Chapter 21

Two-Carbon Compounds and Fatty Acids as Carbon Sources

DAVID P. CLARK and JOHN E. CRONAN JR.

INTRODUCTION

This chapter concerns the transport and degradation of those substrates that are wholly or largely converted to acetyl-coenzyme A (CoA) or glyoxylate as the first stage of metabolism. Entry as acetyl-CoA has two important general consequences. First, generation of energy from acetyl-CoA requires operation of both the tricarboxylic acid (TCA) cycle and the respiratory chains to oxidize the NADH produced. Hence, acetyl-CoA serves as an energy source only during aerobic growth or during anaerobic respiration supported by such alternative electron acceptors as nitrate or trimethylamine oxide. In the absence of a suitable oxidant, acetyl-CoA is converted to a mixture of acetic acid and ethanol by the pathways of anaerobic fermentation (16). Catabolism of acetyl-CoA via the TCA cycle releases both carbon atoms of the acetyl moiety as carbon dioxide and therefore requires the operation of the glyoxylate bypass (Fig. 1) (56, 57).

Fig. 1Glyoxylate bypass and related reactions in E. coli.

Source:

For these reasons, acetate, acetoacetate, and fatty acids (of assorted chain lengths) may be considered as two-carbon substrates; all are converted to acetyl-CoA and the carbon atoms enter central metabolism in that form. Escherichia coli and Salmonella typhimurium regard these compounds as substrates of low status, and most other carbon sources (sugars, sugar derivatives, glycerol, lactate, TCA cycle acids, etc.) are preferred growth substrates. Consequently, the expression of the genes for two-carbon substrate utilization is usually subject to strong catabolite repression.

GENERAL FEATURES OF TWO-CARBON METABOLISM

Most of the substrates discussed in this chapter enter central metabolism as acetyl-CoA. In the simplest case, growth of wild-type E. coli on acetate as the carbon source requires the successive action of acetate kinase and phosphotransacetylase to convert acetate to acetyl-CoA (10, 62). Growth on the more complex substrates, acetoacetate and fatty acids, occurs by cleavage of these molecules to acetyl-CoA. Growth on the reduced substrate ethanol by adhC mutants of E. coli (which are constitutive for alcohol dehydrogenase) proceeds by oxidation of ethanol to acetyl-CoA (13, 17). The glyoxylate bypass (Fig. 1) provides the means to replenish the TCA cycle with those intermediates lost to amino acid and heme biosynthesis (31, 57). The net effect of the glyoxylate bypass is the formation of one molecule of dicarboxylic acid from two molecules of acetyl-CoA. The glyoxylate bypass is the only anaplerotic pathway allowing growth on acetyl-CoA and is present in all organisms that utilize fatty acids or acetate as sole carbon source (57). The glyoxylate bypass is described in detail in chapter 16 of this volume, but an overview is provided in the present context.

The two unique enzymes of the glyoxylate bypass (Fig. 1), isocitrate lyase and malate synthase A, are induced only when E. coli is grown on acetate or fatty acids (57, 109). The structural genes for isocitrate lyase (aceA) and malate synthase A (aceB) map at min 90 on the E. coli K-12 chromosome and are transcribed together with a third gene, aceK. An adjacent gene, iclR, encodes a repressor that regulates aceBAK transcription, and iclR null mutants produce constitutive levels of isocitrate lyase and malate synthase A. The fadR gene (see below) also regulates aceBAK transcription. Transcriptional regulation is only one aspect of the regulation of the glyoxylate bypass. A major regulatory role is played by posttranslational modification of isocitrate dehydrogenase catalyzed by the bifunctional kinase/phosphatase encoded by the aceK gene. As detailed in chapter 16, the function of AceK is to partition the flow of carbon between the glyoxylate bypass and the TCA cycle.

Growth on Acetate as Carbon Source

We will consider both growth on acetate as sole carbon and energy source and incorporation of acetate when provided as a supplement for growth.

The utilization of acetate, whether for oxidation via the TCA cycle (chapter 16), for replenishing intermediates of the TCA cycle via the glyoxylate shunt (chapter 16), or for lipid (chapter 37) and leucine (chapter 27) biosynthesis, requires that acetate first be activated to acetyl-CoA. Two mechanisms are known that bring about this conversion. In the more studied pathway, two enzymes successively catalyze first the conversion of acetate to acetyl phosphate with cleavage of ATP to ADP (reaction 1) and then the transfer of the acetyl moiety from acetyl phosphate to CoA, with liberation of P_i (reaction 2):

(1)  acetate + ATP → acetyl phosphate + ADP

(2)  acetyl phosphate + CoA → acetyl-CoA + P_i

Acetate kinase (ATP acetate phosphotransferase; EC 2.7.2.1), encoded by the ackA gene, catalyzes reaction 1, and phosphotransacetylase (acetyl-CoA[CoA]:orthophosphate acetyltransferase; EC 2.3.1.8), encoded by the pta gene, catalyzes reaction 2. These two genes have been mapped in both S. typhimurium and E. coli near purF, at 48.5 to 49 min on the respective linkage maps (1, 62). The pta gene has recently been cloned (49a, 119) and sequenced (49a). The ackA gene of E. coli has been cloned and sequenced (72). The acetate kinase of E. coli shows substantial similiarity to the corresponding proteins from Methanosarcina thermophila (60) and Bacillus subtilis. Curiously, a second gene, related in sequence to ackA but of unknown function and map location, has also been detected on the E. coli chromosome. Another anomaly is the ackB gene (88). Mutants in ackB lack acetate kinase activity and grow poorly on acetate as sole carbon source, but the lesion maps to min 39 on the E. coli chromosome, far from ackA. In S. typhimurium, ackA and pta seem to be transcribed in the order ackA-pta, since insertions of Mud1 (59) and Tn10 (108) into the ackA gene also decrease phosphotransacetylase activity consistent with transcriptional polarity. In E. coli two transcripts are found for pta, one of which includes ackA (49a). The presence of the pta-specific transcript explains the phosphotransacetylase activity found in strains (B. L. Wanner, personal communication) having insertions or deletions within the ackA gene.

The levels of acetate kinase and phosphotransacetylase in extracts of wild-type E. coli and S. typhimurium vary little with different carbon sources (10, 59), and expression of the ackA and pta genes is thus neither induced by acetate nor catabolite repressed by glucose. The levels of both enzyme activities are similiar (vary less than twofold) in aerobic and anaerobic cultures (10, 59). The ackA-pta pathway is used both during anaerobic growth on sugars when acetate is a major fermentation product (see chapter 18 of this volume) and during aerobic growth on acetate as carbon source (10, 59). Under aerobic conditions, mutants defective in either the ackA or the pta gene are severely impaired in the utilization of acetate as sole carbon source, and when grown on glucose, these strains fail to incorporate labeled acetate (10). However, ackA and pta mutants grown on glycerol are capable of incorporating labeled acetate (10). These results indicate the presence of a second acetate uptake system present in glycerol-grown cells but lacking in glucose-grown cells. Brown et al. (10) showed that E. coli contains an inducible acetyl-CoA synthetase that enables ackA and pta mutants to incorporate labeled acetate. Acetyl-CoA synthetase (acetate:CoA ligase [AMP forming]; EC 6.2.1.1) catalyzes reaction 3:

