Transport of Long-chain Fatty Acids in Escherichia coli EVIDENCE FOR ROLE OF fadL GENE PRODUCT AS LONG-CHAIN FATTY ACID RECEPTOR*

Transport of long-chain fatty acids (LCFA) across the cytoplasmic membrane of Escherichia coli requires functional fadL and fadD genes. The fadD gene codes for an acyl-CoA synthetase (fatty acid: CoA ligase (AMP forming)) which has broad chain length specificity and is loosely bound to the cytoplasmic membrane. The fadL gene codes for a 43,000-dalton cytoplasmic membrane protein which, acting by an unknown mech-anism, is needed specifically for LCFA transport. As a first step to define the role of the fadL gene product, studies were performed to determine if it functions as a LCFA receptor. The LCFA-binding activity was quantitated in intact cells in the absence of LCFA transport by comparing the binding of LCFA in fadD fadL and fadD fadL+ strains. These studies revealed that (i) fadD fadL+ strains bind 6-fold more LCFA than fadD fadL strains; (ii) fadD fadL strains harboring a plasmid containing the fadL gene bind 16-fold more LCFA than fadD fadL strains harboring only the plasmid vector; and (iii) the fadL-specific LCFA-binding activity is regulated by the fadR gene and catabolite repression. Studies with fadL strains harboring fadL plasmids containing in vitro constructed deletions indicate that mutations which alter the physical properties of the 43,000-dalton fadL gene product also affect fadL gene product-specific LCFA-binding activity. Overall, these studies suggest that one role of the fadL gene product in the LCFA

fadL gene product; SDS, sodium dodecyl sulfate. which has broad chain-length specificity and is loosely bound to the cytoplasmic membrane (10, 11). Klein and co-workers (11) have shown that fudD mutants are unable to accumulate exogenous fatty acids of any chain length into the cytosol and/or their membrane phospholipids. These investigators suggested that the acyl-CoA synthetase may be required for a group translocation step in the transport of fatty acids (i.e. vectoral thiol esterification).
The fudL gene, which codes for a 43,000-dalton cytoplasmic membrane protein, has been found by our laboratory to be essential for the transport of LCFA into the cell (7-9,12, 13). Kinetic analyses of transport in wild-type and fudL strains indicate the transport system for LCFA is a specific, active, carrier-mediated process which accumulates fatty acids against a concentration gradient (7). When the fudL transport system, which also actively transports medium-chain fatty acids (C7-Clo) is nonfunctional, medium-chain fatty acids are still able to diffuse into the cell (7). The synthesis of the fadL gene product is induced by LCFA and catabolite repressed by glucose (12). At present, all of the available evidence suggests that the product of the fudR gene, a 29,000-dalton repressor protein, negatively controls the expression of fudL and fudD genes (14)(15)(16).
During an analysis of the expression of the fudL gene product (FLP) using the fadL clone, we observed that the electrophoretic mobility profile of the wild-type FLP was influenced by the temperature at which it is heated in sodium dodecyl sulfate (SDS) (13). FLP has an apparent molecular weight of 43,000 when solubilized at 100 "C in SDS uersus 33,000 when solubilized at 50 "C in SDS (13). The heatmodifiable nature of the fudL gene product in the presence of SDS suggests that it behaves similarly to several outer membrane proteins described by Nakamura and Mizushima (17) and deGeus et ul. (18).
At present it is not clear whether FLP is a receptor and/or translocase for LCFA. As a working model consistent with our studies (7,9,10,12,13) and those of Klein et ul. (11), we have proposed that the transport of LCFA into E. coli requires at least two intrinsic membrane components, the nonspecific outer membrane porins and the inner membrane FLP, to deliver dissociated monomeric LCFA at sites in the inner surface of the cytoplasmic membrane where they are activated and then released into the cell by the peripheral membrane component, the acyl-CoA synthetase (Fig. 1).
As a first step toward defining the function of FLP, we were interested in determining whether the fudL gene product acts as a receptor for LCFA by comparing LCFA binding activity in fudL+ and fadL strains. To avoid LCFA transport and/or LCFA binding contributed by the acyl-CoA synthetase, these studies were performed with fudD derivatives of fudL+ and fadL strains. Overall, the studies presented in this paper suggest that FLP is required for binding of LCFA prior to their translocation into the cell.

