Quantification of Cholesteryl Esters in Human and Rabbit Atherosclerotic Plaques by Magic-Angle Spinning ¹³C-NMR

Materials

CEs (>99% pure) were purchased from NuChek. The purity of the CEs was checked by differential scanning calorimetry (DSC) and thin-layer chromatography before and after NMR experiments. No significant impurities were found.

Human carotid plaque tissues were obtained by endarterectomy, fixed in 10% formalin/Tris-buffered saline immediately after excision (except as noted below), and shipped at ambient temperatures from Baylor College of Medicine (Houston, Tex) to Boston University School of Medicine (Boston, Mass).

Animal tissues were obtained from 4-year-old Watanabe heritable hyperlipidemic (WHHL) rabbits,²¹ which were a gift from Dr Thomas Parker of Cornell University, Ithaca, NY. No dietary or surgical interventions were performed on these rabbits. The rabbit plaque tissue samples were fixed in 10% formalin/saline solution and stored at −80°C.

Sample Preparation

A mixture of CEs with acyl chains composed of 20% palmitate (C16: 0), 30% oleate (C18:1), 35% linoleate (C18:2), 5% palmitoleate (C16:1), 5% linolenate (C18:3), and 5% myristate (C14:0) was used to model the mixture of CEs in plaque.^{10 22} This model sample exists in a liquid-crystalline phase between 38°C and 44°C and in a liquid phase at temperatures >44°C. In real plaque samples, the CE pool usually contains small fractions of impurities, including other lipids, such as triglycerides and cholesterol. These solute molecules generally lower the melting temperature of the CE pool to below or near body temperature.⁹ To avoid complications, we chose to use the model CE mixture without adding these minor components. Hence, the melting temperature of our model system is higher than are CE mixtures from typical plaque samples. Because the molecular motions (viscosity) near the melting point may be anomalous, we performed our MAS-NMR experiments at temperatures 5°C higher than the corresponding phase-transition temperatures, where the motions are representative of a pure liquid phase.²⁰

The CE components were added together in the crystalline or oil phase and then heated to 5°C above the highest melting CE (cholesterol palmitate) and cooled to ambient temperatures. After 5 cycles of heating and cooling, the CE mixture was considered to be homogenous, based on reproducible DSC thermograms. Such a protocol was required because the samples were mixed without solvents.

Plaque samples were cut into small segments (≈2.5- to 5-mm long) to fit into 7-mm ZrO₂ sample rotors used for MAS-NMR measurements. Because rabbit tissues were stored at low temperature, which results in the crystallization of CEs, these plaque samples were heated to 60°C and then cooled to the desired temperature for NMR experiments, generally 37°C. Previous studies had shown that this preheating will melt the crystalline CE phase, which, after cooling, will remain in either the isotropic or liquid-crystalline phase as it was in vivo.³ Thus, the original phases of plaque CEs are restored after preheating of the frozen plaque tissues.²³

To validate that our temperature protocol (freezing, then heating to 60°C and recooling) did not affect spectroscopic properties, we performed the following experiments. Two human carotid plaques (additional to those reported in the Table⇓) were shipped at ambient temperature in PBS/glycerol buffer (50%) without fixation, and ¹³C MAS-NMR spectra were obtained at 37°C before they were frozen. The plaques were then kept at −80°C for 18 hours and thawed at room temperature. After being placed in the NMR probe, each plaque was heated to 60°C for ≈30 minutes and cooled to 37°C. The spectra obtained after this protocol were the same as the initial spectra, indicating that our temperature protocol did not affect the quantification of lipids. Previously, we have also shown that fixation in formalin does not affect the NMR spectra of plaques.⁷

NMR Measurements

MAS-NMR experiments were performed on a Bruker AMX-300 NMR spectrometer equipped with solid-state NMR accessories. For experiments with model CE mixtures, samples were sealed in a small Pyrex glass sphere to avoid leakage of the liquid CE mixture. The glass sphere was then placed in a 7-mm ZnO₂ rotor and balanced with KCl powder in the rotor to avoid vibration during sample spinning. Off-MAS-NMR experiments were performed under conditions identical to those of MAS NMR, except that the sample rotation axis was changed to 7° off the magic angle, as described previously.¹⁸ We have observed that the off-MAS experiment is more advantageous than the nonspinning experiment for detecting the liquid phase because for the same sample size used, rapid sample spinning achieves better field homogeneity and thus, higher signal sensitivity. When MAS and off-MAS-NMR experiments are used together, the CE components in liquid and liquid-crystalline phases can be identified individually for lipids in model mixtures and plaques in situ.^{16 18 22 24}