(3)  acetate + ATP + CoA → acetyl-CoA + AMP + PP_i

Acetyl-AMP (also called acetyl-adenylate) is formed as a fairly stable intermediate that is largely enzyme bound. The lack of acetate incorporation and acetyl-CoA synthetase activity in cells grown on glucose (10) suggests that expression of this enzyme is regulated by catabolite repression. A difficulty in interpreting these data is that mutants lacking acetyl-CoA synthetase have only recently been isolated (58a; L. Daniels and W. Metcalfe, personal communication). Systematic sequencing (9) of the E. coli chromosome revealed an open reading frame (called acs) encoding a putative protein with strong similiarity to known acetyl-CoA synthetases. Other workers (118) have reported a second partial open reading frame at min 48 (adjacent to the ubiG gene) that has strong sequence similarity to acs. However, this partial open reading frame and acs differ by only a single nucleotide, suggesting the possibility of a cloning artifact that linked normally nonadjacent DNA segments. Another possibility (9) is that recombination between two repetitive extragenic palindromic sequences (called REP or P.U.S.) during clone propagation is responsible for linking nonadjacent DNA segments. However, it remains possible that E. coli has two acetyl-CoA synthetase genes of very similar sequence. The chromosomal acs gene has recently been disrupted (58a; Daniels and Metcalfe, personal communication), and the resulting strains grow poorly on acetate as sole carbon source despite a functional ackA-pta system. Mutants having lesions in acs, ackA, and pta are completely unable to grow on acetate as sole carbon source (58a; Daniels and Metcalfe, personal communication). This result indicates that acetyl-CoA synthetase plays a more central role in acetate metabolism than previously supposed and that no third pathway of acetate assimilation is present. It should also be noted that the acs-encoded acetyl-CoA synthetase is active with propionate (Daniels and Metcalfe, personal communication) and hence may be the enzyme that allows this compound to enter metabolism.

Taken together, these data suggest that the acetate kinase and phosphotransacetylase are required for optimal growth of E. coli and S. typhimurium on acetate as carbon source and for the incorporation of acetate under catabolite-repressing growth conditions. Furthermore, the studies with the ackA and pta mutants suggest that the acetyl-CoA synthetase(s) provides these organisms with a second route for converting exogenous acetate to acetyl-CoA. This second route may not be present (or not as active) in S. typhimurium, since ackA mutants of this organism show much less residual growth on acetate (62). This area of metabolism clearly will reward more attention.

Other Roles of Acetyl Phosphate

Several roles have been proposed for acetyl phosphate, in addition to its merely being an intermediate in the conversion of acetate to acetyl-CoA by the acetate kinase/phosphotransacetylase pathway (10). Acetyl phosphate was once proposed as the energy source for those active transport systems using periplasmic binding proteins (44). However, these observations were not confirmed, and strains carrying deletions of the pta-ack operon are not noticeably deficient in the uptake of nutrients accumulated by these active transport systems. The observation of Fox et al. (30) that a phosphoryl group may be transferred from acetate kinase to enzyme I of the phosphotransferase system may be relevant to the energization of transport. This route for phosphoryl transfer implies that acetyl phosphate can donate its high-energy phosphate to the transport proteins of the phosphotransferase system via acetate kinase and enzyme I and hence provide an alternative energy source to phosphoenolpyruvate (30). Such a situation seems likely to occur only during anaerobic conditions when acetyl phosphate is being generated rapidly by sugar fermentation (chapter 18).

Another role proposed for acetyl phosphate is as a global signal of energy metabolism acting through two-component phosphorylation-dependent signal transduction switches. This proposed role stems from the finding that acetyl phosphate (as well as carbamyl phosphate and phosphoramidate) can act as an in vitro phosphate donor to the conserved key aspartate residue of most response regulator proteins (73). These autophosphorylation reactions can be considered as phosphatase mechanisms which proceed through a phosphorylated enzyme intermediate possessing signaling properties. The presence of acetyl phosphate can thus bypass the need for the cognate sensor histidine kinase. Evidence supporting such bypass phenomena has been obtained in vitro and in vivo for the CheY (19, 64), PhoB (61, 113, 114), and NtrC (29) response regulators of E. coli.

In all these cases, evidence for an acetyl phosphate effect comes from the use of mutants lacking the cognate sensor histidine kinase(s) and also defective in acetate kinase (ackA) or phosphotransacetylase (pta) or with both genes deleted (73). Since the Pta-AckA pathway can run in either direction, cells grown on glucose accumulate higher acetyl phosphate levels in strains defective in ackA (which prevents acetyl phosphate breakdown to acetate), whereas a pta defect prevents synthesis of acetyl phosphate from acetyl-CoA. Conversely, supplementation with acetate or pyruvate yields the highest acetyl phosphate levels, especially when AckA is functional and Pta is inactive. Glycerol-supported growth gives the lowest levels of acetyl phosphate. Nonetheless, in all the cases so far examined, signaling in wild-type strains with normal two-component systems is unaffected by loss of activity of AckA or Pta (or both) (73).

DEGRADATION OF FATTY ACIDS

Growth on fatty acids results in their complete conversion to acetyl-CoA, consequently, the glyoxylate cycle plays the same essential role as in growth on acetate. The generation of one acetyl-CoA from each two carbon segment of a fatty acid molecule results in the concomitant production of one equivalent each of FADH₂ (reduced flavin adenine dinucleotide) and NADH (Fig. 2), which may be used in ATP generation. This ATP is not forthcoming during acetate (or acetoacetate) degradation, and more ATP (on a per-carbon-atom basis) must also be consumed to activate these molecules. Hence, growth on fatty acids is energetically more favorable (per carbon atom) than growth with either acetate or acetoacetate.

Fig. 2The ?-oxidation pathway of fatty acid degradation. The transport components and the ? and ? subunits of the fatty acid degradation multienzyme complex are shown together with their component activities and structural genes.

Source:

Growth of wild-type E. coli on fatty acids as sole carbon sources occurs only when the fatty acid is 12 or more carbons long and then only after a distinct lag period required for induction of the fad regulon. Fatty acids of 6 to 10 carbon atoms may be utilized by E. coli cultures which have been previously induced with longer-chain fatty acids or by strains in which expression of the fad regulon is constitutive due to mutations in the fadR gene (78, 85, 86, 99, 115). For this reason we refer to fatty acids of >C₁₂ as long-chain fatty acids and those with C₆ to C₁₁ as medium-chain fatty acids. Carboxylic acids shorter than C₆ are referred to as short-chain fatty acids and cannot be metabolized solely via the fad system.

Most of our knowledge of the genetics and biochemistry of fatty acid degradation has been derived from studies with E. coli. The synthesis of at least five proteins involved in fatty acid β-oxidation (Fig. 2) is coordinately induced when long-chain fatty acids are present in the growth medium. The genetic and enzymological studies of Overath and coworkers, published over 20 years ago, have provided the basis for more recent studies. It should be noted that Black and DiRusso (6) published a review of E. coli fatty acid degradation that appeared following preparation of this chapter.

The genes encoding the enzymes of β-oxidation (the fad regulon) are scattered around the E. coli chromosome (1) (Table 1; Fig. 3), and fadBA is the only known operon. The fad regulon is primarily responsible for the transport, activation, and β-oxidation of both medium-chain (C₇ to C₁₁) and long-chain (C₁₂ to C₁₈) fatty acids, and their expression is specifically controlled by the fadR gene product. Growth of E. coli on short-chain fatty acids (C₄ to C₆) requires, in addition to enzymes of the fad system, two degradative enzymes encoded by the atoD, atoA, and atoB genes which are positively regulated by the atoC gene product.

Fig. 3Genetic map of E. coli K-12 showing locations of the genes of two-carbon metabolism. The locations are given in map units.

Source:

Table 1Gene-enzyme relationships in fatty acid utilization

First, we will examine the genetic and enzymological information regarding the degradative pathways for both long- and short-chain fatty acids. Then we will describe how fatty acids are taken up by E. coli, and last we will consider the mechanisms by which the fad and ato structural genes are regulated.

