Characterization of FadL-specific fatty acid binding in Escherichia coli

97

Biochimica et Biophysics Acta, 1046 (1990) 97-105 Elsevier

BBALIP

53462

Characterization

of FadL-specific fatty acid binding in Escherichia coli Paul N. Black

Department of Biochemisiry, College of Medicine, University of Tennessee, Memphis, Memphis, TN (U.S.A.)

(Revised

Key words:

Fatty acid transport:

(Received 5 January 1990) manuscript received 5 April 1990)

Fatty acid binding

protein;

Fatty acid transfer;

(E. cob)

The product of the fudL gene (FadL) is a central component of the long-chain fatty acid transport system of Escherichiu coli. When fatty acid activation is blocked by a mutation in the structural gene for acyl CoA synthetase (f&D) transport is inhibited allowing a FadL-specific fatty acid binding activity to be measured. This binding activity was 4- to i-fold greater in the fadL+fadD strain LS6928 when compared to the A fadLfadDstrain LS6929. With long-chain fatty acids, this binding activity was saturable and it was estimated that there were approx. 35000 FadL-specific oleic acid binding sites per cell in the find + strain LS6928. The FadL-specific fatty acid binding affinity was highest for oleic acid (18: 1) and palmitic acid (16: 0) giving apparent K, values of 2.3 10 -’ M and 8.8 10 -’ M, respectively. FadL-specific binding affinity of myristic acid (14:0) was nearly an order of magnitude less and no FadL-specific binding of decanoic acid (10: 0) could be measured. Two lines of evidence suggest that FadL-fatty acid binding occurs by a hydrophobic interaction: (1) There was a preference for the long-chain substrates oleic acid and palmitic acid; and (2) oleic acid binding activity was not significantly changed over the pH range 5.0 to 8.0. The FadGspecific binding of oleic acid in the fudL+ strain LS6928 could be blocked by preincubation with antisera raised against purified FadL providing a clear correlation between the activity and identity of FadL. The binding activity associated with FadL was measured in vesicles of the outer membrane following passage over the hydrophobic resin Lipidex 1000. The K, of oleic acid binding attributable to FadL in outer membrane vesicles (6.0 10-' M) was in close agreement with that determined in whole cells. Overall, these studies demonstrated that FadL binds long-chain fatty acids with a relatively high affinity prior to their transport across the outer membrane. l

l

l

Introduction

chain fatty acids (C,,-C,,)

traverse the cell envelope of saturable process [l-4]. The mechanism of long-chain fatty acid transport is not well understood but requires at least the products of the fadL and fadD genes. The product of the fadL gene (FadL) is localized in the outer membrane [6] and is essential for the binding and transport of exogenous long-chain fatty acids [l-4]. This protein also functions as a receptor for the bacteriophage T2 [7]. The fadL gene has been cloned and its protein product purified and preliminarily characterized [5-71. FadL spans the outer membrane [7] and we propose this protein forms a channel specific for long-chain fatty acids. Following transfer across the outer membrane, these molecules traverse the periplasmic space and inner membrane by an undefined process that may involve a periplasmic fatty acid binding protein and/or an inner membrane fatty acid permease. Concomitant with transport, exogenous long-chain fatty acids are delivered to, acyl CoA synthetase (fadD gene product) or acyl-acyl carrier E. coli by a specific, high-affinity,

The acquisition of metabolically valuable nutrients by Gram-negative bacteria such as Escherichia coli requires the passage of these molecules through the cell envelope. In many instances, this process requires proteins found within the outer membrane, periplasmic space and inner membrane. The presence of specific proteins for a given transport system in each compartment of the cell envelope provides the cell with an efficient and remarkably refined system for acquiring nutrients from the environment. These transport systems are highly diversified allowing the specific transport of vitamins, amino acids, nucleosides and sugars. As a class of metabolically valuable nutrients, long-

Correspondence: P.N. Black, Department of Biochemistry, of Tennessee, Memphis, 800 Madison Avenue, Memphis, U.S.A. 00052760/90/$03.50

0 1990 Elsevier Science Publishers

University TN 38163,

B.V. (Biomedical

Division)