All spectra were obtained with a 5-μs 90° pulse width, 12K scans, and a spinning rate of 4 kHz. The ¹³C chemical shifts were referenced to the carbonyl carbon resonance of an external glycine sample (176.06 ppm from tetramethylsilane). Peak assignments were based on previous studies.^{19 25 26} Because the T₁ values for C5 and C6 carbons in noncrystalline CEs are <1 second,²⁵ a 5-second pulse interval time was used to allow full magnetization relaxation, which is important for quantification studies at NMR intensity. The sample temperature was controlled within 1°C by the Bruker B-VT-1000 variable temperature unit. The actual probe temperature was calibrated with the known phase-transition temperatures of pure CEs.¹⁸ The mild baseline “roll” in some spectra results from the plastic insert in the MAS-NMR rotor. After completion of the NMR experiments, the plaques were homogenized and their lipid composition was determined by standard chemical assays.²⁷

DSC Measurements

Measurements were made with a Perkin-Elmer DSC-7 instrument over the temperature range of 0°C to 95°C at heating and cooling rates of 0.5°C/min. DSC confirmed that the CE mixture existed as an isotropic liquid phase above 44°C and as a liquid-crystalline phase between 34°C and 44°C after being cooled from the liquid phase. For pure cholesterol oleate or the CE mixture, we used 2 to 10 mg of sample. Larger sample sizes (wet weight 10 to ≈120 mg) were used for plaque tissues. The phase-transition temperature was determined from the position of the transition peak on the DSC thermograms.

Polarizing Light Microscopy

Microscopy experiments were performed on a Leitz-Dialux microscope fitted with a polarizer and heating stage and a Nikon Microflex UFX camera system. For observing the phase transition of plaque lipids, a heating rate of 1°C/min was used.

Statistical Analysis

Standard statistical analysis with the origin program was used to compare NMR and chemical analysis results. The F factor was calculated as the ratio of the variance of chemical analysis results to the variance of the NMR results: F=S₁² /S₂² , where S₁ represents the chemical analysis variance and S₂ is the NMR analysis variance. The F factor of 1.10 indicates close agreement between these 2 groups of results. Errors are attributable mainly to baseline error and the signal-to-noise ratios of peaks that are integrated. The variability of the NMR integration was determined from ≥3 rounds of integration of the C5 and C6 peaks for each sample.

Identification of CE Phase Transitions in Rabbit Plaques by DSC and Polarizing Light Microscopy

Storage of plaque at low temperature (≤−80°C) results in crystallization of its CEs, thereby altering the phases that were present in vivo. The existence of crystallized CEs can be detected by DSC. The first heating trace for a rabbit aortic plaque after being stored at −80°C (please see Figure I at http://atvb.ahajournals.org) showed a broad endothermic transition (26°C to 39°C) representing the melting of crystalline CEs; no additional phase transitions were detected between 40°C and 95°C. The crystalline CE phase was not detected in subsequent cooling/heating cycles so long as samples were not frozen again, as expected. The broad transition reflects poor cooperativity of melting, probably because CEs with different acyl chains in plaques are inhomogeneously distributed and the CEs are in different lipid pools with different phase-transition temperatures.³ Another possibility is that after rapid freezing, the CE formed a supercooled solid amorphous phase with imperfect crystalline structures.²² The melting of such a solid mixture exhibits a broad transition-temperature range. Because of limitations in instrument sensitivity, we did not detect transitions between liquid-crystalline and liquid phases, which are much less endothermic. Previous studies have shown that after this preheating of a plaque sample, the original phases of CEs at physiological temperature are restored.³

