Journal of Molecular Neuroscience Copyright © 2001 Humana Press Inc. All rights of any nature whatsoever reserved. ISSN0895-8696/01/16:99–108/$12.50

Fatty Acid Transport The Diffusion Mechanism in Model and Biological Membranes

James A. Hamilton,*,1 Rebecca A. Johnson,2 Barbara Corkey,3 and Frits Kamp1,4 1

Department of Biophysics, Boston Medical Center, 715 Albany St., Boston, MA 02118; Institute of Human Nutrition, Columbia University, New York, NY 10032; and 3Obesity Research Center, Evans Department of Medicine, Boston Medical Center, 715 Albany St., Boston, MA 02188; 4 Institute of Physical Chemistry, University of Munich, Schillerstrasse 44, 80336 Munich, Germany 2

Received September 21, 2000; Accepted November 1, 2000

Abstract The transport of fatty acids (FA) across membranes can be described by three fundamental steps: adsorption, transmembrane movement, and desorption. In model membranes, these steps are all rapid and spontaneous for most fatty acids, suggesting that FA can enter cells by free diffusion rather than by protein-mediated mechanisms. Here we present new fluorescence approaches that measure adsorption and transmembrane movement of FA independently. We show that FA adsorb to the plasma membrane of adipocytes and diffuse through the membrane by the flip-flop mechanism within the time resolution of our measurements (~5 s). Thus we show that passive diffusion is a viable mechanism, although we did not evaluate its exclusivity. Important implications of the diffusion mechanism for neural cells are that all types of FA could be available and that selectivity is controlled by metabolism. Studies of FA uptake into brain endothelial cells and other brain cell types need to be performed to determine mechanisms of uptake, and metabolism of FA must be separated in order to understand the role of membrane transport in the overall uptake process. Index Entries: Blood-brain barrier; fluorescence; transmembrane diffusion; fatty acid uptake; adipocyte; phospholipid bilayers.

Introduction Unesterified fatty acids (FA) are essential nutrients for human cells. They can be oxidized to provide the primary fuel for many cells, such as cardiac myocytes, they can be stored in adipocytes in the form of triglycerides, and they can be used in remodeling of membrane phospholipids. The brain requires large amounts of polyunsaturated FA, particularly the ω-3 FA DHA, which must enter brain cells either as the essential FA precursor or as the desired FA itself. In addition to passing the blood-brain barrier (BBB) in the unesterified form, the FA could in principle be transported in as esterified lipids in lipoproteins. When deuterated FA are included in the diets

of suckling rats, and the brain is examined days later, there is no evidence of labeled palmitate in the brain whereas significant amounts of arachidonate are found (Marbois et al., 1992). Does this difference reflect selective transport through the blood-brain barrier or metabolic factors? In contrast, FA injected into the blood, where they bind mainly to albumin, are taken up rapidly (within seconds) by the rat brain and either oxidized (palmitate) or incorporated into phospholipids (arachidonate) (Robinson et al., 1992). Does this imply that all types of FA can be taken up by diffusion? How FA are transported through membranes is a fundamental issue for the uptake of FA into the brain. Although specific studies relevant to the BBB

*Author to whom all correspondence and reprint requests should be addressed. E-mail [email protected]

Journal of Molecular Neuroscience

99

Volume 16, 2001

100

Hamilton et al.

Fig. 1. Simplified scheme of FA transport in a phospholipid bilayer. Adapted with permission from Hamilton (1999).

or to brain cells are relatively scarce, studies with other cells have generated great controversy. Aquestion under debate is whether FA pass through membranes by diffusion mechanisms, or whether specialized protein transporters are required to bind and/or move FA through a membrane. Is a completely diffusive mechanism based on the fundamental properties of FA and their interactions with the lipid components feasible, or might these fundamental properties require proteins to move FA through an unfavorable environment? If both mechanisms are judged feasible, is a combination of free diffusion and protein transport used by certain cells to achieve optimal uptake of FA, or selective uptake of specific FA? The answers to these questions will not be found in the most current textbooks on Cell Biology and Biochemistry. In fact, one is likely to find schematic illustrations of a model membrane (phospholipid bilayer) together with lists of molecules that can or cannot permeate the lipid bilayer (e.g., (Darnell et al., 1990,) p. 532) without mention of FA. Into which general category do they fit: uncharged or charged? Certainly it is appreciated that long-chain FA, on which we will focus, are highly insoluble in aqueous media. It is well known that rather high concentrations of FAare transported through the plasma as a lipid-protein complex with albumin (Spector, 1986). However, the physical properties of FA are generally not understood in depth, and the uptake of FA into cells is often analyzed (we believe inappropriately) in the same way as for water-soluble molecules such as glucose (Hamilton and Kamp, 1999), without proper consideration of the avid partitioning of FA into phospholipid bilayers.

