except that A,, = A,,
`= 480 nm so that the
`measured signal is a product of the solution's light scattering properties.
`The increases were not immediate, and those plotted (S and F) are
`the values 30 min after Gd3+ addition.
`P-Kyn emission combined with a blue shift in its maximum
`(Table I; Figures 4 and 5). The effect was the same when
`calcium acetate was used instead. The order of size of effect
`was again Ca2+ >> Cd2+ > Mg2+ (Table I).
`Effect of Lanthanide Ions. When their use as possible
`quenching agents was investigated, it was discovered that the
`lanthanide ions Eu3+ and Gd3+ also showed very specific in-
`teractions with LY146032. As shown in Table I, GdCl, has
`the opposite effect to CaCl, on the free peptide, causing an
`increase in P-Kyn's relative quantum yield. When followed
`simultaneously by light scattering measurements, the
`fluorescence increase was seen to be accompanied by the
`formation of aggregates (Figure 6), which may render P-Kyn
`into a more hydrophobic environment. At saturating con-
`centrations, in egg PC SUV, GdCl, and EuCI, cause smaller
`increases and blue shifts in P-Kyn emission than does CaC1,
`(Table I; Figure 7). If GdC1, is added to
`M LY146032,
`1 mg/mL egg PC, and 50 mM CaCl,, the fluorescence in-
`creases again with a second saturation at 5 mM GdCl, (Figure
`3). A like result is obtained when calcium ions are added to
`a suspension containing 15 mM GdCl,. Thus, it appears that
`the action of one species does not impede the activity of the
`other.
`The data from Figures 5 and 7, when analyzed as described
`earlier, gave association constants (K,) for the Ca2+ and Gd3+
`effects as 151 M-' and 6.6 X lo3 M-I, respectively. This
`assumes that the fluorescence enhancement is proportional to
`the quantity of bound ion.
`
`F
`
`400
`
`Emission wavelength
`FIGURE 4: Emission spectra of LY 146032 kynurenine fluorescence
`in 1 mg/mL egg PC SUV. F is the relative fluorescence after sub-
`traction of the SUV base line. Spectra show the increase in
`fluorescence as the calcium ion concentration is increased stepwise
`from 0 to 10 mM CaCI2 in standard buffer. Excitation wavelength,
`365 nm.
`
`and Kyn absorption spectrum (Weinryb & Steiner, 1971).
`Chemical Modification of Kyn. In order to test this hy-
`pothesis and perhaps in so doing increase the available Trp
`signal, we changed the absorption characteristics of the Kyn
`residue by acetylation of the free NHz group. This resulted
`in a decrease in the Kyn absorbance (Figure 2), which together
`with the Kyn fluorescence was blue shifted by 20 nm. As the
`Kyn absorption band was initially at a longer wavelength
`( ~ 3 6 0 nm) than the Trp emission (-350 nm), this shift alters
`only slightly the overlap of these two functions. However, the
`combined changes in the Kyn absorption result in a 25% in-
`crease in the Trp emission, which although confirming the
`energy-transfer hypothesis does little to enhance Trp's use-
`fulness for studies in vesicles. Other modifications (carba-
`mylation, benzoylation) were tried, but acetylation proved the
`most effective. All subsequent experiments were carried out
`on the unmodified peptide.
`Kynurenine Fluorescence. Kyn shows only a very weak
`fluorescence as a free amino acid (L-Kyn), but when incor-
`porated into the peptide (P-Kyn), its relative quantum yield
`is 7.6 times greater. As such it is slightly more luminescent
`than P-Trp at
`M, although at higher peptide concentra-
`tions the ratio of fluorescence yield P-Kyn/P-Trp increases,
`possibly due to aggregation effects (IH NMR in water at pH
`6.0 reveals significant line broadening above 5 mM peptide).
`In 1 mg/mL egg PC vesicles the P-Kyn
`M) yield in-
`creases by 2.5-fold and thus becomes useable in studies of
`lipopeptide/vesicle binding. This is further helped by the
`>80-nm difference in its absorption and emission wavelengths,
`which reduces problems due to vesicle light scattering.