Enzymes of Fatty Acid Oxidation

The pathways by which E. coli degrades fatty acids are substantially similar to the β-oxidative pathways present in the mitochondria of mammals and other eukaryotic organisms. This pathway is the classic example of the oxidation of a series of homologous substrates through a series of homologous intermediates. Certain features of this pathway are illustrated in Fig. 2. With each turn of the β-oxidation cycle, the fatty acyl-CoA loses a two-carbon fragment as acetyl-CoA and reduces one molecule of FAD (during the acyl-CoA dehydrogenase reaction) and one molecule of NAD (during the 3-hydroxyacyl-CoA dehydrogenase reaction). Acetyl-CoA, produced in the CoA-dependent thiolytic cleavage, is further metabolized in the TCA. The other product of the cleavage step, the shortened fatty acyl-CoA molecule, reenters the degradation cycle without the need of activation (78).

The first step of fatty acid degradation is the activation of the free fatty acid to an acyl-CoA thioester by acyl-CoA synthetase (fatty acid:CoA ligase [AMP forming]; EC 6.2.1.3). This reaction requires two high-energy phosphate equivalents per molecule of free fatty acid activated. The prediction of Overath and coworkers (54, 85) that E. coli contains a single acyl-CoA synthetase with broad specificity for medium- and long-chain fatty acids, based on analysis of fadD mutants, has been confirmed by purification of the enzyme to homogeneity. Although acyl-CoA synthetase was originally reported as a membrane-associated protein (97), Kameda and Nunn (51) found that over 90% of this enzyme was present in cytoplasmic fractions. More recently, Mangroo and Gerber (71) reported that the degree of membrane association of the acyl-CoA synthetase depends on the state of energization of the cell membrane, although the physiological relevance of this finding is obscure (see below). The molecular mass of the native FadD enzyme is approximately 130 kDa, and the subunit molecular mass determined by polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate is 47 kDa, suggesting that the enzyme might be a dimer or a trimer composed of identical subunits (51). The fadD gene has been cloned and sequenced (7, 34). FadD appears more closely related to a series of other AMP-producing CoA ligases than to the long-chain acyl-CoA synthetases of eukaryotes (34). A single open reading frame encoding a protein of 62 kDa was found, implying that the native protein is a dimer. The upstream regulatory region contained two operator sites for the fadR repressor, as suggested by comparison with other fad regulon genes and directly confirmed by DNase footprinting (7).

In E. coli, very little is known of the enzyme responsible for the next step, acyl-CoA dehydrogenase, and most of these data are available only as a doctoral thesis (K. Klein, Ph.D. thesis, Universität zu Köln, Köln, Germany, 1973). Mutants (fadE) lacking this enzymatic activity have been isolated and their mutations have been mapped (54), and this defect has been attributed to inactivation of a flavoprotein required in the reaction (Klein, thesis). The proteins catalyzing the dehydrogenase reaction per se have been reported to be the products of two genes, fadF and fadG, and to differ in the chain length of the fatty acid substrates utilized (Klein, thesis). Clearly, further work is needed in this area.

The other E. coli β-oxidation enzymes are cytosolic proteins (Fig. 2) and are part of a multienzyme complex having broad substrate specificity. Several groups have purified this complex (2, 83, 91, 92), which has an overall molecular mass of 260 kDa, and found it to have an α ₂ β ₂ structure (Fig. 4). The subunit molecular masses are 79 kDa for the α subunit and 42 kDa for the β subunit. Five enzyme activities, 3-ketoacyl-CoA thiolase (EC 2.3.1.16), enoyl-CoA hydratase (EC 4.2.1.17), l-3-hydroxyacyl-CoA dehydrogenase (EC 1.1.1.35), cis- Δ3-trans-Δ2-enoyl-CoA isomerase (EC 5.3.3.3), and 3-hydroxyacyl-CoA epimerase (EC 5.1.2.3), are associated with this multienzyme complex (2, 83, 91, 92).

Fig. 4The fadBA operon. The relationships between the structural genes and the multienzyme complex of fatty acid degradation are depicted. Negative regulation by the FadR protein and induction in the presence of acyl-CoAs are shown schematically.

Source:

The 3-ketoacyl-CoA thiolase activity is associated with the smaller subunit, whereas the remaining four enzyme activities are associated with the larger 79-kDa subunit (Fig. 2 and 4) (2, 83, 91, 92). Recently, the enoyl-CoA hydratase and 3-hydroxyacyl-CoA epimerase activities were found to utilize the same active site on the 79-kDa α subunit. The cis- Δ3-trans-Δ2-enoyl-CoA isomerase activity is adjacent to the hydratase/epimerase but has distinct catalytic residues (120). However, it remains possible that the activities share a common CoA binding site. Such regions that function in several different reactions provide an explanation for the large number of activities resident in a protein of modest size (75). For saturated fatty acid oxidation, three final steps of the pathway are needed: enoyl-CoA hydratase, 3-hydroxyacyl-CoA dehydrogenase, and 3-ketoacyl-CoA thiolase, which are all catalyzed by the complex (Fig. 2 and 4). When unsaturated fatty acids are degraded, two additional activities also carried by the complex (122), cis-2-enoyl-CoA isomerase and 3-hydroxyacyl-CoA epimerase, are required (Fig. 2).

Overath et al. (54, 85) predicted that 3-ketoacyl-CoA thiolase, 3-hydroxyacyl-CoA dehydrogenase, enoyl-CoA hydratase, and possibly 3-hydroxyacyl-CoA epimerase and cis-2-enoyl-CoA isomerase form an operon. Their evidence was based on the high coordinate induction of the first three of these enzymes as well as on genetic mapping results with mutants deficient in (i) all five enzymes (fadBA), (ii) 3-ketoacyl-CoA thiolase (fadA), and (iii) 3-hydroxyacyl-CoA dehydrogenase (85). This hypothesis was confirmed by cloning and analysis of the genes (24, 76, 121, 123). The fadA gene was found to encode the small β subunit, whereas fadB encoded the larger α subunit.

Sequencing and insertional mutagenesis of the fadBA operon by several groups (24, 76, 121, 123). showed that transcription proceeds from fadB to fadA (Fig. 4) rather than in the opposite direction as originally reported (78, 103). The fadB gene encodes a polypeptide of 729 amino acids, and the fadA gene encodes a polypeptide of 387 amino acids. The two reading frames are separated by only 10 nucleotides, suggesting possible translational coupling (24, 121, 123). Substantial similarities were found between the FadB and FadA proteins of E. coli and several of their eukaryotic counterparts involved in fatty acid β-oxidation (24, 121, 123). It seems that FadA may be either unstable or inactive in the absence of FadB. Plasmids carrying the wild-type fadA gene expressed 3-ketoacyl-CoA thiolase activity in fadA mutants but not in fadBA mutants (104). Moreover, fadA mutants carrying wild-type fadBA on multicopy plasmids express both the 79- and the 42-kDa proteins and give amplified levels of 3-ketoacyl-CoA thiolase (FadA) activity, whereas strains containing wild-type fadA on multicopy plasmids express only the 42-kDa fadA product and have only wild-type levels of 3-ketoacyl-CoA thiolase (FadA) activity (103). Pawar and Schulz (91) found that the short-chain enoyl-CoA hydratase is associated with the fadBA multienzyme complex and argued that the product of this enzyme is channeled to the 3-hydroxyacyl-CoA dehydrogenase also present on the FadB subunit without equilibration with the bulk solvent. These authors report that E. coli also contains a separate long-chain enoyl-CoA hydratase that may be membrane bound (91), but no mutants lacking this activity are known.

Although the possibility of a separate long-chain enoyl-CoA hydratase remains unresolved, You et al. (124) found another gene in the fad regulon, fadH, located at min 71 to 75 on the E. coli chromosome, far removed from all previously known fad loci. The fadH gene encodes a 2,4-dienoyl reductase which is required only for the degradation of unsaturated fatty acids whose double bond extends from an even-numbered carbon atom (124). Mutants with lesions in fadH were isolated as unable to use petroselinic acid (cis-6-octadecenoic acid) but capable of growth on oleic acid (cis-9-octadecenoic acid).

