A method is presented for measurement of triglyceride (TG) synthesis that can be applied to slow-turnover lipids. The glycerol moiety of TG is labeled from 2H2O, and mass isotopomer distribution analysis (MIDA) is applied. Mice and rats were given 4-8% 2H2O in drinking water; TG-glycerol was isolated from adipose and liver during ≤12-wk of 2H2O labeling. Mass isotopomer abundances in the glycerol moiety of TG were measured by GC-MS. The combinatorial pattern of isotopomers revealed the number of H atoms in glycerol incorporating label from 2H2O (n) to be 3.8–4.0 of a possible 5 for adipose tissue and 4.6–4.8 for liver TG. Hepatic TG-glycerol in fact reached 97% predicted maximal value of label incorporation (4.4–4.6 × body 2H2O enrichment), indicating near-complete replacement of the liver TG pool. Label incorporation into adipose tissue revealed turnover of mesenteric TG to be faster (k = 0.21 day–1) than other depots (k = 0.04–0.06 day–1) in mice. TG isolated from subcutaneous depots of growing adult rats plateaued at 85–90% of calculated maximal values at 12 wk (k = 0.05 day–1), excluding significant dilution by unlabeled α-glycerol phosphate. Turnover of plasma TG, modeled from 2H incorporation over 60 min, was 0.06 min–1 (half-life 11.5 min). In summary, use of 2H2O labeling with MIDA of TG-glycerol allows measurement of new α-glycerol phosphate-derived TG synthesis and turnover. The hypothesis that mesenteric TG is more lipolytically active than other depots, previously difficult to prove by isotope dilution techniques, was confirmed by this label incorporation approach.
- mass isotopomer distribution analysis
- stable isotopes
the synthesis and breakdown rates of acylglycerides, such as triacylglycerol (TG), play a role in a number of important conditions including obesity, lipodystrophy, and hyperlipidemia as well as processes such as the biogenesis of cellular organelles (3, 26). The turnover of TG, the major component of adipose tissue, is of particular interest in view of the increasing prevalence of obesity (23, 32). A reliable method for measuring all-source synthesis and turnover of acylglycerides in vivo could provide insights into these and other lipid disorders. To date, however, no such method has been available.
One of the methodological barriers to measuring all-source TG turnover has been that the turnover of fatty acid constituents of TG is not synonymous with the turnover of the intact acylglyceride molecule. The fatty acids in TG may derive from de novo lipogenesis (DNL), or nonesterified fatty acids released from adipose or circulating lipoproteins. The latter, in turn, can be assembled from dietary or de novo synthesized fatty acids in liver. Each of these fatty acid sources makes a variable contribution to any particular TG pool, with each exhibiting its own regulation and kinetics. TG may also undergo partial hydrolysis and reesterification (4, 5), further dissociating the turnover of fatty acid from TG kinetics. Labeling of the fatty acid moiety as a measure of intact acylglyceride kinetics thereby introduces complexity and potential error.
The central component of all acylglycerides is the glycerol moiety; therefore, TG-glycerol kinetics in principle best represent the kinetics of the intact TG molecule. Labeling of adipose tissue TG-glycerol has been problematic, however. Cytosolic α-glycerol phosphate is the immediate precursor for TG synthesis. Adipocytes lack significant glycerol kinase activity (2, 21), and as a result labeled free glycerol is not directly incorporated into TG in adipose tissue. [13C]- or [14C]glucose does label adipose acylglycerides (10) but is an inefficient precursor and is expensive, especially for the long-term studies needed to achieve measurable labeling of the very large, slow-turnover, whole body adipose TG pools.
A potential labeling strategy for α-glycerol phosphate-derived acylglyceride synthesis became apparent to us from the observation that [2H5]glycerol is an inefficient labeling precursor compared with [13C]glycerol for TG made in the liver (27, 31), a tissue containing glycerol kinase. The nearly complete loss of 5[2H5]glycerol label to body water before conversion to the glycerol moiety of acylglycerides in liver suggested that the opposite process must also occur, namely, incorporation of 2H2O from body water into C-H bonds of glycerol in acylglycerides.
The basic principle of this method is that acylglycerides synthesized from intracellular α-glycerol phosphate during a period of 2H2O exposure will contain labeled hydrogen atoms incorporated from tissue water. In contrast, acylglycerides that were synthesized from α-glycerol phosphate before labeled water was present will not contain covalent C-H label in their glycerol moiety, thereby allowing the proportion of newly synthesized vs. preexisting acylglyceride molecules to be calculated. The biochemistry of hydrogen atom incorporation from solvent H2O into gluconeogenic and glycolytic intermediates has been characterized in detail (29) and is shown schematically in Fig. 1. Importantly, C-H bonds in metabolic intermediates such as α-glycerol phosphate do not exchange hydrogen with tissue water outside of enzyme-catalyzed metabolic reactions (i.e., they are not labile in solution). The same principle applies for the C-H bonds in the glycerol moiety of acylglycerides. Accordingly, any incorporation of 2H into C-H bonds of glycerol in acylglycerides requires passage of the glycerol moiety through intermediary metabolic reactions (Fig. 1) during the period of 2H2O labeling and thereby implies new assembly of the acylglycerol from α-glycerol phosphate. Quantitatively, the degree of incorporation of 2H into C-H bonds of α-glycerol phosphate is proportional to two factors: the fraction of labeled hydrogen atoms in tissue water (i.e., the enrichment of tissue 2H2O) and the fraction of C-H bonds in glycerol that are incorporated from body water. Thus, if the body water enrichment and the number of C-H bonds in glycerol that derive from body water are known, the incorporation of 2H2O into H atoms in the glycerol moiety of acylglycerides reveals the fraction of newly synthesized acylglyceride molecules present, regardless of the metabolic source of the fatty acid moiety. Because labeled water is well suited for the measurement of slow-turnover molecules, e.g., DNA in long-lived cells (24), this approach is potentially attractive for measurement of α-glycerol phosphate-derived lipogenesis in adipose tissue and in phospholipids of long-lived organelles such as mitochondria. The absence of radioactivity or other toxicities with stable isotope-labeled water (2H2O) makes this technique particularly attractive for human studies.
