Cloning, Expression, and Characterization of the Human Mitochondrial β-Ketoacyl Synthase

A human β-ketoacyl synthase implicated in a mitochondrial pathway for fatty acid synthesis has been identified, cloned, expressed, and characterized. Sequence analysis indicates that the protein is more closely related to freestanding counterparts found in prokaryotes and chloroplasts than it is to the β-ketoacyl synthase domain of the human cytosolic fatty acid synthase. The full-length nuclear-encoded 459-residue protein includes an N-terminal sequence element of ∼38 residues that functions as a mitochondrial targeting sequence. The enzyme can elongate acyl-chains containing 2–14 carbon atoms with malonyl moieties attached in thioester linkage to the human mitochondrial acyl carrier protein and is able to restore growth to the respiratory-deficient yeast mutant cem1 that lacks the endogenous mitochondrial β-ketoacyl synthase and exhibits lowered lipoic acid levels. To date, four components of a putative type II mitochondrial fatty acid synthase pathway have been identified in humans: acyl carrier protein, malonyl transferase, β-ketoacyl synthase, and enoyl reductase. The substrate specificity and complementation data for the β-ketoacyl synthase suggest that, as in plants and fungi, in humans this pathway may play an important role in the generation of octanoyl-acyl carrier protein, the lipoic acid precursor, as well as longer chain fatty acids that are required for optimal mitochondrial function.

Over the last decade, it has become clear that fungal and plant mitochondria are capable of synthesizing fatty acids de novo (1)(2)(3)(4)(5)(6). The enzymes involved in the pathway are freestanding, monofunctional proteins of the type II variety and are distinct from the multifunctional polypeptide type I FASs 1 present in the cytosol of fungi and animals. Thus far, the preponderance of evidence suggests that these mitochondrial FAS pathways function to produce octanoyl-ACP, the precursor of lipoic acid (4) and longer chain-length fatty acids that may be utilized for the remodeling of mitochondrial membrane phospholipids (7). The presence of a similar mitochondrial FAS system in animals has been suspected for some time, based initially on the discovery of an ACP-like protein in animal mitochondria (8,9). However, only recently have other components of a putative mitochondrial FAS been identified, cloned, and characterized; they include the human ACP and malonyl transferase (10) and enoyl reductase (11). In continuation of the search for other components, we have now identified and characterized a single candidate for a type II human mitochondrial ␤-ketoacyl synthase, the critical enzyme required for catalysis of the chain-elongating condensation reaction.

EXPERIMENTAL PROCEDURES
Cloning of the Human Mitochondrial ␤-Ketoacyl Synthase-Using the sequences of several authentic type I and type II ␤-ketoacyl synthases as probes to BLAST search the human genome sequence data base, we identified a sequence (accession NP_060367) that also was represented in a human EST clone (IMAGE clone 3446492, ATCC MGC-4956) and appeared likely to encode a full-length human ␤-ketoacyl synthase, together with an N-terminal mitochondrial targeting sequence. A cDNA from this EST clone was used as template to generate a construct for expression in Escherichia coli. Specific primers (KS mit qeT/B, Table I) with appropriate restriction sites for cloning were designed, and the cDNA encoding the putative ␤-ketoacyl synthase was amplified using PCR, essentially as described earlier (12). The E. coli expression construct carried a 37-residue N-terminal deletion that we deduced, by analysis of multiple sequence alignments, might represent a mitochondrial targeting sequence. The amplified fragment was purified using a QIAquick PCR purification kit (Qiagen), cleaved with appropriate restriction enzymes and cloned into the pQE80 L (QE) E. coli expression vector (Qiagen). To facilitate purification of the recombinant protein by affinity chromatography, a His 6 -tag was encoded at the N terminus of the construct. The authenticity of the cloned, amplified fragments was confirmed by DNA sequencing.
