Evaluating spin diffusion in MAS-NOESY spectra of phospholipid multibilayers

SOLID STATE Nuclear Magnetic Resonance ELSEVIER

Solid State Nuclear Magnetic Resonance

7 (1996) 239-246

Evaluating spin diffusion in MAS-NOESY phospholipid multibilayers Zhen-jia Chen ‘, Ruth E. Stark

spectra of

*

Department of Chemistry, College of Staten Island and the Graduate School of the City University of New York, 2800 Victory Boulevard, Staten Island, NY 10314, USA

Received 17 October 199.5;accepted 22 February

1996

Abstract Biological semisolids such as aqueous phospholipid dispersions are amenable to experiments that combine magic-angle spinning (MAS) to yield high-resolution ‘H NMR spectra and two-dimensional nuclear Overhauser spectroscopy (NOESY) to estimate proton-proton distances. Using several selectively deuterated dimyristoylphosphatidylcholines, a systematic investigation has been made into the origin of an anomalous NOESY crosspeak between methyl groups of the lipid headgroup and the acyl chains. Although the spin-relaxation and line-narrowing behavior in these multilamellar systems argue against efficient spin diffusion, interdigitated or chain-bendback structural models that bring these ‘H nuclei into close proximity are not implicated by the MAS-NOESY results for this phospholipid system. A mechanism has been proposed for spin communication between the two types of methyl protons within a conventional lipid bilayer structure: through-space interactions link the headgroup and backbone protons on adjacent molecules, and then spin diffusion occurs along the acyl chains. Keywords: Bilayer;

Biological

membrane;

Magic-angle

spinning;

1. Introduction

Because of their importance to the structure and function of biological membranes, phospholipids have been the subject of numerous NMR investigations during the last 25 years. Until less than a decade ago, the bulk of this work fell into two distinct categories: (a> solution-state NMR of small unilamellar vesicles or micelles [1,2] and (b) static solid-state NMR of extended multibilayers in the liquid-crystalline or the gel state [3,4]. Although the

* Corresponding author. ’ Present address: Wyeth-Ayerst 10965, USA. 0926-2040/96/$15.00 Copyright PII SO926-2040(96)01237-4

Research,

Pearl River,

NY

Nuclear Overhauser

effect; Phospholipid;

Spin diffosion

smaller phospholipid aggregates usually yield wellresolved spectra, the semisolid multibilayers formed spontaneously in aqueous media are viewed as better model systems for most physiological membrane phenomena. The distinction between liquid- and solid-state NMR investigative approaches has been blurred by the (re)introduction of magic-angle spinning (MAS) methods to obtain high-resolution natural-abundance ‘H (and 13C) spectra of lipid dispersions [5]: as both lateral diffusion and axial rotation cause significant averaging of proton-proton dipolar interactions in liquid-crystalline model membranes [6], it is possible to obtain quantitatively reliable high-resolution spectra with MAS alone at very modest spinning speeds.

0 1996 Elsevier Science All rights reserved.

240

Z.-j. Chen, R.E. Stark/

Solid State Nuclear Magnetic Resonance 7 (1996) 239-246

The potential of using MAS-assisted data acquisition in concert with “liquid-state” experiments such as two-dimensional correlated spectroscopy (COSY) and nuclear Overhauser spectroscopy (NOESY) was recognized early on for phospholipid dispersions [5]. The goals of such hybrid experiments include making spectral assignments of ‘H and 13C spectra and elucidating the bilayer organizations of phospholipid membranes. A number of useful applications of MAS-assisted NMR have been demonstrated recently, including heteronuclear and homonuclear chemical-shift correlation spectroscopy [7-91 and NOESY for structural studies of bilayer interdigitation [lo] or hydration [11,12]. Nevertheless, many investigators have exercised caution in the interpretation of NOESY results for both vesicle and multilayer phospholipid aggregates, since the crosspeaks may be due to spin diffusion rather than to dipoledipole interactions of ‘H nuclei that are close in space [ 13- 151. In order to address this dilemma systematically, the current study extends prior work [ 11,16- 181 with MAS-assisted NOESY experiments on a series of selectively deuterated multilayers formed by liquid-crystalline dimyristoylphosphatidylcholine (DMPC). A model is proposed for spin communication between headgroup and acyl-chain protons within the lipid assembly.