Our objective here was to test the underlying assumptions of the method and optimize the practical application of this approach. In particular, the relationship between body 2H2O enrichment and the maximum achievable enrichment in TG-glycerol had to be established to use this approach for kinetic calculations. Some H atoms may also derive from dietary glucose and remain unlabeled; the effect of this potentially confounding pathway was evaluated in vivo. We also addressed a long-standing question concerning the turnover of mesenteric fat relative to other adipose depots in rodents. Portions of this work have been published in abstract form (34).
Studies of Adipose Tissue TG Kinetics
Sprague-Dawley rats (200–300 g; Simonsen Labs, Gilroy, CA) and C57Bl/6J mice (16–18 g; Jackson Laboratories, Bar Harbor, ME) were used. The 2H2O-labeling protocol was as described previously (24). Briefly, an initial priming dose of 99.8% 2H2O was given via intraperitoneal injection to achieve ∼2.5% body water enrichment (assuming 60% body wt as water) followed by administration of 4% 2H2O in drinking water for up to 12 wk. Some animals received higher doses of 2H2O (8% drinking water) after priming to 5% body water enrichment. Rodents have previously been maintained on chronic 2H2O intake at up to 30% enrichment in drinking water without effects on growth, food intake, behavior, activity, or fertility (16). Our results are consistent with these reports. Some animals had biopsies (<20 mg of adipose tissue) taken periodically from alternating inguinal fat pads under isoflurane anesthesia. Samples from liver tissue and mesenteric, epididymal, retroperitoneal, and inguinal fat pads were collected postmortem after an overnight fast.
Studies of Plasma TG Kinetics
Seven-week-old, male C57/BL6 mice (20.5 ± 1.0 g; Jackson Laboratories) were used. After a 4-h fast, drinking water was removed, and a single dose of 99.8% 2H2O was given by intraperitoneal injection to achieve ∼2.5% body water enrichment. Groups of five mice were anesthetized with isoflurane and exanguinated at 15-min intervals for 60 min.
Diets were Purina mouse chow. All protocols received approval from the University of California Berkeley Animal Use Committee.
Isolation of TG-Glycerol from Adipose Tissue and Liver
All tissue samples were placed in Kontes dual-glass tissue grinders (Kimble Kontes, Vineland, NJ) with 1 ml of methanol-chloroform (2:1), ground until homogenous, and then centrifuged to remove protein. The solution was extracted with 2 ml each of chloroform and water. The aqueous phase was discarded, and the lipid fraction was transesterified by incubation with 3 N methanolic HCl (Sigma-Aldrich) at 55°C for 60 min. Fatty acid methyl esters were separated from glycerol by the Folch technique, with the modification that pure water rather than 5% NaCl was used for the aqueous phase. The aqueous phase containing glycerol was then lyophilized, and glycerol was converted to glycerol triacetate by incubation with acetic anhydride-pyridine (2:1) as described elsewhere (8). Some samples were extracted, and then TG was separated from other acylglycerides by thin-layer chromatography (TLC), as described elsewhere (13), and then analyzed as we will describe.
Isolation of TG-Glycerol from Plasma
Plasma was isolated from fresh whole blood, and total lipids were extracted by the Folch technique. TG was further isolated by TLC (13). Glycerol isolation and derivitization were then performed as described above.
Measurements of 2H2O Enrichments in Body Water
2H2O enrichments in body water (from plasma or urine) were measured by GC-MS as described previously (24). Briefly, 15–20 μl of plasma were reacted in an evacuated GC vial with calcium carbide to produce acetylene. The acetylene gas was then removed with a syringe and injected into a GC vial containing 10% bromine in carbon tetrachloride and incubated at room temperature for 2 h to produce tetrabromoethane. Excess bromine was neutralized with 25 μl of 10% cyclohexene, and the sample was suspended in ethyl acetate.
Glycerol triacetate was analyzed for isotope enrichment by GC-MS as previously described (8). Mass isotopomer abundances were analyzed by selected ion monitoring: mass-to-charge ratios (m/z) 159–161 (M0-M2) for MIDA calculations; m/z 159–162 (M0-M3) for total 2H label calculations; and m/z 159–160 for fractional synthesis calculations. Body 2H2O enrichments were analyzed as tetrabromoethane by a GC-MS method as recently described (24). An Agilent model 6890 GC with 5973 mass spectrometer (Agilent Technologies, Palo Alto, CA) fitted with a DB-225 fused silica column (J&W, Folsom, CA), and chemical ionization was used for all analyses.
Body 2H2O enrichments. Isotope enrichments of body 2H2O were determined by comparison with standard curves using 2H2O mixed in known proportions with unlabeled water, conversion to acetylene, and GC-MS analysis, as described previously (24).
