MICROBIOLOGICAL REVIEWS, June 1986, p. 179-192 Vol. 50, No. 2 0146-0749/86/020179-14$02.00/0 Copyright C) 1986, American Society for Microbiology A Molecular View of in WILLIAM D. NUNN Department of Molecular Biology and Biochemistry, University of California, Irvine, Irvine, California 92717

INTRODUCTION ...... 179

GENETIC AND BIOCHEMICAL ANALYSIS OF FATTY ACID OXIDATION ...... 179

Studies with the FAO System ...... 179

Studies with the ATO Enzyme System ...... 181

TRANSPORT OF FATTY ACIDS ...... 182

MCFA and LCFA Transport ...... 182

SCFA Transport ...... 185

REGULATION OF FATTY ACID DEGRADATION ...... 185

fad Regulation ...... 185

ato System ...... 186

GENETIC AND BIOCHEMICAL FEATURES OF ...... 187

Glyoxylate Shunt ...... 187

Other Ace ...... 188

Regulation of ace System(s) ...... 189

CONCLUSIONS ...... 189

ACKNOWLEDGMENTS ...... 190

LITERATURE CITED ...... 190

INTRODUCTION GENETIC AND BIOCHEMICAL ANALYSIS OF FATTY ACID OXIDATION Gram-negative such as Escherichia coli can uti- Most of our knowledge of the genetics and biochemistry of lize fatty acids and acetate as sole carbon and energy sources fatty acid degradation is derived from studies with E. coli. (30, 51, 76). After entry into the cell, fatty acids are either Growth of wild-type E. coli strains on fatty acids as sole catabolized or directly incorporated into complex . carbon sources occurs only when the chain length of the Fatty acid degradation occurs by the cyclic 1-oxidation and fatty acid is 12 or more carbon atoms (long-chain fatty acids thiolytic cleavage of fatty acids, yielding several moles of [LCFAs]), and then only after a distinct lag period. The acetyl- (CoA). The acetyl-CoA produced is synthesis of at least five fatty acid-oxidative (FAO) enzymes further metabolized, yielding energy and precursors for (Fig. 1) is coordinately induced when LCFAs (C12 to C18) are cellular biosynthesis. A considerable body of information present in the growth media (51, 52, 76). The genetic studies regarding the physiology, genetics, and molecular biology of of Overath and co-workers (25, 30, 52) provided the first fatty acid and acetate metabolism has been amassed from evidence that the structural genes encoding the FAO en- studies with E. coli. These studies have enabled workers to zymes are located at several sites on the E. coli chromosome define the structural and regulatory genes required for the (Fig. 2) and make up a regulon (referred to as the fad and acetate. Studies the regulon). The fad regulon is primarily responsible for the catabolism of fatty acids involving transport, acylation, and 1-oxidation of medium-chain fatty use of recombinant DNA technology, coupled with studies acids (MCFAs) (C7 to C1l) and LCFAs. Growth of E. coli on that involve the purification and characterization of en- short-chain fatty acids (SCFAs) (C4 to C6) requires, in zymes, have enabled workers to correlate the structural addition to the FAO enzymes, two degradative enzymes genes (fad) with their respective gene products. One conse- (Fig. 3) encoded by the atoA and atoB genes (56). These quence of these efforts is that our understanding of fatty acid genes appear to be regulated by the atoC gene (56). transport processes has been significantly enhanced. These In this section we shall examine the genetic and molecular studies have also uncovered a regulatory interaction among information regarding the FAO and SCFA-degradative fatty acid degradation, acetate metabolism, and unsaturated (ATO) enzymes. In addition, we shall describe how SCFAs, fatty acid biosynthesis. MCFAs, and LCFAs are taken up by E. coli. Following the The purpose of this review is to illuminate studies that description of the FAO and ATO enzymes and transport have involved the examination of the molecular details of systems, we shall relate what is known about the mechanism fatty acid and acetate catabolism in E. coli. Although no by which the fad and ato structural genes are regulated. major reviews of the catabolism of fatty acids or acetate or both by this organism have been written, this article will not Studies with the FAO Enzyme System provide a comprehensive review on all aspects of these The basic features by which E. coli degrades fatty acids processes in E. coli. Instead, we will examine the literature are substantially similar to the ,B-oxidative pathways present on fatty acid and acetate catabolism, with emphasis on in mammalian and other eucaryotic organisms. This pathway studies in which genetic and biochemical manipulations were is a classic example of the oxidation of a series of homolo- used to define the structural and regulatory components of gous substrates through a series of homologous intermedi- these reactions. ates. Certain features of the pathway are illustrated in Fig. 1. With each turn of the ,8-oxidation cycle, the fatty acyl-CoA 179 180 NUNN MICROBIOL. REV.

one molecule each of (ATP) and CoA per molecule of free fatty acid activated. Overath et al. (51) suggested that E. coli has one acyl-CoA synthetase with out _ acd (X L) in _ ty anspor ---_ broad specificity for MCFAs and LCFAs. As supporting evidence, they showed that afadD mutant isolated in their '., I R _ OH laboratory lacks acyl-CoA synthetase activity for MCFAs ATP4X9 and LCFAs (51). In contrast to the hypothesis of Overath et AMP.PPtfacylCoA syfih (M D) al. (51), Samuel et al. (63), on the basis of their studies with R SCo partially purified acyl-CoA synthetase, suggested that the FAD y-O_A @hydrag (g E) acyl-CoA synthetase is a complex of two enzymes, one that activates MCFAs and another that activates LCFAs. To resolve whether E. coli has a single acyl-CoA synthetase cids-A3-trans-A'-en oAC heme with broad specificity or multiple acyl-CoA H,anoyl-CoA hydra0aM (g B) synthetases with limited substrate specificity, Kameda and Nunn (29) purified to homogeneity the E. coli acyl-CoA OH 0 j R_.LS~WOH(D)-E ______-RRSCOA synthetase and found that the purified enzyme had broad substrate specificity for MCFAs and LCFAs. Although 3-hyW*C .dmgosue ( B) acyl-CoA synthetase was previously believed to be a mem- brane-associated , Kameda and Nunn demonstrated that over 90% of this enzyme was present in cytoplasmic CLASH 3-ke"*C bail (b A) fractions. The molecular weight of the native enzyme was approximately 130,000 and the subunit molecular weight O O determined by polyacrylamide gel electrophoresis in the -R_%-- SCtA+-Q presence of sodium dodecyl sulfate was 47,000. These ex- FIG. 1. Cyclic pathway of fatty acid degradation. Principle en- periments suggested that the enzyme may be a dimer or a zymes of the pathway are listed on the right, along with the of identical subunits respective structural genes ofthefad regulon. Acetyl-CoA is further trimer composed apparently (29). metabolized in the TCA Very little is known about the enzyme in E. coli respon- cycle. sible for the next step, acyl-CoA dehydrogenase. Although mutants (fadE) lacking this activity have been isolated and loses a two-carbon fragment as acetyl-CoA and reduces one their mutations have been mapped (30), it is not certain molecule of flavin dinucleotide (FAD) (during the whether the loss of the dehydrogenase protein or that of an acyl-CoA dehydrogenase reaction) and one molecule of associated flavoprotein is responsible for the observed loss nicotinamide adenine dinucleotide (NAD) (during the 3- of activity. hydroxyacyl-CoA dehydrogenase reaction). Acetyl-CoA, The other ,3-oxidation enzymes are also cytoplasmic in E. produced in the CoA-dependent thiolytic cleavage, is further coli. The remaining 1-oxidation enzymes (Fig. 1) are part of metabolized in the tricarboxylic acid (TCA) cycle. The other a multienzyme complex that has broad substrate specificity. of the cleavage step, a shortened fatty acyl-CoA In E. coli two in the ,B-oxidative pathway are molecule, reenters the degradation cycle without further associated with a multienzyme complex which has a molec- activation (51). ular weight of 260,000 (4, 18, 50, 57, 58). Five FAO enzyme The first step of fatty acid degradation is the activation of the free fatty acid to an acyl-CoA by acyl-CoA synthetase (fatty acid:CoA , adenosine monophos- 0 0 phate [AMP]-forming; EC 6.2.1.3). This reaction requires * U 0 CH3CCH2C. CH3CH2CH2C0O Acetoacetate Butyrate out, ace - in short chain fatty acid transport? .

fad_ ABe Acetoacety-CoA (atoA& ato K)

R CHkCSC CA

'I I 0 On v CH3CcHH2CSc CH3CH,cH2 Cc (IQ B) 1IVCoASH11:CaAS-oxidation, enzymesc 0 ( EE, f B) u- FIG. 2. Genetic linkage map ofE. coli K-12 showing the location offad, ato, and ace structural and regulatory genes. Adapted from 2C3C"SCOA --a TCA cycde -.-ATP the revised linkage map of Bachmann and Low (1). FIG. 3. Pathways of SCFA degradation in E. coli. VOL. 50, 1986 FATTY ACID CATABOLISM IN E. COLI 181