98 protein synthetase (acyl-ACP synthetase) [g-11]. In the former case, long-chain fatty acids are derivatized to CoA thioesters which can be broken down by cyclic P-oxidation or incorporated into phospholipids via glycerol-3-phosphate acyltransferase. In the latter case, long-chain fatty acids are incorporated exclusively into the l-position of phosphatidylethanolamine [lo]. The major route of entry for exogenous long-chain fatty acids into the intracellular metabolic pools occurs via acyl CoA synthetase (approx. 98%) although entry via the acyl-ACP synthetase is significant (approx. 2%) [lo]. The requirement for metabolic energy in the transport of long-chain fatty acids is likely due to the activity of the acyl CoA synthetase. FadL is required whether exogenous long-chain fatty acids are derivatized to CoA thioesters via acyl CoA synthetase or incorporated into phosphatidylethanolamine via acyl-ACP synthetase [lO,ll]. Furthermore, mutants containing a fudL lesion have acyl CoA synthetase, palmitoyl CoA dehydrogenase. enoyl CoA hydratase, /3-hydroxy acyl CoA dehydrogenase and /3-keto CoA thiolase levels comparable to wild-type and can P-oxidize long-chain fatty acids in vitro [3]. Collectively these data demonstrate that FadL is a central component of the long-chain fatty acid transport apparatus. Our goal is to define the biochemical events that govern the transport of long-chain fatty acids across the cell envelope of E. co/i. The long-chain fatty acid oleate (18 : 1) binds saturably and reversibly to the cell of fudL+fudD fadR strains [7] yet cannot be transported or P-oxidized [3]. We interpret these data to suggest that FadL acts, in part, to specifically sequester long-chain fatty acids at sites on the outer membrane prior to transport and metabolic utilization. This interpretation is based on our studies demonstrating the outer membrane localization of FadL [6] and is in contrast to earlier studies which claim this protein resides in the inner membrane and acts as a fatty acid permease [2-5,121. The outer membrane localization of FadL has been confirmed more recently in our laboratory by demonstrating this protein (1) acts as a receptor for the bacteriophage T2, (2) is, like OmpF, associated with the peptidoglycan layer, and (3) is proteinase-sensitive in whole cells [7]. As much of the initial work describing the fadL gene was carried out in the context of an incorrect membrane localization of FadL [2-4,121, it has become important to re-examine and reevaluate these earlier data. There are two central questions concerning FadL that must be addressed: (1) how does this protein bind long-chain fatty acids prior to transport?; and (2) how does this protein mediate the passage of long-chain fatty acids across the outer membrane? The present study was undertaken in order to address the first question by characterizing the fatty acid binding properties attributable to FadL in order to understand how

this outer membrane protein specifically binds longchain fatty acids from the extracellular milieu. This work is especially important as the binding of hydrophobic compounds to the outer membrane has received little experimental attention. This is in contrast to studies involving hydrophilic compounds, many of which have been extensively studied. In this report, we describe equilibrium binding constants and specificity of FadL towards fatty acids as binding substrates within whole cells blocked at the level of fatty acid activation. Furthermore, we demonstrate that the long-chain fatty acid binding activity attributable to FadL can be specifically blocked with a polyclonal antibody generated against this protein and can be measured in vesicles generated from the outer membrane.

Methods Bacterial strains and growth conditions The bacterial strains used in this study were RS3010 (fadR), LS6164 (fadR A fadLS), LS6928 (fadR fadD27 zea::TnlO) and LS6929 (fadR AfadL5 fadD27 zea: :TnlO) [4,13]. Cells were routinely grown to mid-log phase (5 * 10’ cells/ml) in tryptone broth (TB) [14] or TB supplemented with 5 mM oleate (18 : 1) and 0.5% Brij 58 at 37°C in a Lab Line gyratory shaker. Strains LS6928 and LS6929 contained fudR mutation to maintain constitutive expression of the fad genes as strains harboring fadD mutation are not inducible by longchain fatty acids [15]. Growth was monitored with a Klett-Sumerson calorimeter using a blue filter. Fatty acid binding assays Two different experimental regimes were employed to evaluate the long-chain fatty acid binding capacity in strains LS6928 (AfadL fadD) and LS6929 (fadD): (1) cells were grown to mid-log phase in TB, harvested, washed once with minimal medium M9 (14) and resuspended in M9 containing 100 pg/ml chloramphenicol to a final cell density of 1 . lo9 cells/ml. The resuspended cells were starved for a carbon source with shaking at 30°C for 15 min prior to assay. For assay, cells (1 ml) were diluted into a 2 x reaction buffer containing 2 pM [3H]oleic acid, 0.33 pm fatty acid free bovine serum albumin (BSA) in M9 containing 100 pg/ml chloramphenicol. The concentration of BSA was varied as specified in the figure legends to give free fatty acid/ bovine serum albumin (FFA/BSA) ratios of 0.5 to 16. Cells were incubated 15 min at 30°C to allow maximal fatty acid binding. Following incubation, cells were collected by centrifugation (3400 X g, 15 min, room temperature), washed once with M9 and resuspended in 2 ml M9. Triplicate samples (100 ~1 each) were taken and dpm determined. Protein concentration was estimated using the method of Lowry et al [16] on total sonicated cell extracts. Data were converted into pmol