[2H]glycerol enrichments. Isotope enrichments of [2H]glycerol derived from acylglycerides were calculated by subtraction of mass isotopomer abundances in unlabeled glycerol standards (7) (1) where EMx represents the excess fractional abundance of mass isotopomer x, Ax represents the abundance of the +x mass isotopomer, Ai represents the abundances of all mass isotopomers monitored expressed relative to the parent ion (e.g., A2 for the +2 mass isotopomer, etc.), S represents the labeled sample, and B represents the unlabeled baseline glycerol molecule. Enrichments were determined by adjusting injection volumes to match M0 abundances in samples and unenriched standards. Only GC-MS runs for which the unenriched (baseline) values of M0-M2 mass isotopomers were within 2% of theoretical (expected) abundances were used. EM1 in MIDA calculations was calculated using isotopomers M0-M2. EM1 for fractional synthesis calculations was calculated using only M0 and M1.
Fractional synthesis of TG-glycerol (f). The fraction of TG that is newly synthesized (f) was calculated on the basis of the precursor-product, or rise-to-plateau, approach (2) where EM1 glycerol represents the mass + 1-labeled species of TG-glycerol in excess of natural abundance, and represents the theoretical plateau or asymptotic value for fully labeled TG-glycerol (see below).
The precursor-product or rise-to-plateau approach for measuring fractional synthesis rates (ks) is based on the following model and assumptions, as discussed previously (7). 1) A two-pool model is assumed, characterized by a well-mixed precursor that can be labeled (intracellular α-glycerol phosphate) and a well-mixed metabolic product pool that is derived from the precursor (TG). 2) The product is derived either exclusively from the labeled precursor or at a constant contribution from the labeled precursor during the labeling period, so that label in the product pool approaches an asymptotic or plateau value determined by label content in the precursor pool. 3) In a two-pool model of this type, the rate at which label in the product pool approaches its asymptotic value is determined by the replacement rate, or fractional turnover rate, of the product, which is typically expressed as a rate constant (units of time–1) and which can be calculated by simple exponential formulas (35, 36). 4) The isotopic enrichment of the precursor pool can be maintained at a relatively constant value during the labeling period, so that a single plateau value can be used (i.e., the exponential formula can be used in its integrated form, Eq. 3). 5) Label is incorporated into the product only via biosynthesis (i.e., no exchange of hydrogen atoms into C-H bonds of acylglyceride occurs independently of metabolic conversions leading to free α-glycerol phosphate and subsequent biosynthetic assembly of the acylglyceride).
Application of the precursor-product approach for turnover of acylglycerides by use of 2H2O requires consideration of several possible models by which 2H might enter α-glycerol phosphate and acylglycerides from 2H2O (see Fig. 2 and below). The experimental simplicity and ease of long-term 2H2O intake and the unique ability to maintain extremely stable body water enrichments (25) over long periods of time make 2H2O administration an ideal approach for applying the precursor-product (rise-to-plateau) method.
Calculation of ks. The theoretical plateau or asymptotic value () in TG-glycerol during 2H2O labeling was determined in two ways: by mass isotopomer distribution analysis (MIDA) of the combinatorial labeling pattern in glycerol () (6, 7) and by measurement of plateau enrichments reached in “fully replaced” TG depots (; see below). The standard precursor-product equation was then applied (3) (4) where ks represents the fractional replacement or synthesis rate constant, and t is time of labeling.
Absolute synthesis rate of TG. The absolute synthesis rate of adipose TG was calculated by multiplying the measured ks over the period of labeling times the pool size of TG. For the purpose of this calculation, we assumed TG content to be 10% of body weight in nonobese young rodents. The absolute synthesis rate of adipose tissue TG can be calculated as (5)
Net lipolytic rate. The net lipolytic (TG breakdown) rate in individual fat depots was calculated from the difference between the absolute rate of TG synthesis and the net rate of TG accumulation, where the latter was determined from the change in weight over time in a fat pad or in the whole body (6)
This value of net lipolysis represents a minimum estimate, to the extent that selective TG cycling occurred during the labeling period, i.e., if the labeled TG pool was part of a biochemical subpopulation that was preferentially broken down and resynthesized (see below).
Calculation of plateau enrichment in TG-glycerol. Accurate determination of plateau or maximal enrichment achievable in TG-glycerol () is required for the determination of fractional synthesis (f; see Eq. 2). One method to determine is to perform a full label incorporation study to isotopic plateau from 2H2O in the TG-glycerol pool of interest. Another method is to apply a formula based on the enrichment and number of labeled hydrogen atoms entering newly formed α-glycerol phosphate. By either method, the plateau value will be a function of two parameters: first, the probability that hydrogen atoms entering α-glycerol phosphate were 2H labeled (p); and second, the number (n) of hydrogen atoms in the C-H bonds of α-glycerol phosphate that were incorporated from body water. In the simplest case, p would equal the 2H enrichment of body water, but other possibilities must also be considered. Metabolically distinct pools of hydrogen might be present (e.g., water and NADH, Fig. 1), and p could be lower than the enrichment of body water, to the extent that NADH is not in equilibrium with water (see appendix for detailed discussion). For this reason, p does not technically refer to the enrichment of H atoms in the body water pool but to the enrichment of H atoms that actually entered newly synthesized α-glycerol phosphate and acylglycerides. With regard to n, the maximal theoretical n of exchanging H atoms from H2O into C-H bonds of glycerol is 5 (i.e., all the C-H positions; Fig. 2). The actual value of n (i.e., the number of H atoms that are actually incorporated from body water, representing the number of hydrogen positions under physiological conditions that are capable of being labeled with deuterium) must be established experimentally. The ability to measure easily body 2H2O enrichments potentially simplifies the calculation of n by use of MIDA, as discussed previously by Lee et. al. (19) and by us (7). Starting