Multilamellar Liposomes Formed by PhosphatidylNucleosides: An NMR-HR-MAS Characterization

Oscar Cruciani,§ Luisa Mannina,*,†,§ Anatolij P. Sobolev,§ Annalaura Segre,§ andPierluigi Luisi‡

Istituto di Metodologie Chimiche, C.N.R., Area della Ricerca di Roma, C.P. 10,I-00016 Monterotondo Stazione, Rome, Italy, Facolta di Agraria, Dipartimento S.T.A.A.M.,

Universita degli Studi del Molise, via De Sanctis, I-86100 Campobasso, Italy, andDipartimento di Biologia, Universita “Roma Tre” , Viale Marconi 446, I-00146 Rome, Italy

Received September 26, 2003. In Final Form: December 5, 2003

We present an NMR investigation of multilamellar vesicles (MLVs) obtained from phosphatidylnucleosides, 5′-(1-palmitoyl-2-oleoyl-sn-glycero(3)phospho)cytidine (1), 5′-(1-palmitoyl-2-oleoyl-sn-glycero-(3)phospho)inosine (2), and their mixtures. Because of the lower stability of liposomes obtained from 2,studies have been preferentially performed in this case with mixed liposomes2/POPC (4:1). The investigationis conducted mostly via the HR-MAS technique and the general observation is that the resolution achievedin this way is superior to that obtained in the past with small unilamellar vesicles (SUVs). A full assignmentis now possible, which includes the spectral region of the ribose ring and part of the glycerol moiety. Alsoin the case of MLVs, both for 1 and 2, a stacking between the aromatic bases of the same liposome layerseems to be ruled out, although in both cases the nucleobases appear to be exposed to the aqueous phase.The splitting of both aromatic H-5cyt and H-6cyt is ascribed to the presence of two aggregate populationsthat may correspond to the two syn and anti conformations observed for cytidine monophosphate in aqueoussolution. On the basis of NOESY cross-peaks, it is not always possible to discriminate between inter- andintramolecular interactions; however, the distances found for 1 appear to be compatible with theintramolecular contacts in the anti conformation of the cytidine and also with intermolecular interactionsbetween neighboring molecules of 1. We also find that the glycerol moiety does not seem to interact withthe cytidine; however, part of the ribose ring seems to be close to the glycerol moiety. More generally, theinteraction of one base with the sugar moiety of a neighboring base, previously observed for SUVs, stillappears to be true for MLVs. Studies have been performed also for mixed liposomes obtained from themixture of 1 and 2, where it is observed that the HR-MAS spectra of the corresponding MLVs are not simplythe sum of the spectra of the two isolated components. In particular, there is the presence of a NOESYcross-peak between the aromatic protons H-6cyt and H-2ino, and this permits us to rule out large patchworkdomains containing only one nucleoside components in the mixed liposomes. Finally, a study is performedon the time evolution of the system obtained by mixing the previously prepared liposomes of 1 and 2. Nointeraction is obtained in this case, i.e., the spectra are constitutive, which is consistent with the generalpicture of liposomes as kinetic traps that are not fusing with each other.

Introduction

Over the last 10 years we have been investigatingliposomes obtained from phosphatidyl nucleosides, inparticular 5′-(1-palmitoyl-2-oleoyl-sn-glycero(3)phospho)-cytidine (1) and 5′-(1-palmitoyl-2-oleoyl-sn-glycero(3)-phospho)inosine (2), which can be obtained enzymaticallyfrom the corresponding phosphatidylcholines and thenucleosides. These liposomes are interesting since, inprinciple, they display the information chemistry of nucleicacids combined with the compartment features of vesicles.The main driving idea for these studies was to investigatewhether and to what extent the chemical and bindingproperties of nucleic acids can be expressed also on thespherical structure of the liposomes.

A modest recognition between “complementary” lipo-somes was previously obtained,1 and also observed on theconfiguration of a Langmuir-Blodgett film made withphosphatidyl nucleosides.2

However, in our previous NMR studies3 on these sys-tems we suggested that the presence of the interaction ofone given base with the sugar moiety of a neighboringbase could explain why the bases are not available forcomplementary base pairing. Moreover, in the case ofliposomes formed using complementary phosphatidylnucleosides, no evidence of the existence of mixed lipo-somes was found.

An analogous NMR study was also carried4 on micellesformed by hexadecylphosphoryl nucleosides and, in par-ticular, hexadecylphosphoryladenosine (C16-AMP) andhexadecylphosphoryluridine (C16-UMP). In these su-pramolecular aggregates the polar heads of the twosurfactants lie almost perpendicular to the micellar radius.It is important to observe that in the case of mixed micellesa significant interaction between the two complementaryaromatic bases is present. Therefore, in the case of micellesformed using complementary hexadecylphosphoryl nu-cleosides, the existence of mixed surfaces formed by bothbases has been proved.

Note that in small liposomes formed by the mixture 1/2a patchwork structure has been evoked since no experi-mental evidence of mixed surfaces was ever found.3 Hence

* To whom correspondence should be addressed. E-mail:mannina@imc.cnr.it.

† Universita degli Studi del Molise.‡ Universita “Roma Tre”.§ C.N.R., Area della Ricerca di Roma.(1) Berti, D.; Baglioni, P.; Bonaccio, S.; Borsacchi-Bo, G.; Luisi, P. L.

J. Phys. Chem. B 1998, 102, 303.(2) Berti, D.; Franchi, L.; Baglioni, P.; Luisi, P. L. Langmuir 1997,

13, 3438.