Journal of Molecular Neuroscience

Here we summarize evidence for the diffusion mechanism of membrane transport of FA and present new data testing the hypotheses of this mechanism in both model and biological membranes. We begin with a simple model membrane consisting of a phospholipid bilayer without proteins (Hamilton, 1999). As shown in Fig. 1, the membrane transport of FA can be dissected into three essential steps: adsorption, transmembrane movement, and desorption. Considering the physical properties of FA and phospholipid bilayers, can each of these three steps proceed spontaneously and rapidly without the assistance of a specialized protein?

Methods Chemicals ADIFAB, 2′,7′-bis-(2-carboxyethyl) -5-(and-6)carboxyfluorescein, acetoxymethyl ester (BCECF, AM), pyranin and parinaric acid were purchased from Molecular Probes. Stock solutions of 10 mM K+ oleate were prepared by adding 2.38 mg pure oleic acid to 1.0 mL distilled water, and solubilizing the oleic acid at pH > 10.0 by adding 5–10 µL of 1 M KOH and vortexing until clarity. Parinaric acid was solubilized in DMSO. Small Unilamellar Vesicles Small unilamellar vesicles (SUV) were prepared as previously described (Kamp and Hamilton, 1992). Egg lecithin (phosphatidylcholine, PC) was hydrated (25 mg PC/mL buffer) at 4°C for 1 h in 100 mM HEPES-KOH buffer, pH 7.40. The lipids were subsequently sonicated for 1 h on ice under a stream of nitrogen gas. For vesicles with internally trapped Volume 16, 2001

Fatty Acid Diffusion Through Membranes pyranin, 0.5 mM of the probe was added to the buffer prior to hydration and untrapped probe was removed after sonication by gel-filtration (Sephadex G25, medium grade).

Rat Adipocytes Rat adipocytes were prepared as previously described (Civelek et al., 1996). Epididymal fat pads from two 200 g male rats were digested in 4 mL Krebs buffer with 6 mg collagenase and 3% (w/w) albumin at 37°C, for 20 min with vigorous shaking. The fat cells were sieved and subsequently washed 2 times with the same Krebs buffer not containing collagenase. BCECF acid was trapped by incubation with 1 µM BCECF-AM for 30 min, at 37°C with gentle shaking. Untrapped BCECF as well as albumin was then removed by washing 3 times with Krebs buffer. TwomL aliquots of fat cells (~106 cells/mL) were distributed into several cuvets containing a stirring bar, which were subsequently used for fluorimetric experiments. Fluorescence Fluorescence measurements were carried out in a Spex fluoromax fluorimeter with stirring, either at room temperature or at 37°C. Pyranin fluorescence was measured at 1.0-s intervals using excitation and emission wavelengths of 455 and 509 nm, respectively. The fluorescence of parinaric acid was measured at an excitation wavelength of 324 nm, and an emission wavelength of 416 nm. For experiments when internal pH and external unbound oleic acid were measured simultaneously, ADIFAB (0.2 µM) was added to the external solution of cells with trapped BCECF. The fluorescence of BCECF (excitation at 439 and 505 nm, emission at 535 nm) and ADIFAB (excitation at 390 nm, emission at 432 and 505 nm) were sampled simultaneously at 4.0-s time intervals. With the aforementioned excitation and emission wavelengths, there was no overlap between the fluorescence of the two probes. The fluorescence ratio of BCECF was related to internal pH and calibrated independently by permeabilizing the cells with nigericin and measuring external pH with a pH electrode after sequential additions of NaOH. The ADIFAB fluorescence ratio was related to external unbound FA by protocols described (Richieri et al., 1992).