99 fatty acid bound/mg cell protein; and (2) a second method was employed in which the BSA was omitted in order to define FadL-specific fatty acid binding parameters. Cells were grown, harvested and washed as in (1). Cells were resuspended to a density of 1 *log cells/ml in M9 containing 100 pg/ml chloramphenicol and starved at 30°C for 15 min. Following starvation, cells were diluted into a 2 X reaction buffer containing 3Hlabelled fatty acid (at concentrations varying from 10 nM to 2000 nM) and incubated for 15 min at 30°C with gentle shaking. Cells were collected and washed once as in (1). Final cell pellets were resuspended in 2 ml and dpm determined in triplicate (100 ~1 each). Data were converted to pmol fatty acid bound/mg cell protein and analyzed using the statistical analysis of enzyme kinetic data programs described by Cleland [17]. All hyperbolic binding data were fit to the following equation:

where b = fatty acid bound at fatty acid concentration L, &, = maximal fatty acid binding (binding capacity), and Ko = equilibrium binding constant. All data presented represent the average of at least three independent experiments. Fatty acid competition

experiments

[3H]Oleic acid binding was determined in the presence of unlabeled oleic acid essentially as described as method (2). [3H]Oleic acid was held constant at 200 nM and competing fatty acids (unlabeled) varied from 0 to 2000 nM. Maximal FadL-specific fatty acid binding in the absence of competing fatty acid was 225 + 50 pmol oleate/mg protein and was defined as 100% binding (see Results). All data were expressed as percent maximal oleic acid bound and plotted versus concentration of competing fatty acid. Antibody

blocking experiments

Oleic acid binding to fadL+ceUs preincubated with preimmune IgG or immune IgG (anti-FadL) [6] was determined as detailed in fatty acid binding method (1). Cells were grown to mid-log phase, harvested, washed and resuspended in M9 containing 0.5 PM fatty acid free BSA and chloramphenicol at 100 pg/ml to a final cell density of 1 * lo9 cells/ml. Anti-FadL antiserum (enriched for IgG by ammonium sulfate fractionation; 18) was added and the samples incubated for 45 min at 30 o C with gentle shaking. After preincubation, a 2 x reaction buffer (see method (1)) was added giving a final FFA/BSA ratio of 6 and incubation continued for 15 min. Samples were centrifuged, cells washed with M9 and total oleate bound determined as detailed above as fatty acid binding method (1). Control experiments

used: (1) the AfadL strain LS6929; (2) preimmune antibody; and (3) no antibody. Membrane

isolation and fractionation

Cells were routinely grown to mid-log phase from an overnight culture in TB supplemented with 5 mM oleate (18 : 1) and 0.5% Brij 58. Following growth, cells were harvested, washed in minimal salts and resuspended in 0.75 M sucrose, 10 mM dithiothreitol (DTI’), 10 mM Hepes (pH 7.4) at a cell density of 5 . lo9 cells/ml. Lysozyme was added to 0.5 mg/ml, EDTA to 10 mM and DNaseI to 10 pg/rnl. Spheroplasts were formed at room temperature (45-60 min), cooled on ice and lysed by sonication (3 x 15 s cycles). Lysed samples were clarified by a 10000 x g 10 min centrifugation to remove unbroken cells. Total cell envelope was isolated from the clarified extract at 250000 x g, resuspended by gentle sonication in 10 mM Hepes (pH 7.4) 10 mM DTT, 5 mM MgCl,, applied to a discontinuous sucrose gradient (70% : 53% : 17%) and centrifuged at 107 000 X g for 14-16 h [19]. Outer and inner membrane fractions were collected with a 3 ml syringe and pelleted at 250000 x g for 1 h. The purity and composition of the membrane fractions was monitored using NADH oxidase activity (as a marker for the inner membrane) [19] and measurement of 2-keto-3-deoxyoctanoic acid (KDO) levels (as a marker for the outer membrane) [20]. Vesicles of the outer membrane were tested for long-chain fatty acid binding activity as described below. Long-chain fatty outer membrane

acid binding

activity

in vesicles of the

Following fractionation from sucrose gradients, outer membranes were pelleted at 250000 X g and resuspended in 10 mM Hepes (pH 7.4) 5 mM MgCl,, 85 mM NaCl by sonication to a final protein concentration of 7-10 mg/ml (estimated using the method of Lowry et al [16]). Outer membrane vesicles were incubated at 37°C for 5 min at varying concentrations of [3H]oleate and protein as described in the figure legends. Following incubation, the reaction mixture was cooled to 0°C in a salt-ice bath for 2 min and centrifuged through a 10 mm X 15 mm Lipidex 1000 column equilibrated with the same buffer at 0°C for 4 min at 3500 X g to separate bound and free [3H]oleate. The amount of [3H]oleate in the eluate, representing the oleate bound to the outer membrane vesicles, was determined by scintillation counting. Data were transformed into pmol fatty acid bound/mg protein and analyzed as described above for whole cells. All data presented represent the average of at least three independent experiments. SDS polyacrylamide blotting

gel electrophoresis

and

immuno-

SDS polyacrylamide gels were prepared and run according to Laemmli [21]. Immunoblotting was done

100 essentially as described by Bumette [22] as modified by Black et al [6]. Materials

[3H]Oleic acid (10.0 Ci/mmol), [3H]palmitic acid (30.0 Ci/mmol) and [3H]myristic acid (39.3 Ci/mmol) were purchased from New England Nuclear. [14C]Decanoic acid (9.9 Ci/mmol) was purchased from Sigma. Lipidex 1000 was obtained from Packard. Antibiotics and other supplements for bacterial growth were obtained from Difco and Sigma. All other chemicals were obtained from standard suppliers and were of reagent grade. Results FadL as a long-chain fatty acid binding protein