(3) Bonaccio, S.; Capitani, D.; Segre, A. L.; Walde, P.; Luisi, P. L.Langmuir 1997, 13, 1952.

(4) Zandomeneghi, G.; Luisi, P. L.; Mannina, L.; Segre, A. L. Helv.Chim. Acta 2001, 84, 3710.

1144 Langmuir 2004, 20, 1144-1151

10.1021/la035804h CCC: $27.50 © 2004 American Chemical SocietyPublished on Web 01/23/2004

it seems possible that the curvature of the system mayplay a major role in determining the surface distributionof the complementary bases.

In this paper we continue this investigation with specialregards to multilamellar liposomes. In our previous NMRstudies3 performed on liposomes, i.e., on small unilamellarvesicles (SUVs), from 1 and 2 the lack of resolution, dueto the intrinsic anisotropy, was the main obstacle to thestructural characterization of these systems and to thestudy of binding availability for the bases of the phos-phatidyl nucleosides in liposomes.

In the present work, we decided to study a systemcharacterized by a very low curvature i.e., multilamellarvesicles (MLVs) obtained from the same compounds.Because of their low curvature, MLVs constitute a morereliable membrane model than SUVs, that are character-ized by a high curvature and by a marked packingasymmetry of the inner and outer monolayer.5-7 In MLVsthe use of magic angle spinning (HR-MAS) allows a realimprovement of the resolution.8,9 It is important tounderline that before the introduction of the HR-MASmethod, due the to intrinsic anisotropy of the multila-mellar systems, MLVs spectra were extremely broad andwere not informative.

In the HR-MAS technique the sample is placed in arotor spinning around an axis, which is oriented at theso-called magic angle of 54.7° with respect to the magneticfield B0.10 Spinning the sample at a sufficient rate allowsan averaging of the spin dipolar interactions and sus-ceptibility distortions. In particular in the case of mem-branes the averaging of the spin like dipolar interactionsby the use of magic angle spinning (MAS) is possible dueto the presence of an internal fast axial rotation9,11 thatreduces the intramolecular dipole-dipole interactions.Therefore, when an internal motion already reduces thesedipolar couplings, the spectral broadening present understatic conditions can be reduced by the use of MAS. Theresult is an NMR spectrum with resolution approachingthat of liquid samples. Interesting applications of theHR-MAS technique in study of interactions have beenreported.12,13 However this method does not work if thedipolar broadening is too severe (stiff systems) or if thevesicles present a motion that interferes with the coherentperturbation applied by the MAS as in the case of SUVs.14,16

The present paper reports the clarification, obtainedby the use of the HR-MAS technique, about some aspectsof the conformation of the phosphatidyl nucleosides andabout the interaction of the bases with the sugar andglycerol moieties and the lipidic bilayer on a multilamellarstructure.

In particular, the present study will deal with MLVsobtained from the two phosphatidyl nucleosides indicatedabove as 1 and 2. Moreover we investigated the possibilityof obtaining mixed MLVs either producing liposomesdirectly from the mixture 1/2 or starting with separateMLVs of 1 and MLVs of 2, which are then mixed together.

To investigate intramolecular interactions betweengroups within the same lipid molecule and intermolecularinteractions between neighboring lipid molecules, we willutilize here 1D- and 2D-1H-HR-MAS.

Materials and Methods

Materials. 1-palmitoyl-2-oleoyl-sn-glycero-3-phospatidylcho-line (POPC) was purchased from Avanti Polar Lipids (Alabaster,AL) and used without further purification.

1 and 2 have been synthesized enzymatically from POPC, andcorresponding nucleosides by the help of phospholipase D asdescribed before.16 Cytidine and inosine were obtained from Fluka(Switzerland). Phospholipase D from Streptomices sp. AA 586

was purchased from Genzyme Diagnostics (West Malling, Kent,U.K.). D2O (99.9% atom D) was purchased by Fluka. All theother reagents of the highest-purity grade were from Fluka,Sigma, or Merck.

Sample Preparation. Liposome preparation has been de-scribed extensively elsewhere.15,16 Multilamellar liposomes from1, from the mixture 2/POPC (4:1, mol ratio), and from the mixture1/2 (1:1 mol ratio) have been prepared by dispersing a dry filmof the lipids, deposited from a CHCl3 solution on a glass wall, inan appropriate buffer solution (20 mM deuterated sodiumphosphate buffer, 20 mM KCl, 0.2 mM EDTA, pD ) 8). The finallipid concentrations were: 18.1 mg/mL (20 mM) for MLVs of 1;18.6 mg/mL (20 mM) in 2 and 3.8 mg/mL (5 mM) in POPC forMLVs of 2/POPC (4:1); 7.7 mg/mL (8.5 mM) in 1 and 7.9 mg/mL(8.5 mM) in 2 for MLVs of 1/2 (1:1). A further sample was preparedby mixing an equal volume of the multilamellar dispersion of 1and of the multilamellar dispersion of the mixture 2/POPC.

Methods. HR-MAS NMR. All HR-MAS experiments wereperformed on a Bruker AVANCE 600 spectrometer operating at600.13 MHz, using a Bruker HR-MAS probe with an externallock. The samples were loaded in 4 mm ZrO2 cylindrical rotorswith spherical inserts (internal volume of 12 or 45 µL) and spunat different spinning rates in the range from 3 to 12 kHz.

