Is Membrane Transport of FFA Mediated by Lipid, Protein, or Both?
2007; American Physiological Society; Volume: 22; Issue: 1 Linguagem: Inglês
10.1152/physiol.00011.2006
ISSN1548-9213
AutoresJ. Patrick Kampf, Alan M. Kleinfeld,
Tópico(s)Lipid Membrane Structure and Behavior
ResumoPOINT/COUNTERPOINTIs Membrane Transport of FFA Mediated by Lipid, Protein, or Both?J. Patrick Kampf, and Alan M. KleinfeldJ. Patrick KampfTorrey Pines Institute for Molecular Studies, San Diego, California. , and Alan M. KleinfeldTorrey Pines Institute for Molecular Studies, San Diego, California. Published Online:01 Feb 2007https://doi.org/10.1152/physiol.00011.2006This article has been correctedMoreSectionsPDF (2 MB)Download PDF ToolsExport citationAdd to favoritesGet permissionsTrack citations An Unknown Protein Mediates Free Fatty Acid Transport Across the Adipocyte Plasma Membrane J. Patrick Kampf Torrey Pines Institute for Molecular Studies, San Diego, CaliforniaDownload figureDownload PowerPoint Alan M. Kleinfeld Torrey Pines Institute for Molecular Studies, San Diego, California [email protected]Download figureDownload PowerPoint The mechanism of free fatty acid (FFA) transport between aqueous phases on either side of a biological membrane is intensely debated. Here, we critically analyze features of the methodologies that have contributed to this debate. We discuss our studies indicating that transport across lipid vesicles is limited by the rate of translocation (flip-flop) between the hemileaflets of the lipid bilayer. These results raise the possibility of a significant barrier to transport through the lipid phase of biological membranes. Our results in adipocytes suggest that FFA transport is mediated by an as yet unidentified membrane protein pump.Circulating Free Fatty Acids and Membrane TransportFree fatty acids (FFA), particularly the long-chain FFA, serve a variety of essential functions; they provide a major portion of physiological energy needs (42), are important constituents in the synthesis of complex lipids, and play critical roles in cell signaling (18). Most FFA are obtained from circulating plasma FFA, the levels of which are regulated by the rate at which FFA are generated, predominantly by adipose tissue (43), and the rate at which they are metabolized, predominantly by liver, muscle, and adipose tissue. Regulation of circulating FFA levels is critical because FFA play essential roles in homeostasis; deviations from normal levels reflect pathology and may adversely affect health (5, 6, 23, 31, 33, 34, 37, 40, 44, 45, 55). Although most FFA in circulation are bound to albumin (48, 56), the small fraction of unbound FFA (FFAu) in the aqueous phase mediates physiological activity and is most sensitive to changes in health and disease (5, 49, 66). Moreover, the FFAu, not FFA-albumin complexes, are transported across membranes.In the course of their storage, production, and consumption, FFA must cross the plasma and mitochondrial membranes of many different cells. Because these membranes might be involved in the regulation of FFA trafficking, the mechanism by which FFA are transported across membranes has been an area of considerable interest for over 40 years (57).Why the Mechanism of FFA Transport Across Membranes is ControversialEven after 40 years of study, there remains considerable disagreement about the mechanism by which FFA are transported across membranes. This controversy stems from a lack of agreement about 1) the rate-limiting kinetic step, 2) the magnitude of the transport rate, 3) the need for membrane proteins in cellular transport, and 4) the function of the several membrane proteins that have been found to play a role in the cellular uptake of FFA.Three general views of the mechanism of FFA membrane transport have emerged (FIGURE 1). At one end of the spectrum, it has been proposed that transport of long-chain FFA would involve extremely rapid (<5 ms) passive transport through the lipid phase of a membrane followed by much slower dissociation from the membrane into the aqueous phase (14, 26, 27, 58, 62, 67). The basis for this interpretation was the finding of apparently extremely rapid flip-flop of FFA across pure lipid vesicles leading to the conclusion that protein-mediated transporters would be irrelevant for FFA transport across biological membranes. At the other end of the spectrum are reports that identify four different proteins that in some way facilitate FFA transport in mammalian cells: FABPpm (53), CD36/FAT (1), FATP (52), and caveolin (63). These studies also report that, in addition to the protein-mediated mechanism, FFA are transported in parallel via a