The link between body weight, lipid metabolism, and health risks is poorly understood and difficult to study. Magnetic resonance spectroscopy (MRS) permits noninvasive investigation of lipid metabolism. We extended existing two-dimensional MRS techniques to permit quantification of intra- and extramyocellular lipid (IMCL and EMCL, respectively) compartments and their degree of unsaturation in human subjects and correlated these results with body mass index (BMI). Using muscle creatine for normalization, we observed a statistically significant (P < 0.01) increase in the IMCL-to-creatine ratio with BMI (n = 8 subjects per group): 5.9 ± 1.7 at BMI < 25, 10.9 ± 1.82 at 25 < BMI < 30, and 13.1 ± 0.87 at BMI > 30. Similarly, the degree of IMCL unsaturation decreased significantly (P < 0.01) with BMI: 1.51 ± 0.08 at BMI < 25, 1.30 ± 0.11 at 25 < BMI < 30, and 0.90 ± 0.14 at BMI > 30. We conclude that important aspects of lipid metabolism can be evaluated by two-dimensional MRS and propose that degree of unsaturation measured noninvasively may serve as a biomarker for lipid metabolic defects associated with obesity.
- magnetic resonance spectroscopy
- lipid unsaturation
- intramyocellular lipid
- extramyocellular lipid
the amount of body fat is a risk factor for several obesity-related disorders. Obesity is a known risk factor for the development of insulin resistance and diabetes and is a key component of metabolic syndrome. The causal relationship between increased dyslipidemia and adiposity and impaired glucose homeostasis is unclear, although it is known that lipid oversupply to the organs primarily involved in glucose homeostasis, that is, muscle, liver, and pancreas, leads to impaired insulin function in those tissues (22).
Previous studies showed that intramyocellular lipid (IMCL) is increased with obesity and in non-insulin-dependent diabetes mellitus (9, 11, 15). It has been suggested that increased visceral adiposity and reduced lipid oxidation might contribute to the increase in IMCL (18). Thus the ability to monitor the IMCL pool and its properties noninvasively by magnetic resonance spectroscopy (MRS) has been an important development (4, 28). The correlation between the magnitude of the IMCL pool, as determined by MRS studies, and insulin resistance, diabetes, and disorders of lipid metabolism has been previously demonstrated (2, 3, 32, 33). Nevertheless, quantification of IMCL and extramyocellular lipid (EMCL) by MRS remains highly problematic (32). The ability to distinguish IMCL from EMCL is based on their different bulk magnetic susceptibility effects due to their geometric arrangements within muscle, which leads to a spectroscopic frequency separation between the two pools; this separation is 0.15 ppm in the soleus and 0.2 ppm in the tibialis muscle and has been validated in vivo (33, 34). At a clinical field strength of 1.5 T, these values are 9.6 and 12.8 Hz, respectively; at 3 T, the frequency separations are twofold greater (32). This permits the IMCL and EMCL signals to be individually identified in MRS but is not adequate for complete separation.
The degree of lipid unsaturation within the IMCL and EMCL pools is also of substantial clinical importance (1, 12). Fatty acids are derived from dietary sources and intracellular biosynthesis, with lipids being composed of three fatty acid chains and a glycerol molecule. The effects of fatty acid on metabolic signaling and energy metabolism are modulated by degree of unsaturation (38). Monounsaturated fatty acids and polyunsaturated fatty acids (PUFAs) are synthesized by the desaturation of saturated fatty acids by an oxidative reaction catalyzed by the iron-containing microsomal enzyme stearoyl CoA desaturase. Therefore, this enzyme determines the degree of desaturation, and its activity, accordingly, has profound effects on lipid metabolism (38). Unfortunately, separation of the olefinic protons (-CH=CH-) from IMCL and EMCL lipid pools in skeletal muscle by conventional MRS techniques is particularly problematic because of the small spectral chemical shifts involved (2, 3).