`Effect of Calcium Ions. When added to the free peptide
`in solution, divalent cations reduced the P-Kyn fluorescence,
`and the size of this quenching diminished in the order Ca2+
`>> Cd2+ > Mgz+ (Table I). There was no observable effect
`on the wavelength of maximum emission, the absorption
`spectrum, or the extent of aggregation as viewed by light
`scattering. When mixed with egg PC vesicles, the result is
`the reverse with 50 mM CaCl,, causing a 10-fold increase in
`
`
`3 of 7
`
`

`
`4642 B I O C H E M I S T R Y
`
`4 6 0 J
`
`5 T
`
`0
`
`1
`
`5
`
`
`
`4
`3
`2
`[Gdfl/mM
`FIGURE 7: LY 146032 kynurenine fluorescence enhancement in
`phosphatidylcholine SUVs by gadolinium ions. Samples contained
`M LY146032 in standard buffer.
`1 mg/mL egg PC SUV and
`Gadolinium was added from 0.5 M GdCI3 stock solutions and mixed,
`and the resultant mixture was analyzed as in Figure 5. The
`fluorescence increase was instantaneous on this time scale.
`
`FIGURE 8: Influence of surface charge on LY146032 kynurenine
`fluorescence enhancement in phosphatidylcholine SUVs by calcium
`ions: ( 0 ) 2.5 mL of 1 mg/mL egg PC SUV and M LY146032
`in standard buffer; (A) 2.5 mL of 1 mg/mL egg PC/stearylamine
`(9:l) SUV and
`M LY146032 in standard buffer; (B) 2.5 mL
`of 1 mg/mL egp PC/PI (9:l) SUV and lV5 M LY146032 in standard
`buffer. For other details see Figure 5 .
`The influence that surface charge has on the Ca2+ effect
`is indicated in Figure 8, which shows the results of adding a
`10% mole fraction of phosphatidylinositol (PI) or stearylamine
`(ST) to the egg PC used to make the vesicles. The presence
`of negative charges (PI) reduces the response to Ca2+ addition,
`and the form of the curve is changed, with no observed satu-
`ration below 50 mM. The enhancement in positively charged
`bilayers is greater than in PC alone. The initial value of F
`has been normalized to 1 in all cases, and so differences due
`to charge in the absence of Ca have been ignored. On average
`(four experiments), the P-Kyn fluorescence at zero Ca2+ in
`negative bilayers was 88% that of the positive preparations.
`Gadolinium chloride was similarly affected by the charges,
`and its supplementary increase of P-Kyn emission in 50 mM
`CaCl, remained.
`As the P-Kyn yield is dependent upon both Ca2+ and PC,
`it is important to consider whether the peptides' affinity for
`PC is calcium dependent. In this respect it is possible to
`measure an association constant between fluorophores and
`membranes by fluorometric titration of the molecule in solution
`with phospholipid vesicles (Kauffman et al., 1983; Kolber &
`Haynes, 1981). The data shown in Figure 9 were analyzed,
`like the ion binding data, according to the method of Lehrer
`and Fasman (1966), and this gave a K , of 5.0 X lo3 M-'
`[application of the analogous method of Kauffman et al.
`(1983) and Kolber and Haynes (1981) gave results within 0.3
`X lo3 M-'1. Unfortunately, however, a comparison with the
`binding of PC in the absence of Ca2+ proved to be impossible
`due to the feeble fluorescence enhancement involved.
`
`L A K E Y A N D P T A K
`
`0.3
`
`FIGURE 9: Effect of phosphatidylcholine and calcium ion titrations
`on LY 146032 kynurenine fluorescence emission and fluorescence
`polarization anisotropy (FPA). DMPC titration (left-hand figure):
`(m) fluorescence increase (F) of
`M LY 146032 in standard buffer
`plus 50 mM CaCI, on addition of aliquots of 10.4 mg/mL DMPC
`SUV, M LY146032, and 50 mM CaCI2 in standard buffer; (0)
`FPA increase under the same conditions. Calcium titration (right-hand
`M LY146032 in standard
`figure): (B) fluorescence increase of
`buffer plus 0.83 mg/mL DMPC on addition of aliquots of 3 M CaCI,;
`(0) FPA under the same conditions. For calculation of FPA, see text;
`T = 30 OC.