activities, 3-ketoacyl-CoA thiolase (thiolase), enoyl-CoA TABLE 1. FAO enzyme activities in E. coli strains containing hydratase (crotonase), 3-hydroxyacyl-CoA dehydrogenase different fadAB plasmids (HOADH), cis-A3-trans-A2-enoyl-CoA (isomer- Plasmid and strain Genes Sp act of: ase), and 3-hydroxyacyl-CoA epimerase (epimerase), are complemented associated with this multienzyme complex (4, 5, 57, 58). (genotype) by plasmid Thiolase HOADH Crotonase Schulz and co-workers (4, 57, 58) purified the complex and pBR322 None found it to have a2,42 subunit structure (a, 78,000 daltons; P, LE392 (wild type) 14 304 1,613 42,000 daltons). Through biochemical characterization of the LS6749 (fadAB) ob 17 16 multienzyme complex from an E. coli B strain, Schulz and LS6595 (fadA) 0 280 1,680 co-workers determined that the thiolase activity is associ- ated with the 42,000-dalton subunit and that the remaining pK52 fadAB four enzymes activities are associated with the larger, LS6749 (fadAB) 308 1,745 7,004 78,000-dalton, subunit (4, 5, 57, 58). When unsaturated fatty acids are degraded, pKl fadA two additional activities, isomerase and LS6749 (fadAB) 0 0 0 epimerase, are required. Although no mutants lacking these LS6595 (fadA) 12 two activities have been isolated, pleiotropicfadAB mutants NDc (ND) (51, 52) and biochemical studies (57, 58, 66) demonstrated a Specific activities are expressed in nanomoles per minute per milligram of that the 78,000-dalton protein also contains both isomerase protein. b Zero implies activity of <0.05. and epimerase activities. c ND, Not determined. Overath et al. (51) suggested that the genes for the enzymes thiolase (EC 2.3.1.16), HOADH (EC 1.1.1.35), crotonase (EC 4.2.1.17), and, possibly, epimerase (EC fadAB genes are part of an and showed that the 5.1.2.3) and isomerase (EC 5.3.3.3) form an operon. Their direction of transcription is fromfadA tofadB (66). ThefadA evidence for suggesting that the fadAB genes constitute an andfadB gene products were identified by maxicell analysis. operon was based on the high coordinate induction of The fadA gene was found to code for the 42,000-dalton a thiolase, HOADH, and crotonase as well as on the mapping subunit, and the fadB gene was found to be associated with properties of mutants deficient in (i) all five enzymes the 78,000-dalton a subunit. A hybrid plasmid containing the (fadAB), (ii) thiolase (fadA), and (iii) HOADH (fadB) (51). fadA gene expressed thiolase I activity in fadA mutants but Since there was no genetic evidence that the fadAB genes not in fadAB mutants (Table 1). These results were inter- were organized in an operon, Spratt et al. (66) cloned the preted to mean that the 42,000-dalton a subunit (representing fadAB genes from E. coli to determine the organization of thiolase I) may not be functional in the absence of the fadB thesefad structural genes (Fig. 4). The approximate location gene product, the 78,000-dalton a subunit. Some support for and orientation of thefadA andfadB genes were determined this contention came from studies that show that fadA by subcloning and Tn5 mutagenesis (66). These studies mutants carrying multicopyfadA+B+ plasmids express both supported the hypothesis of Overath et al. (51) that the the 78,000- and 42,000-dalton proteins and have amplified levels of thiolase activity, whereas those containing multicopy fadA+ plasmids express only the 42,000-dalton protein and have wild-type levels of thiolase activity (66). It appears that in the latter case, the fadA mutant, containing the multicopy fadA+ plasmid and high levels of the 42,000- dalton a subunit, may be restricted from expressing ampli- fied levels of thiolase activity because it has only wild-type levels of the 78,000-dalton a subunit. Additional support for this view comes from studies showing that wild-type strains crsubunk harboring the multicopy fadA+ plasmid have comparable

...... levels of thiolase activity to those of wild-type strains which ::: contain either no plasmid or the plasmid vector pUC9 (66)...... I...... Pawar and Schulz (57) also showed that the medium-chain 4 but not the long-chain enoyl-CoA hydratase is associated with thefadAB multienzyme complex. Although its not clear where the gene that codes for the soluble long-chain enoyl- ....~~~~~~~~~~~~....fad A * .:.:.::::::::::::::::::...... CoA hydratase maps, it is most likely linked to the fadAB genes. The reason for suggesting the latter is that during their Sal Bg11i Bgi11, Cla Bgl 11, Bgl ii, Pstl characterization of clones carrying the fadAB genes, Spratt et al. (66) observed an additional ca. 60,000-dalton protein encoded by the plasmid. The synthesis of this protein, like 1 kb that of the fadA and fadB gene products, was prevented by FIG. 4. Structural organization and direction of transcription of polar transposon Tn5 insertions in the fadA gene (66). thefadAB region. The relative locations of thefadA andfadB genes Needless to say, additional studies that correlate the 60,000- and the direction of transcription are indicated. The endpoints of dalton protein with long-chain CoA hydratase activity must these genes are not precisely defined and are indicated by a broken line. The protein products (expressed in kilodaltons [KD]) specified be performed before definitive proof of the role of this by the genes are indicated at the top of the figure. The two protein protein can be established. subunits encoded by the fadAB genes form a multifunctional en- Studies with the ATO Enzyme System zyme complex which catalyzes at least five fatty acid-degradative enzyme activities (see text). Restriction sites PstI, BglIl, ClaI, and Wild-type E. coli can utilize exogenous LCFAs as a sole SalI are indicated. carbon and energy source because LCFAs induce the syn- 182 NUNN MICROBIOL. REV.

OM IM

PL biosynthesis

G7-Gl acyl-CoA synthetase (~tad D) - _ Long chain . (3-oxidation _-, acyl-CoA CoA+ATP

C, 2-C18

LPS PL PG PL PL FIG. 5. Proposed model of fatty acid transport in E. coli K-12. MCFAs traverse the outer membrane via FLP (encoded by thefadL gene) and via a diffusional process. LCFAs traverse the outer membrane via FLP. To enter the cell, MCFAs and LCFAs must be activated by the acyl-CoA synthetase (fadD gene product). At present, it is not known whether a protein is required to facilitate the transport of these fatty acids across the cytoplasmic membrane. Abbreviations: PL, phospholipid; OM, outer membrane; IM, inner membrane; PG, peptidoglycan; PS, periplasmic space; and LPS, lipopolysaccharide. Note that very little phospholipid is distributed in the outer leaflet of the outer membrane (43). thesis of the FAO enzymes (51, 52, 76). MCFAs can serve as tetrameric protein composed of two a and two 1 subunits substrates for the FAO enzymes but cannot induce the and that thiolase II is a tetrameric protein composed of four synthesis of these enzymes. Therefore, only fatty acids identical subunits. longer than 12 carbons may be used as sole carbon sources by the wild type (fadR+). Strains that constitutively synthe- TRANSPORT OF FATTY ACIDS size the FAO enzymes grow on MCFAs and LCFAs. The constitutive strains have been given the designation MCFA and LCFA Transport fadR. Prior to 1-oxidation, fatty acids must enter the cell via fadR+ and fadR strains cannot utilize short-chain fatty uptake systems which translocate them across the mem- acids (SCFAs) as sole carbon sources (55, 60, 61, 74). For brane. Before any rigorous kinetic analyses were done, the growth to occur on the SCFAs butyrate (C4) and valerate prevailing thought was that fatty acids diffuse through mem- (Cs), E. coli must have constitutive levels of three FAO branes without requiring a protein carrier. However, physi- enzymes (crotonase, HOADH, and acyl-CoA dehydroge- ological and kinetic studies performed in several laboratories nase) and at least two enzymes involved with the degrada- (20, 40, 51) suggest that a saturable carrier mechanism tion of the 1-keto SCFA acetoacetate (AA) (55, 73). The facilitates the entry of LCFAs into E. coli. Furthermore, degradation of AA to acetyl-CoA is a two-step reaction (Fig. genetic and biochemical studies have implicated at least two 3) that first results in the activation of AA to acetoacetyl- proteins, encoded by the fadD and fadL genes, that are CoA, which is catalyzed by acetyl-CoA:acetoacetyl-CoA required to deliver exogenous LCFAs across the cell mem- transferase (AA-CoA transferase), and then involves subse- brane to the cytosolic fatty acid-degradative enzymes. The quent cleavage of acetoacetyl-CoA to acetyl-CoA, which is fadD gene codes for a peripheral membrane-bound protein, catalyzed by thiolase II (19, 20, 56, 68, 73, 74). These the acyl-CoA synthetase, which has broad specificity for enzymes are inducible by AA and have substrate specifici- MCFAs and LCFAs (29). ThefadL gene codes for a 43,000- ties for ,B-keto SCFAs. dalton membrane protein (FLP) which has been implicated Earlier studies performed by Pauli and Overath (56) iden- as being essential for LCFA transport. Although the exact tified the loci responsible for AA degradation as atoA, which role of the fadL gene product in the uptake of LCFAs is encodes AA-CoA transferase, and atoB, which encodes unknown, fatty acid-binding studies have demonstrated a thiolase II. These structural genes are closely linked and are correlation between the presence of FLP and LCFA-binding located at the 47-min region on the revised E. coli chromo- activity (45). The fadL gene product, FLP, is the first somal map (56). The biochemical work of Sramek and membrane protein which has been shown both genetically Frerman (68, 69) has shown that AA-CoA transferase is a and biochemically to be involved in the uptake of fatty acids VOL. 50, 1986 FATTY ACID CATABOLISM IN E. COLI 183