Utilization of Acetoacetate and Short-Chain Fatty Acids

The genetics of the degradation of acetoacetate to acetyl-CoA was first elucidated by Pauli and Overath (90). The utilization pathway is composed of two steps (Fig. 5). First, acetoacetate is activated to acetoacetyl-CoA in a reaction catalyzed by acetyl-CoA:acetoacetate-CoA transferase, and then, the acetoacetyl-CoA is cleaved to two molecules of acetyl-CoA by 3-ketoacyl-CoA thiolase II (22, 32, 33, 90, 94, 95, 105, 106, 111, 112). These enzymes are highly inducible by acetoacetate and are specific for 3-keto short-chain fatty acids.

Fig. 5Pathways for the degradation of four-carbon fatty acids and alcohols. The interaction of components of the fad,ato, and adh systems is required for the degradation of acetoacetate, butyrate, and butanol.

Source:

The earliest studies identified the loci responsible for acetoacetate degradation as two closely linked genes: atoA encoding acetoacetyl-CoA transferase and atoB encoding 3-ketoacyl-CoA thiolase II (90). Subsequent biochemical work showed that acetoacetyl-CoA transferase is a tetrameric α ₂ β ₂ protein (105, 106), whereas 3-ketoacyl-CoA thiolase II is a tetramer composed of four identical subunits (21). These results indicated that a second acetoacetyl-CoA transferase gene encoding the second subunit must exist. Molecular cloning of the known gene cluster (48) showed that this missing gene (called atoD) was linked to the other ato genes and that the three structural genes make up an operon transcribed from atoD to atoB (48) (Fig. 5).

Acetoacetate serves as metabolic inducer for the ato system. When acetoacetate is used as sole carbon source, a 200- to 300-fold induction of both acetoacetyl-CoA transferase and 3-ketoacyl-CoA thiolase II is seen. Neither fad ⁺ nor fadR strains can utilize saturated short-chain fatty acids (C₄ to C₆) as sole carbon sources (85, 90, 94, 95, 112), although both types of strain grow on acetoacetate itself. For growth to occur on butyrate (C₄) or valerate (C₅), E. coli must have constitutive levels of three fad-encoded β-oxidation enzymes (enoyl-CoA hydratase, 3-hydroxyacyl-CoA dehydrogenase, and acyl-CoA dehydrogenase) as well as two enzymes involved with the degradation of the 3-keto acid acetoacetate (90, 116). Although constitutive levels of the three required fad system enzymes are present in fadR strains, no growth on the saturated C₄ or C₅ acids occurs, because these substrates fail to induce the ato enzymes (95, 112). Only strains having constitutive levels of the needed ato- and fad-encoded enzymes utilize C₄ or C₅ acids as sole carbon source.

Overath and coworkers were the first to show that mutants (called But⁺) that grow on butyrate or valerate are readily isolated by plating fadR mutants on minimal medium containing butyrate (85, 87). Most such But⁺ mutants are constitutive for the ato enzymes owing to a mutation in the atoC regulatory gene (49, 85), and the genotype of strains which utilize saturated C₄ or C₅ fatty acids as sole carbon source is therefore fadR atoC.

FATTY ACID TRANSPORT

Before catabolism, fatty acids must enter the cell via uptake systems which translocate these hydrophobic compounds across both the outer and inner membranes. Prior to kinetic analyses, the prevailing thought was that fatty acids diffuse through membranes without requiring a protein carrier. This is probably true for acetate, short-chain fatty acids, and medium-chain fatty acids, for which no transport proteins or genes required for transport have been identified.

However, in the case of long-chain fatty acids, physiological and kinetic studies from several laboratories (4, 33, 54, 58, 67, 81) indicate that a carrier mechanism (Fig. 6) facilitates the entry of these molecules into E. coli. Furthermore, genetic and biochemical studies have shown at least two proteins, encoded by the fadD and fadL genes, to be required to deliver exogenous long-chain fatty acids across the cell membranes to the cytosol. The fadD gene encodes acyl-CoA synthetase, a peripheral membrane-bound protein that has broad chain-length specificity (51, 85). The fadL gene encodes a 48-kDa outer membrane protein essential for long-chain fatty acid transport (3, 4, 5, 8, 58). Although the exact role of FadL in the uptake of long-chain fatty acids is unknown, fatty acid binding studies have demonstrated a correlation between the presence of FadL protein and long-chain fatty acid binding activity (4, 58, 81, 82). When the fadL transport system, which can also transport medium-chain fatty acids (C₇ to C₁₁), is nonfunctional, these medium-chain acids are still able to diffuse across the outer membrane to the acyl-CoA synthetase in the inner membrane, where they are activated to their CoA derivatives and released into the cell. It should be noted that long-chain fatty acids can enter strains that lack FadL, although the level of accumulation cannot support growth on the fatty acid as sole carbon source. Knoll and Gordon (55) have shown that myristate is incorporated into acylated protein in E. coli strains that express a functional Saccharomyces cerevisiae protein acylation system and that the lack of FadL has no effect on this process. However, if palmitate rather than myristate was used, the loss of FadL significantly inhibited protein acylation (55). These results argue that the requirement for FadL depends on the length of the long-chain fatty acid.

Fig. 6Fatty acid transport in E. coli. Note that long-chain fatty acids require FadL protein to cross the outer membrane, whereas medium-chain fatty acids may either utilize FadL or, in the absence of FadL, simply diffuse through the outer membrane. Acyl-ACP synthetase provides a minor pathway for the incorporation of fatty acids into membrane phospholipids.

Source:

The FadL Outer Membrane Protein

The FadL protein was originally reported to be located in the inner membrane (36, 78). However, Morona and Henning (74) found that mutants in the ttr locus abolished growth on long-chain fatty acids, as well as conferring resistance to bacteriophage T2, and were genetically indistinguishable from fadL mutants. Further work by Black (3) showed that ttr and fadL are identical and that the FadL protein is indeed an outer membrane protein, as indicated by its function as a phage receptor (40). The FadL protein was heat modifiable, a characteristic of outer membrane proteins (3, 74). The nucleotide sequence of the fadL gene revealed a single open reading frame encoding a 48-kDa protein of 448 amino acid residues (5). Maturation removed a signal peptide of 27 amino acids (5). Given a native molecular mass of approximately 130 kDa, this suggests a trimeric structure reminiscent of that found for OmpC, OmpF, and other porins (77; chapter 5). Therefore, FadL may be a porin specific for long-chain fatty acids analogous to the LamB maltodextrin-specific porin of maltose catabolism. The FadL protein of E. coli is highly similar (42% amino acid identity) to protein P1 of Haemophilus influenzae, a heat-modifiable outer membrane protein of unknown function (5). No significant sequence similarities were seen between FadL and either OmpA or phospholipase A₁ (both heat-modifiable outer membrane proteins) or OmpF (which also acts as a receptor for phage T2) (5).

The FadL protein has been specifically labeled using a photoaffinity-labeled fatty acid, and its outer membrane location has been confirmed (35, 70). No fatty acid binding protein was detected in the inner membrane by this approach (70), consistent with the idea that fatty acids cross the cytoplasmic membrane by a simple diffusion mechanism (53). However, it remains possible that a putative inner membrane carrier fails to bind the photoaffinity probe (which was derived from a medium-chain fatty acid). Binding of fatty acids to FadL occurs at the C-terminal end of the protein, and certain single-residue changes eliminate fatty acid binding (58).

An interesting recent observation is the finding by Higashitani et al. (43) that expression of FadL is regulated in response to osmolarity. Transcription of the fadL gene is repressed at high external osmotic pressure, and transport of long-chain fatty acids is abolished under these conditions (43). Osmoregulation of fadL is due to binding of OmpR protein to four sites around the fadL promoter, and fadL hence seems subject to the OmpR/EnvZ two-component regulatory system (chapter 77).