slower, nonspecific, presumably lipid phase mechanism. In addition, it has been proposed that FFA transport occurs through a combination of caveolin and CD36/FAT within lipid rafts (46). However, the only membrane protein that has been clearly demonstrated to be an FFA transporter is FadL, an E. coli outer membrane protein (16). The exceptionally strong evidence supporting this finding includes genetic, structural, and transport characteristics. This protein, which transports long-chain FFA through the polar lipopolysaccharide outer membrane, has no eukaryotic homolog.Our view of FFA membrane transport lies somewhere between these opposing views and is based on our studies with lipid vesicles (13, 28, 30, 31, 35, 59), erythrocyte ghosts (36), and cultured adipocytes (29, 32). Unlike the first view, our studies in lipid vesicles indicate that flip-flop, not dissociation, is the rate-limiting step for transport across lipid membranes, suggesting a potential need for a membrane transport protein in those cells with significant FFA metabolic demand. The results of our adipocyte studies are consistent with a facilitated transport mechanism, which is presumably protein mediated as in the second view (29, 32). However, our results are not consistent with a parallel lipid-mediated pathway or the reported characteristics of the putative transport proteins. In fact, we propose that actual FFA transport proteins have not yet been identified in mammalian cells because appropriate experimental methods have not been used in most studies. In the following Point/Counterpoint, we will identify key issues in the controversies surrounding FFA transport, focusing on our own studies to illustrate these issues.Definition of FFA Membrane TransportThe disagreement concerning the mechanism of FFA transport arises in part because there is no consensus on the definition of FFA membrane transport (for example is it between aqueous phases on either side of the membrane or is it from the outer aqueous phase to an esterified product). This ambiguity has led to disagreement regarding the appropriate experimental methods for investigating the transport mechanism. We therefore begin our discussion with a definition of FFA membrane transport, which will be the basis for a critique of the methods used to study FFA transport across lipid vesicles and cell membranes.We define transport of FFA across membranes as the movement of FFAu from the aqueous phase on one side to the aqueous phase on the other side of the membrane (FIGURE 2). This definition implies that FFA are initially transported across the membrane and into the cytoplasm. They may thereafter be activated to acyl coenzyme A (acylCoA) or, as FFA, partition into intracellular membranes and lipid droplets and bind to intracellular proteins. This process is distinct from transport in which activation of the FFA to acylCoA occurs concurrently with translocation across the membrane (16, 17, 47), generating no significant amount of nonactivated FFAu in the cytosol. The studies by DiRusso et al. in bacteria and yeast (16, 17) and recently by Richards et al. in 3T3-L1 adipocytes (47) have shown clearly that the physiologically important step of FFA activation to acylCoA involves the interaction of long-chain acylCoA synthetases (ACSLs) with membrane-bound FATP. In these studies, virtually all the transported FFA was found to be irreversibly associated with the cells (20), as would be expected if activation of FFA to acylCoA was tightly coupled to the transport step. This process in which activation of FFA to acylCoA is tightly coupled to transport is termed vectorial acylation. FFA activation to acylCoA is of course a necessary step for storage and metabolism of FFA. Nevertheless, as we discuss later, our results in adipocytes indicate that the major portion of FFA transported into the cells remains unesterified (29, 32), and we will explain how methodological issues account for these discrepant conclusions.The Overall Transport Process Can be Decomposed into Discrete Steps.Using the above definition of transport, we can enumerate discrete kinetic steps that constitute aqueous-to-aqueous transport of FFA (FIGURE 1). Influx of FFA, from the extracellular to the intracellular aqueous phases, involves five separate kinetic steps: 1) dissociation [koff(A)] from albumin, 2) diffusion through the outer aqueous phase, 3) insertion into the outer hemileaflet of the membrane (kon), 4) flip-flop from the outer to the inner leaflet (kff), and 5) dissociation from the inner leaflet (koff). If FFA transport is facilitated by a membrane protein, then step 4 (and possibly steps 3 and 5) will