Previous work [4] demonstrated FadL acts as a longchain fatty acid binding protein when fatty acid utilization is blocked by mutation (fadD). We confirmed this observation by measuring oleic acid binding at FFA/BSA ratios of 3 and 6 in strains LS6928 (fadL+ fadD) and LS6929 (AfadL fadD). Under these conditions the fadL+ strain bound nearly 4-times more fatty acid than the isogenic AfadL strain. These data suggested that FadL was binding oleate and/or allowing oleate to be transported across the outer membrane. As one of the primary goals of the present work was to

0

Iuo

200

300

4cm

500

e 0

[OLEIC ACID] (nM) Fig. 2. FadL-specific oleic acid binding using fatty acid binding method (2). Closed circles represent the fudL+ strain LS6929 and closed boxes represent the fadL strain LS6928; open circles represent FadL-specific binding (refer to text for details).

4w

determine the FadL-specific fatty acid binding capacity associated with the fadL+ strain, the FFA/BSA was varied from 0.5 to 16 to determine whether oleic acid bound saturably (Fig. 1). Under these conditions a maximal oleic acid binding of approx. 300 pmol/mg protein occurred at a FFA/BSA ratios of 6 and higher. The conclusion for these data was that there are a finite number of oleic acid binding sites on the cell envelope in fadL+ strains.

300

200

Kinetics of FadL-specific long-chain fatty acid binding

loo

5

10

IS

20

OLEATEIBSA Fig. 1. Oleic acid binding in the f&Z.+ strain LS6928 as a function of increasing free fatty acid/bovine serum albumin ratios (FFA/BSA).

Data from experiments described above provided clear evidence that FadL acted as a long-chain fatty acid binding protein. These data were difficult to interpret as the precise concentration of free fatty acid at the different FFA/BSA ratios could not be accurately determined. We decided to employ an alternative fatty acid binding regime in the absence of BSA (method [2]]. Data presented in Fig. 2 demonstrated that oleic acid bound both fadL+ and AfadL strains. The fadL+ strain bound approximately 3- to 4-fold more fatty acid over the range of fatty acid concentrations used. The fatty acids bound by the AfadL strain was appreciable, and in fact at a substrate concentration of 500 nM, bound approx. 250 pmols oleic acid/mg protein. We

101 TABLE

I

Equilibrium

binding constants of FadL-specific fatty acid binding a

102

Fatty

acid

K,

(M)

(SE (M)) b

Binding c

(3.5.10-7) (l.l.lO-‘) (5.1.10-s)

d 1505 (+280) 610 (583)

c

C 10 C 14 C 181Pl

1.2.10-6 8.8.10-7 2.3.10-’

capacity

(pmol/mg)

SO

a Binding constants have been correct to account for the non-specific binding measured in the fadL strain. b S.E. - standard error (see Ref. 16). ’ No detectable binding observed. d Calculated value exceed the solubility of myristic acid under aqueous conditions.

60

4(

2( 1

10

[COMPETING

1cMl

loo0

lowI

OLEIC ACID](nM)

Fig. 3. Competition of FadL-specific [3H]oleic acid binding by unlabeled oleic acid. [3H]Oleic acid was held constant at 200 nM. Data has been corrected to account for the non-specific binding in the fudL strain LS6929.

interpret the binding of fatty acids to the AfudL strain to represent background and not a residual FadLspecific fatty acid binding activity. By subtracting the level of fatty acid binding measured in the AfudL strain from the level of fatty acid binding measured in the fadL strain, a FadL-specific oleic acid binding profile was obtained (Fig. 2). Maximal oleic acid binding occurred at approx. 600 pmol/mg protein and gave an apparent equilibrium binding constant (K,) of 2.33 . 10e7 M oleate. These data suggest there were approx. 35 000 oleic acid binding sites per cell in the fudL+ strain. Bound oleate could be removed from the fudL+ strain by single washes with 50 PM BSA or 0.5% Brij 58 in medium M9 following binding demonstrating that oleic acid binding was reversible (data not shown). Furthermore, competition experiments, in which the concentration of [ 3H]oleic acid was held constant at 200 nM while the concentration of the unlabeled ligand was varied from 10 nM to 2000 nM, illustrated that the binding of [3H] oleic acid could be competed out with unlabeled ligand. These experiments confirmed the approximate K, of FadL-specific oleate binding (Fig. 3). Specificity of FadL-specific fatty acid binding The FadL-specific binding of other long-

medium-chain

and fatty acids was measured with [3H]pal-

mitic acid (16 : 0), [3H]myristic acid (14 : 0), and [‘4C]decanoic acid (10 : 0). As was the case with oleate, both palmitate and myristate showed FadL-specific binding (Table I). These results were interesting in that palmitate bound with a slightly lower affinity than oleate giving an apparent K, of 8.84 . 10V7 M but with an increased binding capacity. With myristate, there was a decrease in affinity by nearly an order of magnitude (Table I). The FadL-specific binding of decanoate (C,,) could not be measured due to low levels of fatty acid binding in both fadL+ and fadL strains. pH dependence of FadL-specific oleic acid binding