All NMR spectra (1D and 2D) have been referenced to thepalmitoyl and oleoyl ω-CH3 set at 0.90 ppm from TMS (tetra-methylsilane), as previously indicated for similar compounds.19,20

To optimize the experimental conditions (temperature, spin-ning speed, and NMR acquisition parameters), a set of experi-ments on three different samples of POPC liposomes wasperformed as extensively shown elsewhere.8 The 2D experimentshave been performed in the experimental conditions (tempera-ture and spinning speed) to ensure a sufficient resolution andwhile maintaining the integrity of the rotor inserts during longtime experiments like NOESY (10 h).

All 1H HR-MAS NMR experiments on MLVs samples havebeen performed at the temperature and the spinning speed lateron indicated for each sample accumulating 128 scans with 32Kdata (time domain) and using a spectral width of 6.0 kHz.

A standard pulse sequence with water suppression by apresaturation of the HDO resonance during the relaxation delaywas used. A 4-6 µs 90° excitation pulse was used with a relaxationdelay of 3s. Prior to Fourier transformation, the data were zerofilled to 32K points and apodized using an exponential linebroadening of 1 Hz.21 All the 2D COSY experiments22 have beenperformed, using a spectral width of 6 kHz in both dimensions,256 increments, 20 scans, 2K data points, and a relaxation delayof 3 s. All the 2D-1H-NOESY experiments22 have been performedin TPPI phase-sensitive mode, using 6 kHz spectral width in

(5) Schuh, J. R.; Muller, L.; Chan, S. I. Biochim. Biophys. Acta 1982,687, 219.

(6) Longmuir, K. J.; Dahlquist, F. W. Proc. Natl Acad. Sci. USA 1976,73, 2716.

(7) Tauskela, J. S.; Thompson, M. Biochim. Biophys. Acta 1992, 1104,137.

(8) Cruciani, O.; Mannina, L.; Sobolev, A. P.; Cametti, C.; Segre, A.L. Colloids Surf. A: Physicochem. Eng. Asp., in press.

(9) Oldfield, E.; Bowers, J. L.; Forbes, J. Biochemistry 1987, 26, 6919.(10) Mehring, M. Principles of High-resolution NMR in solids;

Springer-Verlag: Heidelberg, Germany, 1983.(11) Davis, J. H.; Auger, M.; Hodges, R. S. Biophys. J. 1995, 69, 1917.(12) Handel, H.; Elke, G.; Gottschall, K.; Klaus, A. Angew. Chem.

2003, 42, 438.(13) Winter, W. T Polym. Prepr. 2003, 44, 299.(14) Maricq, M. M.; Waugh, J. S. J. Chem. Phys. 1979, 70, 3300.(15) Long, J. R.; Sun, B. Q.; Bowen, A.; Griffin, R. G. J. Am. Chem.

Soc., 1994, 116, 11950.(16) Bonaccio, S.; Walde, P.; Luisi, P. L. J. Phys. Chem. 1994, 98,

6661.(17) Szoka, F.; Papahadjopoulos, D. Annu. Rev. Biophys. Bioeng.1980,

9, 467.(18) Vemuri, S.; Rhodes, C. T. Pharm. Acta Helv. 1995, 70, 95.(19) Halladay, H. N.; Stark, R.E.; Ali, S.; Bittman, R. Biophys. J.

1990, 58, 1449.(20) Volke, F.; Pampel, A. Biophys. J. 1995, 68, 1960.(21) Ernst, R. R.; Bodenhausen, G.; Wokaun, A. Principles of Nuclear

Magnetic Resonance in One and Two Dimensions, Clarendon Press:Oxford, England, 1987.

(22) Braun, S.; Kalinowski, H. O.; Berger, S. 150 and More BasicNMR Experiments. A practical course. Wiley-VCH: New York, 1998.

HR-MAS on Liposomes from Phosphatidyl Nucleosides Langmuir, Vol. 20, No. 4, 2004 1145

both dimensions, 512 increments in F1, 32 scans, 1K data pointsin F2, a relaxation delay of 3 s, and a mixing time of 15, 20, 25,or 100 ms as indicated later on for each sample.

MLVs of 1. 1D experiments at different temperatures (298,300, 305, 318 K) and at different spinning speeds (3, 4, 5, 6, 8,10, and 12 kHz) have been performed. The following 2D ex-periments have been performed at 318 K and with a spinningspeed of 5 kHz: COSY; NOESY with a mixing time of 15 ms;NOESY with a mixing time of 25 ms.

MLVs of 2/POPC. 1D experiments at different temperatures(300, 318 K) and at different spinning speeds (3, 4, 5, 6, 8, 10,and 12 kHz) have been performed. The following 2D experimentshave been performed at 300 K and with a spinning speed of 8kHz: COSY; NOESY with a mixing time of 15 ms; NOESY witha mixing time of 25 ms.

MLVs of the 1/2 Mixture. 1D experiments at 300 K and atdifferent spinning speeds (4, 5, 8, and 12 kHz) have beenperformed. The following 2D experiments have been performedat 300 K and with a spinning speed of 8 kHz: NOESY with amixing time of 15 ms; NOESY with a mixing time of 20 ms;NOESY with a mixing time of 25 ms.

MLVs of 1 + MLVs of 2 /POPC. 1D experiments at 300 K andat different spinning speeds (3, 8, 10, and 12 kHz) have beenperformed. The following 2D experiments have been performedat 300 K and with a spinning speed of 10 kHz: NOESY with amixing time of 25 ms; NOESY with a mixing time of 100 ms.