(40, 47, 48). A simplistic working model for fatty acid uptake a b c d in E. coli is illustrated in Fig. 5. This model, derived from genetic and biochemical studies (6, 20, 40, 45, 47, 48, 51), suggests that LCFAs are initially adsorbed to FLP, which T-. - m functions as an outer membrane receptor and, possibly, a pore. Once the LCFAs are transferred across the outer ~~~~_.66.2 membrane, they are somehow transferred across the cyto- plasmic membrane to the peripheral membrane-bound acyl- CoA synthetase, where they are activated and released into _ <~.45.0 the cytosol (Fig. 5). When FLP, which also actively medi- ates the uptake of MCFAs, is nonfunctional, MCFAs are still able to diffuse across the cell membrane to the acyl-CoA synthetase, where they are activated and released into the ''.* '00,01E.- cell (Fig. 5). The transport of SCFAs appears to be carrier _.01. X mediated (19), but does not require the functioning of either 0 the fadD or fadL gene product. Several observations demonstrate that the LCFAs enter E. coli via an active unidirectional carrier-mediated mecha- nism(s): (i) the transport of LCFAs into wild-type strains has been shown to be a saturable process (20, 40); (ii) inhibitors which prevent electron transport or uncouple oxidative phosphorylation have been shown to completely block LCFA transport (20, 40); (iii) no efflux oftransported LCFAs 14.4 occurred when wild-type E. coli strains were washed with unlabeled LCFAs (40, 51); and (iv) both the energy of activation and the temperature coefficient Qlo of LCFA transport are representative of enzyme-mediated processes (40). The evidence for suggesting that the acyl-CoA synthetase FIG. 6. Fluorographs of [35S]methionine-labeled proteins from is required for MCFA and LCFA transport into the cell was maxicells of pACC in strain LS6922 with different protein solubili- obtained from the studies of Klein et al. (30) and Frerman zation temperatures (6). Lanes: a, 25°C; b, 50°C; c, 70°C; d, 100°C. and Bennett (20), who showed thatfadD mutants are unable to accumulate exogenous fatty acids of any length into the linkage map, was identified by Nunn and Simons in 1978 as cytosol or membrane lipidF, or both. These investigators being essential for LCFA transport (47). The evidence for suggested that the acyl-CoA synthetase may be required for the latter suggestion was as follows: (i) fadL strains that are a group translocation step in the transport of fatty acids (i.e., constitutive for the synthesis of the FAO -enzymes (termed vectoral thiol esterification). They based their translocation fadR) can oxidize MCFAs but not LCFAs; (ii) in vitro hypothesis on studies showing that (i) the chain length extracts offadR fadL strains oxidize LCFAs at comparable specificity of the acyl-CoA synthetase for fatty acids could rates to those of extracts fromfadR fadL+ strains; (iii) toxic be correlated with chain length specificity for the uptake LCFA analogs inhibit the growth of fadL+ but not fadL system; (ii) no efflux of labeled fatty acids occurred when strains; and (iv) afadL mutation preventsfabA(Ts) mutants preloaded cells were diluted into excess unlabeled fatty from satisfying their unsaturated fatty auxotrophic pheno- acids; and (iii) no free labeled fatty acids were detected type at nonpermissive temperatures in the presence of intracellularly. It should be mentioned, however, that the unsaturated fatty acids. The fabA(Ts) gene codes for a putative translocation product, fatty acyl-CoA, has not been thermolabile ,B-hydroxydecanoyl thioester dehydrase (14). detected intracellularly either. In contrast to Klein et al. (30), ThefadL gene product was first identified by Ginsburgh et Rock and Jackowski (59), on the basis of studies showing al. (23), who compared the membrane proteins offadL+ and that fadD strains incorporate approximately 2% of the fadL strains. These workers demonstrated a close correla- amount of labeled fatty acids as do wild-type cells, con- tion betweenfadL+-encoded transport activity and the pres- cluded that the fadD gene product is not required for fatty ence of a 33,000-dalton inner membrane protein (23). How- acid transport. These studies are difficult to assess because ever, it is unlikely that the fadL gene product is an inner Rock and Jackowski (59) only measured the incorporation of membrane, because the fatty acid-binding studies of Nunn et labeled fatty acid into membrane lipids. Since these workers al. (45) suggest that it is located in the outer membrane (see did not show that theirfadD strain was capable of delivering below). To show that the fadL gene was the structural gene LCFAs into its cyosolic fraction (59), their studies have not for the 33,000-dalton protein, the fadL gene was cloned by definitively disproved the contention of Klein et al. that the Black et al. (6) and translated in vitro. Further analysis fadD gene product is essential for delivering fatty acids into indicated that thefadL gene product (FLP) has an isoelectric the cytosol of E. coli (30). Obviously, more studies, espe- point (pl) of 4.6 and its molecular weight, as determined by cially in vitro reconstitution experiments, will be required to sodium dodecyl sulfate-gel electrophoresis, was dependent resolve the role of the acyl-CoA synthetase in fatty acid upon sample treatment before electrophoresis. The apparent transport. The fact that mutations in the fadD gene disrupt molecular weight was 33,000 when membrane fractions were fatty acid transport more drastically then mutations in any of heated to 50°C, whereas treatment at 100°C modified the the other fad genes except fadL warrants a more thorough protein such that it migrated more slowly, at an apparent inspection of the involvement of this gene in the transport molecular weight of 43,000 (Fig. 6). The heat-modifiable process. nature of FLP suggests that it may assume different confor- The fadL gene, which is located at 50 min on the E. coli mations, as do many outer membrane proteins (43). Black 184 NUNN MICROBIOL. REV.