Transport across the Cytoplasmic Membrane

Several observations demonstrate that long-chain fatty acids enter E. coli via an active unidirectional mechanism (Fig. 6). (i) The transport of long-chain fatty acids into wild-type strains is a saturable process, implying the existence of a binding protein (33, 67). (ii) Inhibitors which prevent electron transport or uncouple oxidative phosphorylation (or both) have been shown to block long-chain fatty acid transport completely (33, 67). (iii) No efflux of transported long-chain fatty acid occurred when wild-type E. coli strains were washed with unlabeled long-chain fatty acids (67, 79). (iv) Both the energy of activation and the Q10 of long-chain fatty acid transport are representative of enzyme-mediated processes (67). As indicated above, both the fadD and the fadL gene products are required to deliver long-chain fatty acids across the cell envelope to the cytosol of E. coli. Since functioning of the fadL gene product does not require energy and FadL is located in the outer membrane, the energy-requiring component appears to be the acyl-CoA synthetase encoded by fadD. Acyl-CoA synthetase is a peripheral membrane protein and is easily dissociated from the cytoplasmic membrane during purification (51). Mangroo and Gerber (71) recently suggested that association of acyl-CoA synthetase with the membrane is energy dependent. In energy-depleted cells, fatty acid uptake is inhibited. Re-energization, by provision of acetate, succinate, or lactate, restores fatty acid uptake, and under these conditions, the acyl-CoA synthetase apparently reassociates with the membrane (71). The basis for this effect is unclear. Changes in conformation and binding of inner membrane proteins such as adenylate cyclase are known to occur in response to the magnitude of the proton gradient. Respiration of substrates such as succinate would indeed replenish the proton gradient, but the energy source required by acyl-CoA synthetase is ATP. Another reason to question the physiological relevance of these results is that an acyl-CoA synthetase from Saccharomyces cerevisiae can functionally replace FadD in the incorporation of exogenous myristate into E. coli membrane phospholipids in vivo (55). It seems unlikely that this heterologous enzyme would mimic the energy-dependent membrane association reported for FadD.

In principle, the physical transfer of fatty acids across the cytoplasmic membrane, even for long-chain fatty acids, does not require any energy or the involvement of any protein. Kamp and Hamilton (53) have shown that nonionized long-chain fatty acids can transit a phospholipid bilayer by a flip-flop process with a half-life of <2 s. In contrast, ionized fatty acids cross the bilayer much more slowly (half-life of several minutes). Thus, trapping of the fatty acid on the inside of the membrane by FadD-catalyzed conversion to the CoA thioester could account for overall unidirectional uptake (54). However, an oleic acid binding protein distinct from FadL has been purified from cell envelopes and partially characterized (50). This protein has been postulated to be an essential inner membrane H⁺/fatty acid transporter (52), but no genetic data support the existence of such a component.

The early evidence that acyl-CoA synthetase is required for medium- and long-chain fatty acid transport was based on the finding that fadD mutants are unable to accumulate exogenous fatty acids of any length into either the cytosol or the membrane lipids of E. coli (54, 78). Acyl-CoA synthetase was therefore required for a group translocation step in the transport of fatty acids (i.e., vectoral thioesterification). The additional evidence supporting this hypothesis included (i) the fatty acyl chain-length specificity of acyl-CoA synthetase, which is very similar to that of the uptake system (54), (ii) the lack of efflux of labeled fatty acids noted above, and (iii) the lack of detectable labeled intracellular free fatty acid (54, 67). Moreover, recent work has demonstrated that the translocation product, fatty acyl-CoA, is present intracellularly (C. O. Rock, personal communication) and that E. coli strains expressing an acyl-CoA-dependent protein acylation system derived from Saccharomyces cerevisiae require FadD for protein acylation in vivo (55).

However, it should be noted that a second fatty acid transport system of much lower capacity (ca. 2% of the FadD pathway) is also found in E. coli (93). This FadD-independent (but FadL-dependent) transport was detected by the incorporation of labeled fatty acids into membrane lipids and is due to acyl-ACP synthetase, which incorporates fatty acids into membrane phospholipids via their acyl-ACP derivatives (93). However, this is a tightly coupled system that seems unable to deliver long-chain fatty acids to the cytosol and can be considered a second, more specialized example of vectoral thioesterification.

Mutants lacking any of acyl-CoA dehydrogenase (fadE), 3-ketoacyl-CoA thiolase (fadA), or 3-hydroxyacyl-CoA dehydrogenase plus enoyl-CoA hydratase (fadB) activities have decreased transport rates for medium- and long-chain fatty acids relative to wild-type strains (54, 78). These findings imply that fatty acid transport is coupled to fatty acid oxidation. Although the mechanism(s) is unknown, it is conceivable that transport is reduced in these fad mutants as a consequence of feedback inhibition. For example, the accumulation of fatty acyl-CoA intermediates in such fad mutants may be inhibitory to acyl-CoA synthetase. β-Oxidation is not an absolute requirement for transport, because unsaturated fatty acid (UFA) auxotrophs which also carry fadAB or fadE mutations can still incorporate exogenous unsaturated fatty acids into phospholipid (54, 78). In contrast, conditional UFA auxotrophs [fabA(Ts)] carrying fadD or fadL mutations (or both mutations) are nonviable at nonpermissive temperatures because they cannot incorporate sufficient UFA into phospholipids to support growth (78). Since fadE and fadAB mutations do not restrict fabA(Ts) strains from incorporating sufficient UFA into their lipids (87), it is clear that defects in the cytosolic fatty acid degradation enzymes do not reduce the uptake of long-chain fatty acids as strongly as do defects in the membrane-bound FadL and FadD proteins.

DeVeaux et al. (21) have characterized and mapped a mutation in a novel gene, fatA, which allows UFA auxotrophs to use trans-UFAs such as elaidate instead of the more natural cis-isomers. UFA auxotrophs are unable to make UFAs but can take up and incorporate exogenous cis-UFAs such as oleate or cis-vaccenate into their phospholipids. However, wild-type E. coli transports elaidate sufficiently for use as sole carbon source. Thus the fatA mutation does not appear to affect fatty acid transport but instead enables the cell to tolerate phospholipids containing trans-UFAs.

Transport of Short-Chain Fatty Acids

Acetoacetyl-CoA transferase, encoded by the atoDA genes, is required for transport of short-chain fatty acids (Fig. 5). There is recent evidence for a very hydrophobic membrane transport component encoded by an open reading frame in the ato operon, called atoE (M. S. Shanley, personal communication). The short-chain acids presumably cross the outer membrane via porin channels and then cross the cytoplasmic membrane via AtoE (Shanley, personal communication). Once the short-chain fatty acids enter the cytosol, they are converted to the CoA thioesters by acetoacetyl-CoA transferase in a trapping mechanism similar to that involving FadD in long- and medium-chain fatty acid uptake. Acetoacetyl-CoA transferase utilizes acetoacetate as well as C₄, C₅, and perhaps to some extent C₆ saturated carboxylic acids (105, 106). Frerman (32) showed that membrane vesicles prepared from cells induced by acetoacetate translocated C₄ acids. Uptake was stimulated by ATP and acetyl-CoA and did not occur in membrane vesicles from noninduced cells (32). Frerman (32) also found that significant amounts of acetoacetyl-CoA transferase were associated with the membrane and that uptake was rapidly inhibited by butyryl-CoA and acetate, the products of the reaction catalyzed by acetoacetyl-CoA transferase (Fig. 3). Although these results show the importance of acetoacetyl-CoA transferase in short-chain fatty acid transport, the possibility that other components (e.g., the atoE product; see below) play a role is not excluded. Genetic studies have shown that not only atoA mutants but also atoB and fadBA mutants (in the case of butyrate) have reduced short-chain fatty acid transport, suggesting a coupling between metabolism and transport similar to that observed for long-chain fatty acids (90).