EXPERIMENTAL PROCEDURES AND RESULTS~
The binding studies revealed that strains containing a functional fadL gene bind significantly more LCFA than strains containing a defective fadL. Furthermore, whereas a fadL+ fadD strain has 6-fold more LCFA-binding activity than a fadD fadL strain (Table 111), a fadL fadD+ strain has the same LCFA-binding activity as the fadD fadL strain. The latter results imply that, in the LCFA transport process, it's the fadL gene product rather than the fadD gene product (acyl-CoA synthetase) that is responsible for sequestering significant quantities of LCFA at sites in the cell membrane for transport. Studies with fadL+ plasmids and in vitro constructed plasmids derived from them suggest that a 43,000dalton protein, designated FLP, is involved in the fadLspecific LCFA binding.

DISCUSSION
Our prior genetic, physiological and kinetic studies have indicated that the fadL gene product, in conjunction with the fadD gene product (acyl-CoA synthetase), is required in order for LCFA to traverse the hydrophobic membrane barrier of the E. coli cell (7-9). After identifying the fadL gene product (FLP) as a 43,000-dalton inner membrane protein and showing that the acyl-CoA synthetase is loosely bound to the membrane (12), we proposed that the pathway for LCFA transport requires that LCFA interact with FLP prior to being esterified by the acyl-CoA synthetase (Fig. 1). The function of FLP was speculated to be a LCFA receptor and/or translocase (13). To determine if FLP can function as a LCFA receptor, LCFA-binding activity was measured in a fadD strain which had a functional fadL gene. The fadD fadL+ strain enabled us to focus specifically on the fadL gene product's interaction with LCFA in the absence of fatty acid transport. By comparing the LCFA-binding activity between the fadD fadL+ strain and a fadD fadL strain, we have shown that the strain containing a functional fadL gene binds 6-fold more LCFA than an isogenic strain containing a nonfunctional fadL gene. The LCFA-binding studies with the fadD fadL+ strain suggest it has approximately 2.3 X lo4 copies of the fadL specific binding components/cell. The fadL-specific LCFA-binding capacity of a strain which contains a multicopy fadL+ plasmid (Table V) was elevated almost 3-fold over that of a strain containing only one copy of the fadL gene (Table  111). Due to the amphipathic nature of oleic acid, it was not surprising to find that strains defective in the fadL gene (LS6164 and LS6929) bind nonspecifically considerable amounts of this fatty acid (-65 pmol/mg of protein in the presence of bovine serum albumin and 185 pmol/mg of protein in the absence of bovine serum albumin). The fact that the fadL strain LS6164 is unable to transport the nonspecifically bound oleate into the cell suggests that one function of the fadL gene product may be to sequester LCFA at sites in the cell membrane where translocation into the cell is facilitated.
To determine the order of steps in the LCFA transport process, we compared the LCFA-binding capacities of fadL, fadD, and fadD fadL strains. The rationale for these studies was that if the step catalyzed by the fadL gene product precedes the step catalyzed by the fadD gene product, then we would expect the double mutant fadD fadL to have quantitatively the same LCFA-binding phenotype as the fadL mutant. Conversely, if the step catalyzed by the fadD gene product precedes the step catalyzed by the fadL gene product, then the fadD fadL mutant would have the binding phenotype of the fadD mutant. The results of these studies strongly support our contention that LCFA encounter the FLP prior to being esterified with CoA by the fadD gene product (acyl-CoA synthetase) because the double mutant had the same phenotype as the fadL mutant (Table 111). Further evidence for the latter conclusion was obtained when an analysis of the lipids bound to the membranes of the fadD and fadL mutants revealed that the LCFA remained unesterified (data not shown).