TABLE 2. [3H]oleate binding by fadL, fadD, and fadD fadL The binding studies of Nunn et al. (45) suggested that FLP strainsa may be located on the outer membrane, contrary to the [3H]oleate bound findings of Ginsburgh et al. (23). Since the binding studies Strain Genotype (pmol/mg of protein) showed that bovine serum albumin (BSA) removed over 95% of the LCFAs bound to fadD fadL+ strains (45), no LS6164 fadL 65 has been presented to indicate how the LS6928 fadDb 442 logical explanation fadD fadL 63 67,000-dalton BSA can penetrate the outer membrane to LS6929 strip LCFAs off a putative inner membrane protein. This a Binding of [3H]oleate was assayed as descirbed by Nunn et al. (45). dilemma has provoked more research on the location of b LS6928 contains a wild-typefadL gene. FLP. Preliminary experiments by Black and Nunn indicate that antibody to FLP cross-reacts with an outer membrane protein which has the same molecular weight as FLP (un- data). If the monomeric molecular weight of FLP is 33,000, published data). Furthermore, these workers have not been these results would suggest that the native form of FLP may able to repeat the findings of Ginsburgh et al. (23). Given the be a tetramer of 33,000-molecular-weight subunits. Alterna- fact that the lipopolysaccharide-rich outer membrane has an tively, if the monomeric molecular weight of FLP is 43,000, unusually low permeability toward hydrophobic molecules the native form of FLP may be a trimer of 43,000-molecular- (43), it is not unrealistic to assume that FLP is an outer weight subunits. membrane LCFA receptor which may also function as a Early studies to discern the role of the fadL gene product pore to increase the permeability of LCFAs across this in the fatty acid transport process entailed establishment of membrane. the kinetic parameters of MCFA and LCFA transport in As indicated above, the LCFA-binding studies with fadD fadL+ andfadL strains. Maloy et al. (40) showed thatfadL+ fadL+, fadD fadL, and fadL strains suggest that LCFAs strains were capable of transporting LCFAs by an active encounter the fadL gene product prior to being activated by saturable process. MCFA uptake by fadL+ strains charac- the fadD gene product. Further evidence supporting this teristically showed Michaelis-Menten curves that suggested contention was obtained when an analysis of the lipids the presence of both a saturable and a nonsaturable compo- bound to the membranes of the fadD and fadL mutants nent (40). In contrast, kinetic analysis of MCFA transport by revealed that over 96% of the LCFAs were not esterified fadL strains indicated that it was a nonsaturable process. (45). A more extensive analysis of the complex compo- These results suggest that thefadL gene product is required sition offadD and fadL strains by Rock and Jackowski (59) for LCFA transport and can function in MCFA transport, has revealed that fadD strains, but not fadL strains, incor- but is not essential. Since thefadL strains were incapable of porate about 2% as much exogenous LCFAs into their transporting LCFAs (40), these kinetic studies confirmed the membrane lipids as does their wild-type parent (fadD+ contention of Nunn and co-workers (40, 47, 48) that thefadL fadL+). The incorporation of exogenous fatty acids into gene product was essential for LCFA transport into E. coli. phospholipids of the wild type originates with the activation Since both thefadD andfadL gene products are required to of fatty acids by the acyl-CoA synthetase, followed by the deliver LCFAs into the cytosol, the kinetic studies did not distribution of the acyl moieties into all phospholipid classes reveal an unambiguous role for thefadL gene product in the via the sn-glycerol-3-phosphate (59). This LCFA transport process. Therefore, following the identifi- pathway does not function in fadD mutants. Instead, fadD cation of the fadL gene product as a membrane protein mutants, with the aid of a functional fadL gene product(s), (referred to as FLP), the next question that was addressed translocate exogenous fatty acids to an inner membrane was whether FLP is a receptor, a , or both. phospholipid enzyme system, the acyl-acyl carrier protein To examine the function of FLP as a receptor, a study synthetase-2-acyl-glycerophosphoethanolamine acyltrans- which involved the comparison of LCFA-binding activity in ferase system (59), which introduces the fatty acid exclu- fadDfadL+ andfadDfadL strains was performed (45). The sively into the 1 position of phosphatidylethanolamine (59). fadD mutation was included in these strains to avoid LCFA SincefadL strains are incapable of incorporating exogenous transport and LCFA binding contributed by the acyl-CoA fatty acids into phospholipids via any route, Rock and synthetase. The binding studies revealed that strains con- Jackowski (59) suggested that the fadL gene product must taining a functional fadL gene bind significantly more play a role in translocating these fatty acids to intracellular LCFAs than do strains containing a defective fadL gene (Table 2). In addition,fadDfadL strains harboring multicopy fadL+ plasmids bind a 16-fold-higher amount of LCFAs than TABLE 3. Binding of [3H]oleate in afadDAfadL strain do fadD fadL strains harboring only the plasmid vector containing various plasmids (Table 3). Furthermore, although a fadD fadL+ strain has [3H]oleate bound (pmol/mg of protein) Plasmida sixfold more energy-independent LCFA-binding activity Without BSAb With BSA With Brij 58C than a fadD fadL strain, a fadD+ fadL strain has the same LCFA-binding activity as a fadD fadL strain (Table 2). pACYC177 185 64 71 These results imply that, in the LCFA transport process, the pACC 3,174 995 845 fadL gene product rather than the fadD gene product (acyl- pAEV 3,025 968 817 CoA is for sequestering significant pACS 1,725 68 82 synthetase) responsible pACK 1,550 69 78 quantities of LCFAs at sites in the cell membrane for pACP, pBEB, pABC, <180 -<64 <68 transport. Studies with fadL strains harboring plasmids pAPP, or pLSS which encode defective fadL gene products indicate that mutations altering the physical properties of FLP also alter a The plasmids were harbored in the fadD fadL strain LS6929. b The binding assay was performed as described by Nunn et al. (45) in the the fadL-specific LCFA-binding activity (45). Collectively, absence of BSA. the binding studies suggested that one role of FLP in the c The assay was as described by Nunn et al. (45), except that BSA was transport process is to function as a LCFA receptor. replaced with the detergent Brij 58 at a final concentration of 0.5%. VOL. 50, 1986 FATTY ACID CATABOLISM IN E. COLI 185 pools which are accessible to both phospholipid biosynthetic and (iii) SCFAs do not competitively inhibit the uptake of pathways. Since these studies do not unequivocally show MCFAs and LCFAs (60, 61). Therefore, it appears that the that exogenous fatty acids can enter the cytosol of fadD SCFA uptake system is distinct from the uptake system(s) mutants, it is more likely that thefadL gene product delivers that concentrate MCFAs and LCFAs. Evidence suggesting exogenous fatty acids to inner membrane pools that are that AA-CoA transferase may play a role in the transport of accessible to these pathways. SCFAs was obtained from the biochemical studies of Frer- Mutants which lack acyl-CoA dehydrogenase (fadE), man (19). By comparing the uptake of C4 into membrane thiolase (fadA), or hydroxyacyl-CoA dehydrogenase and vesicles of E. coli strains taken from noninduced and in- enoyl-CoA hydratase (fadB) activities transport significantly duced (i.e., in the presence of AA) cells, Frerman showed less MCFA and LCFA than do their wild-type parents (40, that membrane vesicles from induced cells translocate C4. 51). These findings imply that fatty acid transport is coupled This uptake was stimulated by ATP and acetyl-CoA and did to fatty acid oxidation. Although the mechanism(s) is un- not occur in membrane vesicles from noninduced cells (19). known, it is conceivable that transport is reduced in these Frerman also found that significant amounts of AA-CoA fad mutants as a consequence of feedback inhibition. For transferase were associated with the membranes and that example, the accumulation of fatty acyl-CoA intermediates uptake was rapidly inhibited by butyryl-CoA and acetate, in the fad mutants may be inhibitory to the fatty acid the products of the AA-CoA transferase-catalyzed reaction transport components (i.e., FLP or acyl-CoA synthetase). (Fig. 3). Although these results suggest a role for AA-CoA Needless to say, P-oxidation is not an absolute requirement transferase in SCFA uptake, they do not unequivocally for transport, because unsaturated fatty acid auxotrophs establish whether this enzyme is solely responsible for which are unable to oxidize fatty acids take up and incorpo- SCFA transport. The fact that genetic studies showing that rate exogenous unsaturated fatty acids into their phos- mutations not only in atoA but also in atoB andfadAB genes pholipids (17, 53). In contrast, conditional unsaturated fatty reduce SCFA transport suggest that SCFA uptake is also acid auxotrophs [fabA(Ts)] which have fadD orfadL muta- contingent upon subsequent SCFA metabolism. Obviously, tions become nonviable at nonpermissive temperatures be- more studies must be performed to unravel the molecular cause they cannot incorporate sufficient unsaturated acids details of the SCFA transport process. into phospholipids. The results with the fabA(Ts) fadD and fabA(Ts) fadL strains are further testimony that defects in REGULATION OF FATTY ACID DEGRADATION LCFA transport components (encoded by fadL and fadD) The fatty-acid degradative (fad) system is primarily re- not only restrict E. coli from transferrring exogenous LCFAs sponsible for the transport, acylation, and ,-oxidation of across the membrane into the cytosol where they are de- MCFAs and LCFAs. Thefad structural genes, which map at graded, but also from delivering these molecules to sites at no fewer than four distinct loci on the E. coli chromosome which bulk membrane phospholipid synthesis occurs. Since (Fig. 2), are regulated by the fadR gene (49, 51, 64, 65). fadE and fadAB mutations do not restrict fabA(Ts) strains SCFA metabolism requires, in addition to the FAO en- from incorporating sufficient unsaturated fatty acids into zymes, at least two degradative enzymes (Fig. 3), which are their lipids to satisfy their unsaturated fatty acid auxotrophy, encoded by the atoA and atoB genes. The ato structural it is clear that defects in the cytosolic fatty acid-degradative genes appear to be regulated by the atoC gene (56). enzymes do not affect the uptake of LCFAs as adversely as do defects in the membrane FLP and peripheral membrane acyl-CoA synthetase. fad Regulation The process of fatty acid uptake requires that fatty acids The fadR+ gene, a multifunctional regulatory gene map- traverse both the outer and inner membranes of E. coli. In ping at 25.5 min, appears to exert negative control over the addition to the fadL mutation, at least two other mutations fad regulon (51, 64, 65) and the ace operon (39, 41, 42). The (ompC and ompF) reduce the entrance of LCFAs into the fadR+ gene is also required for maximal expression of cell. The ompC and ompF genes encode proteins that make unsaturated fatty acid (UFA) biosynthesis (fab) (46). Con- up part of the outer membrane (43). Hydrophilic molecules siderable evidence has accrued which suggests that thefadR are thought to pass through the outer membrane by specific gene product is a repressor that exerts control over the fad carriers or via nonspecific pores formed by the outer mem- regulon and ace operon at the level of transcription (see brane porins (43). Kinetic studies by Maloy et al. (40) have below). Overath and co-workers (51, 52) and Weeks et al. shown that strains defective in ompC and ompF have a (76) first established that LCFAs can induce FAO enzymes. reduced ability to transport MCFAs and LCFAs. These MCFAs can be degraded by the FAO enzymes, but do not studies suggested that there is a nonspecific fadL- induce their synthesis. Mutants able to utilize MCFAs as independent mechanism which allows fatty acids to traverse sole carbon sources have been isolated by the plating of the outer membrane. Alternatively, the outer membrane wild-type cells onto minimal medium containing the MCFA proteins may be required for FLP to be correctly positioned decanoate as sole carbon source. Overath and co-workers in the outer membrane. If so, the lower transport activity first showed that mutants obtained in this way were consti- observed in ompC and ompF strains (40) might be due to tutive for the FAO enzymes and could rapidly oxidize both inefficient receptor or pore activity, or both, of a displaced MCFAs and LCFAs (51, 52). These mutants have been FLP. termedfadR mutants, and Overath et al. (51) suggested that the fadR gene codes for a regulatory protein, possibly a SCFA Transport repressor. Several lines of evidence now support the original hy- E. coli can also transport SCFAs. However, the mecha- pothesis of Overath et al. (51) that the fadR gene product is nism(s) by which SCFAs are transported differs from that a diffusible repressor protein. ThefadR gene maps at a locus involved in the uptake of MCFAs and LCFAs as follows: (i) distinct from all other fad loci (64, 72). Mutants that harbor the acyl-CoA synthetase is not required for the uptake of transposon-mediated insertion mutations in fadR are consti- SCFAs (19); (ii) SCFAs are not transported byfadR strains; tutive for FAO enzymes. ThefadR' allele is trans-dominant 186 NUNN MICROBIOL. REV.