A fundamental parameter in any 2D-1H NOESY experimentis the mixing time. As previously pointed out,3,23,24 due to thepresence of strong spin diffusion, long mixing times producespurious cross-peaks, which in turn may lead to a wronginterpretation. In micelles and other supramolecular lipidicsystems, significant spin diffusion may be present even atrelatively short mixing times. To ensure correct 2D-1H-NOESYexperiments on liposomes, the mixing time must be chosen insuch a way to rule out the possibility of spin diffusion effects.The criterion used was the absence of any residual cross-peakbetween the resonance due to the terminal methyl group of thefatty acid chains at 0.90 ppm and both the resonances due to themethylene protons of the glyceride moiety near 4.5 ppm and tothe fatty acid olefinic protons at ∼5.3 ppm.

Only with a mixing time as short as 15-25 ms were no spuriouscross-peaks found. It must be clear that, due to this very shortmixing time, only strong cross-peaks, corresponding to rathershort interatomic distances, can be observed, while interatomicdistances of the order of 4.5 Å or more cannot be detected. How-ever, in the case of the sample obtained by mixing MLVs of 1with MLVs of 2/POPC, to verify the eventual presence of NOEcontacts between distant protons, a mixing time of 100 ms wasalso used.

Results and Discussion

MLVs of 1. In the given experimental conditions, aresolution sufficient to perform a full assignment wasobtained (see Table 1 and Figure 2). It is worth of noticingthat, in the case of solution NMR experiments on SUVsfrom 13, a full spectral assignment was not possible. Thefull assignment obtained by 2D experiments is given inFigure 2 and Table 1, encompassing the spectral region(3.8-4.6 ppm) relative to the ribose ring and part of theglycerol moiety. In particular, the 2D COSY experiment(data not shown) allows us to assign the broad signal at1.3 ppm to different CH2 groups of fatty chains.

The chemical shifts observed for the aromatic protonsin the case of MLVs of 1 agree with those previouslyfound for cytidine monophosphate in aqueous solutions.3Therefore, also in the case of MLVs, a possible stackingbetween the aromatic bases of the same layer must beruled out, as seen previously in the case of SUVs.3 In thepresence of stacking, the magnetic anisotropy of thearomatic ring would cause a strong upfield shift on the

aromatic 1H in the neighboring ring and vice versa.25 Inthis case this upfield shift is not present.

As is apparent from Figure 2 (1D spectrum) and Figure3 (2D-1H-NOESY map, an expansion of the aromatic andsugar spectral region) the NMR signals of both aromaticprotons H-5cyt and H-6cyt are split. The same splitting ispartially present in the resonance due to the anomericH-1′cyt while all the other signals are not split. Moreoverin the COSY spectra (not shown) the absence of H-6acyt-H-6bcyt cross-peaks and H-5acyt-H-5bcyt suggests thatthese a and b resonances are from different molecules. Inthe 2D-1H-NOESY maps, a cross-peak between the signals6acyt and 5acyt is evident as well as one between the signals6bcyt and 5bcyt.

We interpret this splitting and the cross-peaks previ-ously mentioned as due to the presence of two differentpopulations, a (lower field) and b (higher field), which canbe evaluated in the ratio 5:2. It is then possible that thetwo conformations syn and anti observed for cytidinemonophosphate in aqueous solution3 and supposed in thecase of SUVs from compound 1 are both present also whenthe same nucleoside is involved in a multilamellarstructure. Because of the poor resolution obtainable inthe static NMR experiments, this important evidence wasnot previously observed.

In the 2D-1H-NOESY spectra cross-peaks between thesignals 6acyt and 6bcyt and between the signals 5acyt and5bcyt are present. These NOESY cross-peaks are due to aslow interconversion between the two forms syn and anti.Therefore, due to the presence of this interconversion, allsignals due to the two populations present the sameNOESY cross-peaks. Therefore, specific contacts for eachconformer population cannot be obtained.

As reported in Table 1, the resonances due to aromaticprotons H-5cyt and H-6cyt and those due to some of thesugar ring show observable NOESY cross-peaks. In par-(23) Keepers, J. W.; James, T. L. J. Magn. Reson. 1984, 57, 404.

(24) Olejniczak, E. T.; Gampe, R. T.; Fesik, S. W. J. Magn. Reson.1986, 67, 28. (25) Waugh, J. S.; Fessenden, R. W. J. Am. Chem. Soc. 1957, 79, 846.

Table 1. Assignment of 600 MHz 1H-HR-MAS Spectrum,Performed with a Spinning Rate of 5 kHz and at 318 K,of MLVs Obtained from 1 (20 mM in 20 mM DeuteratedSodium Phosphate Buffer, 20 mM KCl, 0.2 mM EDTA,

pD ) 8)a

resonance1H ((0.02

ppm) NOESY contacts

6acyt 8.05 5acyt, 1′cyt, 2′cyt, 4′cyt, 5′cyt6bcyt 7.98 5bcyt, 1′cyt, 2′cyt, 4′cyt, 5′cyt5acyt 6.21 6acyt, 1′cyt, 2′cyt, 4′cyt5bcyt 6.15 6bcyt, 1′cyt, 2′cyt, 4′cyt1′cyt 5.95 6acyt, 6bcyt, 5acyt, 5bcyt, 2′cyt, 5′cyt, 5′′cyt2′cyt 4.31 6acyt, 6bcyt, 5acyt, 5bcyt, 1′cyt3′cyt 4.30 B4′cyt 4.17 6acyt, 6bcyt, 5acyt, 5bcyt, B5′cyt 4.08 1′cyt, 5′′cyt, B5′′cyt 3.97 1′cyt, 5′cyt, BA 4.10 BB 5.32 C, C′, A, 3′cyt, 4′cyt, 5′cyt, 5′′cytC 4.52 B, C′C′ 4.26 B, CW 5.32 F, T′′P 2.41 F, T, J, P′P′ 2.36 F, T, J, PQ 2.31 F, T, JQ′ 2.26 F, T, JF 2.02 W, T, T′′, J, P, P′, Q, Q′J 1.61 F, T, T′, P, P′, Q, Q′T′′ 1.35 F, WT′ 1.32 JT 1.31 S, J, F, P, P′, Q, Q′S 0.90 T

a In the third column the contacts registered in the 2D-1HNOESY experiments are also reported.