Results and Discussion Adsorption The adsorption of FAto a lipid bilayer occurs spontaneously when FA alone is presented in an aqueJournal of Molecular Neuroscience

101 ous medium, because of the higher solubility of FA in the lipid phase compared to water. The distribution of FA between water and the lipid membrane is quantified by the partition coefficient Kp . At a fixed degree of unsaturation, Kp increases with increasing FA chain length. Kp is about 107 for palmitic acid, a common dietary FA, in a bilayer composed of PC, which is typical of phospholipids in most cell membranes (Hamilton, 1998). It is important to appreciate that because of this high Kp, levels of FA in the membrane can easily reach mM concentrations with respect to the volume of the membrane phase, while unbound FA concentrations in the water phase are in the nM range (Hamilton and Kamp, 1999). The adsorption step of membrane transport of FA is illustrated by the fluorescence experiment of Fig. 2. Here we have used a natural (though rare) fluorescent polyunsaturated FA, parinaric acid. Its fluorescence is higher in a lipid environment than in an aqueous environment; thus the fluorescence reports directly the binding of the FA to phospholipid bilayers. Figure 2 shows that upon injection of parinaric acid into a suspension of PC vesicles (spherical bilayers enclosing a small aqueous volume), the fluorescence immediately increased, reaching a maximum value within the time resolution of the experiment (2 s). The actual rate of adsorption is diffusion-limited and is much faster, on the order of 107/s for a simple system such as this (Hamilton, 1998). Note that the fluorescence reaches a stable plateau reflecting the lipid environment, and gives no information about the steps of membrane transport beyond adsorption; specifically, this experiment does not reveal whether the FAhas crossed the bilayer to equilibrate between the two sides of the membrane. Whereas adsorption is straightforward and not a rate-limiting step of membrane transport in a model bilayer, this may not be the case for a biological membrane. The unstirred water layer around a cell slows diffusion of FAinto the membrane and becomes more significant for increasing chain length above 12 carbons (Dietschy, 1978). In addition, the glycocalyx surrounding some membranes could also make the adsorption step more difficult and slower (Hamilton et al., 1994). These issues require further investigations with approaches that differentiate the adsorption step in cells from the subsequent steps of membrane transport. Below we present new data addressing this issue in adipocytes. What if FA is presented to model membranes in the form of complexes with albumin? Adsorption to the membrane then proceeds through the monomers

Volume 16, 2001

102

Hamilton et al.

Fig. 2. Adsorption of parinaric acid to phospholipid vesicles. At 120 s, 100 nmoles of parinaric acid were added to a suspension of phospholipid vesicles (0.5 mM) and the fluorescence of parinaric acid was monitored. The rapid increase reflects rapid binding of the acid to a lipid environment.

in the aqueous medium in equilibrium with albumin. Net transfer of FA from albumin will proceed spontaneously until an equilibrium distribution between the membrane and albumin is reached (Daniels et al., 1985; Hamilton and Kamp, 1999). Although the monomer concentration is low, desorption of FA from albumin-binding sites occurs rapidly, and the equilibrium distribution can be reached quickly (within sec for typical FA [Daniels et al., 1985; Hamilton, 1998]). On the basis of our understanding of the physical properties of FA and their interactions with albumin and membranes, there is no a priori requirement for specialized mechanisms or additional proteins to facilitate the transfer of FA between albumin and membranes. As discussed earlier, two key considerations in the adsorption step are whether FA partition favorably into a lipid membrane, and whether this can occur very rapidly. A third essential consideration is how binding to the phospholipids affects the ionization of the FA. Because the FA carboxyl group must be positioned at the aqueous interface near the charged phosphate group of phospholipids, this local environment of the carboxyl is expected to be much different from its environment as a monomer in water. The ionization of the carboxyl group within the membrane environment can be readily determined by nuclear magnetic resonance (NMR) spectroscopy, and it changes significantly. The apparent pKa of FA with chain lengths of 8–26 carbons is uniformly around 7.5, a remarkable but predictable

Journal of Molecular Neuroscience

increase in the pKa of about 3 units (Kantor and Prestegand, 1978; Hamilton, 1995). This indicates that at typical physiological pH values, about half of the FA is un-ionized in the phospholipid bilayer. The interfacial ionization property of FA has a profound effect on its transmembrane transport.

Transmembrane Movement Of the three essential steps of FAtransport in membranes, the transmembrane step has given rise to the most controversy. The controversy flourished in part because of the difficulty in measuring the transfer of a FA molecule from one side of a membrane to another, even in a simple lipid bilayer. Does the phospholipid bilayer present a barrier to the transmembrane movement of FA, therefore necessitating the presence of a specialized protein transporter? If FA in a membrane were fully ionized, then a protein transporter might act to remove the barrier to the negative charge of the carboxyl group, a mechanism hypothesized in some studies of protein-mediated uptake of FA into cells (Stremmel et al., 1986). The presence of the un-ionized form of FA in a membrane presents another logical route of transmembrane movement, that of diffusion of the uncharged FA, as in the case of other weak electrolytes. This mechanism had been clearly established for short- and medium-chain FA, but the insolubility of long-chain FA created obstacles for extending these same experimental approaches (Herman, 1964). Our laboratory used a newer Volume 16, 2001