with the assumption that the isotopic enrichment (p) of hydrogen in each actively incorporated C-H of α-glycerol phosphate is equal to the 2H enrichment in body water, then n can be calculated by use of a straightforward MIDA calculation from the combinatorial labeling pattern in glycerol. This is the case because the combinatorial pattern in any polymer is a function of n and p only (6, 19) (see appendix), so that if either variable is known the other is calculable. The existence of a simple n and p for α-glycerol phosphate may not be metabolically accurate, however. In the case of hydrogen incorporation into glycerol, two distinct labeling scenarios can be conceived. In model 1, all five hydrogen atoms in glycerol are derived from a pool that is only partially equilibrated with body water (i.e., p is lower than body water enrichment); in model 2, fewer than five (e.g., four) hydrogens are derived from a pool that is in equilibrium with body water, and the remaining hydrogen(s) (e.g., one) is from a completely unlabeled source. Some combination of the two models may also exist (model 3). As discussed in detail in appendix, model 3 is a more physiologically reasonable model. In model 3, some of the glycerol in newly synthesized TG may arise from glycolysis of unlabeled glucose, which thereby incorporates only three labeled hydrogens, and the remaining may arise from glyceroneogenesis, which incorporates label in all five C-H bonds. Accordingly, we performed calculations using a model where p (the true 2H enrichment of hydrogen atoms entering the newly synthesized glycerol) is not identical to body water but is unknown, and n is assumed to be 5. This involves a standard MIDA calculation (7) based on the labeling pattern in glycerol (see Results). Mixtures of the two models were also simulated to represent intermediate scenarios (wherein n is less than 5 and p is less than body 2H2O enrichment). The calculations revealed that the models give virtually identical results for ks values of acylglycerides, for p = 0.02–0.04, and for n = 3–5 (see appendix). Summation of total 2H label in all the C-H positions of glycerol combined [i.e., equivalent to a radioactive specific activity (19)] was also performed. This provides an external test of the MIDA-calculated value of n in tissues where TG is nearly completely turned over, because if the acylglyceride is 100% newly synthesized, then the total 2H label present in the molecule will be equal to, and thereby reveal, n × p (19). Total 2H label was calculated as described by Lee et al. (19) (7) (8) where mi represents the true fraction of single-, double-, or triple-labeled glycerol molecules present, and ni represents the number of labeled hydrogens on each of the M1-M3 molecular species (i.e., 1, 2, and 3, respectively). M1-M3 captures 99.9% of the 2H isotope in glycerol at the enrichments of p present here. The true fractions of single-, double-, or triple-labeled molecules were calculated from the experimentally measured enrichment (EMx), corrected for natural 13C abundance skew by a modification of the equations of Lee et al. (19).
It should be noted that the peracetylated derivative of glycerol that we analyze removes hydrogens from all -OH positions. Only the C-H enrichments in glycerol are measured by the GC-MS analyses, thereby allowing experimental measurements that establish n and p of C-H bonds only.
ANOVA was used to compare groups with P < 0.05 as the criterion for significance. Curve fitting of label incorporation data was performed using Delta Graph (Delta Point).
Establishing the Number of Exchanging Hydrogens and Maximal Enrichments of C-H Bonds in TG-Glycerol
MIDA calculations (combinatorial analyses). An estimate of n (the number of hydrogens in C-H bonds of α-glycerol phosphate that are incorporated from body H2O) was made by combinatorial analyses of the labeling pattern in TG-glycerol by use of the MIDA approach (6).
The calculated values for n from adipose TG ranged between 3.9 and 4.1 (Table 1), whereas values from 10-wk-labeled hepatic TG ranged from 4.6 to 4.8 (Table 2). Consistent values were observed among different adipose depots and over sequential time points (Table 1).
To validate the determination of n from combinatorial analysis of acylglycerides, we compared the value from MIDA to that from summation of all labeled species (single- to triple-labeled) in a TG pool that is expected to be completely or nearly completely turned over (i.e., hepatic TG from mice given 2H2O for 70–80 days and fasted overnight). The number of exchanging hydrogens was similar (Table 2) whether calculated from the plateau value reached for total 2H labeling (reflecting n × p) or from the internal ratios (combinatorial analysis) of mass isotopomers (n = 4.4–4.5 vs. n = 4.6–4.8, respectively). Total 2H labeling confirmed that fractional replacement (f) of hepatic TG was ∼95% (Table 2).
If a value of 4 is used for n in adipose TG (Table 1) and the measured 2H2O enrichments are used to represent p of hydrogen atoms entering α-glycerol phosphate, maximal EM1 () can be calculated using standard MIDA tables (7). The best-fit equation was (9) where x equals measured body water enrichment, and only ions M0 and M1 of glycerol are included in the calculation of . The fraction of adipose TG that is newly synthesized (f) is then calculated as described above, by using A1 as the denominator of Eq. 2 (7).
Long-term labeling studies of adipose TG pools. Adult rodents were exposed to 2H2O for prolonged periods to test the maximal isotope enrichment in various storage TG pools. Body water enrichments reached a steady-state value within 2–3 days and remained stable for the subsequent weeks (Fig. 3). Body water enrichments were not significantly different between animals within any group. Long-term 2H2O intake had no effects on food intake, weight gain, or behavior and did not alter fertility or full-term pregnancy, consistent with previous reports (16). In serial-biopsied mice (given 8% 2H2O in drinking water), 2H2O enrichment in plasma was 5.13 ± 0.06% (n = 5, mean ± SE); in rats that were biopsied or sequentially killed (given 4% 2H2O), plasma 2H2O enrichments were 3.0 ± 0.09%, and 2.9 ± 0.03%, respectively. EM1 isotope enrichments in glycerol from epididymal adipose TG reached 84 ± 3 and 91 ± 4% of calculated after 8–10 wk of 2H2O labeling in growing rats and mice, respectively. Inguinal (85 ± 4%, 93 ± 4%; see Fig. 5) and mesenteric (85 ± 3%, 88 ± 5%, Fig. 4) adipose TG reached similar values. Hepatic TG values reached 97 ± 1% of the calculated . Total 2H label summed in all mass isotopomers (∑mini, reflecting n × p) plateaued at ∼3.3–3.4 (adipose TG) and 4.4–4.6 (hepatic TG) times body water enrichment.