TABLE 4. Transcriptional control of X'F(fadE-lacZ+) by different of whether free fatty acids or their acyl-CoA derivatives fadR plasmids (15) are the inducers of the fad regulon must await in vitro ,B-Galactosidase activitya studies. fadR genotype of with the following growth Growth offadR mutants in the present of D-glucose causes Plasmid and strain X4 (fadE-lacZ+) conditions': a severe repression offad transcriptional activity (55). Pauli TB TB lac TB ole et al. (55) reported that this repression is partially relieved by the addition of cyclic adenosine 3',5'-monophosphoric acid pACYC177 (cyclicAMP) to the growth medium and that mutants lacking LS6926 Wild type 173 82 743 a functional cyclicAMP receptor protein cannot be induced LS6927 fadR::TnJO 932 287 837 for the FAO enzymes. The expression of the FAO enzymes pACfadR thus requires both cyclicAMP and its receptor protein. The LS6926 Wild type 32 24 877 ability of other class A catabolites (54) to repress the fad LS6927 fadR::TnlO 24 22 150 genes has not been reported, although glycerol does repress to some extent (52). a Values are in nanomoles per minute per milligram of protein. Escherichia coli K-12 mutants constitutive for the synthe- bTB is TB medium supplemented with 0.5% Brij 58; TB lac is TB medium sis of the enzymes of fatty acid degradation synthesize supplemented with 0.5% Brij 58 and 0.4% lactose; TB ole is TB medium significantly less unsaturated fatty acids (UFA) than do supplemented with 0.5% Brij 58 and 5 mM oleate. All were supplemented with wild-type (fadR+) strains (46), both in vivo a,nd in vitro. This ampicillin or kanamycin. defect is asymptomatic unless the fadR strain also carries a lesion in fabA, the structural gene for P-hydroxydecanoyl- thioester dehydrase. Unlike fadR+ fabA(Ts) mutants, fadR to fadR (65). Furthermore, although inducible, fatty acid fabA(Ts) strains synthesize insufficient UFA to support their oxidation in fadR strains harboring multicopy fadR+ plas- growth even at low temperatures and therefore must be mids is at least twofold lower than in wild-type strains supplemrented with UFA at both low and high temperatures containing one copy of the fadR+ gene (15). (46). The low levels of UFA in fadR strains are not due to Evidence suggesting that the fadR gene product regulates their constitutive level of fatty acid-degrading enzymes (46). the expression of the fad regulon at the level of transcription Thus a functional fadR gene seems necessary for the maxi- was obtained from studies with lac fusions of the fad mal expression of UFA biosynthesis. These studies did not structural gene (13, 62). The studies of Clark (13) with strains shed any light on the mechanism(s) by which fadR controls harboringfadAB-lac andfadE-lac operon fusions and Sallus UFA synthesis. The fadR gene may act as an activator to et al. (62) with fadL-lac operon fusions revealed that - positively regulate the transcription of the fabA gene prod- galactosidase activity is constitutive in fadR-lac strains and uct. Alternatively, the fabA gene product may be allosteri- inducible by LCFAs infadR+ fad-lac strains. Furthermore, cally inhibited by an as yet unidentified factor(s) expressed the expression Qf p-galactosidase was repressed in these in fadR but not fadR+ strains. strains under catabolite-repressing growth conditions. Not- ing that FAO enzyme activity in strains harboring a ato System multicopy fadR+ plasmid is fourfold less than in the wild type under noninducing conditions, DiRusso and Nunn (15) The atoA and atoB genes encode enzymes responsible for determined whether fad gene transcriptional activity was SCFA degradation (Fig. 3). These genes are clustered at the regulated similarly by the multicopyfadR' plasmids. Using 47-min region on the revised E. coli linkage map (1, 56). The strains with fadE-lac operon fusions, these workers showed P-keto SCFA AA serves as metabolic inducer for the ato that 0-galactosidase activity in strains harboring the system. When AA is used as the sole carbon source, a 200- multicopyfadR' plasmid was fivefold lower than that found to 300-fold induction of both AA-CoA transferase and in wild-type strains (one copy offadR+) under noninducing thiolase II is observed in wild-type strains (56). conditions (Table 4). These studies suggest that fad tran- Although AA may be utilized byfadR strains,fadR strains scriptional activity can be further repressed by the presence cannot grow on the saturated SCFA C4 or C5, because these of more copies of thefadR gene product. Overall, thefad-lac substrates do not induce the ATO enzymes (56, 60, 73). Only operon fusion studies strongly suggest that the fadR gene strains with constitutive levels of the ATO and FAO en- regulates the fad regulon at the level of transcription. zymes can utilize C4 and C5 as sole carbon source. Consti- In studies on 24 rnerodiploid strains harboringfadR alleles tutive levels of the atoA and atoB gene products result from on the chromosome and an episome, Simons et al. (65) found a mutation(s) in a regulatory gene, atoC, which also maps at that all displayed the fadR phenotype. These studies sug- the 47-min region (56). Mutants (l3ut') able to utilize C4 or gested that only one polypeptide was encoded by the fadR C5 as sole carbon sources are readily selected by plating gene. The latter presumption was confirmed when Di Russo fadR strains onto minimal medium containing butyrate (56, and Nunn (15) cloned the fadR gene and identified a 29,000- 60, 61, 73). Pauli and Overath (56) first showed that most dalton protein as its product. But+ mutants obtained in this way were constitutive for the At present, the identity of the inducer of thefad regulon is ATO enzymes. The mutation causing constitutivity of the not clear. Mutants defective in the fadD and fadL genes are ATO enzymes has been termed atoCc (56), and the geno- unable to induce the other FAO enzymes, whereas mutants type of strains which utilize C4 and C5 as sole carbon sources with mutations in the fadA, fadB, and fadE genes remain is fadR atoCc. inducible for the remaining functional FAO enzymes and All available evidence suggests that the atoC gene codes transport activities (30, 47). When Klein et al. (30) found that for a regulatory protein that exerts positive control ofthe ato fadD mutants were noninducible, they postulated that long- structural genes. Pauli and Overath (56) examined the effect chain acyl-CoA are the in vivo inducers of thefad of a mutation in atoC (atoCc49) on the activity of the ato regulon. However, fadD mutants, likefadL mutants, cannot structural gene products in cells merodiploid for atoC (56) transport LCFAs into their cytosol. Therefore, the question and found that all merodiploids which had atoC+A+B+ on VOL. 50, 1986 FATTY ACID CATABOLISM IN E. COLI 187 structural gene products in cells merodiploid for atoC (56) and found that all merodiploids which had atoC+A+B+ on the episome and atoCcA+B- or atoCcA-B+ on the chromo- some had the But' phenotype and constitutive levels of AA-CoA transferase and thiolase II activities. Because atoCc appeared to be trans-dominant to atoC+, Pauli and Overath (56) hypothesized that atoC codes for a positive regulatory element. Comparisons of the ato system with other well-characterized positive regulatory systems (i.e., arabinose [ara] operon [16] and D- deaminase [dsd] system) reveal many differences. For example, the fre- quency of obtaining constitutive mutants in the positive regulatory elements encoded by araC and dsdC is very low; in contrast, the frequency of obtaining spontaneous atoCc mutants is relatively high (10-5 to 10-6). The majority of dsdCc and arac mutants express low constitutive enzyme levels (7, 11, 16, 24), and araC+ or dsdC+ are partially dominant to araCc or dsdCc, respectively (7, 11, 16, 24); in contrast, the single atoCc49 mutant investigated (56) dis- played high constitutive levels of the ATO enzymes (higher than inducible levels), and the atoCc allele was fully trans- dominant to atoC+. It must also be noted that the FIG. 7. Glyoxylate shunt in E. coli and related reactions. merodiploid studies of Pauli and Overath were done with atoCc on the chromosome (56). Since these mutants were heavily mutagenized to get atoCc atoA or atoCc atoB geno- of these pathways in SCFA metabolism is not known at this types, there may have been host chromosomal mutations time. which caused secondary effects on the ato regulatory sys- To degrade odd-chain SCFA C3 and C5, E. coli must be tem. able to metabolize C3. Although Wegener et al. have pro- An interesting aspect of SCFA metabolism that requires posed pathways by which odd-chain fatty acids are de- further investigation is the finding that at least two types of graded, the precise pathway by which this occurs is un- spontaneous But+ mutants can be selected by plating fadR known. One gene, prp, which maps at 97 min, is known to be mutants onto minimal media containing butyrate (56, 60, 61). required for C3 and C5 metabolism (67). The function of the As indicated above, one type is the atoCc mutants, which prp gene has not been established. constitutively synthesize the ATO enzymes (56). The second type of But+ mutant, first isolated by Salanitro and Wegener GENETIC AND BIOCHEMICAL FEATURES OF (60, 61), can grow on C4 or C5 after a 4- to 6-h lag. These ACETATE METABOLISM workers showed that C4 uptake was inducible in the latter type of mutants, whereas C4 uptake was constitutive in Glyoxylate Shunt atoCc mutants (60, 61). To explain the phenotype of their In addition to the FAO enzymes, the expression of the inducible But+ mutant, Salanitro and Wegener (60) proposed glyoxylate shunt enzymes (Fig. 7) is also required for the that it might have a regulatory mutation that alters the growth of E. coli on fatty acids and acetate as a sole carbon inducer of the regulatory proteins. Since the source. In wild-type E. coli, repression of the ace operon is mutation(s) causing the inducible But+ phenotype was never under the control of the two genes fadR and iciR. The mapped by Salanitro and Wegener (60, 61), it is not clear studies of Maloy and Nunn (41, 42) suggest that both the iciR whether this mutation affects the atoC gene directly or and fadR genes regulate the glyoxylate shunt in a trans- whether it affects some, as yet, undefined ato regulatory dominant and synergistic manner at the level of transcrip- locus. tion. The above studies suggest that the regulatory system(s) Growth on fatty acids as a sole carbon source requires not for SCFA metabolism may be considerably more complex only the transport and degradation of fatty acids, but also than that proposed by Pauli and Overath (56). In addition, utilization of the two-carbon acetyl-CoA units generated via studies by Pauli and Overath (56) and Salanitro and Wegener ,-oxidation (26, 32). Acetyl-CoA is mainly catabolized by (60, 61) indicate that the SCFA atoCc regulatory defect the TCA cycle. However, since each turn of the TCA cycle somehow affects . Observations in involves the loss of two carbon atoms as CO2, no net support of the latter suggestion have come from studies assimilation of carbon from acetyl-CoA can occur by this showing that (i) fadR atoCc strains grow considerably more means. Therefore, growth on substrates such as fatty acids slowly than their fadR parent on MCFAs and LCFAs (56, or acetate, which are catabolized solely to acetyl-CoA, 60), and (ii) fadR atoCc strains have considerably less requires the operation of a separate anaplerotic pathway to crotonase and HOADH activity than theirfadR parent (60). replenish necessary intermediates for cellular biosynthesis. The question arising from these phenomena is whether these This is accomplished by the glyoxylate shunt (26). The net alterations in fatty acid metabolism are primary or secondary effect of the glyoxylate shunt is the formation of 1 mol of consequences of the atoCc mutation. dicarboxylic acids from 2 mol of acetyl-CoA. The glyoxylate Alternative routes of SCFA metabolism have been dis- shunt is the only known anaplerotic pathway allowing cussed by Wegener et al. (77) in a review. Basically, these growth on acetyl-CoA, and it thus occurs in all other workers have summarized in vitro studies which show that organisms that utilize fatty acids or acetate as a sole carbon there are enzymes which appear to condense glyoxylate with source (26). various SCFAs. Although very interesting, the significance The two unique enzymes of the glyoxylate shunt, isocit- 188 NUNN MICROBIOL. REV.