REGULATION OF FATTY ACID DEGRADATION

The fatty acid degradation (fad) system is primarily responsible for the transport, acylation, and β-oxidation of medium-chain (C₇ to C₁₁) and long-chain (C₁₂ to C₁₈) fatty acids. Wild-type E. coli strains grow using long-chain (>C₁₂) fatty acids as sole carbon and energy source but fail to grow on medium- or short-chain fatty acids. Although the enzymes of β-oxidation work effectively on substrates of medium and short chain lengths, such fatty acids are unable to induce the genes of the fad system. Consequently, growth on medium or short fatty acids requires constitutive expression of the E. coli fad regulon. In contrast, many bacteria (e.g., pseudomonads) grow on fatty acids of all chain lengths, and it has been shown that in these cases, fatty acids of C₆ or longer are effective inducers of the β-oxidation system (98). The fad structural genes, which map to at least four loci on the E. coli chromosome (Fig. 3), are regulated by the fadR gene (6, 78, 85). When consecutive cycles of β-oxidation have shortened a fatty acid to the four-carbon stage, acetoacetyl-CoA induces the ato operon, whose products are required for the final step of converting this intermediate to two molecules of acetyl-CoA (90). Growth on saturated short-chain fatty acids requires the action of enzymes of the fad system and also operation of the ato system (see above).

Control of fad Regulon Expression

The level of the fatty acid enzymes in E. coli depends on at least three regulatory systems: the global Crp/cyclic AMP (chapter 85) and ArcAB (chapter 95) systems and the fatty acid-specific fadR system. The Crp/cyclic AMP system exerts its normal positive control of carbon utilization (89) (putative Crp binding sites being found in the promoter regions of the fad genes), whereas both ArcAB and FadR negatively control both the fad regulon (6, 7, 25, 26, 42, 85, 86) and the aceBA operon (66, 68, 69). The ArcAB system has been shown to strongly (>20-fold) repress expression of 3-hydroxyacyl-CoA dehydrogenase (encoded by the fadB gene) and to weakly repress acyl-CoA dehydrogenase (encoded by the fadE gene) and is the system through which anaerobic regulation of β-oxidation is exerted (47). The details of ArcAB repression have not yet been explored, but repression is thought to proceed by the same mechanism that represses the pathways into which the β-oxidation-derived acetyl-CoA flows: the glyoxylate and TCA cycles (chapters 6 and 95).

The FadR protein has a dual role in fatty acid metabolism as a repressor of the β-oxidation pathway and an activator of unsaturated fatty acid biosynthesis (7, 25, 26, 41, 42, 80, 110, 115). The role in fatty acid synthesis is to decrease the activity of the synthetic pathway, since exogenous fatty acids are available for membrane lipid synthesis (chapter 37), whereas in the fad regulon, FadR functions as a LacI-like transcriptional repressor (Fig. 4). The fadR gene was discovered by seeking mutants constitutive for expression of the β-oxidation pathway (85, 115). Early studies showed that long-chain fatty acids induce the β-oxidation enzymes and that although medium-chain fatty acids can be degraded by the same enzymes, these shorter homologs fail to induce the pathway. Selection for mutants able to utilize medium-chain fatty acids by plating wild-type cells onto minimal medium containing decanoate as sole carbon source results in selection of fadR mutants (85, 115). Overath’s original hypothesis that the fadR gene product was a diffusible repressor protein has been supported by several lines of evidence. First, transposon-generated fadR null mutations confer constitutive expression of β-oxidation (100). Second, genetic studies with strains merodiploid for the fadR gene showed that the fadR ⁺ allele is trans-dominant to fadR (100). Moreover, fatty acid oxidation in fadR strains harboring multicopy fadR ⁺ plasmids is decreased relative to that in wild-type strains containing a single chromosomal copy of the fadR ⁺ gene (although such strains remain inducible) (27). A third line of evidence is the isolation of genetically dominant fadR mutants that super-repress β-oxidation (45).

The first evidence that the fadR gene product regulates the expression of the fad regulon at the level of transcription was the lacZ transcriptional fusion study of Clark (15). LacZ expression in strains carrying such fusions was inducible by long-chain fatty acids in wild-type strains and constitutive in fadR strains. Furthermore, the expression of β-galactosidase was repressed in these strains under catabolite-repressing growth conditions (as are the levels of the fatty acid degradative enzymes), and overexpression of FadR gave increased repression, presumably due to increased occupation of operator sites (15).

In complementation studies with several different mutant fadR alleles, Simons et al. (100) found that all of the merodiploids displayed the constitutive fadR phenotype. These studies suggested that only a single polypeptide was encoded by the fadR locus, a prediction confirmed by cloning of the fadR gene and identification of the encoded 29-kDa protein (25). More recently, the fadR gene has been sequenced (23) and the FadR protein has been purified in a form that remains active in vitro (25). FadR contains a putative helix-turn-helix DNA binding motif (23, 39).

Overath and coworkers (85) found that fadD mutants fail to induce the other β-oxidation enzymes, whereas mutants with lesions in the other fad genes remain inducible. They thus postulated that long-chain acyl-CoA thioesters are the in vivo inducers of the fad regulon. In vitro studies have recently confirmed this postulate. DiRusso et al. (25) demonstrated binding of purified FadR protein to a fragment from the fadBA promoter and localized binding to a region just downstream of the transcriptional start site (25). FadR binding also inhibited in vitro transcription of the fadD gene in a system composed of purified components (26). DNA binding by FadR is inhibited by long-chain fatty acyl-CoA esters (25, 42), consistent with the Overath proposal. The K_i values found were approximately 5 nM for acyl-CoAs with C₁₆ or C₁₈ fatty acids, whereas myristoyl-CoA showed a K_i of 250 nM. Decanoyl-CoA and free fatty acids inhibited binding only at millimolar concentrations (25). Later work also demonstrated binding to the promoter regions of the fadL, fadD, and fabA genes (7, 26, 42). The affinity of binding was greatest for fadBA, next for fadL, and least for fabA (where FadR acts as an activator, not a repressor; see below) (26). A complication of the original report (25) that long-chain acyl-CoAs release FadR from its specific DNA binding sites is that these compounds are powerful ionic detergents (comparable to sodium dodecyl sulfate). Thus, the observed release could have been due to protein denaturation rather than the postulated specific ligand-protein interaction. Nonionic detergents were used in an attempt to control for detergent action (25), but since nonionic detergents are very weak protein denaturants, this control was not convincing. However, Henry and Cronan (42) showed that acyl-CoA-mediated release of FadR from the fabA operator was fully and rapidly reversed by thioesterase cleavage of the acyl-CoA. Since detergent denaturation of proteins is characteristically reversed only slowly and inefficiently, this result, coupled with the low levels of long-chain acyl-CoA needed for release, plus the correlation of the acyl-CoA chain length dependence with the fatty acid induction specificity observed in vivo, provides strong evidence that acyl-CoA is the fad regulon inducer.

Henry and Cronan (41, 42) showed that the FadR protein activates transcription of a gene (fabA) that plays a key role in fatty acid biosynthesis in E. coli. The fabA gene is largely repressed in vivo by the presence of exogenous fatty acids. (The function of FadR as a positive activator is discussed more fully in chapter 37.)