Time Course of Adipose TG-Glycerol Enrichments During 2H2O Administration
The kinetics of adipose TG replacement were determined during 2H2O administration in rats and mice. Ad libitum chow-fed rats were killed sequentially (n = 4/time point) after 0–12 wk of 2H2O administration in drinking water. Sequential biopsies of inguinal fat in mice and rats were also performed during 2H2O administration.
The time course of [2H]glycerol labeling in adipose TG from various fat depots in rats is shown (Fig. 4). The replacement rate constants (ks) in adult rats were 0.04–0.06 day–1 for nonmesenteric fat depots and 0.21 day–1 for mesenteric fat, representing half-lives of 12–15 and 3.3 days, respectively. Remarkably, 60% of mesenteric TG was newly synthesized by day 7 (Fig. 4), reflecting this very high replacement rate.
The rate constant for turnover of inguinal adipose TG from sequential biopsies was 0.046 day–1 for mice and 0.064 day–1 in rats (Fig. 5).
Changes in body fat pool size over the labeling period must be considered when adipose TG isotope incorporation is interpreted. If there is an increase in the pool size of body TG, then all the label incorporation does not represent turnover (i.e., replacement), as would be the case if the pool size were constant. The average body weight increase over the 80-day labeling period was 21% for mice (18.8 ± 0.7 to 22.7 ± 0.5 g). Assuming that percent body fat remained relatively constant during these studies, total body fat accrual can be estimated to be 10% of total weight gained (or 5 mg fat/day). Estimated net lipolysis rate, calculated from the total TG synthesis after correction for TG accumulation, was 95 mg · mouse–1 · day–1 based on a ks of ∼0.05/day, a 20-g mouse, and 5 mg/day fat accrual (5%/day × 2 g body fat ≈ 100 mg made –5 mg net retained).
Interestingly, for both rats and mice, the plateau values reached for adipose TG-glycerol labeling reflected 85–90% replacement. This finding is consistent with a small, non-turning-over component of adipose TG stores that persists in growing rodents.
Comparison of Enrichments in Total Acylglycerides to TG
Fractional synthesis rates were calculated for total adipose acylglycerides and for TG isolated by TLC (Fig. 6). Incorporation rates and plateau values were nearly identical by the two methods.
Plasma TG-Glycerol Kinetics after 2H2O Administration
Fractional replacement of plasma TG in normal mice after a single intraperitoneal injection of 2H2O is shown (Fig. 7A). The calculated rate constant based on curve fitting of the rise to plateau was 0.06 min–1, corresponding to a half-life of 11.5 min. By use of a plasma TG concentration of 70–90 mg/dl, plasma volume of 0.6 ml, and body weight of 20 g, TG production equals 75–96 mg · kg–1 · h–1. The stability of body water enrichment over the 60-min labeling period is illustrated (Fig. 7B). Because the hepatic TG values for n observed were 4.7, this value of n was used to calculate , as described above. The resulting equation for maximal EM1 () is calculated from body water (x) as (10)
The availability of a method for measuring total acylglyceride synthesis from α-glycerol phosphate in adipose tissue and liver opens many possibilities for experimental studies. The approach described here is based on 2H2O labeling of the glycerol moiety of acylglycerides and overcomes both general and specific problems that had constrained previous methods for measuring lipid kinetics.
The 2H2O labeling approach overcomes the general difficulty of measuring the turnover of any extremely large, slow-turnover biochemical pool in the body, such as adipose tissue TG. Adipose TG represents the largest pool of any molecule in most adult mammals, even in nonobese animals (i.e., 10% body wt in normal rodents, compared with <4% for muscle myosin and <2% for bone collagen). Although the adipose TG turnover rate had not previously been measured accurately, indirect methods indicated a relatively slow fractional replacement rate. Studies of the effect of altering dietary fatty acid composition on fatty acids in adipose TG (9), for example, suggested that 6–12 mo were required to replace adipose TG stores in humans. Comparison of measured lipolytic rates, from the dilution of plasma glycerol or fatty acids to body fat stores in humans (∼12 μmol · kg body fat–1 · min–1, or 10 mg TG · kg body fat–1 · min–1) (30) also indirectly suggests a relatively low fractional replacement rate for adipose TG (∼1.5% TG turnover/day). Accurate measurement of adipose TG turnover with isotopes must therefore overcome the intrinsic challenge of introducing a sufficient amount of label for measurement into a slow-turnover pool.
Use of labeling approaches to establish total TG synthesis from α-glycerol phosphate in adipose tissue also faces some specific complexities. The relative lack of glycerol kinase activity in fat cells (2, 21) precludes the use of glycerol as a direct precursor to label adipose TG. Carbon-labeled glucose can be used but is difficult to administer in a constant fashion over the circadian cycle and is likely to undergo variable dilution in both the blood and glycolytic pathways in fat cells (the latter pathway may account for our observation that the observed n was not equal to 5; see Fig. 1). Nor is administration of glycerol-labeled TG useful, because circulating TG are hydrolyzed to free fatty acids and free glycerol before uptake by adipocytes, and the free glycerol is not used for TG resynthesis in the adipocyte (2, 21).