Verification That the Putative N-terminal Mitochondrial Recognition Elements Direct the ␤-Ketoacyl Synthase into Mitochondria-A PCRamplified fragment (primer set KS mit1-38 gfpT/B, TableI) encoding the first 38 amino acid residues (putative targeting sequence) of the human mitochondrial ␤-ketoacyl synthase was directionally cloned into the EcoRI and KpnI sites of the pEGFP-N3 vector (Clontech) to facilitate expression as the N-terminal partner of a KS mit1-38 -EGFP fusion protein. Authenticity of the cloned fragment was confirmed by DNA sequencing. HeLa cells, grown in Dulbecco's modified Eagle's medium supplemented with 10% bovine fetal calf serum, were cotransfected with two plasmids using FuGENE 6 reagent; one plasmid encoded the KS mit1-38 -EGFP fusion protein (or as a control the pEGFP-N3 parental vector), the other encoded pDsRed2-Mito (Clontech), which served as a mitochondrial marker. Cells were cultured for 60 h, fixed with paraformaldehyde, and analyzed for green and red fluorescence using a Sony confocal microscope.
Tissue Specificity of Expression of Human Mitochondrial ␤-Ketoacyl Synthase-A 32 P-labeled probe was synthesized by the random priming method using [␣-32 P]dCTP (3000Ci/mmol), a PCR-amplified mitochondrial ␤-ketoacyl synthase template cDNA and NEBlot kit (New England Biolabs). A human multiple tissue Northern blot (Clontech) was exposed to the purified radiolabeled probe (ϳ1 ϫ 10 7 dpm/blot), using ExpressHyb hybridization solution (Clontech), and was processed according to the manufacturer's recommendations. The blot was analyzed using a phosphoimager (Strom 840, Molecular Dynamics).
Expression and Purification of Human Mitochondrial ␤-Ketoacyl Synthase-The recombinant pQE 80L vector was expressed in E. coli DH5␣ cells. Cells were grown at 37°C in LB medium to a density equivalent to A 600 of 0.5 and induced with 0.4 mM isopropyl-␤-D-thiogalactopyranoside at room temperature overnight. Cells were harvested and using a microfluidizer (Microfluidics, Newton MA) were lysed in buffer (25 mM Tris-HCl, pH 7.8, 300 mM NaCl) containing protease inhibitors (leupeptin 5 g/ml, trans-epoxysuccinyl-LGB 10 M, pepstatin 1 g/ml, and antitrypsin 5 g/ml). The cell debris was pelleted by centrifugation at 30,000 ϫ g for 40 min at 4°C, and the resulting supernatant was passed through a 0.45-m filter and loaded at 20°C onto a HiTrap Chelating HP column (5 ml bed volume, Amersham Biosciences). The column was washed with 50 mM imidazole in buffer A (25 mM Tris-HCl, pH 7.8, 300 mM NaCl, 10% glycerol) then the tightly bound proteins were eluted with 250 mM imidazole in buffer A. The latter protein fraction was rechromatographed on the same column, essentially in the same manner, except that an additional wash with 100 mM imidazole in buffer A was included prior to elution of the tightly bound proteins. The purified protein was subjected to ultrafiltration, both to remove imidazole and to effect transfer to the storage buffer (25 mM Tris-HCl, pH 7.8, 300 mM NaCl, 2 mM EDTA, 2 mM dithiothreitol, and 10% glycerol). The enzyme was stable for at least 6 months when stored at Ϫ80°C.

␤-Ketoacyl
Synthase Activity Using Acetyl-CoA as Primer-Recombinant mitochondrial ␤-ketoacyl synthase was assayed for condensing activity using [1-14 C]acetyl-CoA as the primer and malonyl-ACP mit as the chain extender. The malonyl-ACP mit was first generated in situ from malonyl-CoA and holoACP mit using human mitochondrial malonyl transferase (10). The reaction mixture contained 83 mM potassium phosphate buffer, pH 6.8, 0.3 mM dithiothreitol, 1 mM EDTA, 200 M malonyl-CoA, 60 M holoACP mit , and 100 ng of human mitochondrial malonyl transferase in a final volume of 30 l and was incubated at 37°C for 15 min. Then [1-14 C]acetyl-CoA (54 Ci/mol, 50 M final concentration) was added. The mixture was aliquotted into two tubes and incubated for 20 min at 37°C without or with 1 g (725 nM) of mitochondrial ␤-ketoacyl synthase. The reaction was terminated by trichloroacetic acid precipitation.