1146 Langmuir, Vol. 20, No. 4, 2004 Cruciani et al.

ticular, the resonance due to the aromatic proton H-5cytshows NOESY cross-peaks with that due to sugar protonsH-1′cyt, H-2′cyt, and H-4′cyt (contact less evident). The reso-nance due to the aromatic proton H-6cyt gives NOESYcross-peaks with that due to protons H-1′cyt, H-2′cyt, H-4′cyt,and H-5′cyt.

It is important to underline that it is not always possibleto distinguish between intra- and intermolecular interac-tion, using the information (cross-peaks and integrals)obtained from the 2D-1H-NOESY maps. This fact hampersa clear model of the 1 conformation in the vesicularaggregate which, however, is consistent with a model suchas the one suggested in our previous paper.4

The distances, calculated by the integration of theNOESY cross-peaks and using as internal reference thedistance (2.45 Å) between the aromatic protons H-5cyt andH-6cyt are only indicative: in fact, due to the presence ofthe spin diffusion, mixing times extremely short have beenused and no build-up curve has been built. In particular,the distances values obtained between the aromaticprotons H-5cyt and H-6cyt, are only partially useful. In fact,the values obtained for the contacts H-6cyt-H-1′cyt andH-6cyt-H-2′cyt seem compatible with intramolecular in-teractions in the anti conformation but also with inter-molecular interactions between neighboring molecules of1. On the other hand, the distances calculated for thecontacts H-5cyt-H-1′cyt, H-5cyt-H-2′cyt and H-6cyt-H-4′cytare consistent only with intermolecular interactions. It isalso evident from the 2D-1H-NOESY experiments that nocross-peaks between the resonances due to the aromaticprotons and either the resonance due to the glycerol moietyor to the fatty acid chains are present.

Therefore, the glycerol moiety does not seem to interactwith the cytidine; this observation is also confirmed bythe chemical shift values of the glycerol backbone protons.In fact in the case of MLVs from 1, these protons havepractically the same chemical shifts previously observedin liposomes of POPC,8,26 and result unaffected by thepresence of the nucleoside.

The resonance due to the proton B of glycerol backbone,besides the obvious cross-peaks with the resonances dueto the other protons (A, C, and C′) of the same molecularfragment, shows cross-peaks with the resonances due toprotons H-5′cyt, H-5′′cyt, and H-3′cyt of the sugar moiety.

No cross-peaks between resonances of the glycerolmoiety and those of the fatty acid chains are present.Therefore, part of the ribose ring seems to be close to theglycerol moiety, whereas it is surely far away from thealkyl chains. Probably the ribose tends to bend towardthe glycerol moiety so as to minimize the contact betweenits hydrophobic parts and the aqueous phase.

The aromatic base seems to be located away both fromthe glycerol and from the alkyl chains and does not seemreclined back toward the bilayer but exposed toward theaqueous phase.

However the information obtained from the 2D-1H-NOESY maps about the contacts between the aromaticprotons H-5cyt and H-6cyt and protons of the sugar moietyindicate that the interaction of one base with the sugarmoiety of a neighboring base, proposed in the previous

(26) Dorovska-Taran, V.; Wick, R.; Walde, P. Anal. Biochem. 1996,240, 37.

Figure 1. Molecular structures of POPC and of the compounds1 and 2.

Figure 2. 600 MHz 1H-HR-MAS spectrum, performed witha spinning rate of 5 kHz and at 318 K, of MLVs obtained from1 (20 mM in 20 mM deuterated sodium phosphate buffer, 20mM KCl, 0.2 mM EDTA, pD ) 8).

HR-MAS on Liposomes from Phosphatidyl Nucleosides Langmuir, Vol. 20, No. 4, 2004 1147

model,3 seems still true. Therefore, the headgroup of 1 isnot suitable for complementary recognition and binding.

MLVs of 2/POPC. Liposomes of 2 are not stable andtend to aggregate;16,27 Then, we chose to characterizestable MLVs obtained from a mixture 2/POPC (4:1). Wereport the full assignment of the HR-MAS spectrumperformed at 300 K and with a spinning speed of 8 kHzin Table 2 and in Figure 4. The values of the chemicalsshift indicate that also in this case a stacking betweenadjacent aromatic bases is absent.

Thesplittingof theNMR signals relative to thearomatic,H-2ino and H-8ino, and anomeric, H-1′ino, protons is presentalso in this case, but less evident than in the case of MLVsof 1 and observable only as an elongation of diagonal peaksin the 2D spectra.

The resonance due to the aromatic proton H-8ino yieldsobservable NOESY cross-peaks with the resonance dueto the other aromatic proton H-2ino and with the resonancesdue to all the ribose protons (see Table 2). The cross-peakbetween the resonances of the aromatic protons H-8inoand H-2ino is clearer in the 2D-1H-NOESY map obtainedusing a mixing time of 25 ms. Moreover a contact, moreintense than the contact H-8ino-H-1′ino, is present be-tween the aromatic proton H-2ino and the anomeric oneH-1′ino.