Obesity, as defined by body mass index (BMI), exhibits well-known correlates with lipid metabolism (43). Our hypothesis is that noninvasive MRS studies can detect variations in the IMCL and EMCL and in the degree of unsaturation within IMCL and EMCL that are consistent with these known features of lipid metabolism. Therefore, the goal of this work is to elucidate the relationship between BMI, IMCL, and EMCL and the degree of unsaturation as determined by noninvasive MRS studies. The MRS analysis was performed using our recently introduced methods for improved quantification of the IMCL and EMCL pools and their degree of unsaturation (36, 37). These techniques separate the overlapping resonances of IMCL and EMCL and the saturated and unsaturated pools within them by means of two-dimensional (2-D) spectroscopy (36, 37), as opposed to the one-dimensional (1-D) spectroscopic approach applied by previous investigators (2). Although this previous work permitted correlation of the combined saturated and unsaturated components of IMCL and EMCL to pathology, our approach, leading to individual quantification of the saturated and unsaturated fatty acids within these pools, permits a substantially more detailed characterization of lipid status.
MATERIALS AND METHODS
A total of 24 male subjects (20–38 yr of age) were studied, with 8 subjects in each of three groups defined by BMI as follows: 18–24.9 (normal weight), 25–29.9 (overweight), and 30–38 (obese). All subjects were healthy and without any diagnosed or suspected metabolic disorders and gave informed consent. The experimental protocol was approved by the local institutional review board. All data were acquired from the soleus muscle.
Subjects were not participants in high levels of exercise training and were instructed to avoid fatty foods for ≥3 days before data acquisition. Female subjects were excluded to avoid introduction of sex and endocrine status as confounding variables (37a).
Magnetic resonance imaging and MRS experiments.
All experiments were performed using a 3-T magnetic resonance imaging (MRI)/MRS Excite HD scanner (GE Healthcare Technologies, Waukesha, WI) with self-shielded gradients (40 mT/m). At the time of MRS data acquisition, subjects were positioned supine, feet first, within the magnet, with the right calf placed inside a vendor-supplied quadrature transmit/receive coil. For standardization of the position and alignment of the subject's leg, a point one-third of the distance from the medial femoral condyle to the medial malleolus was marked and used for all MRS data acquisition. Five repeated measurements on different days were performed on one subject from each of the three groups to assess measurement variability.
Gradient echo scout images were acquired in the axial, coronal, and sagittal planes with 5-mm slice thickness, in-plane spatial resolution of 976 × 976 μm, repetition time (TR) of 14.7 ms, one excitation, and echo time (TE) of 2.25 ms. After these images were used to localize the target volume, MRS data were collected from a 3 × 3 × 3 cm (27-ml) voxel within the soleus with use of the following two protocols.
Localized 1-D MRS was performed using point-resolved spectroscopy (5) in one subject from each of the three subject groups to demonstrate the overlapping resonances from IMCL and EMCL. The radiofrequency pulse bandwidths and details of the sequence were as described in our previous work (36); acquisition parameters included TR/TE of 2 s/24 ms, 64 excitations, acquisition of 2,048 complex points, and spectral width of 5 kHz.
Localized 2-D correlation spectroscopy (35–36), in which cross peaks are formed from indirect spin-spin couplings, was performed on all subjects with a minimal TE of 30 ms and 64 incremented time steps (t1), with increment (Δt1) = 0.8 ms, resulting in a spectral bandwidth of 1,250 Hz along the second, indirectly detected, spectral dimension. The signals for each value of t1 were averaged over eight scans, resulting in a total acquisition time of ∼17 min. The directly detected dimension (t2) had a bandwidth of 5,000 Hz with 2,048 complex points sampled.
Postprocessing was performed using FELIX software (Accelrys, San Diego, CA). Before Fourier transformation, the time domain data matrix was apodized using shifted sine-squared filters in the t1 and t2 dimensions. The length of the windows was adjusted so that the filters reached zero at the last experimental data point in both dimensions. All data sets were linear predicted from 50 t1 points to 100 t1 points along the F1 dimension and zero filled to 256 points. The diagonal and cross peaks were assigned, and the volume integrals were quantified, including corrections for T1 relaxation as described previously (36). The diagonal peak volumes from (CH2)n protons within the IMCL (1.2, 1.2 ppm) and EMCL (1.35, 1.35 ppm) and the creatine (Cr) signal at 3.03 ppm were estimated. The volumes of the cross peaks between olefinic (-CH=CH-) and allylic methylene (CH2CH=CH) protons within IMCL and EMCL at (5.3, 2.0 ppm) and (5.45, 2.15 ppm), respectively, reflect the size of the monounsaturated fatty acid component of the two lipid pools. Similarly, the cross-peak volumes between olefinic and diallylic methylene (-CH=CH-CH2-CH=CH-) protons within IMCL and EMCL at (5.3, 2.7 ppm) and (5.45, 2.85 ppm), respectively, reflect PUFAs. Ratios of these cross-peak volumes define the degree of unsaturation (36, 37).