Synthesis of Acyl-ACP mit Substrates-The C2:0 -C16:0 acyl-ACP mit thioesters were synthesized enzymatically using recombinant human phosphopantetheinyl transferase (15) and C-terminally His 6 -tagged apoACP mit . A typical reaction mixture, 5 ml in volume, containing 20 mM BisTris-HCl, pH 6.5, 10 mM MgCl 2 , 15 M apoACP mit and 75 M acyl-CoA, was incubated with 250 nM phosphopantetheinyl transferase at 37°C overnight. The reaction mixture was then desalted into buffer B (20 mM BisTris-HCl, pH 6.5, 10% glycerol) using a Vivaspin concentrator (5,000-Da cutoff, Vivascience) and repurified at 20°C on a Hi-Trap Q HP anion exchange column. The bound proteins were eluted with a 20-ml gradient of 0 -250 mM NaCl in buffer B, followed by a 5-ml gradient of 250 -500 mM NaCl in buffer B. The anion exchange chromatography procedure effected a partial separation between residual unmodified apoACP mit and the various acyl-ACP mit species, which eluted later. Fractions were analyzed by SDS-PAGE to assess protein purity and by mass spectrometry to identify those fractions that contained authentic acyl-ACP mit thioesters, essentially free of residual apoACP mit . Starting with ϳ1 mg of apoACP mit , the overall yields of acyl-ACP mit thioesters were typically ϳ 60%, with the exception of C-16-ACP mit for which the yield was 30%.
Isolation of Apo and Holo Forms of Human Mitochondrial ACP-The C-terminally His 6 -tagged apoACP was expressed in an Sf9/baculoviral host/vector system and purified as described earlier, and the holo-form was derived from it using human phosphopantetheinyl transferase (10,15).
Mass Spectrometry-Samples were desalted using a C 4 ZipTip (Millipore, Bedford, MS), mixed with the matrix (sinapinic acid, 10 mg/ml in 30% acetonitrile, 0.1% trifluoroacetic acid) in a 10:1 ratio (v/v), and 1 l of the resulting mixture was spotted onto a stainless steel matrixassisted laser desorption ionization plate. Analysis was performed on an Autoflex® matrix-assisted laser desorption ionization time-of-flight biospectrometry work station (Bruker Daltonics, Billerica, MA) in a positive linear mode. The mass scale was externally calibrated utilizing cytochrome c. Both singly and doubly charged ions were used for a molecular mass determination.
␤-Ketoacyl-ACP Synthase Assay-The ␤-ketoacyl synthase activities were assayed according to Garwin et al. (16). The incubation mixture (total volume 20 l) contained 0.2 M potassium phosphate buffer, pH TABLE I PCR primers used in this study T/B in primer names indicate sense/antisense primer, respectively. Uppercase letters indicate that the primer sequence matches cDNA sequence, whereas bases in lowercase are not present in the cDNA and were incorporated into the primers to engineer restriction sites at the ends of amplified fragments. The engineered restriction sites were used to clone individual fragments into pQE80L or pEGFP-N3 plasmids for expression in E. coli or HeLa cells, respectively.  (17)), 10 M ACP mit , 20 M [2-14 C]malonyl-CoA (52 Ci/mol), 1-50 M acyl-ACP mit , and 25 ng (27 nM) of mitochondrial ␤-ketoacyl synthase. The malonyl-ACP mit was first generated from malonyl-CoA and ACP mit , using malonyl/acetyl transferase, by incubation for 15 min at room temperature. Then the reaction was started by addition of acyl-ACP mit and mitochondrial ␤-ketoacyl synthase, continued for 2 min at 37°C, and finally terminated by the addition of 0.4 ml of reducing reagent (0.1 M K 2 HPO 4 , 0.4 M KCl, 30% tetrahydrofuran, and 5 mg/ml NaBH 4 ). The contents of the assay tubes were mixed vigorously and incubated at 37°C for 45 min. Toluene (0.4 ml) was added, the contents of the tubes were vortexed for 2 min, and a portion (0.4 ml) of the upper phase was taken for liquid scintillation spectrometry. Blank reactions were performed without acyl-ACP mit . Kinetic parameters were determined using EnzymeKinetics (Trinity Software); values represent the means Ϯ S.D. for calculations using Lineweaver-Burk and non-linear regression methods.