Accordingly, most methods for measuring adipose lipid synthesis have labeled the fatty acid moiety through the various metabolic routes (e.g., labeled dietary fatty acids, DNL with [13C]acetate, or labeled water). These approaches cannot reveal all-source TG synthesis in the tissue, however, and are complicated by issues such as partial lipolysis-reesterification cycles and potential selectivity in the metabolism of different fatty acids.
The technique described here is based on incorporation of hydrogen atoms from 2H2O into covalent C-H bonds in the glycerol moiety of acylglycerides (Fig. 1). Because formation of C-H bonds in glycerol occurs only during reactions of intermediary metabolism that lead up to free α-glycerol phosphate, but not in glycerol that is already bound to acyl groups (Fig. 1), any label incorporated into acylglycerides reflects molecules that were newly assembled from α-glycerol phosphate during the period of label exposure (i.e., the newly synthesized fraction). We should emphasize that we measure only α-glycerol phosphate-derived TG in adipose tissue by this technique. If there are unlabeled glycerol sources that contribute to adipose TG deposition (e.g., from mono- or diacylglycerides that are recycled to triglycerides), however, our results indicate that this can be, at best, a small fraction (see Fig. 4).
The potential for a useful 2H2O-labeling method for TG-glycerol was suggested to us by the observation that [2H5]glycerol is a very poor label for hepatic VLDL-TG (27), whereas [13C]glycerol is an efficient label (31). Replacement of [2H5]glycerol hydrogens in the liver by hydrogens from body water must therefore be extensive, before entry into hepatic TG. Near-complete loss of hydrogen from C-H bonds into body water implies near-complete replacement of hydrogen into these C-H bonds from body water. Kalhan et al. (14) and Jensen et al. (11) have recently published work describing the use of labeled water to measure glyceroneogenesis, based on similar principles, and Previs et al. (28) recently described in abstract form a similar technique for measuring TG synthesis.
It was necessary first to establish the degree to which C-H positions in α-glycerol phosphate pools used for TG synthesis are in fact derived from hydrogens that ultimately came from cellular water. This value (n) was determined by two independent and complementary approaches: labeling to plateau (i.e., allowing the TG pool to be fully replaced) and combinatorial analysis of the 2H-labeling patterns (MIDA of TG-glycerol). The 2H-labeling combinatorial pattern in a biosynthetic polymer reflects the true precursor isotope enrichment (p) and the number of repeating monomeric subunits (n). Accordingly, the value of n that is calculated by MIDA from partially turned-over molecules in principle should predict the measured value of n that will be reached at plateau in fully replaced TG pools (7).
Our results were consistent with this prediction. The EM2/EM1 ratio in TG-glycerol from mouse livers after 10 wk of 2H2O labeling gave values of n = 4.6–4.8 (Table 2) by using body water enrichments to represent p (models 1 or 3). The sum of labeled H atoms present in all mass isotopomers (reflecting n × p) of fully replaced hepatic TG was similar (4.4–4.6). A slight underestimation using total labeled H atoms is not surprising, to the extent that very low enrichments of high-mass isotopomers species (e.g., EM4) are undetected or the TG pool is less than 100% replaced. Importantly the MIDA-calculated value was relatively stable among depots, animals, and conditions (Table 1). We anticipated that a high proportion, but not all, of the C-H bonds in α-glycerol phosphate would incorporate labeled 2H from body water, as was in fact observed, on the basis of previous observations that there is some labeling of VLDL-TG-glycerol from [2H5]glycerol (27). These previous results, showing retention of partially dedeuterated species in VLDL-TG-glycerol after [2H5]glycerol administration, indicated ∼20% efficiency of [2H5]glycerol labeling in liver compared with [2-13C1]glycerol labeling, suggesting that some C-H positions in hepatic α-glycerol phosphate need not ultimately come from water. Our data are consistent with 80% of the C-H positions in adipose tissue lipogenic α-glycerol phosphate being incorporated from body water. Whether the form of the 80% incorporation is four labeled and one nonlabeled C-H positions (Fig. 2, model 2), 80% incorporation in each C-H position (Fig. 2, model 1), or various combinations (Fig. 2, model 3) turns out to have essentially no effect on calculated TG synthesis parameters (see appendix and Fig. 7). The value of n in liver may be somewhat higher (4.6–4.8), consistent with more active glyceroneogenesis and replacement of C-H bonds (Fig. 1) in α-glycerol phosphate in liver (11, 14). In practice, the actual value in any tissue or physiological state can be confirmed by MIDA of the labeling pattern in the TG-glycerol isolated.
An important practical conclusion from these results is that future measurements of adipose TG turnover from 2H2O can probably use a standard value for n (4.0) and use the measured body 2H2O enrichments for p. Validating these assumptions can always be tested by performing MIDA on the labeling pattern in the glycerol moiety and comparing with the calculated pattern expected (e.g., the EM2/EM1 ratio; Fig. 9), using n = 4 and measured body 2H2O enrichments (see Table 3). If the experimental ratios in a particular setting do not reflect these values, n can be calculated from the data (7, 19).