`
`Io
`
`1 0
`
`fC
`
`FIGURE 10: Dependence of LY 146032 kynurenine fluorescence on
`lipid-phase transition. All samples contained lo-' M LY146032. Top:
`1 mg/mL egg PC SUV and 50 mM CaC1, in standard buffer. Middle:
`1 mg/mL DMPC SUV and 50 mM CaC12 in standard buffer. Bottom:
`1 mg/mL DMPC SUV in standard buffer (Le., no calcium). Results
`are from temperature increases (10-50 "C) only.
`
`In order to gain some insight into the role of Ca2+ in
`LY 146032/PC binding, we carried out measurements of
`fluorescence polarization anisotropy (FPA). In 50 mM CaC12,
`P-Kyn shows a rise in FPA when PC is added that parallels
`the increase in P-Kyn emission (Figure 9). This indicates that
`the move of P-Kyn into a hydrophobic environment is asso-
`ciated with attachment to a larger structure, i.e., the phos-
`pholipid vesicle. At zero calcium, the FPA of LY146032 in
`1 mg/mL PC is already at the maximum value of 0.30 reached
`in the PC titration. This confirms that the peptide is already
`fixed to the vesicles. An addition of Ca2+ causes an increase
`of the P-Kyn emission (Figure 9), which suggests a change
`in its environment but does not modify its FPA.
`The interaction of LY146032 and PC was also followed as
`a function of temperature (Figure 10). In egg PC vesicles
`(50 mM CaC1,) the fluorescence yield fell smoothly with
`increasing temperature due to greater nonradiative energy
`losses. This fall was, however, less than in aqueous solution
`due to the protective shielding of the lipid bilayer. In DMPC
`bilayers (50 mM CaCl,) the fluorescence falls steadily until
`21 OC when it rapidly increases to a value at 26 OC that is
`greater than its yield at 10 OC. It then drops almost linearly.
`In the absence of Ca2+ the P-Kyn yield drops as though in
`aqueous solution, with only a slight flattening of the curve
`around 21 OC.
`If DPPC bilayers are substituted, the
`fluorescence shows a rapid increase between 39 and 45 OC
`
`
`4 of 7
`
`

`
`C A L C I U M - D E P E N D E N T A N T I B I O T I C L Y 1 4 6 0 3 2
`
`(data not shown), while similar results were obtained from
`samples containing gadolinium ions rather than calcium ions.
`The increase in yield is thus associated directly with the
`"melting" of the lipid chains that occurs at the phase transition
`temperature (T,) of the lipid in SUV [T,(egg PC) < 0 OC;
`T,(DMPC) = 20.8 OC (Gaber & Sheridan, 1982); T,(DPPC)
`= 38 OC (Suurkuursk et al., 1976)l. Hence, in the absence
`of calcium the P-Kyn residues bound to SUVs are largely
`unaffected by the gel to liquid-crystalline transition of the fatty
`acyl chains. Whereas when calcium is present, it is clear that
`the P-Kyn fluorophore is very much more sensitive to the state
`of the hydrocarbon interior of the bilayer.
`DISCUSSION
`The results presented in this paper clearly demonstrate that
`there is a calcium-dependent interaction between LY 146032
`and phospholipid vesicles that may aid our understanding of
`this antibiotic's in vivo calcium requirement.
`In all the experiments described here we can assume that
`the lipopeptide is in monomer form. However, our 'H NMR
`studies have shown that aggregation does occur at concen-
`M. When mixed with SUV in the
`trations higher than
`absence of CaC12, the P-Kyn yield is seen to increase by
`2.5-fold, and the FPA study confirmed that the peptide does
`not require calcium in order to bind to these model membranes.