Fatty Acid Diffusion Through Membranes

103

Fig. 3. Flip-flop of FA across the bilayer of phospholipid vesicles by rapid movement of the un-ionized FA. (A) Oleic acid was added to vesicles with entrapped pyranin, resulting in an instantaneous (t1/2 < 1 s) decrease in fluorescence corresponding to a pH drop of 0.3 units, followed by a slow relaxation of the pH gradient caused by the leak of protons. (B) The same experiment was repeated, and when BSA was added to the external medium, an immediate increase in pH to the initial value was seen. Adapted with permission from Hamilton (1999).

approach, NMR spectroscopy, to show that the rate of transmembrane movement of un-ionized unconjugated bile acids in PC vesicles was fast (ms) whereas the rate for the ionized form was very slow (Cabral, 1987). However, the NMR method was not applicable to FA. A new fluorescence approach developed by Dr. Kamp in our lab clearly delineated the transmembrane steps of FA transport from other steps and demonstrated that long-chain FAdiffuse (“flip-flop”) rapidly through a phospholipid bilayer via the unionized form of the acid. Our hypothesis and our key experiments are illustrated in Fig. 3. When FA are added to a suspension of phospholipids vesicles, the FAfirst bind to the monolayer to which they were exposed. Binding and ionization equilibria are reached rapidly. The un-ionized FA move across the membrane to diminish the concentration gradient until the chemical potential of FA on the inside and outside leaflets are equal. Key to our experimental approach is that the newly arrived FA will reach ionization equilibrium quickly and release protons to the small internal aqueous volume. This process can be monitored by a fluorescent pH dye trapped in this internal volume. Figure 3 shows that oleic acid addition to the external volume of PC vesicles results in an immediate and rapid decrease in internal pH, consistent with the predictions of our model. The pH remains low because the permeation of H+ is relatively slow. If albumin is introduced to the external buffer in a suf-

Journal of Molecular Neuroscience

ficient amount to bind all of the added FA, the pH decrease is immediately reversed. Thus, we have “flip” and “flop,” and a bi-directional mechanism of FA transport through membranes. The rate of flipflop of common dietary FA in protein-free phospholipid vesicles has been demonstrated by similar fluorescence measurements (with faster time resolution) to occur on the ms time scale, and to not be strongly dependent on the acyl chain length (Kamp et al., 1987). Although the estimations of FA flip-flop obtained by studies of FA with attached fluorescent labels led to widely conflicting conclusions about the rates of natural FA, there is now good general agreement about the fast transmembrane diffusion of FA in simple model membranes (Hamilton, 1998; Hamilton and Kamp, 1999). New investigations of the diffusion of FA into cells are reported below.

Desorption If FA can rapidly adsorb to and move through a lipid bilayer, can they spontaneously dissociate from the phospholipids bilayer, or is this step dependent on a protein? The fluorescence data of Fig. 3 reveal the answer for oleic acid. The addition of albumin, which provided an external sink for the FA, resulted in rapid removal of oleic acid from vesicles. Thus, both the steps of flip-flop and desorption were fast (half-time <2 s). Indeed, data prior to our studies with vesicles showed that transfer of typical FAfrom vesicles to albumin or to other vesicles was rapid and spontaneous (Daniels et al., 1985). By studying Volume 16, 2001

104

Hamilton et al.

Fig. 4. Dependence of the desorption rate constant (24°C) on the FA chain length. Solid symbols represent saturated FA with 14–26 carbons; the open circle represents oleic acid, and the open square, linoleic acid. Adapted from Zhang et al. (1996).

transfer of FA from donor to acceptor vesicles with an entrapped pH dye, and utilizing our knowledge of the upper limits of the rates of flip-flop of FA, we determined the rate of desorption of FA with chain lengths of 14–24 carbons (Zhang et al., 1996). The data were extended to the 26-carbon saturated FA hexacosanoic acid by NMR measurements (Ho et al., 1995). The kinetic measurements for this series of FA revealed linearity of the log of the rate constant with the chain length of the FA (Fig. 4). With the addition of two CH2 groups to a saturated chain, the rate constant decreased by about a factor of 10. From studies of stearic (18⬊0), oleic (18⬊1), and linoleic (18⬊2) acids, we found that the introduction of each double bond decreased the rate constant by an amount approximately equal to the removal of one CH 2 group. More recent studies of another group confirmed the dependence of desorption on unsaturation, and showed that for each degree of unsaturation, there was a linear dependence with chain length, as with the saturated chains. The rate constants for FA with less than 20 carbons after adjustment for unsaturation all correspond to half times of < 1 s. Thus, the rate of spontaneous desorption of the FA from a simple phospholipid vesicle is the rate-limiting step of the diffusion mechanism of membrane transport of FA. However,