2. Experimental All samples of 1,2-dimyristoyl-sn-glycero-3-phosphocholine (dimyristoylphosphatidylcholine, DMPC) were purchased from Avanti Polar Lipids (Alabaster, AL). In addition to the lipid without isotopic labels, the following partially deuterated species were obtained: DMPC-d,, DMPC-d,,, DMPC-d,,, and DMPC-d, (Scheme 1, la-lf). Deuterium oxide (99.996%) was purchased from Aldrich Chemical Company (Milwaukee, WI). To prepare samples for NMR investigations, the lipids were first cosolubilized in 2:l (v/v) chloroform-methanol and then dried under a stream of nitrogen and overnight at reduced pressure. Aqueous dispersions were made by hydrating the lipid films with 1:l (w/w) D,O that was 150 mM in NaCl. All samples were vortex-mixed and freeze-thawed five times to produce a uniform paste. The lipids remained dispersed as a single

la

lb

lc

Id

le If

phase for the duration of the magic-angle spinning NMR experiments. The NMR measurements were made using a Varian Unitypluswidebore spectrometer equipped for both solid-state and liquid-state NMR. The ‘H resonance frequency was 300.00 MHz, the spinning speed was adjusted to 2.000 f 0.002 kHz unless indicated otherwise, and the sample temperature was regulated at 30 f 1°C or 35 f 1°C. One-dimensional magic-angle spinning ’ H NMR spectra were acquired with 8 ps 90” pulses, 10 kHz spectral widths, and 1 s recycle delays. In a typical experiment, 32 transients consisting of 8K time-domain points were collected, zero-filled to 16K points, and Fourier transformed. Spin-lattice relaxation

Z.-j. Chen, R.E. Stark/

Solid State Nuclear Magnetic Resonance 7 (1996) 239-246

times T, were determined using an inversion-recovery experiment [ 191. Two-dimensional MAS-assisted nuclear Overhauser spectroscopy [20] was conducted using the standard pulse sequence (90”~t,--90”-T,-90”acquire [t2]),, with n = 16 and 128 I, values. The mixing time TV was varied between 50 and 1000 ms in separate 3 h experiments. All data were collected in phase-sensitive mode [21] and analyzed with VNMRX software (Varian Associates, Palo Alto, CA).

?

500

1000

3. Results 3.1. Motional

Spinning averaging

of dipole-dipole

For aqueous dispersions of DMPC above the gelto-liquid crystalline phase transition, magic-angle spinning suffices to narrow the ’ H spectral lines as expected and to resolve all major functionalities of the phospholipid headgroup [POCH 2, CH 2 N, N(CH,),], backbone (CH,O, CHO), and acyl chains [a-CH,, P-CH,, (CH,),, w-CH,]. In comparison with many crystalline or inorganic solids, the diminished magnitude of ’ H- ‘H dipole-dipole interactions in DMPC multilayers is manifest in several ways. First of all, the spectral lines are narrowed to

Chemical shift/ppm

relaxation a

times for aqueous lamellar dispersions Proton type a

o-CH 3

1.30 2.34 3.24 3.68 4.29 5.25

(CH,),, wCH 2 N(CH,), CH,N POCH 2 CHOCOR Average T, Normalized std. dev.

a Assigned following Ref. b ’ H resonance frequency ’ ‘H resonance frequency d ‘H resonance frequency e ‘H resonance frequency

2500

3000

Speed (Hz)

yield high-resolution ‘H NMR spectra once the MAS rate exceeds = 0.6 kHz. Secondly, lowering the rotation rate triples the (CH,), sideband intensities but broadens the centerband linewidth by only 50% (Fig. 1). Finally, distinct spin-lattice relaxation times are observed for each chemically shifted proton type in the spectrum, with values roughly as anticipated

of phospholipids

DMPC b

T,(H)/ms C,,C,oPC

0.90

zoo0

Fig. 1. ‘H NMR spinning sideband intensities and line widths of the centerband plotted as a function of rotation rate for the bulk methylene resonance in a 50% (w/w) DMPC aqueous dispersion maintained at 30°C. The first upfield sideband, which is well resolved from the main spectrum, is measured as a percentage of the centerband peak intensity. Linewidths (Hz) are measured at half maximum peak height.