`The molecule is presumably anchored to phospholipid bilayers
`by insertion of its fatty acyl moiety into the nonpolar hydro-
`carbon layer. The existence of both negative and positive
`charges on the LY 146032 headgroup should not be overlooked,
`and electrostatic interactions may also be involved at this stage.
`The high value of the FPA (0.30) indicates that the mobility
`of the P-Kyn residue is very restricted. However, the small
`change in the emission wavelength, the marked temperature
`dependence, and the insensitivity to the lipid phase transition
`all show that, while the peptide is certainly adhering to the
`membrane, this residue is situated at the lipid-water interface
`and is still accessible to the solvent. When calcium ions are
`added to the peptide/vesicle suspension, the blue shift of 10
`nm and increase in relative quantum yield show that the PO-
`larity of the environment around P-Kyn falls considerably.
`This suggests that the neutralization of the peptide induces
`a motion toward the bilayer interior in which the P-Kyn comes
`in closer contact with the hydrocarbon chains. That such a
`change occurs is confirmed by the closely related yield en-
`hancement that occurs on the melting of these chains at T,.
`The change in the position of the peptide caused by CaZf
`binding does not modify the FPA of its P-Kyn residue. Let
`us recall that, for a simple spherical molecule, the FPA is given
`by the classical Perrin equation
`FPA = 0.40[1/(1 + T/O)]
`where 7 is the fluorescence lifetime and 0 the rotation corre-
`lation time. Usually, the major causes of FPA variations are
`changes in the correlation time 0.
`In the present case, a FPA of 0.30 found in the absence of
`calcium indicates that the fluorescent probe linked to a slowly
`rotating vesicle already has a very restricted residual motion.
`A penetration of the peptide in the bilayer induced by CaZ+
`could still slightly reduce this motion, but at the same time,
`there is an increase in the fluorescence lifetime (7) that could
`compensate for an increase in 0. Qualitatively, one can only
`state that the change in the position of the peptide modifies
`the fluorescence parameters of P-Kyn in a manner that
`maintains the FPA nearly constant.
`In fact, the P-Kyn residue probably has several lifetimes
`T~ and several correlation times 01, depending on its motions
`
`V O L . 2 1 , N O . 1 3 , 1 9 8 8 4643
`on the peptide cycle and in the bilayer environment. The
`binding of Ca*+, which affects the conformation of the peptide
`and its position, modified the values and the distribution of
`such fluorescence parameters in a complex manner, which
`cannot be determined only by static measurements of an-
`isotropy. Measurements of all lifetimes and anisotropies are
`then required and should be the next step of the present study.
`In these experiments calcium ions can interact with two
`classes of site, (a) the negatively charged groups of the lipo-
`peptide and (b) the zwitterionic groups of phosphatidylcholine.
`Calcium does bind to pure phosphatidylcholine bilayers and
`induce changes in organization that can be monitored by a
`variety of techniques [for a review, see Altenbach and Seelig
`(1984)l. Such studies indicate that the binding is weak (1-100
`M-l) and that it provokes a rigidification of the bilayer
`(Chapman et al., 1977) and establishes a positive surface
`potential on these zwitterionic bilayers (Altenbach & Seelig,
`1984; Shah & Schulman, 1965). Much greater Ca2+ inter-
`actions are seen on negatively charged films. This is most
`apparent at low calcium concentrations, where the originally
`negative surface potential is rapidly neutralized. However,
`the Ca2+ ions do not bind directly to individual lipids but lie
`in the membrane potential trough with their effects shared
`among all the surface components (Macdonald & Seelig,
`1987). The degree of selectivity between Ca2+, CdZ+, and
`Mgz+ manifested by LY146032 is not shown by pure PC
`systems alone (Chapman, 1977), while the Ca2+ >> Cd2+ >
`Mg2+ order of both solution quenching and membrane en-
`hancement points to the peptide's involvement in the binding.