The FadL-specific binding of fatty acids showed a preference for oleate and palmitate suggesting this process may occur by a hydrophobic interaction. A series of experiments were performed that evaluated oleic acid binding over a range of pH. Data from these experiments (Table II) demonstrated that between a pH of 5.0 to 8.0, the FadL-specific binding of oleic binding changed only slightly. As the changes in oleic acid binding were small over this range of pH, these data may be more consistent with a hydrophobic interaction rather than an ionic interaction between FadL and the fatty acid.

TABLE

II

FadL-specific PH

oleic acid binding as (I junction of pH a Oleic acid binding @mol/min per mg protein)

( f S.D.) 8.0 7.5 7.0 6.5 6.0 5.5 5.0

157.6 130.9 145.1 181.2 212.6 200.4 173.5

(22.5) (24.1) (18.4) (31.7) (24.7) (13.2) (28.3)

a Binding measured using 100 nM oleic acid; data presented for non-specific binding measured in the fudL strain.

accounts

) i

50

ANTIBODY

1W

I50

ADDED

2w

(~1)

Fig. 4. Blocking activity of anti-FadL. Experiments performed as detailed for method (1) following preincubation of cells with antibody generated against FadL or normal rabbit serum; open circles and strain preincubated with anti-FadL or squares represent the f&L+ preimmune serum, respectively; closed circles and squares represent the fadL strain preincubated with anti-FadL or preimmune serum, respectively. Protein concentration in antibody samples was 9 mg/ml.

Blocking of oleic acid binding using anti-FadL FadL has been purified to homogeneity and preliminarily characterized [6]. In addition mono-specific polyclonal antisera have been raised against the purified protein [6]. This antisera recognizes only FadL on immunoblots, therefore, we were interested in determining whether it could block the FadL-specific binding of oleic acid. Experiments using fadL+ and AfadL+ strains and preimmune IgG or immune IgG at different concentrations demonstrated a FadL blocking activity by anti-FadL (Fig. 4). The blocking activity of anti-FadL illustrated that FadL was responsible for binding the long-chain fatty acids. Furthermore, these data provided a clear correlation between the FadL protein and its function within the outer membrane. Analysis of long-chain fatty acid binding activity in outer membrane vesicles In an effort to understand the role of FadL as a component of the long-chain fatty acid transport machinery, it was necessary to develop an in vitro system to measure the FadL-specific long-chain fatty acid binding activity. Glatz and Veerkamp [23] showed that Lipidex 1000 can effectively remove free fatty acids from solu-

tion at 0°C. It was reasonable to predict that a longchain fatty acid binding activity associated with outer membrane vesicles could be measured by separating the vesicular [ 3H]oleate and free [ 3H]oleate by passage over a Lipidex 1000 column at 0°C. Long-chain fatty acid binding was measured in outer membrane vesicles isolated from fadL+ and fadL strains. The outer membrane vesicles from fadL strains served as a negative control in these experiments. Outer membrane fractions, separated using sucrose gradients, were essentially free of inner membrane contamination (Table III). As a first step for using Lipidex 1000 as an experimental regime, it was necessary to validate the technique. Of particular importance was: (1) the ability to recover all vesicle material applied to the Lipidex column; and (2) the ability of the Lipidex column to bind free [3H]oleate at 0°C. As a measure of vesicle material, total protein was determined on material applied to and recovered from the Lipidex column. Data obtained using this approach demonstrated the same amount of protein applied to the column was eluted (using increasing concentrations of outer membrane vesicle protein). Therefore, no vesicle material bound to the column. A second test that was performed was the retention of free fatty acids in the Lipidex columns at 0°C. [3H]Oleate was applied to the column (25-500 pmol fatty acid) equilibrated at 0°C and the column centrifuged 4 min at 0°C. The eluate was counted and the total pmol oleate recovered was determined. Data from these experiments demonstrated only 0.8 to 2.8 pmol of fatty acid could be recovered. In fact, these experiments demonstrated that well over 90% of [3H]oleate applied to the column was retained at 0°C. Since there was nearly complete recovery of protein from and a nearly complete binding of free [3H]oleate to the Lipidex 1000, we were confident with this experimental approach. Outer membrane vesicles were prepared from and fadL+ strains as described in Methods with the final protein concentration adjusted to 7-10 mg/ml. Outer membrane vesicles, containing 50-250 pg total protein were incubated in a reaction mixture containing 500 nM [3H]oleate for 3 min, cooled to 0°C

TABLE

III

Characteristics of inner (IIU) and outer (OM) membrane fractions from Sucrose gradients IM

NADH oxidase KDO b FadL ’

a

OM

fadL

fadL+

fadL

fadL+

1.71 19 _

1.81 14

0.11 231

0.16 218 +

L Activity expressed in pmol/min per mg protein. 2-keto-3-deoxyoctanoic acid levels (pg/mg protein). ’ FadL detected by immunoblotting.