interac-

tions

Table I Proton spin-lattice

,500

670 620 450 480 350 460

500 400 310 350 310 330 350

505 0.23

364 0.18

[IO] and referenced to w-CH3 set as 0.90 ppm. = 360 MHz; T, = (7’ - T,, j/T, = 0.04. From [5]. = 200 MHz, T, = 0.06. From [lo]. = 200 MHz, T, = 0.10. From [8]. = 300 MHz, T, = 0.022. This work.

c

Egg PC d

DMPC e

1080 517 258 353 212 187 525

759 540

447 0.69

345 306

488 0.43

Z.-j. Chen, R.E. Stark/

242

Solid State Nuclear Magnetic Resonance 7 (1996) 239-246

based on their known motional characteristics (Table 1; [lo]). Although the test of differential T, values is limited by spectral overlap of some of the methylene

protons, when taken together these observations suggest that proton spins in the liquid-crystalline state of DMPC may be sufficiently isolated from one another

F2

so- *

CH,O -

”

d

, CHO

5.0

4.0

3.0

2.0

1.0 Fl (pm)

Fl (pm)

5.0

e

0

5.0

m&J

4.0

0

2.0 3.0 Fl @pm)

e

1.0

Fig. 2. Two-dimensional nuclear Overhauser enhancement (NOESY) contour plots obtained for DMPC multilayers at 300 MHz with 2.000 kHz MAS. Mixing times (rM) of 150-300 ms were employed and Gaussian apodization was used to process the data. The frequency axes refer to acquisition (F2) and evolution (Fl) periods, respectively, of the pulse sequence. The chemical shifts are referenced to o-CH, groups resonating at 0.90 ppm, with all resonance assignments derived from earlier work on related phospholipid systems [IO]. As compared with solution-state NOESY experiments conducted without spinning, typical MAS-NOESY spectra exhibit more serious artifacts near strong resonances. (a) Unlabeled DMPC (la), 7’= 30°C. rM = 150 ms; (b) DMPC-d, (lb) T= 35”C, rt,, = 300 ms; (c) I:1 mixed dispersion of DMPC-d, (le) and DMPC-d&f), T = 30°C TV = 250 ms. The arrows highlight the N(CH,),-o-CH, crosspeak of interest.

Z.-j. Chen, R.E. Stark/

to preclude spin diffusion NOESY experiments.

Solid State Nuclear Magnetic Resonance 7 (1996) 239-246

in suitably designed

3.2. NOESY crosspeaks in unlabeled DMPC Displayed in Fig. 2(a) are typical results of MAS-assisted NOESY experiments conducted on multilamellar DMPC samples. In agreement with prior results obtained at 360 MHz, 26”C, and using a 500 ms mixing time [5], strong dipolar connectivities are exhibited herein for selected pairs of protons even at the shorter mixing time of 150 ms. Many of these interactions are expected based on the bilayer and phospholipid molecular structure [e.g., N(CH,), X CH,N in the headgroup, (CH,),, X w-CH, on the acyl chains]. An unexpected feature that has been observed nonetheless in different phospholipid aggregates is the NOESY crosspeak between N(CH,),‘s of the choline headgroup and w-CH,‘s of the acyl chains [5,10- 12,16- 181. Although traditional bilayer structures would not bring these groups into close proximity, several other possible rationales for magnetization transfer are shown schematically in Fig. 3. Interdigitation and chain bendback [22,23] are two modified bilayer arrangements for which thes: groups may approach intermolecularly to within 5 A of one another, thus making it possible to observe NOESY connectivities.