The resonance due to the anomeric proton H-1′ino showsother two NOESY cross-peaks with the resonance due tothe ribose proton H-4′ino and with a resonance at 4.50ppm. Because of the superimposition at 4.50 ppm of theribose proton H-3′ino resonance and of the glycerol cresonance, an unequivocal interpretation of the NOESYcross-peak in question is not possible. However the lackof further NOESY contacts between the anomeric protonsH-1′ino and the other protons of the glycerol moiety suggests

that this cross-peak might be due to a contact betweenthe anomeric proton H-1′ino and the proton H-3′ino.

The chemical shifts values and the information obtainedfrom the 2D-1H-NOESY maps indicate for the polar headthe same situation reported in the case of MLVs of 1;therefore, again, the nucleobase is not turned back towardthe bilayer but is exposed toward the aqueous phase.

In the case of MLVs of 1 a distinction between intra-and intermolecular contacts does not seem possible; while,for MLVs of 2/POPC, the existence of intermolecularcontacts is confirmed by the presence of a NOE cross-peak between the resonances due to the protons H-8inoand H-2ino. This cross-peak can be due only to anintermolecular interaction; in fact in the same moleculethese protons are too far apart (∼6.5 Å) to allow observableNOE effects. Moreover the contact H1′ino-H-2ino can onlybe due to intermolecular interaction.

MLVs of the Mixture 1/2. It is important to underlinethat the HR-MAS spectra of the MLVs obtained from themixture 1/2 are not simply the sum of the spectraregistered for MLVs from a single phosphatidyl nucleoside.This evidence is not trivial, because in the case ofunilamellarvesicles fromthesamecompounds3 thespectraobtained for the vesicles formed by the 1:1 mixture of 1/2was exactly the sum of the two separate spectra.

The comparison between the 1D HR-MAS spectra ofMLVs obtained from the 1:1 mixture of 1/2 (see Figure 6)and the HR-MAS spectra of MLVs from the singlephosphatidyl nucleoside (carried under the same experi-mental conditions) emphasizes a chemical shift changefor the aromatic and the anomeric protons of cytidineand inosine. In particular, on respect to the cases whereonly one phosphatidyl nucleoside was present, we ob-serve an upfield shift for the resonances due to cytidineand a downfield shift for those due to inosine. This shiftis quite evident for the resonances due to aromatic andanomeric protons; all the other resonance signals areunaffected.

(27) Bonaccio, S.; Wessicken, M.; Berti, D.; Walde, P.; Luisi, P. L.Langmuir 1996, 12, 4976.

Figure 3. 2D-1H-NOESY spectrum, performed with a spinningrate of 5 kHz at 318 K and with a mixing time of 15 ms, of theMLVs obtained from 1: expansion of the spectral region ofaromatic and sugar protons.

Table 2. Assignment of 600 MHz 1H-HR-MAS Spectrum,Performed with a Spinning Rate of 8 KHz and at 300 K,of MLVs Obtained from the Mixture 2/POPC (20 mM in 2

and 5 mM in POPC in 20 mM Deuterated SodiumPhosphate Buffer, 20 mM KCl, 0.2 mM EDTA, pD ) 8)a

resonance1H ((0.02

ppm) NOESY contacts

8ino 8.41 2ino,1′ino, 2′ino, 4′ino, 5′ino, 5′′ino, 4.50 ppm2ino 8.13 1′ino, 8ino1′ino 6.06 8ino, 2ino, 2′ino, 4′ino, 4.50 ppm2′ino 4.63 8ino , 1′ino, 3′ino3′ino 4.50 2′ino4′ino 4.31 8ino, 1′ino, 5′ino, 5′′ino5′ino 3.95 8ino, 4′ino, 5′′ino, B5′′ino 3.89 8ino, 4′ino, 5′ino, BA 4.10 B, CB 5.33 C′, C, A, 5′ino, 5′′inoC 4.50 C′, A, BC′ 4.26 C, BW 5.33 F, TR 3.29M 4.37 NN 3.75 MP 2.41 F, T, J, P′P′ 2.38 F, T, J, PQ 2.29 F, T, JQ′ 2.24 F, T, JF 2.03 W, T, J, P, P′, Q, Q′J 1.60 F, T, P, P′, Q, Q′T (≡T′, T′′) 1.30 S, J, F, W, P, P′, Q, Q′S 0.90 T

a In the third column the contacts registered in the 2D-1H NOESYexperiments are also reported.

1148 Langmuir, Vol. 20, No. 4, 2004 Cruciani et al.

It is important to underline that the above shifts arepresent for all the used spinning speeds and therefore arenot an artifact dependent by the experimental conditions.In Table 4, we report the values of the shifts for thearomatic signals registered at different spinning speeds.

A small splitting of aromatic and anomeric signals, aspreviously pointed out for MLVs derived from a singlephosphatidyl nucleoside, is present also in this case andwith the same pattern previously seen.

In the case of MLVs from the mixture 1/2, in the2D-1H-NOESY maps (see Figure 7) no cross-peaks arepresent between resonances due to the aromatic protonsand those due to the glycerol moiety (see Figure 7).Therefore, it is possible to suppose that the two polar headskeep the outside orientation previously reported.

With respect to the NOESY experiments performed onMLVs obtained from single phosphatidyl nucleosides,some of the cross-peaks due to the resonances of the ribosesprotons are absent, possibly due to the lack of resolutionin this spectral region.

The presence of a cross-peak between the aromaticproton 6cyt of cytidine and the aromatic 2ino of inosineconstitute the main information on the whole 2D map.This cross-peak is present at any mixing time used foracquisition (15, 20, and 25 ms) and for both populations(a, b) of the aromatic proton 6cyt.