FadR exerts different effects on the synthesis and degradation of fatty acids depending on the location of its binding site on the DNA (42). In the fabA case, the FadR binding site is upstream of the promoter elements at the location where positive activator proteins typically bind to assist RNA polymerase action (42). In contrast, the binding sites mapped for the genes of the fad regulon lie within the promoter, where binding of regulatory proteins impedes RNA polymerase action (26, 42). The operator sequence to which FadR binds is a weakly palindromic 17-bp sequence whose consensus is 5'-AGCTGGTCCGAYNTGTT-3'. The fadD and fadL genes have two such sequences each (7, 8, 26), although expression of fadL appears to be only weakly (less than twofold) regulated by FadR (26, 96). Indeed, a difficulty in our understanding of the regulation of β-oxidation is that the levels of induction observed for β-oxidation in E. coli vary markedly among different reports. The variation observed for a given growth medium suggests that regulators other than Crp and FadR are complicating the results. A likely possibility is anaerobic regulation mediated by the ArcAB system, which would vary with the cell density and aeration of presumed aerobic log-phase cultures, and thus it would seem advantageous to assay FadR action in strains defective in ArcAB regulation. It should be noted that S. typhimurium β-oxidation seems to differ from that of E. coli. Wild-type S. typhimurium strains grow slowly on decanoic (C₁₀) acid (117), whereas E. coli wild-type strains fail to grow on this acid. Moreover, E. coli strains grow more rapidly on oleate than on acetate, whereas S. typhimurium strains have the opposite behavior (21).

Regulation of the ato Operon

All available evidence indicates that the atoC gene codes for a positively acting regulatory protein, as originally proposed by Overath on the basis of the trans-dominance of constitutive mutants to the inducible wild-type allele. Further work by Jenkins and Nunn (48, 49) showed that AtoC is indeed a positive transcriptional activator. Shanley and coworkers (12; Shanley, personal communication) have recently completed the sequence of the ato gene cluster and have deduced that the AtoC protein is a member of the NtrC-NifA family of σ ⁵⁴-RNA polymerase transcriptional activators (Shanley, personal communication). A gene located immediately upstream of atoC, called atoS, encodes a putative sensor kinase (Shanley, personal communication) which presumably binds acetoacetate (or a metabolite) and subsequently phosphorylates AtoC (chapter 95). The complexity of this regulatory circuit probably explains both the high frequency (10^–5) with which mutants expressing the ato operon constitutively are isolated (49, 90) and the diversity of constitutive phenotypes (49, 78, 90).

Wegener and coworkers (116) have proposed pathways by which the odd-chain fatty acids propionate and valerate are degraded. Although the proposed pathway is not established, recent ¹³C labeling studies are consistent with this proposal. Evans et al. (28) have shown that [¹³C]propionate can enter metabolism by either of two routes, as acetyl-CoA (presumably after conversion to pyruvate and subsequent decarboxylation) or as succinate (since a vitamin B₁₂-dependent enzyme is involved in carboxylation of propionate to a C₄ compound, this vitamin must be added for growth on propionate). Mutants in the prp gene (104) mapping at min 97 are defective in C₃ and C₅ acid metabolism and could be defective in the Wegener pathway, but no data are available. The incorporation of propionate into the methyl end of fatty acids (46) suggests that propionate is converted to propionyl-CoA (perhaps by acetyl-CoA synthetase) and then utilized in the fatty acid synthetic pathway. It seems likely that the 3-ketoacyl ACP synthetase III is involved in this synthesis of odd-chain fatty acids (see chapter 37).

GROWTH ON ETHANOL AND BUTANOL

The production and excretion of ethanol occur during fermentative growth of E. coli and S. typhimurium by reduction of the acetic acid moiety of acetyl-CoA to acetaldehyde and then to ethanol:

(4)  CH₃CO-SCoA + NADH + H⁺ → NAD⁺ + CoASH + CH₃CHO

(5)  CH₃CHO + NADH + H⁺ → NAD⁺ + CH₃CH₂OH

Reaction 4 is catalyzed by a CoA-linked acetaldehyde dehydrogenase (ACDH), and reaction 5 is catalyzed by alcohol dehydrogenase (ADH). When ethanol is used as a carbon source by aerobic organisms, these two reactions occur in reverse. The resulting acetyl-CoA is oxidized via the TCA cycle, and the NADH produced is oxidized by the oxygen-linked respiratory chain.

Obligate anaerobes such as clostridia catalyze these reactions with two enzymes that are separate polypeptides (102, 107) but that form a complex such that highly toxic free acetaldehyde is not released (65, 101). Facultative bacteria such as E. coli possess the same two enzyme activities (17). However, in this case, a single large polypeptide is responsible for both activities (18, 37). Thus, the adhE gene of E. coli K-12 encodes a polypeptide of 891 amino acids, which is equivalent to the combined length of a typical ADH plus a typical ACDH (37). Temperature-sensitive mutations in adhE result in both thermolabile ADH activity and thermolabile ACDH activity (18, 63). Sequence comparisons suggest that the adhE gene of E. coli may be the product of the fusion of two separate genes encoding ADH and ACDH activities (17).

Although E. coli is a facultative anaerobe and can grow in air by oxidizing such substrates as acetate, succinate, or glycerol, it is unable to grow aerobically at the expense of ethanol or other short-chain alcohols. The reason for this is that the adhE gene, encoding both ADH and ACDH activities, is inadequately expressed under aerobic conditions (13). When constitutive (adhC) mutations were isolated, these resulted in high-level aerobic expression of both enzyme activities and the ability to grow on ethanol or 1-propanol as carbon source (14). The AdhE multifunctional enzyme is highly active with both two- and three-carbon substrates and will therefore convert ethanol to acetyl-CoA and 1-propanol to propionyl-CoA. However, longer-chain substrates are less well metabolized, and adhC mutants of E. coli fail to grow on 1-butanol as sole carbon source. The adhC mutations are cis-acting and map very close to the adhE structural gene at 27 min on the E. coli genetic map (14). AdhE also converts halogenated alcohols and aldehydes to their extremely toxic CoA derivatives. In consequence, mutants which have lost the AdhE enzyme may be selected by resistance to chloroethanol or chloroacetaldehyde (18). The unsaturated alcohol, allyl alcohol (CH₂=CHCH₂OH), is converted to the highly toxic alkylating agent acrylaldehyde (acrolein, CH₂=CHCHO), and propargyl alcohol (CH≡CCH₂OH) is also converted to the corresponding aldehyde. Again, mutants resistant to these alcohols have lost the ADH-ACDH enzyme, and their mutations map at the adhE locus (63).

Wild-type strains of E. coli have two major problems in growth on 1-butanol: (i) these strains cannot convert butanol to butyryl-CoA, and (ii) these strains cannot metabolize butyryl-CoA. However, double mutants able to grow on butyrate are known. Such mutants have lesions in both the fadR and atoC regulatory genes and hence express both the acetoacetate pathway and the β-oxidation system for fatty acids in a constitutive manner (see above).

Butyrate is converted to butyryl-CoA by the atoAD-encoded transferase (Fig. 5). The oxidation of butyryl-CoA to crotonyl-CoA by FadE, the hydration of crotonyl-CoA to acetoacetyl-CoA by FadB, and finally the AtoB-catalyzed thiolytic cleavage of acetoacetyl-CoA yield two molecules of acetyl-CoA. The requirement for lesions in both atoC and fadR is due to the inability of butyrate to induce either the ato operon or the fad regulon. Such atoC fadR strains fail to grow on butanol because they are unable to convert butanol to butyryl-CoA. Since an atoC fadR double mutant can use butyrate, it might be supposed that an atoC fadR adhC strain would grow on butanol. However, this is not the case, apparently because the activity of the adhE-encoded ADH-ACDH is insufficient to effectively convert butanol to butyryl-CoA (14). However, mutants that possess 10-fold further elevated levels of ADH-ACDH can be selected from an adhC constitutive strain. These strains were selected by growth on ethanol in the presence of the ADH inhibitor 4-methyl-pyrazole and contain mutations in a new gene, adhR, which maps at min 72 of the E. coli genetic map, far removed from the adhCE locus (14). Strains carrying the adhR mutation no longer grew well on acetate or ethanol in some media, a behavior due to acetate toxicity rather than the inability to metabolize acetate. A suppressor mutation, aceX, which eliminated these growth defects and sensitivity to acetate, was then selected. An E. coli strain containing the adhR and aceX mutations plus the fadR atoC and adhC mutations was at last able to grow on butanol as carbon source (14). Both the adhC and the adhR mutations are necessary to give enzyme levels of ADH and ACDH sufficient to metabolize butanol to butyryl-CoA, whereas the other mutations are necessary for the metabolism of butyryl-CoA (Fig. 5).