Fig. 3. A schematic view of lipid bilayer arrangements and spin-communication mechanisms that permit NOES between protons at the termini of the choline headgroup and the acyl chains: (a) spin diffusion in a standard bilayer; (b) interdigitation of the acyl chains; (c) and (d) bendback of either acyl chain toward the aqueous interface.

243

On the other hand, the anomalous crosspeaks may be attributed to spin diffusion [13-151, whether viewed as a relay of spin polarization through a network of protons or the establishment of a common spin temperature for these ‘H nuclei. Despite the indications of spin isolation noted above and our use of the shortest possible mixing times for which crosspeaks are discernible, there is no guarantee that particular pairs of protons linked in NOESY experiments are close in space. 3.3. Interrupting intramolecular spin difision

between the choline headgroup and acyl-chain termini

In order to prevent intramolecular magnetization transfer between w-CH, protons at the end of the acyl chains and N(CH,),‘s of the choline headgroup [ 161, the NOESY experiments were repeated using DMPC-d, (Fig. 2(b)). With the CH,CH, segment of the headgroup replaced by a CD,CD, moiety, a possible spin diffusion pathway originating at the choline methyl groups is interrupted. As molecular models show that the closest protons are now situated at least 7.4 A from one another, intramolecular NOES involving the N(CH,),‘s should be absent. Contrary to this prediction, Fig. 2(b) shows no reduction in the N(CH,),-w-CH, crosspeak intensity. In fact, this feature of the NOESY spectrum remains prominent even with mixing times as short as 200 ms (data not shown). We have also followed the strategy of other research groups by conducting MAS-NOESY experiments on a 1: 1 mixture of DMPC-d,, and DMPC-d,. Since half of the lipid molecules are chain-perdeuterated and half are headgroup-perdeuterated, intramolecular spin diffusion should be precluded and only intermolecular NOESY interactions should be permitted between these portions of the phospholipid. In earlier NMR studies of spontaneously-forming vesicles and multilayers formed by various phosphatidylcholines, the N(CH ,I,-w-CH s crosspeaks disappeared upon deuteration [ 17,181. The investigators concluded that, despite a dispersion of ‘H T, values, spin diffusion rather than spatial proximity must have been responsible for the dipolar connectivities observed in phospholipid aggregates. By contrast, similar experiments conducted on sonicated vesicles [ 161 and dilute DMPC multilayers [l l]

244

Z.-j. Chen, R.E. Stark/Solid

State Nuclear Magnetic Resonance 7 (19%) 239-246

showed retention of the NOESY crosspeak. Our MAS-NOESY experiments on DMPC multilayers reveal an w-CH,-N(CH,), crosspeak that is diminished in intensity but is seen clearly (Fig. 2(c)).

F2

(P

3.4. Assessing chain be&back and interdigitation A number of unorthodox bilayer packing arrangements have been proposed during the last 15 years based on physical measurements such as X-ray diffraction [22] and statistical mechanical calculations [23]. As shown in Fig. 3, either chain bendback or interdigitation could bring N(CH,), and w-CH, protons from neighboring molecules close to one another in space. The suggestion of a wide positional distribution of alkyl chain ends i.e., dynamic bendback of the chains toward the water-bilayer interface, has received some support from measurements of N(CH,),-w-CH3 crosspeaks in NOESY experiments performed on sonicated vesicles [ 161. Interdigitated packing arrangements, which are thought to be important for membrane fusion and cellular recognition, are consistent with the NOESY data obtained for aqueous multilamellar dispersions of phospholipids with one chain twice as long as the other [lo]. Gel-phase DMPC is not thought to interdigitate significantly [24], although no analogous determination has been made for liquid-crystalline samples. With selectively deuterated DMPC’s, it is possible to test whether either of these organizational models accounts for intermolecular NOESY crosspeaks in our multilamellar system. The strategy in these experiments is to remove one of the potentially interacting dipoles by replacing hydrogen with deuterium. If the l-chain is perdeuterated, the N(CH&w-CH, NOESY crosspeak should disappear if it stems from either mixed interdigitated bilayers or bendback of this chain (Fig. 3(b) and 3(d)). If the 2-chain is perdeuterated, the N(CH ,),-w-CH 3 NOESY crosspeak should disappear if the intermolecular interactions occur via bendback of this acyl chain (Fig. 3(c)). Phospholipid motions make each of these structural possibilities dynamic rather than static; the NOES then result from the approach of protons that move apart a short time later. Nevertheless, perdeuteration of the bending or interdigitating chain should abolish or strongly attenuate the through-space interaction. In fact, a

5.0

4.0

3.0

2.0

110

F1 (mm)

2.