Furthermore, the NOESY contact between protons 8inoand 2ino, present in the case of MLVs from 2/POPC, istotally absent in the NOESY experiment performed with

mixing time of 15 ms and is very low in the other twocases (20 and 25 ms).

The chemical shift variation of resonances due toaromatic and anomeric protons is possibly due to aninteraction between inosine and cytidine. Since theresonances of the two nucleobases give a NOESY cross-peak between the resonances due to the protons H-6cytand H-2ino, they are also sensitive to a different magneticanisotropy and therefore are shifted.

All these experimental data rule out the formation oflargepatchworkdomainscontainingonlyonephosphatidylnucleoside and enforce a model in which the two phos-phatidyl nucleosides are alternated all over the bilayer.

It is possible to draw some conclusion about theinteraction between inosine and cytidine when these areinvolved in MLVs. The values of the chemical shifts allowthe exclusion of a stacking interaction. In fact in thepresence of stacking, the experimental shifts should belarger. The observed contact between the proton H-2inoand the proton H-6cyt suggests an arrangement of theneighboring nucleobases that is compatible neither witha classic Watson-Crick base-pairing nor with a base-pairing other than Watson-Crick (Hoogsteen, reversedHoogsteen, and reversed Watson-Crick).28 It is worthnoticing that the observed interaction H-2ino-H-6cyt (seeFigure 8) indicates that the molecule side utilized byinosine in the base pairing is always engaged with theside of cytidine not used for the base pairing.

MLVs of 1 + MLVs of 2. The existence of an interactionbetween cytidine and inosine observed in the sample ofMLVs directly obtained from the mixture 1/2 suggestedthat one can verify the possibility that this interactioncould be a sufficient drawing force to obtain MLVs of 1/2starting by MLVs from 1 and MLVs from 2 mixed together.

(28) Landolt, H.; Bornstein, R. Numerical Data and FunctionalRelationships in Science and Technology, New Series, Group VII, Vol.1a: Nucleic Acids. Springer-Verlag: Heidelberg, Germany, 1989.

Figure 4. 600 MHz 1H-HR-MAS spectrum, performed witha spinning rate of 8 kHz and at 300 K, of MLVs obtained fromthe mixture 2/POPC (20 mM in 2 and 5 mM in POPC in 20 mMdeuterated sodium phosphate buffer, 20 mM KCl, 0.2 mMEDTA, pD ) 8).

Figure 5. 2D-1H-NOESY map, performed with a spinning rateof 8 kHz at 300 K and with a mixing time of 15 ms, of the MLVsobtained from the mixture 2/POPC: expansion of the spectralregion of aromatic and sugar protons.

HR-MAS on Liposomes from Phosphatidyl Nucleosides Langmuir, Vol. 20, No. 4, 2004 1149

To verify a possible evolution of the system, the sampleobtained by mixing MLVs of 1 with MLVs of 2/POPC hasbeen monitored over 2 weeks. After the mixing, the samplewas analyzed every hour for the first 2 days and later onevery 8 h.

It is worth of noticing that the spectra did not show anychange during the time interval considered and that the1H HR-MAS spectra appear always as the simple sum ofthe single spectra, i.e., those obtained from a singlephosphatidyl nucleoside performed under the same ex-perimental conditions. This evidence points out that thesystem is stable and does not evolve; therefore, we do notobserve any interaction between cytidine and inosine.

In the two 2D-1H-NOESY experiments performedrespectively with a mixing time of 25 and 100 ms, thesame cross-peaks seen in the case of MLVs from 1 and inthe case of MLVs from 2 are present; no cross-peaksbetween signals due to the cytidine and signals due to theinosine are present and also the contact H-6cyt-H-2ino,seen in the case of MLVs obtained from the mixture 1/2,is totally absent.

Concluding RemarksThe most general observation arising from this study

is the high-resolution achieved by HR-MAS also in thecase of MLVs obtained from 1 and 2. The present papercan therefore be taken as a clear example of the greatpotential offered by the HR-MAS techniques. The fact thatwe are able to study the fine conformational details of

groups hanging from the bilayer may suggest that theNMR HR-MAS technique can be useful for studying thestructure and conformation of drugs and other guestmolecules bound to the lipidic bilayer. In the case of MLVsformed by 1 and 2, indeed this technique allowed one toreach a detailed description of the structural features ofthe cytidine and inosine bases in phosphatidyl nucleosidesthat are involved in multilamellar liposomes.

The present study confirms that in liposomes formedby the phosphatidyl nucleosides 1 and 2, an interactionbetween one given base and the sugar moiety of aneighboring base is present and that the aromatic basesare exposed toward the aqueous phase. It is evident fromthe 2D-1H-NOESY experiments that no cross-peaksbetween the resonances due to the aromatic protons andeither the resonances due to the glycerol moiety or to thefatty acid chains are present. We can say, however, thatpart of the ribose ring is close to the glycerol moiety,whereas it is far away from the alkyl chains. As alreadymentioned, probably the ribose tends to bent toward theglycerol moiety so as to minimize the contact between itshydrophobic parts and the aqueous phase.

In the cases of the considered multilamellar liposomes,the interaction between adjacent aromatic bases is clearlyshown by the above commented NOESY contacts. Theexistence of this intermolecular interaction is evident in

Figure 6. Comparison between 1H-HR-MAS spectra regis-tered with a spinning speed of 12 kHz at 300 K for MLVsobtained from the mixture 1/2, MLVs obtained from 1 and MLVsobtained from the mixture 2/POPC (expansion of the aromaticand anomeric protons spectral region).