103 Discussion

100

0

150

200

[PEOTE~I @a) C

BM)

1B

0

2co

4co

6w

8M)

IZco

[OLEIC ACID](nM) Fig. 5. (A) Protein concentration dependence of oleic acid binding in outer membrane vesicies generated from the fadL+ strain RS3010 (open circles) and the _&z&C. strain LS6164 (closed circles). (B) FadLspecific binding of oleic acid to outer membrane vesicles as a function of oleic acid concentration: refer to text for details.

in a salt-ice bath and passed over Lipidex columns equilibrated at 0°C. Data from experiments using vesicles of the outer membranes from fudL+ and AfudL strains are illustrated in Fig. 5A. It was clear that there was a long-chain fatty acid binding activity specifically associated with the outer membrane vesicles from the fadL+ strain RS3010. The level of fatty acid binding to outer membrane vesicles generated from the AfadL strain LS6164 may have been the result of partitioning into the vesicle and thus provided an ideal internal control to monitor background level of fatty acid binding using this experimental system. In an effort to clarify the FadL-specific binding of long-chain fatty acids, outer membrane vesicles from and fadL fudL+ strains were prepared and incubated with increasing concentrations of [3H]oleate (100-1000 nM). These data demonstrate that the FadL-specific binding of oleate in outer membrane vesicles occurred by a saturable process with an apparent K, of 6.0. lo-’ M (Fig. 5B).

The requirement for a functional fadL gene in the transport of long-chain fatty acids was established by Nunn and co-workers [l-3]. These studies demonstrated the fadl-specific transport system is most specific for myristic acid with a K, of 4.1. 10e5 M (by comparison of K, values for C,, of 6.7 - 10V5, C,, of 7.6. 10e5 M and 18 : 1 of 13.2 * 1O-5 M) [3]. Mediumchain fatty acids (Cs-C,,) enter the cell via a FadL-independent method and with a decrease in affinity [3]. Although these studies provide a solid kinetic background describing fatty acid transport, they did not distinguish the contribution of the fadL gene product independent of fatty acid utilization. Nunn et al [4] demonstrated that the activity of the fadL gene product could be evaluated in strains blocked for fatty acid activation (fadB; acyl CoA s~thet~e-deficient). These studies demonstrated the long-chain fatty acid oleate bound saturably and reversibly to fudL+ cells. These studies, however, did not address the question of affinity or specificity of FadL towards fatty acids as binding substrates. Furthermore, the conclusions reached from these initial studies were that the fadL gene encoded an inner membrane fatty acid permease that, in fadD strains, acted to sequester long-chain fatty acids at specific sites on the inner membrane prior to transport [4]. We have demonstrated that FadL is a transmembrane protein of the outer membrane and not an inner membrane fatty acid permease [6,7]. Given our recent data on the characterization of the fudL gene product, we reinterpret these earlier data to suggest that FadL acts to specifically allow long-chain fatty acids to traverse the outer membrane. In fadD strains, FadL acts as a long-chain fatty acid binding protein of the outer membrane. The present study was undertaken in an effort to understand the long-chain fatty acid binding activity of FadL in the outer membrane. In particular, we were interested in: (1) defining the fatty acid binding capacity and equilibrium binding constants of FadL; and (2) determining whether this activity could be measured in outer membrane vesicles. The fatty acid binding activity attributable to FadL was initially measured in whole cells using an assay regime where oleate was dissolved in fatty acid free bovine serum albumin (giving FFA/BSA ratios of 0.516). Prior to assay, the cells were starved in the presence of 100 ag/ml chloramphenicol to block protein synthesis and the incorporation of exogenous palmitate into phosphatidylethanolamine via acyl-ACP synthetase [ll]. Under these conditions, the activity of FadL could be measured independently of the acyl CoA synthetase and the acyl ACP synthetase. The data from these experiments showed oleate binding in the fudL+ strain was 4-fold greater than in the AfadL strain. This binding