1 3.0

a .---*

v

? 4. v

1

I)

5.01. p* L , , -, 5.0

4.0

/ , , , 3.0

2.0

1.0

F1 Wm)

Fig. 4. NOESY contour plot for multibilayers of DMPC-d,, (lc and Id) under conditions similar to those described in Fig. 2, showing retention of the N(CH,), - o-CH 3 crosspeak with 1-chain perdeuteration (bottom) and Zchain perdeuteration (top).

strong NOESY crosspeak is retained between the w-CH, and N(CH,), protons, regardless of which DMPC-d,, species is examined (Fig. 4). Thus neither the interdigitated nor the bentback bilayer packing model provides a convincing explanation of our NOESY results.

4. Discussion Fig. 5 provides a framework for understanding the foregoing MAS-NOESY data. We invoke the traditional bilayer structure supported by other physical data [24], viewing chain interdigitation and bendback

Z.-j. Chen, R.E. Stark/

Solid Stare Nuclear Magnetic Resonance

as less likely explanations of the NOESY results for DMPC-d,, (Fig. 4). Although the conventional bilayer organization does not favor intermolecular approach of the N(CH,), and o-CH, protons, we can explain their NOESY crosspeak with a mechanism involving close intermolecular contacts combined with intramolecular spin diffusion. If the choline methyl groups are close in space to the glycerol backbone of a neighboring DMPC molecule, then N(CH ,),-CH *O and N(CH ,),-CHO NOESY crosspeaks are expected and observed in all of the multilamellar dispersions. For 1:l mixtures of DMPC-d,, and DMPC-d,, the intermolecular headgroup-backbone interaction should occur for half of the DMPC molecules (cases “b” and “~“1. Subsequent spin diffusion along the phospholipid acyl chain will then lead to a N(CH,),--w-CH3 NOESY crosspeak in case “b” only. Supporting evidence for this mechanism of spinspin interaction comes from several sources. First, the volume of the N(CH,),-o-CH3 crosspeak in 1: 1 DMPC-d,,:DMPC-d, mixtures is predicted and found to be one-fourth of that observed in unlabeled DMPC multilayers when 200 ms NOESY results are compared for the two phospholipid samples at 30°C. (Even if such a spectral feature were present in the earlier experiments [ 17,181, it could have been difficult to detect a decade ago because of limitations in temperature control and pulse-programming hardware.) Secondly, the NOESY buildup curves show a lag time for the N(CH3),-o-CH, crosspeak (Fig. 6). This unusually sluggish cross relaxation is suggestive of spin diffusion [25], though it could also be

Fig. 5. Four possible schemes for intermolecular interactions between DMPC-d,, and DMPC-d,, where blackened headgroups and thickened acyl chains correspond to perdeuterated residues. Only arrangement “b” is consistent with the NOESY results presented in Fig. 2(c).

I

7 (1996) 239-246

/ / 100

I

300

245

I

I

500

/

700

I 900

Mixing Time (ms) Fig. 6. Variation of selected MAS-NOESY crosspeak volumes with mixing time for a I:1 mixed dispersion of DMPC-d,, and DMPC-d,. The volumes are normalized to the N(CH,), resonance.

attributed to more rapid motion [14], less orientational order [ 161, or simply smaller signals than for pairs of interacting N(CH3),-CH,O and N(CH,),CHO groups. The semisolid nature of phospholipid bilayers has enabled biophysicists to use magic-angle spinning to yield high-resolution ‘H NMR spectra [5] and to combine MAS with structurally informative solution-state experiments such as correlated spectroscopy (COSY and HETCOR [7-91) and NOESY [5,10-121. Nevertheless, the partial ordering and sluggish motion that characterize these assemblies also allow more efficient spin communication than is typical in aqueous solutions of macromolecules, and introduce a potentially confounding factor into the interpretation of MAS-NOESY experiments. Although both spin-relaxation and line-narrowing properties will reveal instances of strong spin diffusion, definitive interpretation of NOESY crosspeaks in many lyotropic liquid-crystalline systems requires selectively deuterated materials, laborious experimentation, and supporting physical evidence for their aggregate organization.