Journal of Molecular Neuroscience

the half times are still fast for this group of FA, which includes the common dietary FA and the ω-3 and ω-6 FA, and there appears to be no a priori requirement for a protein to catalyze release of FA. In contrast, the saturated FA with a very long chain (20 carbons) exhibited slow half times of spontaneous desorption. Hexacosanoic acid, a marker for Adreno-leukodystrophy and Zellweger’s syndrome (Moser and Moser, 1989; Moser, 1997), desorbs 10,000 times slower than stearic acid. We have hypothesized that the slow desorption of the very long-chain saturated FA (VLCFA) contributes to deleterious effects in cell membranes (Ho et al., 1995). Our laboratory is addressing the questions of whether intracellular fatty acid binding proteins (FABP) can bind VLCFA and if so, whether FABP decrease the rate constants for desorption of VLCFA.

Passive Diffusion in Cells We have previously discussed in detail the implications for transport of FA in biological membranes that a simple protein-free phospholipid “model” membrane system provides (Hamilton, 1998, 1999; Hamilton and Kamp, 1999). The diffusion hypothesis has specific corollaries that can be tested in cells, some of which have been validated in earlier studies (Hamilton, 1999; Hamilton and Kamp, 1999). However, many investigators have remained unconVolume 16, 2001

Fatty Acid Diffusion Through Membranes vinced of the diffusion hypothesis, and the hypothesis of protein-mediated transport has attracted much attention. It is imperative to assess critically the evidence for these different mechanisms and to develop new approaches to discriminate between them. One new approach is to test in cells the flip-flop hypothesis of transmembrane diffusion of FA developed from model membrane studies. This model predicts that intracellular pH will decrease with inward flux of FA, and increase with outward flux. Such effects might be measured if the buffering capacity of the cytoplasm, and the exchange of H+ by active membrane processes do not negate the perturbations of cytosolic pH by FA flux. In spite of such potential problems, we showed that predicted pH effects are observed in pancreatic β-cells (Hamilton et al., 1994) and in rat adipocytes (Civelek et al., 1996) by entrapping a pH-sensitive fluorophore. The influx of FA (presented in the unbound form or bound to albumin) into the cytosol caused pH decreases. The efflux of FA to the external buffer caused pH increases. To address whether the observed pH changes are due to the diffusion mechanism or might be caused by protein-mediated co-transport of FA anions and H+, we investigated the effects of certain FA analogs on intracellular pH in adipocytes. The addition of tetradecyl (C-14) amine to the external buffer of a suspension of adipocytes resulted in a pH increase, which is explained by the flip-flop of the neutral form of the molecule followed by removal of H+ from the cytosol to reach ionization equilibrium on the inner leaflet of the plasma membrane. The addition of a FA dimer, consisting of two long-chain FA molecules linked covalently across the middle of each chain, caused an acidification that was not reversed by albumin, to which they do not bind readily. (These results are consistent with the predictions of the flip-flop model, and not by protein-mediated transmembrane transport.) Other new support for the flip-flop of FA in cell membranes has come from studies that manipulated the pH gradient across the plasma membrane of adipocytes (Trigatti and Gerber, 1996) and neutrophils (Mastrangelo, 1998). These new studies of diffusion of FAinto cells provide additional validation of this mechanism, but do not demonstrate its exclusivity. In our view the diffusion mechanism should be considered a working hypothesis. In fact, the measurement of intracellular pH reports the combined steps of adsorption and transmembrane movement. As mentioned earlier, because of the unstirred water layer and the presence of other membrane