The 2H2O method has a number of advantageous features for measuring acylglyceride synthesis. The ease of 2H2O administration allows the long-term labeling studies required for measuring adipose TG kinetics. The stability of isotope enrichment in the precursor pool (Fig. 3) is unique for labeled water compared with essentially all labeled organic molecules, due to the slow turnover of body water and its rapid equilibration among almost all body compartments. 2H2O is inexpensive and can be administered to humans or experimental animals for weeks or months at a time. DNL (12) and adipocyte proliferation (1, 24, 34) can also be measured concurrently during 2H2O administration. It should be noted that extensive human studies have been carried out using long-term 2H2O intake to measure adipose TG turnover and DNL (1) without adverse effects. TG synthesis is readily measurable in adipose tissue if body 2H2O enrichments are in the 1% range (1).
In addition to measurement of absolute TG synthesis rates, the 2H2O incorporation technique described here allows some interesting related measurements. The fractional breakdown rate of many other biomolecules has been measured from label incorporation curves, based on the steady-state assumption (i.e., that synthesis is balanced by breakdown), but this approach has not previously been applied to adipose TG because of its very slow turnover rate. In nongrowing or slowly growing animals, fractional breakdown rate can be calculated directly from the ks (35).
Use of this approach for adipose TG pools clearly demonstrated that mesenteric TG has a much higher fractional synthesis rate, and thus fractional breakdown rate, than other fat depots in rats (Fig. 4). Because of the difficulty of sampling portal blood to assess the venous drainage of mesenteric fat, particularly in small animals, the long-standing hypothesis that mesenteric fat has a higher lipolytic rate than other depots (15) remains untested by tracer dilution techniques. The tracer incorporation technique described here provides an alternative experimental strategy that may be applicable in humans as well (e.g., by sampling mesenteric fat at surgery after 2H2O administration).
Other quantitative findings were also of interest. Our observation that plasma TG production rates were 75–96 mg · kg–1 · h–1 are similar to results reported using tritium and radiolabeling methods, e.g., 96 (17), 113 (22), and 98 (33) mg · kg–1 · h–1. In contrast, our calculated net lipolysis rates in adipose tissue (∼95 mg · mouse–1 · day–1) were lower than rates measured by dilution of [13C]palmitate in 3-h-fasted mice (∼250 mg · mouse–1 · day–1) (13). This difference suggests either that “net lipolysis” represents less than absolute adipose lipolysis due to selective recycling of a rapid-turnover TG subpopulation back into adipose stores, that fasting lipolytic rates do not reflect integrated 24-h values over the long term in ad libitum-fed mice, or that a significant proportion of plasma free fatty acid flux is derived from sources other than adipose TG (e.g., plasma lipoproteins or TG stores in tissues such as liver). It will be of interest to investigate these possibilities in future studies using the 2H2O labeling technique.
Finally, the 2H2O-TG-glycerol-labeling approach also proved to be applicable for short-term studies of TG kinetics. Exchange of 2H2O into body water pools is extremely rapid, with plateau enrichments reached in blood in less than 10 min, and the resulting 2H2O enrichment is very stable over subsequent hours (Fig. 7B). 2H2O administration allows the equivalent of an intravenous continuous-label administration protocol to be carried out (Fig. 7A) without the need for intravenous lines or repeated isotope administration. Precursor-product-labeling kinetics can thereby be measured.
In summary, we describe here a method for measuring synthesis and turnover rates of acylglycerides that can be applied to slow-turnover as well as fast-turnover lipid pools. The method takes advantage of the convenience, stability, and diffusability of H2O and the power of combinatorial analysis. Use of this 2H2Olabeling-MIDA approach on TG-glycerol allowed direct testing of a hypothesis that had been difficult to test by isotope dilution techniques, namely that mesenteric TG is more lipolytically active than other fat depots. Many other applications can be envisioned with this approach.
MIDA Calculations for 2H Incorporation into Glycerol of Acylglycerides
Background. To measure the synthesis of new acylglycerides by labeling the glycerol moiety, an administered label must enter the precursor pool (α-glycerol phosphate, α-GP) but not exchange with the formed product (acylglyceride). Hydrogen atoms in C-H bonds of acylglycerides are not labile in physiological solution (27), but α-GP is in communication metabolically with enzymatic reactions that incorporate H atoms into specific C-H positions from cellular H2O (Fig. 1). Hydrogen-labeled water will therefore appear in α-GP to some extent, up to a maximum of all five C-H bonds in glycerol (Fig. 2).
The actual extent of incorporation from 2H2O into α-GP is not known a priori, but is a physiological variable that must be established experimentally. Previous studies have shown extensive but incomplete replacement of C-H labels in hepatic TG-glycerol from plasma glycerol (27).
Different models of hydrogen incorporation into α-GP. Several different models can be envisioned for labeled and unlabeled C-H positions in α-GP (Fig. 2). We will first discuss two mathematical models representing extreme cases, which differ from the simplest case where all C-H bonds can be labeled (i.e., n = 5) and the enrichment of hydrogens entering α-GP is equal to that from body water (i.e., p = 2H20 enrichment). Both alternative models give rise to EM2/EM1 ratios that are lower than what would be predicted if n were 5 and p were equal to body water.
Model 1 (Fig. 2) represents a situation wherein all five hydrogens in α-GP are exposed to a pool of hydrogen that is only partially equilibrated with body water. In this model, n = 5 but p < body 2H2O enrichment. The second model (model 2; Fig. 2) represents a situation wherein a particular C-H bond (such as the α-C-H position) of α-GP might not be exposed to incorporation from 2H2O, because metabolic pathways traversed en route to α-GP do not include any steps that could incorporate hydrogen at that position (Fig. 1). A scenario of this type would reduce the number of labeled C-H positions (n) from 5 to 4, for example. The resultant EM2/EM1 ratio would reflect an n of 4 with a p equal to body water (Table 1). The effect of changing n from 5 to 4 on EM2/EM1 ratios can be calculated as illustrated in Fig. A1. Interestingly, calculations show that the isotopomer patterns resulting from these two models (p = 80% of 2H2O vs. n = 4of5)are almost indistinguishable; these two models result in only a small difference in the predicted within the range of 2H2O enrichments present in these studies (Fig. A1). Fractional synthesis values (f; Eq. 2) are therefore not noticeably affected by selection of either model.