The binding and transport studies with pACC-and pAEVbearing strains (Tables V and VI) demonstrate that both optimal LCFA binding and transport activity is encoded within the 2.6 EcoRV fragment. These studies support the earlier findings of Black et al. (13) which established that the fadL-coding region lies within the 2.6 EcoRV fragment and must contain the SalII, KpnI, BstEII, and PuuII, restriction sites within its domain. Maxicell studies with plasmids pACC, pAEV, PACK, pACS, and the other in vitro constructed deletion plasmids (pACP, pBEB, pABC, pAPP, and pLSS) revealed that the right side of the 2.6 EcoRV fragment (encompassing the EcoRV, to at least the Sa&) encodes for the 43,000-dalton membrane-bound FLP (13). The studies presented in this paper suggest there is a correlation between LCFA-binding activity and the presence of FLP. The studies with fadL strains harboring the plasmids PACK and pACS suggest that these plasmids encode for defective proteins which 1) have lower LCFA-binding affinities than the FLP encoded by the fadL+ plasmid pACC, 2) have isoelectric points that are different from the PI of the FLP encoded by the fadL' plasmid pACC and 3) do not have the heat-modifiable properties exhibited by the FLP encoded by pACC. Although the physiological significance of the latter finding (3 above) is not apparent, it is conceivable that the proteins synthesized by plasmids PACK and pACS have lower affinities for LCFA because they cannot assume the proper conformation for receptor activity. Needless to say, the studies with PACK and pACS suggest that, in addition to the EcoRVl to Sa& fragment, DNA to the left of the SalIl site and an intact KpnI site are essential for the expression of FLP which has optimal LCFA-binding activity.
It is not clear whether FLP functions as both a LCFA receptor and translocase. If there is only one protein encoded by the 2.6 EcoRV fragment as our maxicell studies suggest (Figs. 5 and 6), one could conclude that FLP is both a receptor and translocase. Accordingly, one interpretation from the studies with PACK and pACS may be that the proteins synthesized by these plasmids have defective receptor sites due to deletions in the promotor distal portion of the fudL gene (Fig. 3). The fact that a deletion of the DNA to the left of the Sa& site affects the physical properties of FLP suggests the fudL gene extends beyond the promotor distal Sunl site.
However, these studies do not rule out the possibility that an undetectable component(s) encoded by the EcoRV2 to SulIl fragment is required for LCFA translocase activity. Since preliminary complementation studies3 suggest only one gene is required for LCFA binding and transport and only one gene product (the 43,000-dalton FLP) is expressed by pAEV, it's more likely that the FLP is both a receptor and translocase. However, the present studies do not definitively prove this to be the case.
In summary, the studies presented in this work show that (i) fudl-defective strains bind significantly less LCFA than fadL' strains; (ii) strains harboring a fadL+ multicopy plasmid bind significantly more LCFA than strains containing one copy of the fadL' gene; (iii) fadL' LCFA-binding activity is regulated by the fudR gene and catabolite repression; and (iv) a correlation between the presence of the 43,000-dalton fudL gene product and LCFA-binding activity has been established. Collectively, these studies suggest that the fudL gene product may function as a LCFA receptor. Present efforts in our laboratory are being directed toward determining if the fudL gene product or some, as yet, unidentified component functions as the LCFA translocase component.

Experimental Procedures
The L snll K-12 bacterial s t r a i n s and plasrida used i n t h i n study are li8t.d in Tables 1 and 2 respectively. incubated a t 37' C i n a new BrunPrick gyrotory vater bath shaker.