Journal of Molecular Neuroscience

105 components such as the glycocalyx, it cannot always be assumed that the adsorption step is rapid. An important avenue of our current investigations is to relate the adsorption step to the transmembrane step. When FA is added to a cell: how fast is adsorption; does flip-flop follow immediately; is the amount of FAthat diffuses through the membrane proportional to the amount that binds to plasma membrane? One approach, as illustrated in Fig. 5, is to monitor simultaneously the partitioning of FA between the external buffer and the cell and the flip-flop by use of two independent fluorescent probes. The pH fluorophore BCECF was first incorporated into the cytosol of adipocytes, which were then suspended in a buffer with the fluorescent fatty acid binding protein ADIFAB (Fig. 5, right side). The fluorescence of ADIFAB reports the concentration of unbound FA in the external buffer at any given time, from which the rate and the quantity of FA partitioning into the cell membrane can be determined. The simultaneous responses of ADIFAB and BCECF fluorescence are shown in Fig. 5 (left panels). ADIFAB fluorescence intensity increased immediately upon addition of oleic acid to the suspension of adipocytes, signifying arrival of the oleic acid in the external buffer, and reached a maximal intensity within 2–3 s, the time resolution of the measurement. Our calculation of partitioning showed that > 99% of the added FA was bound to the membrane, and the concentration of unbound oleic acid was in the nM range, well below its aqueous solubility at pH 7.4. The second addition of twice the amount of oleic acid to cells resulted in about twice as much binding to cells. The concentrations in the membrane mM when expressed in terms of the volume of the membrane phase. The BCECF fluorescence shows that intracellular pH decreased to a minimal value within the time resolution of our measurement, and therefore that oleic acid began to arrive on the inner leaflet of the plasma membrane immediately after binding to the outer leaflet. These dual measurements in adipocytes show rapid binding to cells and rapid transmembrane movement by flip-flop. Further studies will investigate whether there is a linear relationship between the adsorption of FA to cells and intracellular pH changes, i.e., whether the diffusion mechanism is effective at both low and high concentrations of FA. Are studies of model membranes and cells as described above relevant to the brain and the BBB? The first barrier for transport of FA into brain cells consists of a layer of endothelial cells with tight junctions between adjacent cells. In principle, FA could

Volume 16, 2001

106

Hamilton et al.

Fig. 5. The right hand side illustrates the strategy of measuring simultaneously changes in internal pH (trapped BCECF) and extracellular FA concentrations (ADIFAB). The left hand side shows the changes in (OA) outside the cells (top) and internal pH (bottom) after addition of 20 and 40 nmol OA to epididymal fat cells at 100 s and 200 s, respectively. The BCECF is in the cytoplasm surrounding the lipid droplet (black circle).

cross the endothelial cell by diffusion processes: desorption from albumin in the blood, adsorption to the luminal side of the endothelial-cell membrane, flipflop, desorption, diffusion through the cytosol, and passive diffusion through the abluminal-surface membrane. Entry of FA from blood into other cells such as muscle cells and adipocytes also proceeds through endothelial cells. These endothelial cells do not have tight junctions, but passage of FA as monomers or as complexes with albumin through the endothelial clefts was considered not to be a viable mechanism for bulk transport of FA to cardiac muscle cells (van der Vusse et al., 2000). Moreover, diffusion of FA through the cytosol was evaluated to be the major mechanism for intracellular transport in cardiac endothelial cells, despite the fact that these cells contain very little intracellular FABP (Van Nieuwenhoven et al., 1994). Therefore, diffusion processes could, in principle also govern passage of FAthrough brain endothelial cells. One of the implications of the passive diffusion mechanism is that membrane and intracellular transport will not be highly selective for different longchain FA. Uptake (in the absence of metabolism) will

Journal of Molecular Neuroscience

depend on the partitioning of FAinto the membrane, which should be similar for various long-chain FA. After partitioning, the FA will spontaneously cross the membrane and desorb into the cytosol. This model is consistent with the uptake measurements by Rapoport and co-workers, who showed a high permeability of the brain to palmitic, arachidonic, and docosahexaenoic acid (Robinson et al., 1992). Their results are not necessarily in conflict with oral feeding studies showing little labeled palmitate after long periods (days; [Marbois et al., 1992]). Palmitic acid is mainly converted into esterified lipids, which have very low uptake into the brain (Robinson et al., 1992). Furthermore, if a specific FA enters the brain and is not utilized, it can leave quickly by the diffusion mechanism. The free diffusion mechanism and proteinmediated mechanisms of transport of FA through membranes should be considered alternative hypotheses, subject to further testing. Postulating that membrane transporters are present in the BBB could rationalize why certain FA are preferentially taken up by the brain, if this is true. However, to date there are no detailed descriptions of the molecular