TABLE 5. Specific actvities of glyoxylate shunt enzymes in iclR and fadR mutantsa Sp act (nmollmin per mg of protein) for: Relevant Isocitrate Malate synthase Strain genotype Succinate Acetate + acetate Succinate Acetate Succinate K-12 Prototrophic 21 244 70 103 410 186 SM6034 iclR 209 320 276 267 594 330 RS3040 fadR TnJO 172 272 204 371 566 389 SM6042 iclRfadR TnJO 283 396 386 396 592 402 a Experimental assay conditions were performed as described by Maloy and Nunn (42). rate lyase and malate synthease A, are normally induced of IDH in excess of 10-fold, maintaining a nearly constant only when E. coli is grown on acetate or fatty acids (31, 71). activity for IDH during growth on acetate. Given the role it The structural genes for , aceA, and malate plays in the glyoxylate bypass reaction, it is not surprising synthase A, aceB, map at 90 min on the E. coli K-12 that the aceK gene is part of the ace operon. chromosome (Fig. 2). Mutations in an adjacent gene, iciR, The utilization of acetate, whether for lipid biosynthesis, produces constitutive levels of isocitrate lyase and malate oxidation via the tricarboxylic acid cycle, or replenishment synthase. Since the aceA and aceB genes were found to be of TCA dicarboxylic acids via the glyoxylate bypass, re- closely linked and coordinately expressed, Brice and quires that it first be activated to acetyl-CoA. Two mecha- Kornberg (9) postulated that the aceA and aceB genes may nisms that bring about this conversion have been elucidated. form an operon controlled by a repressor protein encoded by In one, acetyl-CoA synthetase (acetate:CoA ligase [AMP- the iclR gene. From genetic and biochemical studies with forming]; EC 6.2.1.1) catalyzes the acetylation of CoA mutants containing TnJO insertions in their aceA or aceB concomitant with the cleavage of ATP to AMP and inorganic genes or both, Maloy and Nunn confirmed that the aceAB pyrophosphate (3, 38, 70; N. 0. Kaplan and F. Lipman, genes make up an operon which is transcribed from the aceB Fed. Proc. 7:163, 1948). In the other, two enzymes catalyze to the aceA gene (42). (i) the conversion of acetate to acetyl phosphate, with cleavage of ATP to adenosine diphosphate (ADP), and (ii) Other Ace Enzymes the transfer of the acetyl moiety from acetyl phosphate to CoA, with liberation of inorganic phosphate (12, 27, 38, 70). The glyoxylate shunt diverts isocitrate from the Krebs Acetate kinase (ATP acetate phosphotransferase; EC cycle, bypassing the C02-producing steps (Fig. 7). During 2.7.2.1), which is encoded by the ack gene, catalyzes reac- growth on acetate (or fatty acids), the flow of isocitrate tion (i), and phosphotransacetylase (acetyl-CoA:orthophos- through the glyoxylate bypass is also controlled via the phate ; EC 2.3.1.8), encoded by the pta of (IDH), the gene, catalyzes reaction (ii). The last two genes have been Krebs cycle enzyme which competes with isocitrate lyase (2, mapped near purF, at 48.5 to 49 min on the E. coli linkage 71, 75). The latter control mechanism is accomplished by the map (37). Acetyl phosphate is thought to be an intermediate phosphorylation of IDH by a bifunctional protein, the IDH in the activation of acetate to acetyl-CoA (10). Also, it has kinase-phosphatase (21, 22, 27). Since the phosphorylated been implicated as the energy source for transport systems form of IDH is completely inactive (8, 34), phosphorylation utilizing periplasmic binding proteins (28). The levels of forces isocitrate through the glyoxylate bypass (36, 44). acetate kinase and phosphotransacetylase in extracts of Recently, Laporte et al. presented genetic and biochemical wild-type E. coli do not vary with different carbon sources evidence that suggests that kinase and phosphatase activities (10), suggesting that the expression of the ack and pta genes are catalyzed by a single protein encoded by aceK (36). The is not induced by acetate or catabolite repressed by glucose. latter gene was mapped in the ace operon downstream from Mutants defective in either the ack or pta genes are severely aceB and aceA (36). Through the cloning and characteriza- impaired in utilizing acetate as a sole carbon source and are tion of the aceK gene, Laporte and Chung (33) presented incapable of incorporating labeled acetate when grown on evidence that this gene encodes the 66,000-dalton IDH glucose (10). However, ack and pta mutants grown on kinase-phosphatase. It is not clear whether this bifunctional glycerol are capable of incorporating labeled acetate. Brown enzyme has two distinct active sites per monomer, one et al. (10) performed studies which suggest that an inducible catalyzing the kinase reaction and the other catalyzing the acetyl-CoA synthetase enables ack and pta mutants to phosphatase reaction, or one per monomer which incorporate labeled acetate. Since these mutants are unable catalyzes both reactions. Laporte and Chung (33) proposed to incorporate acetate when grown on glucose (10), the that the two reactions occur at independent sites and that the acetyl-CoA synthetase, like the glyoxylate shunt enzymes, phosphatase domain is produced through a partial duplica- appears to be regulated by catabolite repression. Overall, tion of a primordial IDH kinase gene, leaving a fused these studies suggest that the acetate kinase and product. Evaluation of this hypothesis must await the deter- phosphotransacetylase are required for E. coli to grow mination of the nucleotide sequence of aceK. The phosphor- optimally on acetate as a carbon source and to incorporate ylation system can respond to variations in the intracellular acetate under catabolite-repressing growth conditions. Fur- levels of IDH. From studies which involved the comparison thermore, the studies with the ack and pta mutants suggest of the levels of IDH activity and phosphorylation in strains that the acetyl-CoA synthetase provides these organisms containing either one or multiple copies of the IDH struc- with an alternate route for conversion of exogenous acetate tural gene, Laporte et al. (35) showed that the phosphoryla- to acetyl-CoA. It is not clear whether any other enzymes or tion system can compensate for changes in the cellular level transport proteins are required for the uptake of acetate, VOL. 50, 1986 FATTY ACID CATABOLISM IN E. COLI 189