Polarizing light microscopy is a more sensitive method for observing liquid to liquid-crystalline phase transitions of CEs and allows a semiquantitative measurement of the phases on each thin section of tissue examined. Figure 1A⇓ shows a polarized light microphotograph (×60) of a lipid-rich plaque section (10-μm slice) at 38°C. All of the CEs were melted in a liquid phase, because heating to 60°C did not reveal further phase transitions (loss of birefringence). The weakly birefringent regions (less bright) were observed between 38°C and 60°C may be from phospholipid/cholesterol bilayers or tissue matrix proteins. In contrast, there was a remarkable change in birefringence (bright regions) on cooling from 38°C (Figure 1A⇓) to 25°C (Figure 1B⇓), corresponding to the formation of liquid-crystalline CEs. In Figure 1A⇓, a strongly birefringent component was also detected near the site of plaque fracture (near the fibrous cap). This may be crystalline cholesterol, because the adjacent sample in this plaque was found by cross-polarization MAS spectroscopy⁷ to be rich in cholesterol monohydrate (not shown). The estimation of the total CE in a specific phase in an entire plaque by polarizing light microscopy is not feasible.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

⇓. A, Polarized light photomicrograph of a WHHL rabbit aortic plaque (10-μm slice) with low stenosis but with the appearance of rupture. The black regions represent nonbirefringent (liquid) CEs in the plaque at 38°C. B, Photomicrograph of liquid-crystalline CEs in the same plaque segment at 25°C. The bright regions at 25°C that correspond to black regions at 38°C represent birefringent liquid-crystalline CEs. The photomicrographs were from a plaque thin section obtained before NMR studies (spectrum C in Figure 4⇓).

¹³C NMR Quantification of Pure CEs and a CE Mixture

MAS-NMR can detect and in principle quantify both liquid and liquid-crystalline CEs in plaques with different pathological characteristics.^{22 24} To quantify CEs in plaques, it is necessary to construct a calibration curve based on pure CE treated under the same experimental conditions used to study plaques. Figure 2A⇓ shows the high-resolution, natural abundance proton-decoupled ¹³C MAS-NMR spectrum of a model CE mixture in the liquid phase. The resonances of the C5 and C6 carbons in the sterol ring are well separated from the resonances of other components in plaques, including resonances of cholesterol.^{16 25} The chemical shifts of the C5 and C6 resonances of CE are independent of acyl chain composition²² and potentially suitable for quantification of total CEs in plaques. The inserts in Figure 2A⇓ show increasing intensities of the C5 and C6 peaks with increasing amounts of CEs.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

⇓. A, ¹³C MAS-NMR spectra of different quantities of a model CE mixture (composition as in Methods). Selected resonances are identified as follows: carbonyl carbon (C=O), methylene carbon (C=C), terminal methyl (ω-CH₃), and steroid ring resonances (C, followed by the carbon numbers). All spectra were obtained at 55°C with a spinning rate of 4 kHz; 12 880 spectral accumulations were processed with 5-Hz line broadening. B, ¹³C NMR calibration results based on the integration of C5 and C6 resonances of cholesterol oleate and a CE mixture. The average integrated intensities of the C5 and C6 peaks were plotted against the corresponding sample weights: (•) MAS and (○) off-MAS results for the CE mixture; MAS(▴) and off-MAS(▵) for pure cholesterol oleate. The solid line is the linear regression fit of the data and intersects the x axis at 2.2 mg. The dashed line is the linear regression fit of the open circles and intersects the x axis at 4 mg. It should be emphasized that such calibration curves must be obtained for each individual spectrometer and the chosen experimental conditions.

Similar spectral features were detected for this model mixture in the liquid phase with off-MAS-NMR. The integrated intensities of the C5 and C6 peaks were plotted against the corresponding sample quantities Figure 2B⇑; the filled circles represent the results with MAS, and the open circles, results with off-MAS. The solid line representing the linear regression fit of the MAS-NMR data crosses the x axis at 2.2 mg, implying a detection threshold of ≈2.0 mg by this method. The dashed line is the linear regression fit of the off-MAS-NMR data, and it crosses the x axis at 4.0 mg, implying a detection limit of ≈4.0 mg for the off-MAS experiment. The intensity from the off-MAS spectrum for a given amount of CEs (open circle) is ≈70% of the intensity of that for MAS, which indicates an instrumental sensitivity difference between the MAS and off-MAS measurements, because the model CE mixtures are completely isotropic at 55°C. This lower sensitivity results in a higher threshold of detection for the off-MAS experiment. Therefore, different calibration curves are required for off-MAS and MAS quantification of liquid CEs. Additional data showing the linear relationship between off-MAS-NMR peak intensity and the sample quantity of cholesterol oleate are shown in Figure 2B⇑ (filled circle). The intensities in off-MAS and on-MAS measurements of pure cholesterol oleate at 86°C fell on the same curves as those for the model CE mixture at 55°C (Figure 2B⇑). This indicates that liquid CEs can be quantified by using the same calibration curve, independent of their acyl chain composition.