be determined by binding to, translocation by, and dissociation from the protein transporter rather than a passive flip-flop step through the lipid bilayer phase of the membrane. Flip-flop denotes the process by which the FFA, with its carboxyl head group at the lipid-water interface and its acyl tail inserted into the hydrocarbon region of one hemileaflet of the lipid bilayer, is translocated across the membrane so that it winds up with the same orientation with respect to the opposite hemileaflet (FIGURE 1). Passive flip-flop may occur in the lipid bilayer phase of a membrane without protein assistance. The molecular details of the passive flip-flop mechanism are only beginning to emerge from work in lipid vesicles (e.g., see Ref. 28), and it is possible that this process might be facilitated by the activity of a transport protein in cellular membranes, as discussed with reference to the models of FIGURE 5. Our studies (29, 32) indicate that the kinetic steps 1–5 are reversible; adding fatty acid-free albumin initiates rapid efflux and completely extracts nonactivated FFA from the cell, which, as discussed later, is most of the FFA transported into the cell. In our view, monitoring intra- and extracellular FFAu levels provides the foundation for studying membrane transport of FFA.Transport of FFA Across Lipid VesiclesA fundamental question in FFA transport is whether a biological membrane needs a protein transport pathway given the prediction by Overton's rule that FFA should readily permeate the lipid bilayer phase of the membrane (65). Determination of the transport characteristics of pure lipid vesicles would seem to be the best way to address this issue. However, considerable controversy has characterized such studies (13, 14, 19, 25, 27, 31, 35, 54, 58, 59, 62). The origin of virtually all the controversies that arise in lipid vesicle or cellular transport of FFA stems from the highly insoluble nature of long-chain FFA. During transport, the overwhelming amount of FFA is bound to lipid and protein phases; generally 5 μM) concentrations of FFAu that are present in the absence of BSA (or other carriers) perturbs the bilayer structure (13, 30, 36). In contrast, delivering FFA in complex with BSA exposes the vesicles to 100-fold lower FFAu concentrations and reveals 100-fold slower influx rate constants compared with uncomplexed FFA (13). In our measurements, FFA dissociation from BSA does not affect the determination of the flip-flop rate constants because the time required to transfer FFA from BSA to vesicles or cells, under the highly buffered conditions of our experiments, is much less than flip-flop (13, 15, 29). This lack of dependence on dissociation from BSA is apparent from the two orders of magnitude difference in influx rates we observe from SUV (3 s−1) to adipocytes (0.02 s−1) using the same FFA-BSA complexes (13, 28, 29, 36).Flip-flop can equally well be measured by monitoring FFA efflux from vesicles to an external FFA sink such as BSA (FIGURE 3A). Two kinetic steps must be resolved because FFA transfer to BSA involves both flip-flop and dissociation. Studies in which only the vesicle-trapped pyranine fluorescence was monitored found a single rate, which was interpreted as dissociation because it was assumed that flip-flop was immeasurably fast (62, 67). However, direct measurements of dissociation, by monitoring FFA binding to BSA, reveal rate constants that are 5- to 10-fold faster than those obtained from the pyranine fluorescence (13, 41). The time course of FFA transfer from the vesicle to BSA reveals two rate constants: a slow one that is equal to that obtained from pyranine measurements (flip-flop) and one that is 5-to 10-fold faster (dissociation) (13). It is important to emphasize that the measured rate constants are remarkably accurate; disparate groups have obtained virtually identical values for the rate constants for efflux (13, 67) and dissociation (13, 41). Thus the findings that FFA transport involves extremely rapid flip-flop and rate-limiting dissociation (26) are based on a misinterpretation of the influx and efflux measurements.Different Mechanisms of Flip-flop and Dissociation.To begin to understand the flip-flop barrier, we recently determined the temperature and FFA size dependence of transport in lipid vesicles (28). Both flip-flop and dissociation rates decreased exponentially with FFA size, but only flip-flop was temperature dependent. The results suggest that the barrier to flip-flop is the work needed to create sufficient free volume to allow the FFA to reorient between the lipid-water interfaces on either side of the bilayer. The free volume