TABLE 6. Mu d(Ap lac) operon fusions with the aceA genea TABLE 8. Specific activities of glyoxylate shunt enzymes in strains merodiploid for the fadR gene Strain Relevant genotype Growth 3-Galactosidase conditions"b activity (U) Sp act (nmollmin per mg TL1 iclR fadR AlacZ StrainStrain fadR Growth Isocitrateof protein) of:Malate N <1 genotypea conditions I <1 lyase synthase M11-i-S 0/+ SMUD-1 aceA::Mu d(Ap lac) iclR+ Succinate 11 42 fadR+ AlacZ 198 N 20 Acetate 95 I 310 Oleate 125 253 M2-5-S 0/- SMUD-2 aceA::Mu d(Ap lac) iclR Succinate 81 111 fadR+ AlacZ Acetate 106 315 N 300 I 310 Oleate 128 320 M11-i +/+ SMUD-3 aceA::Mu d(Ap lac) iclR+ Succinate 7 36 fadR AlacZ Acetate 84 183 N 200 I 440 Oleate 83 238 M12-1 +/- SMUD-4 aceA::Mu d(Ap lac) iciR Succinate 12 45 fadR AlacZ Acetate 64 205 N 650 Oleate 121 284 I 700 a Experimental conditions were performed as described by Maloy and M47-1 -/+ Nunn (42). Succinate 18 52 bGrowth conditions: N, noninducible; I, inducible. Acetate 60 210 Oleate 146 277 because no mutants deficient in acetyl-CoA synthetase ac- M23-1 -/- tivity have been isolated or characterized. Succinate 62 125 Acetate 95 329 Regulation of ace System(s) Oleate 122 336 Evidence that the fadR gene plays a role in acetate afadR genotype shown as episomal allele/chromosomal allele. A zero metabolism was first noted when Maloy and Nunn (41) indicates that no episome was present. observed that acetate was incorporated intofadR strains at a considerably higher rate than into isogenic fadR+ strains. Biochemical studies demonstrated that the increased rate of by fadR mutants did require a functional glyoxylate shunt. acetate incorporation by fadR mutants was not due to an Enzyme studies (39) further showed that the levels of increased rate of macromolecular synthesis or degradation glyoxylate shunt enzymes were elevated in fadR mutants or to differences in the acetyl-CoA pool size or in the under noninducing growth conditions (Table 5). These stud- activities of enzymes required for acetate transport or oxi- ies strongly suggested that the activity of the glyoxylate dation. However, the increase rate of acetate incorporation shunt enzymes is regulated by the fadR gene. To determine whether the fadR or iciR genes, or both, regulated the expression of the ace operon at the level of TABLE 7. Specific activity of isocitrate lyase in strains transcription, Maloy and Nunn (42) constructed ace-lac merodiploid for the iclR genea operon fusions and demonstrated that both iciR and fadR negatively regulate the ace operon at the level of transcrip- Isocitrate lyase sp act (nmol/min per mg of protein) tion. Interestingly, iclR fadR mutants have higher levels of Strain for: the glyoxylate shunt enzymes than do iciR fadR+ and iclR+ Relevant genotype fadR mutants. When the same mutants contain ace-lac Succinate + acetate Acetate operon fusions, the iclR fadR double mutants also have higher levels of P-galactosidase activity than do the iciR MM-1 0/aceA+ icOR+ 21 98 316 fadR+ and iclR+ fadR mutants (Table 6). Merodiploid stud- MM-2 0/aceA+ iclR 206 299 355 ies demonstrated that both the iclR and fadR genes regulate MM-4 0/aceB iclR <0.5 <0.5 <0.5 the of the shunt in a trans-dominant MM-5 F' aceA+ icIR+laceA+ iclR+ 31 50 114 expression glyoxylate MM-6 F' aceA+ iclR+/aceA icOR+ 18 45 206 manner (Tables 7 and 8). Although several models could be MM-7 F' aceA+ iclR+/aceA icOR+ 52 79 305 proposed from these studies, the simplest interpretation is MM-8 F' aceA icIR+/aceA+ icIR+ 3 31 209 that the iclR and fadR gene products act independently and MM-9 F' aceA iclR+laceA+ iclR 19 69 276 synergistically to cause repression of the ace operon; acting MM-10 F' aceA+ iclR/aceA iclR+ 23 74 298 together they cause the full repression of the ace operon. MM-11 F' aceA+ iclR/aceA+ iclR 292 326 422 MM-12 F' aceA iclR/aceA+ icOR+ 17 81 310 CONCLUSIONS MM-13 F' aceA iclR/aceA+ iclR 269 285 387 Most of our knowledge regarding the utilization of fatty a Experimental growth and assay conditions were performed as described acids or acetate or both as sole carbon sources stems from by Maloy and Nunn (42). genetic and biochemical studies performed with E. coli. For 190 NUNN MICROBIOL. REV. this organism to metabolize the C2 compound acetate, the transferase has been shown to be required for SCFA trans- glyoxylate shunt enzymes, encoded by the ace operon, must port, but it is not clear whether this enzyme is solely be expressed for the continual replenishment of dicarboxylic responsible for the uptake of SCFAs in this organism. acids drained from the TCA for cellular biosynthesis. In A considerable amount of information regarding the FAO addition, prior to its metabolism the acetate must be acti- enzymes has been learned from studies with E. coli. At least vated to acetyl-CoA by either (i) acetyl-CoA synthetase or five of the FAO enzymes are part of thefadAB multienzyme (ii) acetate kinase and phosphotransacetylase. As indicated complex. Although biochemical studies suggest that the in this review, the synthesis of the glyoxylate shunt enzymes quaternary structure of this complex is not an artifact, it and the acetyl-CoA synthetase are induced by the presence remains to be established whether posttranslational process- of acetate in the growth media of E. coli. The synthesis of ing occurs for the active complex to form. The long-chain these enzymes is also catabolite repressed by glucose. All enoyl-CoA hydratase and acyl-CoA dehydrogenase are not the presently available evidence suggests that induction of part of thefadAB multienzyme complex. The structural gene the ace operon when acetate is present is accomplished by for the long-chain enoyl-CoA hydratase has not been iden- the inactivation of the iciR gene product(s). The nature of the tified to date, and the nature of the only known mutation inducer under these conditions is presumed to be acetyl- (fadE) affecting acyl-CoA dehydrogenase activity has not CoA, but this has not been definitively established. been established. One of the most interesting findings with respect to the ace The most interesting finding, which has not been further operon is the fact that the fadR gene product negatively explored, is the study indicating that the fadR gene plays a regulates its expression. Although the studies described in role in UFA biosynthesis. Conceptually, it makes sense that this review indicate that the ace operon is under the tran- endogenous fatty acid syntheses be restricted when scriptional control of the iclR andfadR genes, the molecular exogenous fatty acids are present in the growth media of a details by which this regulation is accomplished has not been bacterial cell. Therefore, it is not surprising that a regulator elucidated. For instance, it is not known whether the fadR gene like thefadR gene functions not only in the degradation gene product exerts control over the ace operon by directly of fatty acids, but also in their synthesis. Why this gene interacting with cis-acting ace operon regulatory sites or directly or indirectly controls UFA but not saturated fatty interacting with the iciR gene product(s). biosynthesis is not known. Clearly, the most exciting re- The FAO enzymes, coupled with glyoxylate shunt en- search in the future will be directed at the molecular details zymes, enable E. coli to utilize MCFAs and LCFAs as sole by whichfadR exerts its global control on UFA biosynthe- carbon and energy sources. As indicated in this review, sis, fatty acid degradation, and acetate metabolism. wild-type E. coli can only utilize LCFAs as carbon sources because LCFAs induce the synthesis of the FAO enzymes. ACKNOWLEDGMENTS To be capable of utilizing MCFAs, E. coli must have a nonfunctionalfadR gene. All the available data indicate that This work was supported by Public Health Service grant GM22466 the latter gene exerts negative transcriptional from the National Institutes of Health. control of the We thank Rowland H. Davis for helpful comments on the manu- fad regulon. What remains to be determined is the nature of script and Esther Ervin for the figure illustrations. We also thank the inducing substrate and the molecular details by which the Kathy Ruiz for typing the manuscript. fadR gene product interacts with the cis-acting regulatory sites of the fad regulon. To grow on saturated SCFAs as sole carbon sources, E. LITERATURE CITED coli must constitutively express two ATO enzymes and three 1. Bachmann, B. J., and K. B. Low. 1983. Linkage map of FAO enzymes. The regulation of the ATO enzymes appar- Escherichia coli K-12, Edition 7. Microbiol. Rev. 47:180-230. ently requires an activator encoded by the atoC gene (56). 2. Bennett, P. M., and W. H. Holms. 1975. Reversible inactivation To induce or be degraded by the enzymes, of the isocitrate dehydrogenase of Escherichia coli ML308 FAO LCFAs during growth on acetate. J. Gen. Microbiol. 87:37-51. must utilize thefadD andfadL gene products to be delivered 3. Berg, P. 1956. Acyl adenylates: an enzymatic mechanism of to the cytosolic sites at which these events occur. The acetate activation. J. Biol. Chem. 222:991-1013. membrane-bound fadL gene product FLP appears to func- 4. Binstock, J. F., A. Pramanik, and H. Schultz. 1977. Isolation of tion as a LCFA receptor. It is not clear whether FLP also a multienzyme complex of fatty acid oxidation from Escherichia functions as a LCFA permease. Evidence suggesting that coli. Proc. Natl. Acad. Sci. USA 74:492-495. this protein may function as a permease is as follows: (i) 5. Binstock, J. F., and H. Schultz. 1981. Fatty acid oxidation Complementation studies suggest that only one gene is complex from Escherichia coli. Methods Enzymol. 71:403-411. required for LCFA binding and transport; (ii) a hybrid 6. Black, P. N., S. F. Kianian, C. C. DiRusso, and W. D. Nunn. plasmid which solely expresses FLP enablesfadL strains to 1985. Long chain fatty acid transport in Escherichia coli: cloning, mapping and expression of the fadL gene. J. Biol. bind and transport LCFAs; and (iii) mutations in the pro- Chem. 260:1780-1790. moter-distal region of thefadL gene result in the synthesis of 7. Bloom, F. R., E. McFall, M. C. Young, and A. M. Carothers. a physically altered FLP which binds but does not transport 1975. Positive control in the D-serine deaminase system of LCFAs. Escherichia coli K-12. J. Bacteriol. 121:1092-1101. The translocation of MCFAs to the cytosolic FAO en- 8. Borthwick, A. C., W. H. Holms, and H. G. Nimmo. 1984. The zymes can occur actively byfadL-mediated assistance or by phosphorylation of Escherichia coli isocitrate dehydrogenase in a diffusible mechanism. In either case thefadD gene product intact cells. Biochem. J. 222:797-804. is required for MCFAs to be delivered to the other cytosolic 9. Brice, C. G., and H. L. Kornberg. 1968. Genetic control of FAO enzymes. To date, the attempts to isolate mutants isocitrate lyase activity in Escherichia coli. J. Bacteriol. which specifically affect the 96:2185-2186. transport of MCFAs have failed. 10. Brown, T. D. K., M. C. Jones-Mortimer, and H. L. Kornberg. A major difficulty in screening for MCFA mutants is the fact 1977. The enzymatic interconversion of acetate and acetyl- that these fatty acids are more toxic to the cell than are coenzyme A in Escherichia coli. J. Gen. Microbiol. 102: LCFAs. Transport of SCFAs by E. coli does not require 327-336. functional fadD and fadL gene products. The AA-CoA 11. Carothers, A. M., E. McFall, and S. Palchaudhiri. 1980. Physical VOL. 50, 1986 FATTY ACID CATABOLISM IN E. COLI 191