Volume 16, 2001

Fatty Acid Diffusion Through Membranes properties of the putative transport proteins in other cells, except that some of them have enzymatic activity and one is an established receptor for lipoproteins (Hamilton and Kamp, 1999). Although it was originally proposed that protein transporters facilitated the transmembrane movement of FA, this was before recent work showing the spontaneous movement of un-ionized long-chain FA through membranes. It remains to be established what step of membrane transport, if any, is catalyzed by a specific protein. Furthermore, there is no model for highly selective uptake of FA into cells (other than by metabolism), as the putative transporters have low specificity for particular FA (Abumrad et al., 1998). Other known FABPs (albumin, intracellular FABP, and the nuclear peroxisomal proliferator activated receptor [PPARγ]) bind a wide range of FA (Spector, 1975; Paulussen and Veerkamp, 1990; Xu et al., 1999). Clearly, studies of FA uptake into brain endothelial cells and other brain-cell types need to be performed to determine mechanisms of uptake. The diffusion hypothesis can be tested by intracellular pH measurements as previously applied to adipocytes and to β-cells (Hamilton et al., 1994; Cirelek et al., 1996; Fig. 5). On the basis of general parameters governing passive diffusion of a wide range of drugs across the BBB (Fischer et al., 1998), we would predict that all common dietary FA would be able to diffuse across the BBB. Nevertheless, different FA (without attached probes or isotope labeling) can be tested on individual cell types to determine if certain FA are not able to enter brain cells by passive diffusion. In studies of the uptake of FA into the brain an important challenge will be to delineate nonmetabolic steps of membrane transport very carefully from metabolic steps. By monitoring intracellular pH over a period of minutes, it is possible to follow intracellular metabolism of FA, and to discriminate FA that are slowly metabolized from those that are rapidly metabolized (Civelek et al., 1996).

References Abumrad N., Harmon C., and Ibrahimi A. (1998) Membrane transport of long-chain fatty acids: evidence for a facilitated process. J. Lipid Res. 39(12), 2309–2318. Cabral D. J., Small D. M., Lilly H. S., and Hamilton J. A. (1987) Transbilayer movement of bile acids in model membranes. Biochemistry 26(7), 1801–1804. Civelek V. N., Hamilton J. A., Tornheim K., Kelly K. L., and Corkey B. E. (1996) Intracellular pH in adipocytes:

Journal of Molecular Neuroscience

107 effects of free fatty acid diffusion across the plasma membrane, lipolytic agonists, and insulin. Proc. Natl. Acad. Sci. USA 93(19), 10,139–10,144. Daniels C., Noy N., and Zakim, D. (1985) Rates of hydration of fatty acids bound to unilamellar vesicles of phosphatidylcholine or to albumin. Biochemistry 24(13), 3286–3292. Darnell J., Lodish H., and Baltimore D. (1990) Molecular Cell Biology. Scientific American Books, Inc., New York. Dietschy J. M. (1978) General principles governing movement of lipids across biological membranes, in Disturbances in Lipid and Lipoprotein Metabolism (Dietschy J. M., Gotto J., Ontko A. M., and Ontko J. A., eds.), American Physiological Society, Bethesda, MD, pp. 1–29. Fischer H., Gottschilch R., and Seelig A. (1998) Blood-brain barrier permeation: Molecular parameters governing passive diffusion. J. Membrane Biol. 165, 201–211. Hamilton J. A., Civelek V. N., Kamp F., Tornheim K., and Corkey B. E. (1994) Changes in internal pH caused by movement of fatty acids into and out of clonal pancreatic beta-cells (HIT). J. Biol. Chem. 269(33), 20,852–20,856. Hamilton J. A. (1995) 13-C NMR studies of the interactions of fatty acids with phospholipid bilayers, plasma lipoprotein, and proteins, in Carbon-13 NMR Spectroscopy. (Beckman N., ed.), Academic Press, USA, pp. 117–157. Hamilton J. A. (1998) Fatty acid transport: difficult or easy? J. Lipid. Res. 39(3), 467–481. Hamilton J. A. (1999) Transport of fatty acids across membranes by the diffusion mechanism. Prostaglandins Leukot. Essent. Fatty Acids 60(5–6), 291–297. Hamilton J. A. and Kamp, F. (1999) How are free fatty acids transported in membranes? Is it by proteins or by free diffusion through the lipids? Diabetes 48(12), 2255–2269. Herman R., Leaf A., and Schwartz W. (1964) Diffusion of weak acids across the toad bladder. J. Gen. Physiol. 48, 379–389. Ho J. K., Moser H., Kishimoto Y., and Hamilton J. A. (1995) Interactions of a very long chain fatty acid with model membranes and serum albumin. Implications for the pathogenesis of adrenoleukodystrophy. J. Clin. Invest. 96(3), 1455–1463. Kamp F. and Hamilton J. A. (1992) pH gradients across phospholipid membranes caused by fast flip-flop of un-ionized fatty acids. Proc. Natl. Acad. Sci. USA 89(23), 11,367–11,370. Kamp F., Zakim D., Zhang F., Noy N., and Hamilton J. A. (1995) Fatty acid flip-flop in phospholipid bilayers is extremely fast. Biochemistry 34(37), 11,928–11,937. Kantor H. L. and Prestegard J. H. (1978) Fusion of phosphatidylcholine bilayer vesicles: role of free fatty acid. Biochemistry 17(17), 3592–3597. Marbois B. N., Ajie H. O., Korsak R. A., Sensharma D. K., and Edmond J. (1992) The origin of palmitic acid in brain of the developing rat. Lipids 27(8), 587–592. Mastrangelo A. M., Jeitner, T. M. and Eaton J. W. (1998) Oleic acid increases cell surface expression and activ-