ANOVA was used to compare the IMCL-to-Cr ratio, EMCL-to-Cr ratio, and degree of unsaturation among normal-weight, overweight, and obese subjects. Bonferroni's post hoc test was employed to test for pairwise differences. Values are means ± SD, and statistical significance was assigned at P < 0.05.
Figure 1A shows a 1-D spectrum obtained by point-resolved spectroscopy (5) from soleus muscle. The CH3 and (CH2)n from the IMCL and EMCL pools, along with Cr and trimethylamine protons, are seen. Because of the complete overlap of the olefinic protons from IMCL and EMCL resonances between 5.3 and 5.5 ppm, it was impossible to estimate the degree of unsaturation in the two lipid pools by use of these resonances.
Figure 1B shows a localized 2-D correlation spectrum (localized 2-D correlation spectroscopy) (35, 36) recorded from a normal-weight 30-yr-old subject. Resonance assignments were based on our earlier work (36). In addition to the CH3 and (CH2)n groups of IMCL and EMCL, olefinic, allylic methylene, and diallylic methylene groups from the IMCL and EMCL lipid pools can be identified by the cross peaks C1, C2, C3, and C4. The glycerol backbone protons (cross peaks labeled C5) and the imidazole protons of carnosine (cross peaks labeled C6) are also evident. The cross peaks labeled C1 and C3 arise from the indirect spin-spin coupling between olefinic (-CH=CH-) and allylic methylene (CH2CH=CH) protons of IMCL and EMCL, respectively. Cross peaks C2 and C4 arise from the indirect spin-spin coupling between olefinic (-CH=CH-) and diallylic methylene protons (-CH=CH-CH2-CH=CH-) of IMCL and EMCL, respectively. The cross peaks labeled C5 are due to the J coupling between CH2 and CH groups of the glycerol backbone protons, whereas those labeled C6 are due to the residual dipolar coupling between CH groups from the imidazole ring protons of carnosine (2–4, 36, 37). The diagonal resonance frequencies (F1, F2) from the n-methylene protons resonating at (1.2, 1.2 ppm) for IMCL and at (1.35, 1.35 ppm) for EMCL and the signal of the methylene group of Cr resonating at (3.03, 3.03 ppm) were used to calculate the IMCL-to-Cr and EMCL-to-Cr ratios. Figure 1C shows an expansion of the C1, C2, C3, and C4 cross-peak region of spectra generated from normal-weight, overweight, and obese subjects. The reduction in the volume of C2 and C4, corresponding to PUFAs, is reduced in obese and overweight subjects compared with normal-weight subjects, reflecting their lower degree of unsaturation.
Table 1 shows the resonance frequencies employed for estimation of the IMCL-to-Cr ratio, EMCL-to-Cr ratio, and degree of unsaturation within IMCL and EMCL. Average IMCL-to-Cr ratio and corresponding coefficients of variation in the three groups of subjects were 5.9 ± 1.7 (28.8%), 10.9 ± 1.82 (16.6%), and 13.1 ± 0.87 (6.6%) for the normal-weight, overweight, and obese subjects, respectively (Fig. 2A). The differences between values for the normal-weight vs. overweight, normal-weight vs. obese, and overweight vs. obese subjects were statistically significant. Corresponding results from five repeated measurements on a single subject from each group were 5.6 ± 0.37 (6.6%), 11.4 ± 0.64 (5.6%), and 13.7 ± 0.5 (3.6%), respectively. In all cases, the standard deviations of the repeated measurements were smaller than the differences between group means.
Corresponding EMCL-to-Cr ratios were 18.4 ± 2.64 (14.3%), 25.7 ± 1.77 (6.8%), and 31.2 ± 2.3 (7.37%) for normal-weight, overweight, and obese subjects, respectively (Fig. 2A). The value for each group was statistically significantly different from the values of the other two groups. Corresponding results from five repeated measurements on a single subject from each group were 17.5 ± 0.69 (3.9%), 24.9 ± 1.01 (4.0%), and 34.2 ± 1.22 (3.6%). Again, the standard deviations of the repeated measurements were smaller than the differences between group means.