Identification of the Putative Human
Mitochondrial ␤-Ketoacyl Synthase-A single candidate sequence for a freestanding, type II ␤-ketoacyl synthase was identified in both the human and mouse genomic sequence data bases (sequences 85% identical). Compared with ␤-ketoacyl synthase sequences from prokaryotes, the putative 459-residue mammalian mitochondrial counterparts have extensions of ϳ40 residues at the N termini that are rich in basic amino acids and lack acidic residues (Fig.  1). These features are characteristic of mitochondrial targeting sequence elements. The sequence of a recently discovered ␤-ketoacyl synthase from Arabidopsis thaliana also includes an N-terminal extension that functions as a mitochondrial targeting sequence (18). Surprisingly, sequences for fungal mitochondrial ␤-ketoacyl synthases do not include obvious N-terminal extensions, although the actual sequences at the N termini possess the general characteristics of mitochondrial targeting sequence elements. Thus, it is unclear at present how these fungal enzymes are targeted for import into mitochondria. Xray crystallographic analysis and mutagenesis of ␤-ketoacyl synthases associated with type II FAS systems have established the importance of several residues in the functioning of this type of enzyme (19 -27). A multiple sequence alignment revealed that all of these residues were positionally conserved in the putative mammalian mitochondrial enzymes, as well as in the confirmed plant and fungal mitochondrial ␤-ketoacyl synthases (Fig. 1). These residues included the cysteine nucleophile (residue 209), two catalytic histidine residues thought to be required for promoting the decarboxylation of malonyl moieties (residues 348 and 385), an essential lysine of uncertain function (residue 380), a glycine residue that allows entrance into the substrate-binding tunnel (residue 273), two threonine residues that hydrogen bond with the ACP phosphopantetheine moiety (residues 350 and 352), and a glycine-rich motif near the C terminus that contributes to the oxyanion hole (residues 443-449). This analysis provided a strong indication that mammalian nuclear genomes encode a single, type II ␤-ketoacyl synthase having an extended sequence at the N terminus. The coding sequence for the human enzyme is located on chromosome 3 (3p24.2) and that for the mouse counterpart on chromosome 14.
Verification That the Putative Mitochondrial ␤-Ketoacyl Synthase Contains an N-terminal Mitochondrial Targeting Sequence-A cDNA encoding the putative human ␤-ketoacyl syn-thase mitochondrial targeting sequence (residues 1-38) fused in-frame with the coding sequence for EGFP, a red-shifted variant of green fluorescent protein, was coexpressed in HeLa cells together with a fusion protein consisting of the mitochondrial targeting sequence of cytochrome c coupled to the Discosoma sp. red fluorescent protein. Confocal fluorescence microscopy revealed that the KS mit1-38 -EGFP colocalized with the red fluorescent mitochondrial marker protein (Fig. 2). In contrast, the EGFP protein lacking the putative mitochondrial targeting sequence of the ␤-ketoacyl synthase expressed throughout the cytoplasm. Thus, this experiment confirmed that the N-terminal region of the full-length ␤-ketoacyl synthase indeed is a mitochondrial targeting sequence.