UTILIZATION OF GLYCOLATE AND GLYOXYLATE

The compounds discussed thus far are metabolized as acetyl-CoA via the TCA cycle. Utilization of the two-carbon compounds glycolate and glyoxylate is an exception to this picture in that their metabolism by the TCA cycle is of secondary importance. These two compounds are products of plant and algal metabolism and are utilized as sole sources of carbon and energy by a variety of bacterial species including E. coli and S. typhimurium. Glycolate can also be produced by oxidation of glycoaldehyde, which is produced by catabolism of certain carbohydrates (e.g., d-arabinose).

The first step of the pathway is the oxidation of glycolate to glyoxylate (Fig. 7), catalyzed by glycolate oxidase. Glyoxylate is then metabolized by either of two divergent condensation reactions, one leading to the glycolytic pathway and the other to the TCA cycle (84).

Fig. 7Pathways of glycolate and glyoxylate metabolism in E. coli. The enzymes are indicated as follows: 1, glycolate oxidase; 2, glyoxylate carboligase; 3, tartrontic semialdehyde reductase; 4, glycerate kinase; and 5, malate synthase G. Note that reaction 2 involves two molecules of glyoxylate.

Source:

In the first instance, two molecules of glyoxylate are condensed by glyoxylate carboligase to form tartronic semialdehyde plus CO₂. Tartronic semialdehyde is then reduced to glycerate, which is phosphorylated to give the glycolytic intermediate glycerate 3-phosphate (84). In the other condensation, glyoxylate reacts with acetyl-CoA to form malate, the TCA cycle intermediate (84, 109). This reaction is identical to that of the glyoxylate cycle except that the condensation is catalyzed by malate synthase G encoded by the glcB gene rather than by the glyoxalate cycle enzyme malate synthase A encoded by the aceB gene (73a, 84, 111). Malate synthases G and A are regulated by distinct mechanisms. Malate synthase G is strongly induced (1,000-fold) by glycolate (73a), whereas malate synthase A is induced 20-fold by growth on acetate or fatty acids (109).

Mutants lacking glyoxylate carboligase fail to grow on either glycolate or glyoxylate, whereas mutants lacking glycolate oxidase grow on glyoxylate, but not glycolate (84). Malate synthase G is not essential for growth on either compound as sole carbon source, since glyoxylate carboligase gives rise to a glycolytic intermediate. In contrast, glyoxylate carboligase is essential even when malate synthase G is functional, because further metabolism of the carboligase condensation product enters glycolysis and provides the source of the acetyl-CoA needed for the malate synthase G reaction. The secondary and dispensable role of the TCA cycle in the metabolism of these compounds is shown by the finding that mutants deficient in citrate synthase (gltA) grow on glycolate with the same rate and yield as a wild-type strain (84). However, during growth on the more oxidized compound glyoxylate, the presence of malate synthase G allows more rapid growth of the gltA strain (84).

The molecular genetics of glycolate-glyoxylate metabolism is not well developed, although some recent progress has been made. A spontaneously arising deletion at min 64.2 of the E. coli map that results in lack of growth on glycolate has been shown to remove both the glycolate oxidase and malate synthase G genes (109). The later gene (called glcB) has been sequenced (73a), and the deduced amino acid sequence is similiar to that of known malate synthases including malate synthase A (AceB). The gene (called gcl) encoding glyoxylate carboligase has been cloned by a combination of genetic and reverse genetic approaches and was mapped to min 12 on the E. coli map (11). The coresponding protein was known to be one of the class of enzymes that require FAD, although there is no oxidation or reduction step in the catalyzed reaction. The most numerous members of this enzyme class are the acetohydroxy acid synthases catalyzing early reactions of the branched-chain amino acid synthetic pathway (chapter 27). These proteins catalyze a reaction chemically analogous to that of glyoxylate carboligase, and the Gcl protein shows 30% amino acid residue identity with these enzymes (11). The carboligase also shows significant similiarity with E. coli PoxB (pyruvate oxidase), an FAD-linked dehydrogenase that oxidizes pyruvate to acetate. Glyoxylate carboligase contains a vestigial quinone binding site, as do the acetohydroxy acid synthases (11). These data argue that glyoxylate carboligase, like the acetohydroxy acid synthases, was derived from a dehydrogenase similiar to PoxB. Mutants defective in the conversion of tartronic semialdehyde to glycerate have not yet been isolated, although mutants lacking glycerate kinase are available (Y. Y. Chang, personal communication).

The regulation of the genes of the glycolate-glyoxylate pathway is not well understood. Glycolate is known to induce glycolate oxidase and malate synthase G (73a, 84, 109), while only glyoxylate, the product of the oxidase, induces glyoxylate carboligase (11, 84). It seems that glyoxylate produced during operation of the glyoxylate shunt can also induce gcl expression, since increased glyoxylate carboligase activity results from growth on acetate (84; Chang, personal communication). Increased enzyme activities upon addition of glycolate or glyoxylate are believed to be due to increased transcription of the structural genes, but data are available only for the malate synthase G (glcB) gene (73a), which does not appear to be cotranscribed with the neighboring gene (73a; Chang, personal communication) that encodes glycolate oxidase. No regulatory genes have been isolated, but the isolation of point mutants that simultaneously lose two enzymes of the pathway suggests a common regulatory system (84; Chang, personal communication).

CONCLUSIONS AND OUTLOOK

The pathways for uptake and degradation of fatty acids, acetoacetate, and acetate are well investigated, and their regulation is also largely understood except for occasional minor details. The transport of acetate and fatty acids of short and medium lengths is presumed to occur by simple diffusion of the nonionized form of the acid, followed by trapping as the CoA thioester. Only in the case of long-chain fatty acids is an actual transport mechanism required for crossing the outer membrane, and even here the nonionized acid may diffuse unassisted across the lipid bilayer of the cytoplasmic membrane. Since it is very difficult to prove a negative, this assertion must remain technically unproven, and it is perhaps possible that some sort of diffusion facilitator protein(s), analogous to the GlpF protein involved in glycerol transport, may yet be uncovered for fatty acids.

As regards degradation, the breakdown of fatty acid derivatives which generate "awkward" intermediates may yet provide a few more novel enzymes and their corresponding genes, as in the case of the fadH-encoded dienoyl reductase. As far as we know, no systematic study of the ability of E. coli to degrade unsaturated, acetylenic, hydroxylated, branched, or ring-containing fatty acids has yet been made.

Two interesting issues arise in regulation. First, the aceBAK operon specifying the components of the glyoxylate system is under dual regulation by both the IclR and the FadR proteins. Removal of either of these regulatory proteins, either by genetic mutation or by provision of the appropriate inducing metabolite, results in relief of repression and consequent transcription of aceBAK. Although it is easy to understand how either of two positive elements can be sufficient for activation, it is hard to visualize a simple molecular mechanism to explain how the absence of either of two repressors results in induction. However, it seems possible that the FadR protein acts indirectly by regulating expression of the iclR gene. The binding of IclR and FadR to their presumed sites within the aceBAK and iclR promoter regions needs to be clarified.

Second, the significance of acetyl phosphate requires future elucidation. Is the autophosphorylation of response regulators merely an interesting artifact which sheds some incidental light on their mechanisms of operation, or is the level of acetyl phosphate really a physiologically relevant signal?