Acknowledgements This work was supported by grants (to R.E.S.) from the National Institutes of Health (ROl-

246

Z.-j. Chen, R.E. Stark/

Solid State Nuclear Magnetic Resonance 7 (1996) 239-246

DK36888) and the PSC-CUNY Research Award Program of the City University of New York (664200 and 665 151). Funds to purchase the NMR spectrometer were provided by the National Science Foundation (BIR-9214560) and the City University of New York. We thank Dr. Leon Van Gorkom for technical assistance during the course of this work.

References [ll

S.I. Chan, D.F. Bocian and N.O. Petersen, in E. Grell (Ed.), Membrane Spectroscopy, Springer-Verlag, New York, 198 1, pp. l-50. [21R.A. Bums, Jr. and M.F. Roberts, Biochemistry, 20 (1981) 7102. [31 R.G. Griffin, Methods Enzymol., 7, (1981) 108. [41 P.F. Knowles and D. Marsh, B&hem. J., 274 (1991) 625. bl J. Forbes, C. Husted and E. Oldfield, J. Am. Chem. Sot., 110 (1988) 1059. [61H. Wennerstrom, Chem. Phys. Lett., 18 (1973) 41. [71 C.W.B. Lee and R.G. Griffin, Biophys. J., 55 (1989) 355. [81K.-L. Li, C.A. Tiial, M. Guo and R.E. Stark, Biochemistry, 32 (1993) 9926. [91 J.D. Gross, P.R. Costa, J.-P. Dubacq, D.E. Warschawski, P.-N. Lirsac, P.F. Devau and R.G. Griffin, J. Magn. Reson. B., 106 (1995) 187.

I101 H.N. Halladay,

R.E. Stark, S. Ali and R. Bittman, Biophys. J., 58 (1990) 1449. [Ill F. Volke and A. Pampel, Biophys. J., 68 (1995) 1960. [I21 Z. Chen, L.C.M. Van Gorkom, R.M. Epand and R.E. Stark, Biophys. J., 70 (1996) 1412. [131 A. Kalk and H.J.C. Berendsen, J. Magn. Reson., 24 (1976) 343. 1141J.W. Keepers and T.L James, J. Magn. Reson., 57 (1984) 404. [151 J. Fejzo, A.M. Krezel, W.W. Westler, S. Macura and J.L. Markley, J. Magn. Reson., 92 (1991) 651. [I61 Z.-C. Xu and D.S. Catiso, Biophys. J., 49 (1986) 779. 26 (1987) [I71 N.E. Gabriel and M.F. Roberts, Biochemistry, 2432. [I81 J. Forbes, J. Bowers, X. Shan, L. Moran, E. Oldtield and M.A. Moscarello, J. Chem. Sot. Faraday Trans. 1, 84 (1988) 3821. [191 R.L. Void, J.S. Waugh, M.P. Klein and D.E. Phelps, J. Chem. Phys., 48 (1968) 3831. DO1 K. Wuthrich, G. Wider, G. Wagner and W. Braun, J. Mol. Biol., 155 (1982) 311. El1 D.J. States, R.A. Haberkom and D.J. Ruben, J. Magn. Reson., 43 (1982) 286. m J.L. Slater and C.-H. Huang, Prog. Lipid Res., 27 (1988) 325. k31 K.A. Dill and P.J. Flory, Proc. Natl. Acad. Sci. USA, 78 ( 1980) 676. [241 R.H. Pearson and 1. Pascher, Nature, 281 (1979) 499. b51 K. Wuthrich, NMR of Proteins and Nucleic Acids, Wiley-Interscience, New York, 1986.