Contribution of Fatty Acids Released From Lipolysis of Plasma Triglycerides to Total Plasma Fatty Acid Flux and Tissue-Specific Fatty Acid Uptake

C57BL/6 mice, aged 3–6 months, were used for all experiments. All mice were housed under standard conditions (lights on 0700–1900 h) with free access to water and regular lab diet. For the isolation of lymph, a Wistar rat was obtained from Charles Rivers. All experiments were approved by the institutional animal care committee.

A male Wistar rat was orally given a mixture of 0.5 ml olive oil with 1 mCi of [9,10(n)-³H]oleic acid, after which the rat was anesthetized with 50 mg/kg nembutal i.p. A canula was placed in the thoracic lymph duct, and lymph was collected for the following 4 h. Lymph was mixed with a protease inhibitor cocktail (1 mmol/l benzamidine, 2 mg/l antipain, and 2 mg/l leupeptine; Sigma) and stored until the next day.

Lipid emulsions were prepared as described by Rensen et al. (16) and separated into three different fractions (average size 45, 80, and 150 nm) by ultracentrifugation. After preparation of the particles, the small (45-nm) particles were enriched with 1 mg purified recombinant apoE (17) per 50 mg TG by incubation at 37°C for 30 min.

[¹⁴C]oleic acid was complexed to BSA as previously described (18). [¹⁴C]oleic acid was mixed with either lymph or lipid emulsions. Citrate was added to a final concentration of 3 μg/ml.

Mice were fasted overnight; all experiments were performed in anesthetized mice (0.5 ml/kg Hypnorm; Janssen Pharmaceutica, Beerse, Belgium and 12.5 mg/kg midazolam; Roche, Mijdrecht, the Netherlands) between 0800 and 1100 h. An infusion needle was placed in the tail vein, attached to a minipump (Harvard microdialysis low flow 11; Harvard Apparatus, Holliston, MA), and chylomicrons or lipid emulsions were infused at a rate of 0.2 ml/h for 1.5–2 h, after which steady state was achieved. A 150-μl blood sample was taken by tail bleeding into paraoxon-coated capillaries to prevent ex vivo lipolysis (19).

For the bolus injection protocol, a single 0.2-ml injection into the tail vein was applied and small 30-μl blood samples were taken at regular time points after injection of the labeled mixture. At the last time point, a larger sample was taken for determination of TG and FA levels.

TG, FA, and glucose were determined in plasma as previously described (20). Lipids were extracted from plasma according to Bligh and Dyer (21). The lipid fraction was dried under nitrogen, dissolved into chloroform/methanol (5:1 [vol/vol]) and subjected to TLC (LK5D gel 150; Whatman) using hexane:diethylether:acetic acid (83:16:1 [vol/vol/vol]) as mobile phase. Standards for FA, TG, and cholesteryl oleate were included during the TLC procedure to locate the spots of these lipids. Spots were scraped, lipids dissolved in hexane, and radioactivity measured.

One group of mice was fasted overnight and another was allowed to eat ad libitum. A mixture of 45-nm particles labeled with [³H]triolein and [¹⁴C]-labeled oleate was used as infusate. After 2 h of infusion, plasma was taken for specific activity measurements, mice were killed, and organs were quickly removed and frozen in liquid nitrogen. Tissues were dissolved in 5 mol/l potassium hydroxide in 50% (vol/vol) ethanol at 70°C for 1 h. After overnight saponification, an aliquot of the solution was neutralized with equimolar HCl, acidified with 1% trichloroacetic acid, and the FA extracted into methanol/chloroform according to Blight and Dyer. Radioactivity in this fraction was measured and corrected for the corresponding plasma specific activities of [³H]FA and [¹⁴C]FA.

In steady state, the rate of disappearance (R_d) (μmol/min) can be calculated from the steady-state specific activity of the infused tracer (22):

where V_in is the rate of label infusion (in Bq/min), q is the steady-state radioactivity of label in the total plasma pool (in Bq), and M is the poolsize (in μmol), i.e., q/M represent the specific activity of the compound being traced. The fractional catabolic rate (FCR) is defined as the R_d divided by the pool size:

Substituting R_d using equation 1 gives:

The extent to which [³H]FA from lipolysis of [³H]TG escapes immediate uptake was assessed by the assumption that once [³H]FA and [¹⁴C]FA have mixed, the rate of irreversible loss of label is equal for both FAs. The underlying assumption is that equilibration with albumin is much faster than the rate of FA release from LPL. Using the specific activity of [³H]FA and the FCR of [¹⁴C]FA, we can estimate the rate at which [³H]FA entered plasma (the “infusion rate of [³H]FA,” V_in,[3H]FA):

Dividing this rate with the infusion rate of [³H]TG gives the extent to which label that was infused into the circulation as TG ends up in the circulation as FA. This ratio is defined as plasma TGFA recovery:

The plasma TGFA recovery, as calculated in this manner, does not discriminate between FAs being hydrolyzed without mixing with albumin-bound FA and FAs that are taken up as TG, i.e., by endocytosis of the whole particle. Therefore, we included labeled CE as a marker for whole-particle uptake when we used the lipid emulsion and followed the bolus experimental protocol (see below).