Average degree of unsaturation in the IMCL pool and corresponding coefficient of variation in the three groups of subjects were 1.51 ± 0.08 (5.2%), 1.30 ± 0.11 (8.5%), and 0.90 ± 0.14 (15.5%) for the normal-weight, overweight, and obese subjects, respectively (Fig. 2B). The value for each group was statistically significantly different from the values of the other two groups. Corresponding results from five repeated measurements on a single subject from each group were 1.42 ± 0.042 (2.9%), 1.10 ± 0.035 (3.18%), and 0.78 ± 0.031 (4%), respectively. The standard deviation of the repeated measurements was smaller than the differences between group means.
Corresponding values for the degree of unsaturation in the EMCL pool were 1.12 ± 0.037 (3.3%), 1.04 ± 0.045 (4.3%), and 0.82 ± 0.14 (17%) for normal-weight, overweight, and obese subjects, respectively (Fig. 2B). The differences between values for the normal-weight vs. obese and overweight vs. obese subjects were statistically significant. Corresponding results from five repeated measurements on a single subject from each group were 1.00 ± 0.06 (6%), 0.94 ± 0.034 (3.6%), and 0.63 ± 0.03 (4.8%), respectively. The standard deviation of the repeated measurements was smaller than the differences between group means.
The biosynthesis and composition of fatty acids in phospholipids have been found to be abnormal in obesity, diabetes, and metabolic syndrome in human subjects (39, 40) and within plasma, liver, heart, kidneys, and erythrocytes of experimental animals in models of diabetes mellitus (10, 21).
IMCL level as determined by MRI is an established marker for characterization of obesity-related insulin resistance, diabetes, and metabolic syndrome (2, 3) and has more recently been used to evaluate risk for cardiovascular disease (41). It has also been shown that IMCL functions as storage for energy substrate readily accessed during rest and physical activity (1a), whereas EMCL provides long-term energy storage (1a). Nevertheless, the mechanisms by which these pools develop and their potential causal relationship to deficits in glucose homeostasis are incompletely understood and represent the topic of ongoing investigation (24, 26, 30).
Recent biochemical studies have emphasized the roles of saturated and unsaturated fatty acids in obesity and diabetes (8, 19, 20, 42). For example, relatively high levels of saturated fatty acids and low levels of PUFAs are found in individuals with insulin resistance and metabolic syndrome (17). However, the ability to distinguish the roles and metabolic correlates of the degree of lipid saturation has been hampered by the technical challenges involved in such studies. Previous efforts to evaluate unsaturation using MRS have shown limited promise. Here, we have implemented novel localized 2-D MRS techniques to separately assess saturated and unsaturated lipid components and, therefore, to permit an accurate estimate of the degree of unsaturation (36, 37).
Although there is an adequate supply of saturated fatty acids in the diet, synthesis of unsaturated fatty acid is required. One mechanism for this is desaturation of saturated fatty acids, that is, conversion of (-CH2CH2-) groups to (-CH=CH-) by the desaturase enzymes. Desaturation is an oxygen-dependent process that can introduce a double bond at positions nine, six, five, and four from the carboxyl end of a fatty acyl CoA thioester. The corresponding reactions are catalyzed by separate desaturase enzymes, termed the Δ9-, Δ6-, Δ5-, and Δ4-desaturases. The number and position of carbon-carbon double bonds in fatty acids affect their physical and, therefore, their physiological properties. In addition, regulation of cell membrane fluidity through control of the number of double bonds in fatty acids is achieved by desaturases (38).