104 activity was maximal at approx. 300 pm01 oleate/mg protein at FFA/BSA of 6. An accurate assessment of the FadL-specific fatty acid binding could not be determined in a system containing BSA as a fatty acid carrier. Therefore, fatty acid binding was measured in a BSA and detergent-free system (at fatty acid concentrations ranging from 10 to 1000 nM). Under these conditions a FadL-specific long-chain fatty acid binding activity was measured giving an apparent K, of 2.33. lo-’ M oleate and a binding capacity of approx. 600 pmol oleate/mg protein. To obtain these values, the levels of fatty acid binding in the AfudL strain were subtracted from the levels of fatty acid binding in the fadL+ strain. Although there was oleate binding to the AfudL strain, we have interpreted this as background. This interpretation was based on several criteria: (1) the AfudL strain used in this study represents a deletion of the entire gene [4,12] and thus would not result in residual FadL-binding activity; (2) fudL strains in general fail to take up long-chain fatty acids whether they are destined for P-oxidation or incorporation into phospholipids [2,3,8, 10,111; (3) the FadL-specific binding of long-chain fatty acids is reversible; (4) the level of oleate binding in the AfudL strain increased more linearly over the range of concentrations used; and (5) 10 to 20% of the residual binding was due to fatty acids sticking to the borosilicate glass tubes. Competition experiments, in which labeled oleic acid was held constant at 200 nM while the concentration of unlabeled oleic acid was varied, verified the equilibrium binding constant of approx. 2. lo-’ M oleate and verified the reversible nature of FadLspecific fatty acid binding. The substrate specificity of the FadL-specific binding showed a preference for oleic acid (18 : 1) followed by palmitic acid (C,,) and myristic acid (C,,). Although the binding capacity of palmitate was higher than for oleate, the respective affinities for these two fatty acids were of the same order of magnitude. We suspect the lower affinity for myristic acid may be due to the length of the hydrocarbon tail. No FadL-specific binding of decanoic acid (C,,) was detected. These data are only in partial agreement with those of Maloy et al [3] who determined the kinetic constants for fatty acid transport. Like transport, FadL-specific fatty acid binding has highest affinity towards long-chain fatty acid substrates. The substrate specificities for long-chain fatty acid binding were, however, somewhat different, having greatest affinity for oleic acid rather than myristic acid [3]. Maloy et al [3] demonstrated that a fadL+fudR strain transports the medium-chain fatty acid decanoate at a higher rate than a fadL fudR strain, indicating that FadL has some affinity for medium-chain fatty acids. The level of decanoate transport in fudL AfudL strains described by Maloy et al [3] is appreciable and as a consequence may mask a medium-chain fatty acid bind-

ing activity attributable to FadL. This is particularly plausible as the concentration of fatty acid substrates in the transport assays are several orders of magnitude greater than those used in the binding assays. The present study measured fatty acid binding as opposed to utilization and may be more reflective of the specificity of FadL rather than other components of the fatty acid transport or metabolic system. Consistent with the idea that the binding of longchain fatty acids to FadL occurred by a hydrophobic interaction between the protein and the hydrocarbon tail of the fatty acid were the higher affinities for oleate and palmitate. When FadL-specific oleic acid binding was measured over a range of pH, only slight changes in oleic acid binding were observed. At the more acidic pH values, oleic acid binding was 1.5-fold greater than the level seen at physiological pH. While we can not rule out that these data reflect a slight ionic interaction between FadL and the long-chain fatty acids, these data, taken together with fatty acid binding specificity, are more consistent with fatty acid-FadL binding occurring by a hydrophobic interaction. Such a hydrophobic interaction has been suggested between the hydrocarbon tail of fatty acids and the fatty acid binding pockets on BSA [24]. In the fatty acid transport study by Maloy et al [3], the level of long-chain fatty acid transport was measured over a range of pH and showed maximal activity at a pH of 7. The present study evaluated the fatty acid binding activity of FadL independently of metabolic utilization. As a consequence, the pH profile obtained for transport obviously does not reflect the activity of FadL, but rather other component(s) involved in activation or /?-oxidation of longchain fatty acids. Using a mono-specific polyclonal antibody generated against purified FadL, we have shown that oleic acid binding could be blocked by 60%. This observation is particularly significant in that it is the first to show the blocking of a FadL-specific activity. In this respect, this data provided a clear correlation between the identity and the function of FadL. Glatz and Veerkamp [23] used the hydrophobic resin, Lipidex 1000, to measure fatty acid binding to eukaryotic fatty acid binding proteins. Using such a resin, we demonstrated that outer membrane vesicle material was not retained but was very effective at binding free oleic acid at 0” C. Elution of all vesicle material and binding free oleic acid at 0°C were essential parameters in order for this technique to be used. Using Lipidex 1000 and outer membrane vesicles, a FadL-specific oleic acid binding activity was measured. The binding of oleic acid followed saturation kinetics and gave an apparent K, of binding of approx. 6. lo-’ M. The apparent affinity of FadL towards oleic acid under these conditions was similar to that determined in whole cells (approx. 2 . lo-’ M). These studies demon-