model of FFA transport provides a possible explanation for the observed deviations (i.e., slower transport for larger partition coefficients) from Overton's rule. Dissociation is limited primarily by the FFA's aqueous solubility (see, for example, Figure 8 of Ref. 28). In summary, the results of this study reveal distinct mechanisms governing the barriers to flip-flop and dissociation and identify factors that might affect transport of FFA through the lipid phase of biological membranes.Limitations of Lipid Vesicle Studies.The critical finding of the lipid vesicle studies is that the rate-limiting step across simple lipid vesicles is the translocation step (flip-flop) within the bilayer itself. This raises the possibility that more complex lipid phases, as might exist within biological membranes, could generate barriers large enough to prevent significant transport through the membrane's lipid phase. For example, our results imply a virtually impenetrable lipid phase barrier in the adipocyte plasma membrane (29). However, there is little possibility of reproducing in a lipid vesicle the actual lipid phase within a biological membrane given the extremely large number of different lipid species and transverse and lateral asymmetry of such membranes. Nevertheless, if a composition of lipid vesicles can be discovered that generates very slow transport, as observed in adipocytes, it would help support the case for a membrane protein-mediated transport mechanism, and such vesicles could be used for reconstitution studies. Work in progress in our laboratory reveals that specific lipid compositions can significantly reduce rates of FFA transport across lipid vesicles.Transport of FFA Across Cell MembranesAs mentioned above, three views have emerged for how FFA are transported across cell membranes: 1) rapid diffusion through the lipid phase, 2) (known) membrane protein plus lipid phase diffusion, and 3) (unknown) membrane protein only. Because methodology is key to resolving these disparate views, we will mostly focus on our studies in human erythrocytes and adipocytes where the FFAu concentration was monitored.Erythrocyte Ghosts.Our first investigation of cellular FFAu transport was carried out in human erythrocyte ghosts simply because such ghosts have been characterized extensively and because virtually identical stopped-flow mixing methods as used for lipid vesicles could be used for ghosts. In these studies, we trapped ADIFAB or pyranine in resealed ghosts and performed influx studies by adding FFA as complexes with BSA and measured efflux from FFA-loaded ghosts to ADIFAB (36). We found that flipflop was rate limiting, with rate constants similar to those for large lipid vesicles (13). Moreover, we found no effect on FFA transport by a variety of protein-specific reagents. Although we cannot rigorously exclude a protein-mediated process, the simplest mechanism consistent with these results and the uptake measurements of Ref. 10 is a passive lipid phase pathway.Transport Across Adipocyte Membranes.FFA metabolism is not the primary function of human erythrocytes. If regulating FFA transport across membranes is a vital physiological function of other cell types, one would expect adipocytes, whose major function is to store and release FFA, to manifest a high degree of regulation. Not surprisingly, FFA uptake (2–4, 46, 52, 60, 61, 63, 68) and transport (12, 24, 29) have been studied extensively in adipocytes. Uptake studies have linked the four different proteins discussed above to adipocyte transport: FABPpm (53), CD36 (1), FATP (52), and caveolin-1 (63). Support for the identification of these proteins as transporters was obtained by observing that uptake is 1) inhibited by specific reagents (2, 4, 60, 61), 2) correlated with expression levels of the protein (1, 22, 51), and 3) (partially) saturated at high extracellular FFAu concentrations (4, 61). Similar results have been obtained in uptake studies in giant sarcolemmal vesicles of muscle (7, 8).Uptake Measurements Do Not Reveal Transport Kinetics.In our view, the role of these proteins as FFA transporters is uncertain because uptake measurements share several features that may limit their ability to elucidate the membrane transport steps as defined in FIGURE 1. Three criticisms of uptake measurements will be discussed. First, the most important issue relates to the experimental configuration of uptake measurements and whether the quantity measured as transported FFA is reversibly associated with the cell. An uptake measurement involves the determination of the quantity of labeled (generally radioactive or fluorescent) fatty acid associated