A number of studies have examined the role of particular desaturase enzymes in diabetes and related disorders (6, 7, 14, 21, 23). The Δ9-desaturase enzyme converts saturated fatty acids to monounsaturated fatty acids. Conversion of linoleic acid to γ-linolenic acid by Δ6-desaturase is the rate-limiting step in the conversion of linoleic acid to arachidonic acid (6, 7). The Δ5-desaturase converts dihomo-γ-linolenic acid to arachidonic acid and eicosatetraenoic acid to eicosapentaenoic acid, which can be converted to decosahexaenoic acid. Thus these enzymes are necessary for the biosynthesis of arachidonic acid, eicosapentaenoic acid, decosahexaenoic acid, and other unsaturated fatty acids (6, 7). Insulin therapy reverses and overcorrects the diminished Δ9- and Δ6-desaturase activities and restores fatty acid composition to normal, except for the decrease in arachidonic acid (14). Thus alterations in fatty acid composition in tissues from diabetic animals reflect the consequences of insulin deficiency. Despite the interest that would accordingly attach to the development of in vivo approaches to monitor desaturase activity, conventional 1-D MRS has not been capable of resolving the olefinic protons within the IMCL and EMCL pools. In contrast, the 2-D MRS technique described here provides a quantitative measure of unsaturation through the cross peaks generated between the olefinic, allylic, and diallylic methylene protons.
In support of our original hypothesis, we found that the degree of unsaturation was significantly higher within IMCL than EMCL in subjects from all three BMI groups, indicating the relative efficiency of double bond creation within the cells present in muscle fibers compared with the lipid pool present outside these cells. In addition, desaturation was decreased in overweight and obese subjects compared with normal-weight subjects, indicating the reduction of desaturase activity in these subjects.
The reduced degree of unsaturation in overweight and obese subjects may result from increased lipid peroxidation (25). Fatty acids are particularly prone to reactive oxygen species (ROS)-induced oxidative damage. A large proportion of the total intracellular ROS burden originates from mitochondria, so that accumulation of fatty acids in close proximity to muscle mitochondria, that is, within IMCL, may increase the likelihood of ROS-induced lipid peroxidation. Although this process would be more likely to occur in individuals with impaired glucose homeostasis, given their proportionally greater IMCL pool, it was not the purpose of this study to correlate MRS measurements with blood and/or other measurements of insulin sensitivity.
Although diffusion of fatty acids into mitochondria is largely regulated by the transporter enzyme carnitine palmitoyltransferase 1, such diffusion can also occur independent of this mechanism (13). High IMCL levels and high degrees of lipid unsaturation in the muscle of insulin-resistant and diabetic individuals may increase lipid diffusion into the mitochondria in type 2 diabetes (31). This is consistent with observations of higher levels of lipid peroxidation in obese and insulin-resistant subjects than in control subjects (27). In addition, lipid peroxides themselves could lead to mitochondrial damage; indeed, the extent of mitochondrial abnormalities is greater in patients with type 2 diabetes than in control subjects (16).
The ability to further define these processes would be significantly advanced by the availability of methods for noninvasive assessment of IMCL and EMCL pools as well as their degrees of unsaturation. Previously, we developed 2-D MRS approaches to this (36, 37). In the present work, we have demonstrated the applicability of the methodology in human studies by performing evaluations in normal-weight, overweight, and obese subjects. As hypothesized, we found greater accumulation of IMCL and EMCL in overweight and obese subjects, as well as a decrease in desaturation with increasing BMI. To the best of our knowledge, this study is the first to demonstrate the differences in the degree of unsaturation within IMCL and EMCL pools of normal-weight, overweight, and obese human subjects.
Perspectives and Significance
The present study and our earlier work demonstrate the ability of noninvasive 2-D MRS to delineate the saturated and unsaturated triglyceride components within IMCL and EMCL in skeletal muscle and the sensitivity of these measurements to physiological status. We found that the level of IMCL and degree of unsaturation are directly related to BMI, consistent with the known dysregulation of lipid metabolism in obesity. Thus we propose that the MRS-determined degree of unsaturation be further evaluated as a potential biomarker for lipid-based metabolic disturbances and as a therapeutic target for assessment of drug-, nutrition-, or exercise-based interventions for dyslipidemia.
This research was supported by a Health Sciences Center Grant from West Virginia University (S. S. Velan) and in part by the Intramural Research Program of the National Institute on Aging (R. G. Spencer) and National Institute of Diabetes and Digestive and Kidney Diseases Research Grant DK-018777 (V. M. Rajendran).
We thank the staff of the Center for Advanced Imaging and Department of Radiology for support of the 3-T MRI/MRS system.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2008 the American Physiological Society