To test further the feasibility of using NMR to quantify CEs in different phases (liquid and liquid-crystalline states), we prepared a heterogeneous mixture of these 2 phases by placing cholesterol oleate (C18:1; 6.0 mg) and cholesterol myristate (C14:0; 13.2 mg) in the same sample rotor without direct contact between the 2 species. This was achieved by sealing each sample inside a separate, small Pyrex glass sphere and then placing both sealed glass spheres into the same 7-mm rotor for NMR measurements. In this way, the thermotropic phase behavior of each individual component is preserved (details of the phase transitions of each of these CE are found in Reference 28²⁸ ).

Figure 3⇓ shows the off-MAS-NMR spectra at 80°C and 86°C. At 80°C, cholesteryl oleate is completely liquid and cholesterol myristate, completely liquid-crystalline. By using the calibration curve of Figure 2B⇑ for off-MAS experiments, the integrated intensities of C5 and C6 peaks at 80°C correspond to ≈6 mg total CE. This is the amount of cholesterol oleate present in the sample mixture, indicating that only signals of the liquid phase were detected, as expected. At 86°C, cholesterol myristate also melted into the isotropic phase. The integrated intensities of C5 and C6 peaks at this temperature correspond to ≈19.0 mg total CE, equivalent to the total mass of cholesterol oleate and cholesterol myristate. This finding indicates that the off-MAS spectrum at 86°C reflects the combined signals from the cholesterol oleate and cholesterol myristate.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

⇓. A, Off-MAS spectrum at 80°C for cholesterol oleate and cholesterol myristate present in 2 separate Pyrex glass spheres placed into the same MAS-NMR rotor. B, Off-MAS spectrum at 86°C for the same sample configuration as in A.

In contrast to the off-MAS spectrum, the MAS spectrum (not shown) detected the same total intensity of CE signals at both temperatures, 80°C and 86°C (≈19 mg, estimated from the calibration curve for MAS NMR experiments in Figure 2B⇑). By comparing the MAS and off-MAS NMR results at 80°C, we estimated that 13.0 mg of the total CE was liquid-crystalline at this temperature. This corresponds to the amount of cholesterol myristate present in this model system. Hence, we have demonstrated that the combination of MAS and off-MAS NMR can be used to quantify CEs in different phases. This method has potential application to studies of atherosclerotic plaques and other biological samples.

Quantification of CEs in Rabbit and Human Atherosclerotic Lesions

Selected ¹³C MAS-NMR spectra at 37°C of atherosclerotic aorta tissues from 3 different WHHL rabbits are shown in Figure 4⇓. The NMR spectra contain well-defined signals of CEs, as in the model CE mixture (Figure 2A⇑). The quantity of CEs in these and some additional plaques was determined by using the calibration method described above. After the NMR experiments, the plaque tissues were homogenized and the lipids extracted for chemical analysis. The results of chemical analysis and MAS-NMR quantification (the Table⇑) are in good agreement. MAS-NMR spectra of human carotid plaques (not shown) obtained from endarterectomy samples were generally similar to those of the rabbit aortic plaques, as published previously.¹⁶ The MAS-NMR and chemical analysis of the CE content of selected individual human plaques are in good agreement (the Table⇑).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

⇓. ¹³C MAS-NMR spectra of 3 WHHL rabbit aortic plaque tissues under the same experimental conditions used for calibration curves, except temperature (37°C). The well-resolved C5 and C6 peaks of CEs were used for quantification, based on the calibration results of model CE mixtures in Figure 2B⇑. After the NMR experiments, the lipid content of the tissues was determined by chemical analysis (the Table⇑).

The liquid and liquid-crystalline phases of CEs in plaques were identified as described before¹⁶ and quantified by the approach described above for the model CE mixture. The results thus obtained for selected rabbit plaques (the Table⇑) show variable amounts of CEs in the liquid-crystalline phase at 37°C, ranging from 6% to 41% (for additional MAS and off-MAS spectra of these plaques, please see Figure II at http://atvb.ahajournals.org). This indicates that most of the CEs in the plaques used for this study were liquid at or near body temperature. These results agree with the previous NMR studies that show rabbit aorta plaque CEs were either in a liquid phase²⁵ or that liquid and liquid-crystalline phases coexisted near body temperature.²⁹