Phosphatidylcholine ‘Wobble’ in Vesicles Assessed by High-Resolution ¹³C Field Cycling NMR Spectroscopy

Abstract

High resolution ¹³C NMR field cycling (covering 11.7 down to 0.002 T) relaxation studies of the sn-2 carbonyl of phosphatidylcholines in vesicles provide a detailed look at the dynamics of this position of the phospholipid in vesicles. The spin-lattice relaxation rate, R₁, observed down to 0.05 T is the result of dipolar and CSA relaxation components characterized by a single correlation time τ_c, with a small contribution from a faster motion contributing CSA relaxation. At lower fields, R₁ increases further with a correlation time consistent with vesicle tumbling. The τ_c is particularly interesting since it is 2-3 times slower than what is observed for ³¹P of the same phospholipid. However, cholesterol increases the τ_c for both ³¹P and ¹³C sites to the same value, ~25 ns. These observations suggest faster local motion dominates the dipolar relaxation of the ³¹P while a slower rotation or wobble dominates the relaxation of the carbonyl carbon by the α-CH₂ group. The faster motion must be damped with the sterol present. As a general methodology, high resolution ¹³C field cycling may be useful for quantifying dynamics in other complex systems as long as a ¹³C label (without attached protons) can be introduced.

Motions of phospholipids in membranes have been the subject of intensive research for several decades. While chain motions are understood in a quantitative fashion as a result of diverse ²H and ¹³C NMR studies,¹ the dynamic behavior of the interfacial region, including the fatty acyl carbonyls and the phosphorus moiety, is much less well characterized. Spin-lattice relaxation rates (R₁) measured over a range of magnetic fields can in principle be very useful for determining time scales for different motions in the lipid bilayer. However, for phospholipid aggregates only a few studies have measured R₁ (²H and ¹³C) at three or four fields, and those above 1 T, to try and assess lipid motions.^2,3 Recent high-resolution ³¹P field cycling spin-lattice relaxation studies have yielded information on the motions of the phosphorus of different phospholipids in diverse aggregates.^4-6 In this technique, the magnetic field is cycled by mechanically shuttling the sample from the center of the probe to a substantially lower magnetic field above the probe for the delay times normally used in conventional NMR relaxation sequences, then shuttling the sample back to the probe for readout of the relaxation at the lower field. This method, which allows access to a very wide field range (0.002 to 11.7T) for measuring R₁ and interpreting it in terms of spectral density functions,^4,5 is particularly useful for spin ½ nuclei without a directly bonded proton where the relaxation at the high fields of modern spectrometers is moderately long (R₁ ~1 s^-1) and dominated by mechanisms other than dipole-dipole interactions (e.g., chemical shift anisotropy, CSA). For multicomponent phospholipid vesicles, the analysis of the ³¹P R₁ profile as a function of field for each phospholipid molecule yields several correlation times ranging from ps to μs for the motion of this segment of the phospholipid.^4,5

The carbonyl region of phospholipids is likely to be critical for interactions with some peripheral proteins and other membrane components and might sense different motions than the phosphate portion of the molecule. In this report we present the first high-resolution ¹³C field cycling relaxation studies of sn-2 carbonyl ¹³C-labeled phosphatidylcholines (prepared by acylation of 1-palmitoyl-phosphatidyl-choline with either [1-¹³C]-oleic acid or [1-¹³C]-palmitic acid⁷) in small unilamellar vesicles (SUVs) as a direct probe of this interfacial region of phospholipids.

In small unilamellar vesicles (prepared by sonication) at 25°C, 1-palmitoyl-2-[1-¹³C]oleoylphosphatidylchoine (PO[1-¹³C]PC) mixed 1:1 with dioleoylphosphatidylmethanol (DOPMe) exhibits the field dependence profile shown in Figure 1A. Qualitatively, the profile between 0.04 and 11.7 T has features similar to that for ³¹P R₁ of the POPC in these same bilayers.^4,5 The simplest analysis treats this field-dependent profile as the result of dipolar and CSA relaxation components characterized by a single correlation time τ_c, with a small contribution from a faster motion contributing CSA relaxation.⁴ The balance of each of these terms for PO[1-¹³C]PC in the POPC/DOPMe vesicles is shown in Figure 1B. Above 2 T, R₁ is dominated by CSA relaxation, while dipolar relaxation is the major mechanism below 1 T. The small rise in ¹³C R₁ at high fields, due to faster, sub-nanosecond motions that contribute to CSA relaxation of the carbonyl, is much smaller for the ¹³C site compared to the ³¹P interaction.^4,5 More interestingly, the correlation time, τ_c, associated with the dipolar relaxation is 16±2 ns. This value is about 2-3 times the value obtained for the ³¹P nucleus of POPC in POPC/DOPMe vesicles.^4,6 Thus, the 5-7 ns motion that effectively relaxes the phosphate group either does not alter the orientation of the sn-2 carbonyl site or contributes much less to relaxation of the ¹³C-carbonyl compared to a slower motion. The value for R_c(0), the relaxation rate extrapolated to zero field, from this region of the field dependence of R₁ provides an estimate of r_CH, the distance of the ¹³C-labeled carbon to the major proton(s) that relax it,⁴ in this case the sn-2 α-CH₂.⁸ The value obtained, 2.4 Å, is similar to what one would expect for the ¹³C-C-H distance involved.

Figure 1.