Volume 16, 2001

108 ity of CD11b on human neutrophils. J. Immunol. 161(8), 4268–4275. Moser H. W. and Moser A. B. (1989) X-linked adrenoleukodystrophy, in The Metabolic Basis of Inherited Disease. (Scriver C. R., Beaudet A. L., and Sly W. S., eds.), McGraw Hill, New York, pp. 1511–1532. Moser H. W. (1997) Adrenoleukodystrophy: phenotype, genetics, pathogenesis and therapy. Brain 120(Pt 8), 1485–1508. Paulussen R. J. A. and Veerkamp J. H. (1990) Intracellular fatty acid binding proteins: characteristics and function, in Subcellular Biochemistry, vol. 16 (Hilderson H. J., ed.), Plenum Press, New York, pp. 175–226. Richieri G. V., Ogata R. T., and Kleinfeld A. M. (1992) A fluorescently labeled intestinal fatty acid binding protein. Interactions with fatty acids and its use in monitoring free fatty acids. J. Biol. Chem. 267(33), 23,495–23,501. Robinson P. J., Noronha J., DeGeorge J. J., Freed L. M., Nariai T., and Rapoport S. I. (1992) A quantitative method for measuring regional in vivo fatty- acid incorporation into and turnover within brain phospholipids: review and critical analysis. Brain Res. Brain Res. Rev. 17(3), 187–214. Spector A. A. (1975) Fatty acid binding to plasma albumin. J. Lipid Res. 16(3), 165–179. Spector A. A. (1986) Plasma albumin as a lipoprotein, in Biochemistry and Biology of Plasma Lipoptotein. (Scanu,

Journal of Molecular Neuroscience

Hamilton et al. A. M. and Spector, A. A., eds.), Marcel Dekker, Inc., New York, pp. 247–279. Stremmel W., Strohmeyer G., and Berk P. D. (1986) Hepatocellular uptake of oleate is energy dependent, sodium linked, and inhibited by an antibody to a hepatocyte plasma membrane fatty acid binding protein. Proc. Natl. Acad. Sci. USA 83(11), 3584–3588. Trigatti B. L. and Gerber G. E. (1996) The effect of intracellular pH on long-chain fatty acid uptake in 3T3-L1 adipocytes: evidence that uptake involves the passive diffusion of protonated long-chain fatty acids across the plasma membrane. Biochem. J. 313(Pt 2), 487–494. van der Vusse G. J., van Bilsen M., and Glatz J. F. C. (2000) Cardiac fatty acid uptake and transport in health and disease. Cardiovasc. Res. (45), 279–293. Van Nieuwenhoven F. A., Verstijnen C. P., Van Eys G. J., Van Breda E., De Jong Y. F., Van der Vusse G. J., and Glatz J. F. (1994) Fatty acid transfer across the myocardial capillary wall: no evidence of a substantial role for cytoplasmic fatty acid-binding protein. J. Mol. Cell. Cardiol. 26(12), 1635–1647. Xu H. E., Lambert M. H., Montana V. G., Parks D. J., Blanchard S. G., Brown P. J., et al. (1999) Molecular recognition of fatty acids by peroxisome proliferatoractivated receptors. Mol. Cell 3(3), 397–403. Zhang F., Kamp F., and Hamilton J. A. (1996) Dissociation of long and very long chain fatty acids from phospholipid bilayers. Biochemistry 35(50), 16,055–16,060.

Volume 16, 2001