The changes in time of TG- and FA-associated labels were fitted with a compartmental model (Fig. 1), essentially as developed by Hultin et al. (13). Where Hultin used two compartments for the FA label and a variable fraction in blood (or volume of distribution), we found better fits for both [³H]FA and [¹⁴C]FA when we used three compartments for FA. Thus, we explicitly modeled the rapid loss of label by redistribution and set the fraction in blood to 1. A single compartment with a variable volume of distribution sufficed to describe the TG kinetics accurately. The metabolic simulation software package Gepasi (23) was employed to estimate initial values for the kinetic parameters of the model. Gepasi has powerful algorithms to obtain absolute regression minima without the need for good estimates of the initial parameters (24). WINSAAM (25) was subsequently used for fine-tuning by specifying errors in each time point.

The plasma TGFA recovery, as defined in equation 5, can be calculated from the compartmental modeling parameters as:

The contribution of TG-derived FA to the total FA flux was estimated as:

This equation assumes that 1 mol of TG releases 3 mol of FA. Theoretically, this should be considered as the upper boundary because formation of di- and monoacylglycerol are ignored. We cannot discriminate between label being taken up as acylglycerols or FA, but since the plasma TGFA recovery was high (see Table 1), assuming complete lipolysis of TGFA (with release into the plasma according to the plasma TGFA recovery) should quantitatively give a reasonable estimate of the contribution of TGFA lipolysis to the FA flux.

All data are presented as means ± SD of parameters obtained in individual mice. Nonparametric Mann-Whitney U tests were used for all statistical analyses. A P value of 0.05 was taken as boundary of significance.

[³H]TG-labeled rat chylomicrons were mixed with [¹⁴C]-labeled oleic acid, injected intravenously into anesthetized mice, and the behavior of the tracers were followed in plasma in time. [³H]TG activity decayed monoexponentially (Fig. 2). [¹⁴C]FA activity decreased very rapidly. [³H]FA activity initially increased in plasma and then started to decrease slightly. All data were fitted to a four-compartmental model (Fig. 1). Kinetic constants of [³H]FA and [¹⁴C]FA were forced to be identical. The solid lines in Fig. 2 show the fit of the model. The model (and all other models that we have tried) required (almost) all [³H]TG to be transferred to the FA compartments to get an optimal fit (Table 1). In three of five fits, k_0,1 of the model even fell to zero during the fitting procedure (i.e., complete recovery of label in plasma was required). On average, the plasma TGFA recovery was 88%. From the model, FCR and R_d were computed for [¹⁴C]FA and [³H]TG (Table 1). From the R_d and plasma TGFA recovery, an estimate can be made to what extent TG-derived FAs contribute to the total FA flux (see equation 7). For the chylomicrons, it was estimated that 19% of the total FA flux was derived from lipolysis of TG.

Table 1 also shows the results of the steady-state infusion experiment. Steady state was reached within 1.5 h. In line with the result from the compartmental modeling, 81% of the infused [³H]TG activity was recovered as [³H]FA. The FCRs for TG and FA, calculated by the steady-state approach, also agreed well with those found with the compartmental modeling (Table 1).

After establishing that both tracer methods yielded the same results, we tested whether these results were specific for chylomicrons or whether they also apply to smaller TG-rich particles. Therefore, we used lipid emulsions and enriched these particles (average particle size 45 nm [16]) with apoE, which is a good ligand to the VLDL receptor (26). Because of the small size and presence of apoE, we expected that whole-particle uptake could lead to a quantitatively significant underestimation of the plasma recovery of TGFA (because TG-label not being lipolyzed but taken up as whole particle could not be distinguished from TG label being lipolyzed into FA and then being taken up without prior mixing with the plasma FA pool). Therefore, we included [³H]-labeled CE as a marker for particle uptake and used the bolus injection protocol to follow the CE kinetics.