mapping of the Escherichia coli D-serine deaminase region: control. J. Biol. Chem. 259:14068-14075. contiguity of the dsd structural and regulatory genes. J. Bacte- 37. Levine, S. M., F. Ardeshir, and G. F. Ames. 1980. Isolation and riol. 142:174-184. characterization of acetate kinase and phosphotransacetylase 12. Chow, T. C., and IF. Lipman. 1952. Separation of acetyl transfer mutants of Escherichia coli and Salmonella typhimurium. J. enzymes in pigeon extract. J. Biol. Chem. 196:89-103. Bacteriol. 143:1081-1085. 13. Clark, 1). 1981. Regulation of fatty acid degradation in Esche- 38. Lipmann, F. 1944. Enzymatic synthesis of acetyl phosphate. J. richia coli: analysis by operon fusion. J. Bacteriol. 148:521-526. Biol. Chem. 155:55-70. 14. Cronan, J. E., D. F. Silbert, and D. L. Wulif. 1972. Mapping of 39. Maloy, S. R., M. Bohlander, and W. D. Nunn. 1980. Elevated the fabA locus for unsaturated fatty acid biosynthesis in Esch- levels of glyoxylate shunt enzymes in Escherichia coli strains erichia coli. J. Bacteriol. 112:206-211. constitutive for fatty acid degradation. J. Bacteriol. 143: 15. Di Russo, C. C., and W. D. Nunn. 1985. Construction and 720-725. characterization of hybrid-plasmids containing the fadR gene in 40. Maloy, S. R., C. L. Ginsburgh, R. W. Simons, and W. D. Nunn. Escherichia coli. J. Bacteriol. 161:583-588. 1981. Transport of long and medium chain fatty acids by 16. Englesberg, E., and G. Wilcox. 1974. Regulation: positive con- Escherichia coli. J. Biol. Chem. 256:3735-3742. trol. Annu. Rev. Genet. 8:219-242. 41. Maloy, S. R., and W. D. Nunn. 1981. Role of gene fadR in 17. Esfahani, M., T. Ioneda, and S. J. Wakil. 1971. Studies on the Escherichiq coli acetate metabolism. J. Bacteriol. 148:83-90. control of fatty acid metabolism. J. Biol. Chem. 246:50-56. 42. Maloy, S. R., and W. D. Nunn. 1982. Genetic regulation of the 18. Feigenbaum, J., and H. Schulz. 1985. Thiolases of Escherichia glyoxylate shunt in Escherichia coli K-12. J. Bacteriol. coli: purification and chain length specificities. J. Bacteriol. 149:173-180. 122:407-411. 43. Nikaido, H. 1979. Nonspecific transport through the outer 19. Frerman, F. E. 1973. The role of acetyl-coenzyme A in the membrane of bacteria, p. 361-408. In M. Inouye (ed.), Bacterial transferase uptake of butyrate by isolated membrane vesicles of outer membranes, biogenesis and functions. John Wiley & Escherichia coli. Arch. Biochem. Biophys. 159:444 452. Sons, Inc., New York. 20. Frerman, F. E., and W. Bennett. 1973. Studies on the uptake of 44. Nimmo, G. A., and H. G. Nimmo. 1984. The regulatory proper- fatty acids by Escherichia coli. Arch. Biochem. Biophys. ties of isocitrate dehydrogenase kinase and isocitrate dehydro- 159:434- 443. genase phosphatase from Escherichia coli ML308 and the roles 21. Garnak, M., and H. C. Reeves. 1979. Phosphorylation of isocit- of these activities in the control of isocitrate dehydrogenase. rate dehydrogenase of Escherichia coli. Science 203:1111-1112. Eur. J. Biochem. 141:409-414. 22. Garnak, M., and H. C. Reeves. 1979. Purification and properties 45. Nunn, W. D., R. Colburn, and P. N. Black. 1985. Transport of of phosphorylated isocitrate dehydrogenase of Escherichia coli. long chain fatty aci,ds in Escherichia coli: evidence for role of J. Biol. Chem. 254:7915-7920. fadL gene product as long chain fatty acid receptor. J. Biol. 23. Ginsburgh, C. L., P. N. Black, and W. D. Nunn. 1984. Transport Chem. 261:167-171. of long chain fatty acids Escherichia coli: identification of the 46. Nunn, W. D., P. K. Griffin, D. Clark, and J. E. Cronan, Jr. 1983. fadL gene product. J. Biol. Chem. 259:8437-8443. Role for the fadR gene in unsaturated fatty acid biosynthesis in 24. Heincz, M. C., and E. McFall. 1978. Role of the dsdC activation Escherichia coli. J. Bacteriol. 154:554-560. in regulation of D-serine deaminase synthesis. J. Bacteriol. 47. Nunn, W. D., and R. W. Simons. 1978. Transport of long-chain 136:96-103. fatty acids by Escherichia coli: mapping and characterization of 25. Hill, F. F., and D. Angelmaier. 1972. Specific enrichment of mutants in the fadL gene. Proc. Natl. Acad. Sci. USA mutants of Escherichia coli with an altered acyl-CoA synthetase 75:3377-3381. by tritium suicide. Mol. Gen. Genet. 117:143-152. 48. Nunn, W. D., R. W. Simons, P. A. Egan, and S. R. Maloy. 1979. 26. Hillier, S., and W. T. Charnetzky. 1981. Glyoxylate bypass Kinetics of the utilization of medium and long chain fatty acids enzymes in Yersinia species and multiple forms of isocitrate by a mutant of Escherichia coli defective in the fadL gene. J. lyase in Yersinia pestis. J. Bacteriol. 145:452-458. Biol. Chem. 254:9130-9134. 27. Holms, W. H., and P. M. Bennett. 1971. Regulation of isocitrate 49. O'Brien, W., and F. Frerman. 1973. A mutant of Escherichia dehydrogenase activity in Escherich/ia coli on adaption to ace- coli with altered inducer specificity for the fad regulon. tate. J. Gen. Microbiol. 65:57-68. Biochem. Biophys. Res. Commun. 54:697-703. 28. Hong, J., A. G. Hunt, P. S. Masters, and M. A. Lieberman. 1979. 50. O'Brien, W., and F. Frerman. 1977. Evidence for a complex of Requirement of acetyl phosphate for the binding protein- three beta-oxidation enzymes in Escherichia coli: induction and dependent transport systems in Escherichia coli. Proc. Natl. localization. J. Bacteriol. 132:532-540. Acad. Sci. USA 76:1213-1217. 51. Overath, P., G, Pauli, and H. U. Schairer. 1969. Fatty acid 29. Kameda, K., and W. D. Nunn. 1981. Purification and character- degradation in Escherichia coli. An inducible acyl-CoA synthe- ization of acyl coenzyme A synthetase from Escherichia coli. J. tase, the mapping of old-mutations, and the isolation of regula- Biol. Chem. 256:5702-5707. tory mutants. Eur. J. Biochem. 7:559-574. 30. Klein, K., R. Steinberg, B. Fiethen, and P. Overath. 1971. Fatty 52. Overath, P., E. Raufuss, W. Stoffel, and W. Ecker. 1967. The acid degradation in Escherichia coli. An inducible system for induction of the enzymes of fatty acid degradation in Esche- the uptake of fatty acids and further characterization of old richia coli. Biochem. Biophys. Res. Commun. 29:28-33. mutants. Eur. J. Biochem. 19:442-450. 53. Overath, P,, H. U. Schairer, and W. Stoffel. 1970. Correlation of 31. Kornberg, H. L. 1966. The role and control of the glyoxylat'e in vivo phase transitions of membrane lipids in Escherichia coli. cycle in Escherichia coli. Biociem. J. 99:1-11. Proc. Natl. Acad. Sci. USA 67:606-642. 32. Lakshmi, T. M., and R. B. Helling. 1978. Acetate metabolism in 54. Pastan, I., and S. Adbya. 1976. Cyclic adenosine 5' mono- Escherichia coli. Can. J. Microbiol. 24:149-153. phosphate in Escherichia coli. Bacteriol. Rev. 40:527-551. 33. LaPorte, D. C., and T. Chung. 1985. A single gene codes for the 55. Pauli, G., R. Ehring, and P. Overath. 1974. Fatty acid degrada- kinase and phosphatase which regulates isocitrate dehydroge- tion in Escherichia coli: requirement of cyclic adenosine nase. J. Biol. Chem. 260:15291-15297. monophosphate and cyclic receptor 34. LaPorte, D. C., and D. E. Koshland. 1982. A protein with kiqase protein for enzyme synthesis. J. Bacteriol. 117:1178-1183. and phosphatase activities involved in regulatiop of tricar- 56. Pauli, G., and P. Overath. 1972. Ato operon: a highly inducible boxylic acid cycle. Nature (London) 300:458-460. system for acetoacetate and butyrate degradation in Escherichia 35. LaPorte, D. C., P. E. Thorsness, and R. E. Koshland. 1985. coli. Eur. J. Biochem. 29:553-562. Compensatory phosphorylation of isocitrate dehydrogenase: a 57. Pawar, S., and H. Schulz. 1981. The structure of the multi- mechanism for adaptation to the intracellular environment. J. enzyme complex of fatty acid oxidation from Escherichia coli. Biol. Chem. 260:10563-10568. J. Biol. Chem. 256:3894-3899. 36. LaPorte, D. C., K. Walsh, and D. E. Koshland. 1984. The branch 58. Pramanik, A., S. Pawar, E. Antonian, and H. Schultz. 1979. Five point effect: ultrasensitivity and subsensitivity to metabolic different enzymatic activities are associated with the multi- 192 NUNN MICROBIOL. REV.

enzyme complex of fatty acid oxidation from Escherichia coli. properties of Escherichia coli coenzyme A transferase. Arch. J. Bacteriol. 137:469-473. Biochem. Biophys. 171:14-26. 59. Rock, G. O., and S. Jackowski. 1985. Pathway for incorporation 69. Sramek, S. J., and F, E. Frerman. 1975. Escherichia coli of exogenous fatty acids into phosphatidylethanolamine in coenzyme A transferase: kinetics, catalytic pathway and struc- Escherichia coli. J. Biol. Chem. 260:12720-12724. ture. Arch. Biochem. Biophys. 171:27-35. 60. Salanitro, J., and W. Wegener. 1971. Growth ofEscherichia coli 70. Stadman, E. R., and H. A. Barker. 1950. by on short chain fatty acids: growth characteristics of mutants. J. enzyme preparations of Clostridium kluyveri. J. Biol. Chem. Bacteriol. 108:885-892. 184:769-793. 61. Salanitro, J., and W. Wegener. 1971. Growth ofEscherichia coli 71. Vanderwinkel, E., and M. De Vliegher. 1968. Physiologie et on short chain fatty acids: nature of the transport system. J. genetique de l'isocitritase et des malate synthases chez Esche- Bacteriol. 108:893-901. richia coli. Eur. J. Biochem. 5:81-90. 62. Sallus, L., R. J. Haselbeck, and W. D. Nunn. 1983. Regulation of 72. Vanderwinkel, E., M. De Vliegher, M. Fontaine, D. Charles, F. fatty acid transport in Escherichia coli: analysis by operon Denamur, D. Vandevoorde, and D. DeKegel. 1976. Septation fusion. J. Bacteriol. 155:1450-1454. deficiency and phospholipid pertubation in Escherichia coli 63. Samuel, D., J. Estroumza, and G. Ailhaud. 1970. Partial purifi- genetically constitutive for the beta-oxidation pathway. J. Bac- cation and properties of acyl CoA synthetase of Escherichia teriol. 127:1389-1399. coli. Eur. J. Biochem. 12:576-582. 73. Vanderwinkel, E., M. De Vlieghere, and J. Vande Meersshe. 64. Simons, R. W., P. A. Egan, H. T. Chute, and W. D. Nunn. 1980. 1971. Mutation habilitant Escherichia coli a' croitre sur acides Regulation of fatty acid degradation in Escherichia coli: isola- gras moyens. Eur. J. Biochem. 22:115-120. tion and characterization of strains bearing insertion and tem- 74. Vanderwinkel, E., P. Furmanski, H. C. Reeves, and S. J. Ajl. perature-sensitive mutations in gene fadR. J. Bacteriol. 1968. Growth of Escherichia coli on fatty acid: requirement for 142:621-632. coenzyme A transferase activity. Biochem. Biophys. Res. Com- 65. Simons, R. W., K. T. Hughes, and W. D. Nunn. 1980. Regulation mun. 33:902-908. of fatty acid degradation in Escherichia coli: dominance studies 75. Vanderwinkel, E., P. Liard, F. Ramos, and J. M. Wiame. 1963. with strains merodiploid in gene fadR. J. Bacteriol. 143: Genetic control of the regulation of isocitritase and malate 726-730. synthase in Escherichia coli K-12. Biochem. Biophys. Res. 66. Spratt, S. K., P. N. Black, M. M. Ragozzino, and W. D. Nunn. Commun. 12:157-162. 1984. Cloning, mapping, and expression ofgenes involved in the 76. Weeks, G., M. Shapiro, R. 0. Burns, and S. J. Wakil. 1969. fatty acid degradative multienzyme complex ofEscherichia coli. Control of fatty acid metabolism. I. Induction of the enzymes of J. Bacteriol. 158:535-542. fatty acid oxidation in Escherichia coli. J. Bacteriol. 97:827- 67. Spratt, S. K., C. L. Ginsburgh, and W. D. Nunn. 1981. Isolation 836. and genetic characterization of Escherichia coli mutants defec- 77. Wegener, W. S., H. C. Reeves, R. Rabin, and S. J. Aji. 1968. tive in propionate metabolism. J. Bacteriol. 146:1166-1169. Alternate pathways of metabolism of short-chain fatty acids. 68. Sramek, S. J., and F. E. Frerman. 1975. Purification and Bacteriol. Rev. 32:1-26.