(A) Field dependence of ¹³C spin lattice relaxation rates, R₁, for PO[1-¹³C]PC in SUVs composed of POPC (5 mM)/DOPMe (5 mM) in the absence (●) and presence (○) of 5 mM cholesterol; the semi-log plot emphasizes the behavior below 1T). (B) Deconvolution of R₁ in the absence of cholesterol into a dipolar (—) and CSA (- - -) component with correlation time τ_c, and a faster CSA motion (dotted line visible above 6T).

We also examined the ¹³C relaxation profile for an sn-2 ¹³C-labeled saturated chain lipid, dipalmitoyl-PC (PP[1-¹³C]PC), in this case mixed with unlabeled POPC to stabilize the small vesicles for extended observation at 40°C. The temperature was chosen to keep the fluid bilayer for the many acquisitions covering a course of a few days and to avoid getting too close to the maximum temperature of the shuttling system (50°C). As seen in Figure 2A, the PP[1-¹³C]PC profile was very similar to that for the PO[1-¹³C]PC and could be fit with a single τ_c of 16.2±2.5 ns; the average r_CH was 2.51 Å.

Figure 2.

(A) Field dependence of PP[1-¹³C]PC in SUVs of DPPC (5 mM)/POPC (5 mM) at 40°C in the absence (●) and presence (□) of 5 mM cholesterol. (B) Very low field dependence of R₁ on field from 0.07 down to 0.004 T for PO[1-¹³C]PC (mixed with DOPMe) in the absence (●) and presence (○) of 5 mM cholesterol.

If the ¹³C R₁ is measured at much lower fields (down to 0.002 T), there is a further rise in the ¹³C relaxation rate that reflects the vesicle tumbling contribution to R₁.⁵ For PO[1-¹³C]PC mixed with DOPMe, the correlation time is 1.4±0.5 μs (Figure 2B). This is very close to what one would expect as the correlation time for rotation of 250-300 Å vesicles (the average diameter measured by dynamic light scattering for this preparation is 254 Å with 91% of the sizes between 200 and 400Å).⁹ Geometric information can also be obtained from this very low field relaxation.⁵ The area under this dipolar portion of the R₁ versus field curve compared to total dipolar relaxation yielded four values (±45.1° and ±65.7°) for the average θ_CH that the angle of the ¹³C-H vector makes with respect to the bilayer normal. While the orientation of the carbonyl to α-CH₂ vectors in the crystal structure of dimyristoyl-PC might rule out the large negative angles,¹⁰ there may be sufficient segmental motion of the CH₂ so that differentiating among these θ_CH is difficult. However, the ability to measure an averaged angle provides a novel way of characterizing changes induced by additives.

Cholesterol in a bilayer broadens the phase transition and makes membranes more gel-like. This physical change lengthens ³¹P τ_c values dramatically^4,5; it would also be expected to alter the dynamics of the sn-2 carbonyl group of PC. With 33 mol% cholesterol in the PO[1-¹³C]PC/DOPMe vesicles (Figure 1A), the ¹³C-carbonyl of PC shows an increase in τ_c with cholesterol, but it is relatively small – from 16±2 ns to 23±3 ns. For the same amount of cholesterol in the PP[1-¹³C]PC/POPC vesicles, the τ_c increased to 28±4 ns (Figure 2A). Thus, the effect of cholesterol on the carbonyl motions of the two different vesicle systems is essentially the same – a small but significant increase in τ_c. For comparison, with cholesterol present, the ³¹P τ_c increased from 5-7 ns to ~25 ns (data not shown). In the presence of the sterol, the nanosecond correlation times for both the carbonyl and phosphate moieties are essentially the same, whereas in the absence of cholesterol, the two nuclei are sensitive to motions on slightly different time scales.

Another way of looking at differences in the ³¹P and ¹³C motions in the nanosecond regime is to examine the temperature dependence of R₁ at a fixed low field where R₁ is directly proportional to τ_c. Different motions are likely to have different energetics. For the ¹³C-labeled POPC, R₁ was measured at 0.06 T. An Arrhenius plot of the ¹³C R₁ (Figure 3) leads to a slope of 27±7 kJ/mol. For comparison the temperature dependence for the ³¹P of POPC at low field (0.032 T) yields a lower energy barrier, 12±3.8 kJ/mol.

Figure 3.

Fixed low field spin lattice relaxation rates for POPC in small vesicles with DOPMe as a function of temperature: (●) ¹³C R₁ for PO[1-¹³C]PC measured at 0.06 T; (○) ³¹P R₁ for POPC measured at 0.032 T.

These differences in the τ_c extracted from ³¹P and ¹³C in the same sample strongly indicate that the dipolar interactions of each group with its nearest protons do not reflect the same overall motion of the phospholipid. It has been suggested from molecular dynamics simulations that the ~10 ns ³¹P τ_c arises from motions that treat the phospholipid as a cylinder encompassing the ³¹P-glycerol-acyl chains that ‘wobbles’ around an axis perpendicular to the membrane surface.⁶ The shorter τ_c for ³¹P compared to ¹³C suggests that the phosphorus motion also contains some faster local motion along with the motion supplied by wobble (presumably what dominates the more rigid carbonyl relaxation). Interestingly, both ³¹P and ¹³C have the same correlation time when cholesterol is present, suggesting the faster motion of the phosphorus has been dampened by the presence of the sterol. High resolution ¹³C field cycling may also be useful for quantifying dynamics in other complex systems as long as the ¹³C label (without attached protons) can be introduced.