`The association constant for calcium is outside the range
`quoted for Ca-PC binding, but it should be made clear here
`that all the K, values are first approximations only and use
`bulk solution concentrations rather than those existing in the
`double layer. Nevertheless, the binding constant is independent
`of the peptide concentration (results not shown) and is
`therefore a genuine feature of this lipid-peptide interaction
`rather than a function of the number of peptides per unit area
`of membrane.
`The role of CaZf in the membrane interaction is clarified
`by the experiments with Gd3+. It is known from previous
`studies (Akutsu & Seelig, 1981; Hauser et al., 1975) that Gd3+
`interacts more strongly with phospholipids than does calcium,
`and this has been shown to be due to close association of the
`trivalent lanthanides with the phosphate group of the lipid
`headgroup (Hauser et al., 1975).
`Hence, the fluorescence enhancement induced by Gd3+ in
`SUVs already saturated with Ca2+ is to be expected as the
`trivalent ions will replace the divalent calcium at the interface.
`However, this means that the fluorescence enhancement that
`occurs when Ca2+ is added to a Gd3+-saturated sample must
`be due to the existence of a site that is selective for Ca2+ over
`Gd3+. Our knowledge of the phospholipid headgroup pref-
`erence for gadolinium thus confirms that a calcium-selective
`site must exist on the peptide.
`The aggregation of LY146032 in free solution by gadoli-
`nium ions does indicate a degree of direct association that is
`not entirely unexpected between two such species. The rele-
`vance of this result to the membrane effect is however not yet
`clear.
`At this stage it is reasonable to propose a model for the
`observed fluorescence changes in which the peptide, initially
`bound to the membrane by its lipid tail, is drawn further into
`the membrane by calcium bound to its negative residues and
`calcium associated with the neighboring phospholipid interface.
`Such a peptide will provoke a localized negative potential that
`
`
`5 of 7
`
`

`
`4644 B I O C H E M I S T R Y
`
`loosely associated CaZ+ will neutralize. When in competition
`with Gd3+, these will be replaced, and only the tightly bound
`calcium ion(s) will remain. This may explain the slight sig-
`moidal nature of the Ca2+ titration curve with the peptides
`high-affinity binding site effective in the 0-2 mM range and
`the phospholipid binding saturating at 30 mM. Unfortunately,
`the induced fluorescence enhancement does not allow a serious
`analysis of this region, and the calculated binding constants
`assume one type of site. The proposed role for the lipid
`phosphate groups may also explain why LY 146032 is less
`effective in increasing the conductivity of bilayer lipid mem-
`brane (BLM) formed from nonphospholipids (Lakey & Lea,
`1986).
`In the presence of negatively charged lipids, localized neg-
`ative potentials will not be effective in obtaining the calcium
`ions necessary for charge neutralization and the membrane
`penetration will be lessened as seen under Results. The effect
`of cationic lipids may therefore be explained as an enhance-
`ment of the local negative potential well around the peptide.
`Further study is required on the role of the positively
`charged ornithine in these experiments. It may be neutralized
`by a close union with one of the negative residues, but if free,
`it could possibly play an important role in the more electro-
`statically neutral calcium/peptide complex.
`Polymixin, a positively charged lipopeptide antibiotic, might
`be expected to show binding to the phosphates of zwitterionic
`phospholipids but instead is entirely dependent on negatively
`charged lipids for its membrane binding (Hartmann et al.,
`1977). Its lipid chain does enter the membrane (Hartmann
`et al., 1977), but the charge interaction is less specific than
`LY 146032 in which a positive membrane surface charge
`cannot replace calcium.
`Two other calcium-sensitive acidic peptide antibiotics are
`known, bacitracin (Stone & Strominger, 1971) and ampho-
`mycin (Matsui et al., 1963). Bacitracin, a cyclic peptide with
`one Asp residue, is in fact sensitive to a range of divalent
`cations (Mg2+, Ca2+, Cu2+, Zn2+; Stone & Strominger, 1971).
`It has been shown to inhibi