Complementation of the Yeast cem1 Mutant Strain by the Human ␤-Ketoacyl Synthase-The in vivo function of the cDNA-encoded human mitochondrial ␤-ketoacyl synthase was established by complementation of the mitochondrial ␤-ketoacyl synthase defect of a S. cerevisiae cem1 mutant (2). This mutation is characterized by a nuclear-encoded respiratorydefective petite phenotype, resulting in the inability to grow on non-fermentable carbon sources such as glycerol or lactate. Consequently, the four spores originating from a sporulated BY4743 diploid consist of two lactate-positive and two lactatenegative segregants (Fig. 3B). In accordance with the use of kanMX4 as a disruption cassette for CEM1, the two lactatenegative spores are G418-resistant (Fig. 3B). In contrast, sporulation of BY4743 cells transformed with pKS mit -GFP resulted in four respiratory-competent spores, even though two of them were again G418-resistant and hence contained the defective cem1⌬ allele (Fig. 3A). Obviously, the ␤-ketoacyl synthase defect in these cells was compensated by the human ␤-ketoacyl synthase homolog expressed from pKS mit -GFP. Because this plasmid contains the yeast URA3 gene as a selection marker, the uracil-prototrophy of all four spores proved the equal distribution of the multicopy plasmid among the meiotic segregants. The C-terminal fusion of GFP to pKS mit -GFP-encoded human mitochondrial ␤-ketoacyl synthase apparently did not interfere with the ␤-ketoacyl synthase function. Furthermore, the successful complementation of the respiratory defect implied that the mitochondrial targeting sequence at the N terminus of the human ␤-ketoacyl synthase was recognized in the fungal host. This conclusion was supported by the observation that in situ fluorescence elicited by the C-terminal GFP marker was not distributed throughout the cytoplasm but was limited to specific regions of the cell likely corresponding to mitochondria (data not shown).
Tissue Distribution of Human Mitochondrial ␤-Ketoacyl Synthase-Analysis of a multiple tissue Northern blot with a radiolabeled cDNA probe revealed that mitochondrial ␤-ketoacyl synthase transcripts are most abundant in the heart and skeletal muscle, as well as the liver and kidney, tissues that contain high levels of active mitochondria (Fig. 4). Transcripts were present at relatively low levels in the placenta, brain, spleen, and lung and were barely detectable in the colon, thymus, and leukocytes. This relatively broad tissue expression contrasts with that of the type I cytosolic FAS, which is not expressed at significant levels in heart or skeletal muscle and is expressed at high levels in tissues such as liver that are involved in lipogenesis and energy homeostasis.
Expression and Purification of the Human Mitochondrial ␤-Ketoacyl Synthase-The mitochondrial ␤-ketoacyl synthase construct KS mit38 , in which the first 37 residues constituting the mitochondrial targeting sequence were replaced by a His 6tag, expressed well in E. coli, although appreciable amounts of the protein were sequestered in inclusion bodies when the cells were grown at temperatures above 20°C. The protein could be FIG. 1. Sequence alignments of putative mitochondrial, prokaryotic type II, and mammalian type I counterparts. The alignment was performed using ClustalW. Conserved residues known to be important for catalytic function are shown boxed in boldface. Numbering is for the human mitochondrial enzyme. Accession numbers for the sequences are, in order of presentation, NP_060367, XP_127578, AB073746, CAB58180, XP322142, NP_416826, NP_287229, and GO1880. obtained in high purity by two successive metal-ion affinity chromatography steps, and its mobility on SDS-PAGE corresponded closely to that of a species with the expected molecular mass of 45.8 kDa (Fig. 5).
Substrate Specificity and Kinetic Properties of the Human Mitochondrial ␤-Ketoacyl Synthase-The type II FAS systems of prokaryotes and chloroplasts typically employ a specialized enzyme (␤-ketoacyl synthase III) for catalysis of the first elongation step that uses acetyl-CoA, rather than acetyl-ACP, as the primer. Therefore, the possibility that the mitochondrial ␤-ketoacyl synthase could utilize acetyl-CoA directly as a substrate for elongation was evaluated using the malonyl thioester of human mitochondrial ACP (ACP mit ) as the chain extender; however, no activity was observed even when high concentra-tions (0.725 M) of enzyme were used.