Table 3. Assignment of 600 MHz 1H-HR-MAS Spectrum(Performed with a Spinning Rate of 8 kHz and at 300 K)

of MLVs Obtained from the Mixture 1/2a

resonance 1H (ppm) ((0.02 ppm) NOESY contacts

8ino 8.45 1′ino, 2ino, 4.53 ppm2ino 8.19 1′ino, 8ino, 6acyt, 6bcyt6acyt 8.01 1′cyt, 5cyt, 2′cyt, 2ino6bcyt 7.93 1′cyt, 5cyt, 2′cyt, 2ino5cyt 6.14 1′cyt, 6acyt, 6bcyt, 2′cyt1ino 6.09 8ino, 2ino, 4.35 ppm1′cyt 5.94 2′cyt, 6acyt, 6bcyt, 5cyt2′cyt 4.31 1′cyt, 6acyt, 6bcyt, 5cyt3′ino 4.534′cyt 4.185′ino 3.975"cyt 3.975"ino 3.90A 4.11 B, CB 5.33 A, C′, CC 4.53 A, C′, BC′ 4.27 B, CW 5.35 FP 2.42 T, JP′ 2.36 T, JQ 2.31 T, JQ′ 2.27 T, JF 2.03 T, WJ 1.61 T, P, P′, Q, Q′T, T′, T′′ 1.30 S, F, J, P, P′, Q, Q′S 0.90 Ta In the third column the contacts registered in the 2D 1H NOESY

experiments are also reported.

Table 4. Values (in (2 Hz) of the Differences Registeredbetween the Resonances Due to the Aromatic ProtonsH-8ino, H-2ino, H-6cyt and H-5cyt in the Case of MLVs

Obtained from the Mixture 1/2 and in the Case of MLVsObtained from 1 and MLVs Obtained from 2/POPCa

spinning speed,kHz ∆ν8ino ∆ν2ino ∆ν6acyt ∆ν6bcyt ∆ν5acyt

5.0 26 36 468.0 24 36 36 43 44

12.0 24 33 39 47av. 25 35 38 45 45

a All the experiments have been performed at 300 K and withthe spinning speeds indicated in column 1.

1150 Langmuir, Vol. 20, No. 4, 2004 Cruciani et al.

the case of MLVs obtained from 2/POPC and in the caseof MLVs obtained from the mixture 1/2. It is importantto underline that in the case of SUVs obtained from thesame compounds this interaction was excluded.3

Mixed vesicles were obtained mixing the two phos-phatidyl nucleosides at the beginning of the preparationof the lipidic film. The experimental data (chemical shiftsvariation and NOESY contacts) indicate that in thesemixed vesicles the two phosphatidyl nucleosides arealternated all over the bilayer. It is worth noticing thatin contrast to the case of SUVs obtained from the samecompounds, in the mixed MLVs, no large patchworkdomains have been found.

In the mixed MLVs, some kind of interaction betweeninosine and cytidine is present, but this is not a stackinginteraction or a classic H-bonded interaction between thetwo complementary bases. The precise nature of thisinteraction is not easy to rationalize.

The fact that no interaction is observed by mixingliposomes obtained separately from 1 and 2 is unfortunate

(one might have expected a Watson-Crick interactionamong “complementary” liposomes)sbut it is not com-pletely unexpected on the basis of common knowledgeabout the physical behavior of phospholipids vesicles. Infact, it is known that liposomes are not chemical equi-librium systems, actually they can be more properly seenas kinetic traps, i.e., structures that are stabilized andseparated by each other owing to a high activation energythat protect them from transforming into other energetic-ally similar structures.29 In fact, it is very difficult if notimpossible to cause liposomes of different sizes to simplyfuse with each other.30 Thus, “complementary” liposomesof phosphatidyl nucleotides may have no tendency to fusewith each other to form mixed liposomes. However, as wehave seen, stable mixed liposomes can be obtained startingfrom the mixture of monomers.

Several other phosphatidyl nucleosides, bearing forexample adenine and uracil bases, were also prepared.The corresponding MLVs are presently under investiga-tion, and the results will be reported at a later time.

Abbreviations

HR-MAS: high-resolution magic angle spinningMAS: magic angle spinningPOPC: 1-palmitoyl-2-oleoyl-sn-glycero-3-phospatidylcho-

lineSUVs: small unilamellar vesiclesMLVs: multilamellar vesicles1: 5′-(1-palmitoyl-2-oleoyl-sn-glycero(3)phospho)cytidine2: 5′-(1-palmitoyl-2-oleoyl-sn-glycero(3)phospho)inosine

Acknowledgment. We thank Dr. Giovanni Ughettofor helpful comments and discussions.

LA035804H

(29) Luisi, P. L. J. Chem. Educ. 2001, 78, 380.(30) Zhilang, C.; Luisi, P. L. J. Phys. Chem., in press.

Figure 7. Left side: expansion of the spectral region of aromatic and sugar protons of a 2D-1H-NOESY spectrum (T ) 300 K,spinning speed ) 8 kHz, mixing time of 15 ms) of the MLVs obtained from the mixture 1/2. Right side: expansion of the aromaticregion of the 2D map.

Figure 8. (a) Schematization of the contact between the proton2ino and the proton 6cyt registered in the case of MLVs obtainedfrom the mixture 1/2. (b) Schematization of a classical Watson-Crick bases pairing.

HR-MAS on Liposomes from Phosphatidyl Nucleosides Langmuir, Vol. 20, No. 4, 2004 1151