The two models above are mathematical extremes, which could give rise to the results reported here. Intermediate models and their physiological bases can also be considered. One alternative model (not shown) can be envisioned wherein all the C-H positions of α-GP are exposed equally to incorporation from H2O but where a portion of the glycerol molecules contributing to α-GP had not been exposed at any position to incorporation from H2O. This condition might apply if a pool of free glycerol (e.g., from plasma or locally hydrolyzed TG) were channeled directly into α-GP without exposure to intermediary metabolic reactions (Fig. 1), as may be the case in the liver (27). This scenario would not be discernible as an altered labeling pattern (i.e., n or p) on the basis of combinational (MIDA) calculations (6), however. Instead of a lower value for n, a lower percent incorporation from 2H2O would be present in all five C-H bonds of α-GP. The result would be that neither the EM2/EMI ratio nor the calculated p of C-H bonds would be affected, but the plateau value () would be lower than if complete replacement of the TG pool were achieved. Experimentally, the result would therefore be a calculated value of n = 5 but an underestimation of f.
This scenario may contribute to our observation in adipose TG of f < 100% (Fig. 4). The absence of glycerol kinase in adipose tissue makes this less likely to be the explanation, however. It should be noted that, if some of the adipose TG were synthesized in the liver and transported to adipose tissue, a portion of plasma glycerol might be converted to hepatic TG-glycerol without exchanging with body water (27). Our observation that f was close to 100% in hepatic TG makes this unlikely, however.
A final model to consider here (model 3; Fig. 2) may most closely reflect the true physiology. In this model, adipose TG-glycerol is derived from two discrete sources: glycolysis and glyceroneogenesis. In α-GP generated via glycolysis, only the three hydrogens on carbon-1 and carbon-2 have hydrogens incorporated from body water. Conversely, α-GP generated via glyceroneogenesis will have all five hydrogens incorporated from body water. This results in a mixture of two different values of n. These mixtures can be modeled (Tables 3 and 4). The resultant effects on are small for model 3 and lie close to model 2 (see Table 4, comparison of calculations based on a 3 + 5 mixture to direct n = 4 for values). MIDA calculations. 2H incorporation into α-GP alters two aspects of isotopic labeling: the total amount of isotopic label in the α-GP pool and the mass isotopomer distribution present (i.e., the relative proportions of M0, M1, M2, etc., molecules). Indeed, combinatorial analysis, or MIDA, can be used to infer n on the basis of characteristic changes in the mass isotopomer pattern if p is known (7, 19). This approach has been used for estimating the n of H atoms from 2H2O in newly synthesized cholesterol and fatty acids, for example (19).
The same calculation algorithms can be used to establish the relationship between mass isotopomer abundances and n, as we have previously described in detail for calculating relationships between mass isotopomer abundances and p (6). A table of this type (Table 3) is shown for glycerol, calculated for models 1 and 3 (see above). The peracetylated derivative of glycerol was used for calculations. In practice, these tables can be used to calculate n (or p) from measured mass isotopomer ratios [e.g., excess M2/excess M1 (EM2/EM1)]. The resulting value of n (or p) is then used to calculate (the asymptotic label achievable), so that f can be calculated from the precursor-product formula (A1)
Two key points are apparent from Tables 3 and 4 and Figs. A1 and A2. First, the different models of C-H labeling in α-GP have minimal impact on calculated biosynthetic parameters. Second, if the assumed value of n is incorrect in an experimental setting, this will be readily apparent by comparing observed to expected EM2/EM1 ratios, using body 2H2O enrichments to represent p (Fig. A2). The correct value of n can then be calculated directly from the experimental data. Finally, if a reproducible value of n can be established under a variety of experimental conditions, it would allow the M1 mass isotopomer to be measured alone, with measurement of 2H2O enrichment when TG synthesis is measured (i.e., is calculable from body 2H2O enrichment if n is known in advance). Alternatively, both M1 and M2 mass isotopomers in glycerol can be monitored, for application of MIDA to calculate either n or p directly, and from there to calculate and TG synthesis.
Examples of experimental data. Sample experimental data are shown for 2H2O incorporation into adipose tissue TG after 5 wk of 8% 2H2O administration (Table A1). In animal 1, for example, body 2H2O enrichment was stable at 5.2%. EM1 = 0.0495 and EM2 = 0.0082 in TG-glycerol, so that the EM2/EM1 ratio equals 0.1657. By reference to Table 3, this ratio indicates that n equals 4 positions (at p = 5.2%; model 1) or, alternatively, p equals 4.0% (77% of body 2H2O; model 2). By either model, calculated is similar (0.149 and 0.143, respectively, for models 1 and 2; Table 4) so that f is calculated to be 33–35% () for adipose TG after 5 wk of labeling.
These studies were supported in part by National Heart, Lung, and Blood Institute Grant HL-65919, a University of California at Berkeley College of Natural Resources Agricultural Experiment Station grant, and an unrestricted gift from KineMed, Inc. (Emeryville, CA). E. J. Murphy received support from a Pfizer postdoctoral fellowship.
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.
↵* S. M. Turner and M. K. Hellerstein cowrote the appendix.
- Copyright © 2003 by American Physiological Society