For evaluation of catalytic activity using various acyl-ACP mit thioesters as primers, we devised a novel, simple, efficient procedure for synthesis of these substrates by exploiting the broad substrate specificity of the human phosphopantetheinyl transferase. This enzyme will transfer the phosphopantetheinyl moiety from CoA to a variety of apoACPs, essentially independent of whether the donor CoA substrate is offered as the free thiol, or as an acyl-CoA thioester. Thus, acyl-CoAs containing 2-16 C atoms were all efficiently converted to the corresponding acyl-ACPs by this enzyme. The human phosphopantetheinyl transferase has a remarkably broad tolerance not only for the acyl-S-phosphopantetheinyl donor but also for the carrier protein acceptor and can phosphopantetheinylate both peptidyl and acyl carrier proteins from prokaryotes and eukaryotes (15). Thus, the enzyme is ideally suited for routine synthesis of acyl-ACPs of any acyl chain length and ACP origin.
In the kinetic studies, care was taken to validate steady state conditions with each substrate. Typically, only 27 nM enzyme and incubation times of 2 min were used. When 40-fold higher concentrations of enzyme were used, a significant radiolabel from malonyl moieties was recovered in the toluene extract even when no priming substrate was provided, indicating that malonyl decarboxylase activity of the ␤-ketoacyl synthase most likely was generating acetyl moieties that were used as the primer. At steady state, using only 27 nM enzyme, no significant incorporation of radiolabel was observed in the absence of priming substrate indicating that formation of acetyl priming moieties by decarboxylation was negligible. Under these conditions, no more than 20% of the substrate was utilized during the course of the assay.
Acyl-ACP mit thioesters containing acyl moieties with 2-14 C atoms all were elongated by the mitochondrial ␤-ketoacyl synthase; only C-16-ACP mit was a very poor substrate (Fig. 6A). When compared in the acyl-ACP mit concentration range 5-10 M, the activities ranked in the following order: C-12 Ͼ C-10 Ͼ C-6 Ͼ C-8 Ͼ C-4 Ͼ C-2 Ͼ C-14 Ͼ C-16. However, the K m values varied according to the acyl chain-length of the substrate. Thus, the enzyme exhibited K m values in the low micromolar range for C-4 -C-12 ACP thioesters but had markedly lower affinity for C-2 and C-14 substrates. Although it was not possible to estimate actual kinetic parameters for C-2 because of an apparently very high K m value, the V max was Ͼ250 nmol⅐min Ϫ1 mg Ϫ1 for this substrate and values for the other substrates were in the range 100 -400 nmol⅐min Ϫ1 mg Ϫ1 , ex-  2. The N-terminal 38 residues of the putative ␤-ketoacyl synthase contain a mitochondrial targeting sequence. HeLa cells were cotransfected with two constructs, one encoding the red fluorescent protein mitochondrial marker, the other encoding either the green fluorescent protein (EGFP) or a chimera consisting of the first 38 residues of the ␤-ketoacyl synthase fused to the N terminus of the green fluorescent protein (KS mit1-38 EGFP). Cells were fixed then analyzed for green and red fluorescence by confocal microscopy. cept for C-12-ACP, which was Ͼ1000 nmol⅐min Ϫ1 mg Ϫ1 .
Inhibition of the Human Mitochondrial ␤-Ketoacyl Synthase by Cerulenin-Cerulenin, an antibiotic produced by Cephalosporium caerulens, is a potent inhibitor of both prokaryotic type II and eukaryotic type I FAS systems (28) and acts by binding to the hydrophobic pocket formed at the ␤-ketoacyl synthase dimer interface and reacting with the active site cysteine through its C-2 carbon (20,29). The human mitochondrial ␤-ketoacyl synthase was also inactivated by cerulenin (Fig. 7), although sensitivity appears low compared with the prokaryotic ␤-ketoacyl synthases I and II for which IC 50 values below 25 M are typically reported (25). DISCUSSION This study establishes that the human nuclear genome encodes a single mitochondrially targeted ␤-ketoacyl synthase. This is the fourth nuclear-encoded component of a putative human mitochondrial FAS system to be characterized to date; the others are the ACP, malonyl transferase, and enoyl reductase. In common with the other three human mitochondrial FAS enzymes, as well as the fungal and plant mitochondrial ␤-ketoacyl synthases, the animal ␤-ketoacyl synthases appear more closely related to prokaryotic and plastid counterparts than to the cytosolic type I FAS of the same species. Thus, phylogenetic analysis reveals that all of the mitochondrial ␤-ketoacyl synthases originate from a single branch (Fig. 8), consistent with the generally accepted hypothesis that mitochondria originated from free-living prokaryotes.