Por
The bacteria vere routinely grwn in TB broth (19)

C l -o b a t e a~
Was anaayed with the binding asnay mixture described above.  iiiii-wiii &&ea 95% of the laheled oleate vas removed by a single wash containing Lirij 5 8 or BSA and l i p i d a n a l y s i s r e v e a l e d t h a t a l l of the removed oleate was unesterfied (data Mt~BhoWn).
These data suggest that the strain ~~~ containing a functional fadL gene product has LCBA binding a c t i v i t y large fadL deletion AladL5 (Pigure 2 ) , we decided t o determine i f Since the l a t t e r s t u d i e s v e r e p r f o r m e d w i t h a s t r a i n t h a t w n t a i n s ' t h e strains containing smaller f&L deletions (Pigurr 2) exhibited the same 1w LCPA binding a c t i v i t y . The result. in Table   N show t h a t fodD binding a c t i v i t y than the fodD s t r a i n LS6928 (Table 111) Table   I. pL indicates the location of the fad& promoter. This f i ure i s a modification of figure 2 described by Ginsburgh +f dl (12). %inding of LCFA occurs only in the absence of BSA and/or Brij58.  The r e s u l t s i n Table v. sbw. as e G c t e d . t h a t i s 6 9 2 9 harboring the plasmids g C C and/or p& have f u l l y functional LCPA binding activity.
when LS6929 harbors the plamid gCP, Fern, pLPP, and/or pLS8 it n e i t h e r t r a n s w r t s LCFA (Table V I (Table VI) and/or binding a c t i v i t y (Table v) t o 186929. when LCPA binding activity wan measured i n LS6929 harboring the plasmids pACK and WCS. ve observed. serendipitously, that these plautids only rrstored LCPA binding a c t i v i t y when BSA vas excluded frcm m a t o eolubilize the oleate and minimize binding due t o t h e thb assay mixture. BSA iPA/BSA -1) was standardly included i n our a s u y aggregation of fatty acid micelles.
In the absence of BSA. Ls6929 harboring pACS or PACK has approximately 608 t h e LCBA binding a c t i v i t y of LS929 harboring pACC or pABV (Table V )  To test this supposition, we mmpred the LCPA binding a c t i v i t y i n Ls6928 (wild-type for U and 186929 harboring pcs as a function of LCFA concentration. These studies, performed i n t h e absence of BSA, indicatbd the LCPA binding affinity of Ls6q29 containing p C S is considerably lover v i t h La6929 harboring PACK gave similar results as LS6929 harboring g c s than that of the wild-type fadL s t r a i n LS6928 (Pigure 4 ) . Studies WCC. pABV, gCS, p C K . and the other in yitrp wnstructed deletion (data not shavn). Overall, the binding studies v i a s t r a i n s harboring plaemide (Table V)  In previous studies, we have obeerved t h a t t h e plasmide pACK and wcs did not enable fadL e t r a i n s t o grow on and/or transport LCPA (13) Bince t h e l a t t e r s t u d i e s were performed in the presence of the detergent 8 r i j 5 8 (13) and our binding studies shw t h a t this detergent prevents a h d D fadL s t r a i n w n t a i n i n g PACK or g c s from binding LCPA, we decidbd t o re-examine vhether these plasmids wuld restore LCFA transport to a f.dL mutant which has a functional fndp gene, e t r a i n 686164. The transport+studies performed i n the absence of B r i j 5 8 . sbw t h a t u n l i k e LCPA t r a n s p o r t a c t i v i t y t o LE6164 (Table V I ) . thb fnpL plasmid pACC, the plasmids g C K and pACS do not restore pAEV direct the syntbesis of a heat modifiable membrane protein vhich has p a & + plasmids paCC and an i8oelectric point (PI) of 4 6 and an a parent molecular veight of 43.000 daltone when B o~u b i l i z e i a t 100°C f n SI)8 and 33,000 daltons when s o l u b i l i z e d a t 50 c i n SDS. since the binding studies indicated thb plaamids p C K and p C S enwde for defective fnpL gene products that have p a r t i a l LCPA binding a c t i v i t y , an analysis of the proteins enwdbd by these plasmids vas performed. Mxicell studies revealed that both plasmids direct the synthesis of a protein vhich has an apparent molecular veight of 43,000 daltons ( Figure 5 ) . nowever, unlike the protain enmded for by g C C . the molrcular veight of the proteins made by g c s and pACK do not appear t o be heat modifiable in the presence of So8 MCC i s 4.6. the nx of the Droteins wnthesired bv nAcs and DACK a r e 5.0 ( Figure 5 ) .