with, and presumably transported into, the whole cell. In contrast, the ADIFAB measurement of cytosolic FFAu provides a direct measure of the aqueous to aqueous mechanism illustrated in FIGURE 2. In the uptake experiments, FFA is added to the cells, generally as a complex with BSA, and after a specified incubation time the cells are washed with fatty acid free BSA to remove nontransported FFA (for example, see Refs. 4, 39, 64). The amount of radioactivity or fluorescence associated with the washed cells is taken as the transported FFA. To remove the extracellular FFA without disturbing cell-associated FFA, most uptake studies use specific reagents, such as phloretin (29), to inhibit FFA efflux. However, neither our measurements (29) nor those of Faergeman et al. (20) revealed any effect of phloretin on FFA transport.In contrast to uptake measurements, our experimental configuration for measuring influx involves no wash step; the FFA-BSA complexes are present continuously as the intracellular FFAu concentration is monitored in real time (29). Under these conditions, we observe that virtually all the FFA transported into the cells can be extracted rapidly from the cells by washing (29, 32). This suggests that very little of the transported FFA is activated to acylCoA because acylCoA is unlikely to be extracted rapidly from the interior of the cells (9). Yet, as discussed above, virtually all labeled FFA is irreversibly associated with cells in uptake measurements and is probably activated to acylCoA, consistent with vectorial acylation (20, 47). These results suggest that, although uptake measurements detect the small fraction of FFA transported into the cells that becomes activated, most of the transported FFA remains unesterified and partitions reversibly into hydrophobic phases such as lipid droplets. Thus our results suggest that uptake measurements are not sensitive to the actual FFA transport step.Second, saturation of the initial transport rate with increasing extracellular FFAu (FFAo) concentration provides an important confirmation of a protein-mediated transport mechanism. Yet, virtually all uptake measurements reveal saturation only after subtraction of a component that is assumed to be diffusion through the membrane lipid phase. Moreover, without directly measuring [FFAo], it is likely, especially at high [FFAo] where saturation is most apparent, that the actual [FFAo] is significantly smaller than calculated using albumin binding constants, thereby giving the appearance of saturation. The reason for this is that [FFAu] increases exponentially with increasing ratios of FFA bound (FFAb) to albumin (21, 48, 56). In the presence of cells, [FFAb] and therefore [FFAu], will be reduced through FFA transfer to the cells by an amount that depends on the number and type of cells. This reduction in [FFAu] can be quite large at high FFAb-to-albumin ratios because of the exponential dependence, and without directly measuring [FFAu] as described in Ref. 50 the level of reduction cannot be determined.Third, identification of specific proteins as transporters relies on the positive correlation of uptake with protein expression. Uptake studies reveal little or no transport in preadipocytes compared with fully differentiated cells, and this lack of transport has been correlated with the expression of the four proteins identified as transporters, the preadipocytes expressing little or none of these proteins (1, 22, 51). In contrast to uptake studies, our measurements of intra-cellular FFAu reveal no significant difference in FFA transport characteristics in preadipocytes compared with adipocytes, suggesting that the above identified proteins are not directly involved in the membrane translocation step (in preparation).FFAu Measurements Reveal Transport Kinetics.As far as we are aware, only three reports of adipocyte transport satisfy the criteria for transport measurements we have outlined (11, 29, 32). In our own studies, we used two methods to assess actual transport. In the first, we microinjected ADIFAB into living 3T3-F442A adipocytes, clamped the FFAo concentration to a fixed value, and then monitored images of intracellular FFAu using ratio fluorescence microscopy of ADIFAB (29). In the second method (see below), we used multi-isotope imaging mass spectrometry of 13C-labeled oleate to provide an independent test of our intracellular [FFAu] values, also in 3T3-F442A adipocytes (29, 32). The live cell imaging studies provided the first measurements of unbound intracellular FFA (FFAi) concentrations and allowed the time courses for influx and efflux to be monitored under physiological