105 strated that the oleic acid binding activity associated with FadL could be measured in vitro. As importantly, the parameters of oleic acid binding defined for FadL in vivo were very similar to parameters defined using the Lipidex 1000 in an in vitro system. The Lipidex 1000 technique offers a number of advantages for measuring the activity of FadL in vitro. First this resin binds free long-chain fatty acids at O’C while allowing all vesicle material (including bound long-chain fatty acids) to pass through the column. Second, use of this technique avoids the need for detergents to solubilize the fatty acids (and thus disrupting vesicles or cellular membranes). Finally, use of the Lipidex 1000 is rapid and reproducible. With our success using this system with outer membrane vesicles generated from fudL+ strains, the eventual reconstitution of FadL within a lipid-lipopolysaccharide vesicle is likely to yield information further defining the activity of FadL. FadL acts to initially bind fatty acids to the cell envelope of E. coli. The FadL-specific binding affinity of long-chain fatty acids is intermediate between the LamB-dependent binding of maltodextrans [25,26] and the BtuB-dependent binding of vitamin B12 [27]. It is worth noting that the K, of eukaryote fatty acid binding protein (FABP) fatty acid complexes are in the range of 10-7-10-6 M [23,28,29]. We have shown that the FadL-specific binding of long-chain fatty acids occurs between lop7 and lop6 M. Although the relative affinities of FadL and FABPs for fatty acids are of the same magnitude, it seems unlikely these proteins have a high degree of homology. It may be however, that the fatty acid binding domain(s) on these different proteins have sequence and/or structural homologies. Part of the current work underway in our laboratory is designed to define the fatty acid binding domain(s) of FadL in an effort to understand how long-chain fatty acids specifically interact with this protein. Acknowledgements I thank Angela Burnett and William Holden for technical assistance; Anita Hardeman for word processing; and Satoru Ken Nishimoto, Concetta DiRusso, Dennis Kiick and Gerald Carlson for comments on this manuscript. This work was supported by grant DCB881714 from the National Science Foundation.

References 1 Nunn, W.D. and Simons, R.W. (1978) Proc. Natl. Acad. Sci. USA 75, 3377-3381. 2 Nunn, W.D., Simons, R.W., Egan, P.A. and Malay, S.R. (1979) J. Biol. Chem. 254, 9130-9134. 3 Maloy, S.R.,’ Ginsburgh, C.L., Simons, R.W. and Nunn, W.D. (1981) J. Biol. Chem. 256, 3735-3742. 4 Nunn, W.D.. Colbum, R. and Black, P.N. (1986) J. Biol. Chem. 261, 167-171. 5 Black, P.N., Kianian, S.K., DiRusso, C.C. and Nunn, W.D. (1985) J. Biol. Chem. 260, 1780-1789. 6 Black, P.N., Said, B., Ghosn, C., Beach, J.V. and Nunn, W.D. (1987) J. Biol. Chem. 262, 1412-1419. 7 Black, P.N. (1988) J. Bacterial. 170, 2850-2854. 8 Klein, K., Steinberg, R., Fiethen, B. and Overath, P. (1971) Eur. J. B&hem. 19,443-450. 9 Frerman, F.E. and Bennett, N. (1973) Arch. B&hem. Biophys. 159, 434-443. 10 Rock, CO. and Jackowski, S. (1985) J. Biol. Chem. 260, 1272012724. 11 Jackowski, S. and Rock, C.O. (1986) J. Biol. Chem. 261, 1132811333. 12 Ginsburgh, C.L., Black, P.N. and Nunn, W.D. (1984) J. Biol. Chem. 259, 8437-8443. 13 Simons, R.W., Egan, P.A., Chute, H.A. and Nunn, W.D. (1980) J. Bacterial. 142, 621-632. 14 Miller, J.H. (1972) in Experiments in Molecular Genetics, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY. 15 Overath, P., Pauli, G. and Schairer, H.U. (1969) Eur. J. B&hem. 7, 559-574. 16 Lowry, O.H., Rosebrough, N.J., Farr, A.L. and Randall, R.J. (1951) J. Biol. Chem. 193, 265-275. 17 Cleland, W.W. (1979) Methods Enzymol. 63, 103-138. 18 Jaton, J.C., Brandt, D.C. and Vassalli, P. (1979) in Immunological Methods (Lefkovits, I. and Perris, B., eds.), Academic Press, New York. 19 Osbom, M.J., Gander, J.E., Parisi, E. and Carson, J. (1972) J. Biol. Chem. 247, 3962-3972. 20 Osborn, M.J. (1963) Proc. Natl. Acad. Sci. USA 50, 499-506. 21 Laemrnli, U.K. (1970) Nature (Lond.) 227, 680-685. 22 Bumette, W.N. (1981) Anal. Biochem. 112, 195-203. 23 Glatz, J.F. and Veerkamp, J.H. (1983) Anal. B&hem. 132, 89-95. 24 Peters, T., Taniachi, H. and Anfinsen, C.B. (1973) J. Biol. Chem. 248, 2447-2451. 25 Heine, H.-G., Francis, G., Lee, K.S. and Ferenci, T. (1988) J. Bacterial. 170, 1730-1738. 26 Ferenci, T., Schwentorat, M., Ullrich, S. and Vilmart, J. (1980) J. Bactetiol. 142, 521-526. 27 White, J.C., DiGirolamo, P.M., Fu, M.L., Preston, Y.A. and Bradbeer, C. (1973) J. Biol. Chem. 248, 3978-3986. 28 Bass, N.M. (1985) Chem. Phys. Lipids 38, 95-114. 29 Sacchettini, J.C., Meininger, T.A., Lower, J.B., Gordon, J.I. and Banaszak, L.J. (1981) J. Biol. Chem. 262, 5428-5430.