Again, TG-associated label was lost from plasma in a monoexponential manner, whereas FA kinetics needed at least two compartments (Fig. 3). CE label decayed much slower than the TG label, indicating that in the first 15 min, lipolysis of TG dominated over whole-particle uptake. The initial volume of distribution was equal for both labels, which should be the case because they are part of the same particle. [³H]FA label rapidly appeared in plasma and subsequently disappeared. In accordance with the chylomicron data, 89% of the label had to enter the plasma FA compartment to obtain a good fit. It was estimated that 11% of the total FA flux was originating from TG-derived FA (Table 1).

To estimate the relative contribution of TG-lipolysis and plasma FFA to the total amount of FA taken up by tissues, we measured label retention after a 2-h infusion of [³H]TG-labeled lipid emulsion and [¹⁴C]FA in the liver, adipose tissue, and skeletal and heart muscle. [³H] counts originating from TG and [¹⁴C] counts from FFA were measured in tissue extracts and taken as representative of the uptake of the respective substrates. This experiment was carried out in fed animals and in mice that were fasted overnight. Large differences in retention of label existed between tissues, and significant differences between the fasted and fed state were observed for liver, skeletal muscle, and WAT but not for heart muscle (Fig. 4). Relative uptake of [³H]TG versus [¹⁴C]FA was highest in heart muscle (Fig. 4C) and lowest in liver (Fig. 4D), irrespective of nutritional status. In adipose tissue, absolute [³H] label in the TG store was twofold higher in the fed state (Fig. 4A), but because [¹⁴C]FA label retention was also increased, the relative [³H]TG retention did not significantly change. In skeletal muscle, [³H]TG retention was higher in the fasted state, both absolutely (Fig. 4B) and relative to [¹⁴C]FA retention (Fig. 4E).

In the present study, we have studied the fate of FA derived from lipolysis of plasma TG. We observed that 80–90% of infused TG tracer could be recovered as plasma FA, irrespective of tracer method or TG-rich particle used. It may appear puzzling that the small amount of [³H]FA label found in plasma in Figs. 2 and 3 indicates that most of the [³H]TG label entered plasma as [³H]FA. This is caused by the very rapid turnover of plasma FFA (see [¹⁴C]FA curves in Figs. 2 and 3). With the lower turnover of TG, the “infusion” rate of [³H]FA is relatively slow compared with the turnover of [³H]FA, and therefore much of the label is rapidly cleared. We have not included the individual rate constants of the FA kinetics, as they are very dependent on the model structure, whereas the overall kinetic constants, the FCR and R_d, are related to the area under the curve of the tracer and are therefore robust to details of the model. It is the FCR of FA that determines the plasma TGFA recovery, not the individual rate constants, an argument also made by Hultin et al. (13). Other studies have probed FA kinetics in mice in greater detail (27).

The high recovery of TGFA in plasma is in agreement with the compartmental modeling studies in rats (13). It does not agree, however, with the steady-state data presented by Wolfe et al. (11). Their conclusion was that only 5% of the total FA flux originated from VLDL-derived FA. This implies that the plasma TGFA recovery in their system was very small. However, the authors used the specific activity of the infusate (i.e., of FA in the VLDLTG infusate) to calculate the flux from VLDL to FFA rather than the specific activity of plasma VLDLTG itself. In our view, the latter is the proper precursor enrichment to calculate rate of production of FFA from TG.

When >80% of FA generated by LPL is lost in plasma, one may raise doubts as to the usefulness of tissue-specific regulation of LPL activity. Yet, we have measured enormous differences in the ratio of [³H]FA and [¹⁴C]FA retention between tissues, which were also dependent on the fed or fasting state (Fig. 4). Hultin et al. (13) found similar results, as well as observed that the ratio of [³H]FA and [¹⁴C]FA depended on the nutritional status. The observed [³H]/[¹⁴C] ratios and the changes therein upon fasting reflect well-documented differences in LPL activity in these tissues (3,5). Thus, liver does not have LPL activity, and its relative retention of [³H]TG is low (Fig. 4). The uptake of [³H] label in liver most probably reflects uptake of particle remnants that occurs primarily in liver. Skeletal muscle LPL, however, is upregulated during fasting and the relative amount of [³H]TG label measured was indeed higher in the fasted state (Fig. 4E). In WAT, the uptake of [³H]TG was expected to be higher in the fed state because LPL is insulin stimulated in WAT (5,28). This is also reflected in the total amount of [³H] counts (Fig. 4A) but not in the ratio of [³H] to [¹⁴C] label retention (Fig. 4E). In WAT, FA kinetics are complicated by the fact that in the fasted state, there is substantial intracellular FA hydrolysis of TG stores. It is clear from these data, however, that there is differential uptake of TGFA and FFA, which clearly follows the activity of LPL in these tissues. It is also important to note that in extrahepatic tissues, TGFA is (at least) equally important as plasma FFA, a conclusion that is in line with previous work (11,29).