conditions (29). To monitor influx, we increased [FFAo] rapidly from zero and used microinjected ADIFAB to show that [FFAi] increased quite slowly (kin = 0.02 s−1), reaching a steady-state level in ~150 s (FIGURE 4). [Similar slow influx rates were also observed by monitoring intracellular pH in human and rat adipocytes from different fat depots (11).] Once steady state had been reached, [FFAo] was returned to zero, and efflux was measured by monitoring [FFAi] and found to be twice as fast as influx (kout = 0.04 s−1). As indicated in the previous section, an important feature that distinguishes these studies from uptake measurements is that, during these transport measurements, the extracellular source of FFA is only altered when switching from influx to efflux.This unique ability to generate spatial and temporal images of FFAi has allowed us to determine the characteristics of actual FFA transport in adipocytes, the major features of which include the following, as described in Ref. 29. First, and perhaps most extraordinarily, FFAi concentrations at steady state are larger (two- to fivefold) than the FFAo concentrations. Depletion of cellular ATP abolishes the steady-state concentration gradient ([FFAi] > [FFAo]), which indicates that FFA transport involves an ATP-dependent pump. It is important to recognize that this ATP dependence is unrelated to the ATP dependence of acylCoA formation because 1) FFA are rapidly extracted (29, 32), 2) inhibiting acylCoA formation with Triacsin C has no effect on the transport rates or the gradient (29), and 3) transport occurs with the same rate constants in ATP-depleted cells. Second, measurements with increasing [FFAo] reveal a single and saturable transport mechanism, which implies that transport occurs through a membrane protein with no significant transport through the lipid phase (Figure 4 of Ref. 29). Third, the transporter possesses an FFAo-sensing gate that modulates efflux so that, at elevated FFAo levels, the efflux rate constant decreases and becomes equal to or less than influx (Figure 3 of Ref. 29). Fourth, FFA transport properties of the adipocyte plasma membrane are distinct from lipid vesicles. In addition to the characteristics described above, which do not occur in lipid vesicles, adipocyte rate constants are ~100-fold slower than in lipid vesicles (21). Last, we observed no effect of reagents that have been shown to modulate uptake either at the metabolic level (insulin, triacsin C) or through inhibition (phloretin, DIDS, and proteases), which suggests that the effect of these reagents on measurements of uptake are at a different step than is revealed by measurements of [FFAi].Given the unprecedented nature of these results, it was important to test key features using an independent method, albeit one that could provide direct information about FFAi. Consequently, we used multi-isotope imaging mass spectrometry (MIMS) to study FFA transport in 3T3-F442A adipocytes (32, 38). This is a new technology that allowed us to quantitatively image the distribution of [13C]oleate with a spatial resolution of 40 nm within the plane of the adipocyte. In these studies, cells were treated as in the live cell/ADIFAB method except [13C]oleate:BSA complexes were used to deliver FFA. After incubation with [13C]oleate:BSA for 20 min, 13C/12C ratio images of the cells generated by MIMS revealed that most of the [13C]oleate was in the lipid droplets. The 13C/12C ratios were used to determine the concentration of 13C in the droplets, which in turn was used to estimate the aqueous (cytosolic) concentrations of the [13C]oleate. The results yielded FFAi concentrations that were approximately fivefold greater than FFAo concentrations, consistent with an energy-dependent FFA transport pump, as found in the ADIFAB/live cell studies. This study also demonstrated that virtually all the 13C label was reversibly associated with the cells, consistent with our ADIFAB results, showing that [FFAi] rapidly returns to zero when [FFAo] is clamped to zero (FIGURE 4). This result indicates that, even after the 20-min incubation at 37°C, only a tiny (<0.1%) fraction of the added FFA is metabolized, suggesting that vectorial acylation may not play a significant role in uptake of FFA, at least for 3T3-F442A adipocytes (17, 47).Summary.Contrary to most studies with pure lipid vesicles, we find that flip-flop is the rate-limiting step for transport across lipid bilayers (13). The flip-flop rate depends on both FFA structure and vesicle type and may be related to the free volume within the bilayer (28). Because flip-flop is slow and dependent on the membrane structure,
Referência(s)