The combination of near complete recovery of TGFA label in plasma and yet considerable differential uptake of these labels in the different tissues appears paradoxical. The paradox can be solved from the perspective that the labels actually enter the system at different sites. The simplest model that can solve the paradox is the capillary model (22). This model consists of three compartments: an arterial compartment (A), a venous compartment (V), and a compartment representing the capillary bed (C) (Fig. 5). In this model, we assume complete mixing of labels in compartment C, i.e., those (labeled) FA molecules that are released by LPL activity enter the same compartment, compartment C, as FFAs arriving from the circulation. The only difference for [¹⁴C] and [³H] label is the site of injection: for [¹⁴C]FA, label enters the general circulation via the vein; for [³H]TGFA, the site of “injection” is the capillary bed (where LPL is active).

The steady-state solution of this system for [¹⁴C]FA tracer is:

where q/M is the specific activity of tracer (see research design and methods), V_in the tracer infusion rate, and R_d is the rate of disappearance into tissues, in this system k_0CM_C. This is the classical solution to estimate R_d from specific activities of infusion experiments (i.e., equation 1).

For [³H]TGFA tracer, however, “injection” is at the capillary bed, where LPL is located, and the steady-state solution of such a system is different:

Here, V_LPL rather than V_in is used to stress the fact that the “infusion rate” is determined by the local LPL activity. It follows that the capillary specific activity (q_C/M_C) differs from the one at the sampling site (q_A/M_A) with a factor that depends on the relative amount of label escaping immediate uptake. It should be noted that this ratio corresponds to the plasma TGFA recovery defined in equation 5. It is also related to the FA entrapment, as defined by Evans et al. (15).

In a system with multiple organs, the differences between q_c/M_c and q_A/M_A become even larger because the local “infusion rates” V_LPL are determined by local LPL activity and may therefore differ greatly from the overall systemic “infusion rate” that determines the arterial specific activity. This model can therefore explain how a steady-state infusion model with complete local mixing of [¹⁴C] and [³H] label can give rise to differential uptake of labels according to LPL activity.

Figure 6 illustrates a proposal of the physiology underneath the behavior of the tracers. The activity of LPL is proposed to increase the local pool of FA close to the endothelium (Fig. 6). The relative sensitivity of the uptake rate toward this increase in local FA therefore decides how much of the newly formed FA will be extracted and metabolized and how much will end up in venous plasma. Thus, tissue-specific regulation of LPL would allow for the production of a local increase in FA that then drives the (facilitated) diffusion of FA into the underlying cells.

This mechanism has implications for the current methods in experimental animals to measure tissue-specific uptake of FA, i.e., by FA analogs (BMIPP, bromopalmitate, or palmitate only [30,31]). These methods do not take into account the tissue-specific contribution of LPL to the local FA pool that may affect the local rate of label transfer. The proposed mechanism may have further implications for the interpretation of traditional tracer studies of plasma FFA turnover and oxidation. The release of FAs from adipose tissue and plasma fatty oxidation has been studied intensively by primed continuous infusions of labeled palmitate. Our data indicate that the release of FAs by adipose tissue in these studies may have been overestimated by the extent to which TGFA have contributed to the plasma FFA pool. Accordingly, the data on plasma FA oxidation include the contribution of unlabeled plasma TG-derived FAs to the plasma FA pool. However, the quantitative impact of the present data on these parameters in humans is uncertain.

In conclusion, our data indicate that the distinction between TGFA and plasma FFA is artificial. Traditionally, plasma TG metabolism and FFA metabolism are being studied in separate research areas; TG metabolism is mainly being studied in the context of cardiovascular disease and FFA metabolism in the context of diabetes. We propose a model in which FAs released from LPL are locally in rapid exchange with plasma FFAs and serve the important purpose of maintaining a driving force for (facilitated) diffusion. This model therefore embraces the importance of LPL and combines it with that of other factors involved in FA metabolism, such as acylation stimulating protein (32) and FAT/CD36 (6,30). Understanding these interactions between lipoprotein metabolism and FFA metabolism is crucial for understanding complex metabolic diseases such as the insulin resistance syndrome.

This work was conducted in the framework of the Leiden Center for Cardiovascular Research LUMC-TNO and supported by the Dutch Heart Foundation, (grant NHS 96011) and the Netherlands Organization for Research NWO (grants 903-39-194 and 903-39-291).

The authors thank anonymous referees for useful comments and discussion.