Present day prokaryotes and plant plastids employ three ␤-ketoacyl synthases with different specificities to fulfill their fatty acid requirement. They are ␤-ketoacyl synthase III, which catalyzes the initial condensation step using acetyl-CoA as primer, and ␤-ketoacyl synthases I and II, which catalyze the remaining elongation reactions using acyl-ACPs as primers. All three types of enzyme share a similar overall fold and differ primarily in the structure at the active site, reflecting their different specificities (31). The ␤-ketoacyl synthases I and II have similar specificities, except that ␤-ketoacyl synthase II is able to elongate 16:1-18:1. Because only one type II ␤-ketoacyl synthase was identified in the human genome, the substrate specificity of this enzyme was of particular interest. The hu-man mitochondrial enzyme, which appears more closely related to prokaryotic ␤-ketoacyl synthases I and II than to ␤-ketoacyl synthase III enzymes (Fig. 7), is unable to utilize acetyl-CoA as the primer. However, in contrast to the prokaryotic ␤-ketoacyl synthases I and II, it is able to catalyze the initial condensation reaction using a 2-carbon primer, presented as acetyl-ACP. Otherwise, the specificity of the mitochondrial enzyme (Fig. 6) is quite similar to that of the prokaryotic ␤-ketoacyl synthases I and II (32) in that, with all three types, enzyme activity falls off significantly with substrates containing more than 12 C atoms. However, the specific activity of the human mitochondrial enzyme is ϳ20-fold higher than that reported for the E. coli ␤-ketoacyl synthases I and II (32). It is unclear whether the disparity reflects a true difference in catalytic efficiency or can be attributed to suboptimal assay conditions used for the prokaryotic enzymes.
Recently, the mitochondrial ␤-ketoacyl synthase from A. thaliana has been identified, cloned, and characterized (18). This enzyme too exhibits a high sequence similarity with ␤-ketoacyl synthases I and II of prokaryotes and is able to restore growth of the E. coli CY244 strain, which lacks both ␤-ketoacyl synthases I and II. The authors concluded that the A. thaliana mitochondrial enzyme does not use either acetyl-CoA or acetyl- Multiple sequence alignments were performed using ClustalW, and the unrooted tree was constructed using TreeView (30). Mitochondrial ␤-ketoacyl synthases lineages are shown in heavy lines with boldface labels, type I ␤-ketoacyl synthases associated with cytosolic FASs (cFAS) and modular polyketide synthases (mPKS) are shown as dashed lines, and the distantly related thiolases that catalyze the reverse reaction in fatty acid oxidation are shown as dotted lines. All other lineages represent type II ␤-ketoacyl synthases and the three types are distinguished by the notation KSI, KSII, or KSIII.
ACP as primer but instead uses acetyl moieties derived by decarboxylation of malonyl-ACP. In this respect, the enzyme appears to differ from the human mitochondrial ␤-ketoacyl synthase, which clearly is able to utilize acetyl-ACP as primer.
The major products of the prokaryotic type II FAS pathway are long chain acyl-ACPs containing 16 and 18 C atoms, but one of the intermediates, octanoyl-ACP, is siphoned off for the production of lipoyl-ACP. Experiments with plant mitochondria indicate that the major products of this mitochondrial type II FAS system also are octanoyl and long chain length acyl-ACPs (5). Furthermore, the profile of fatty acids synthesized by soluble extracts of E. coli CY244 cells supplemented with the A. thaliana mitochondrial ␤-ketoacyl synthase also showed a bimodal distribution with maxima at C-8 and C-14 -16. The hypothesis that the mitochondrial FAS system might be responsible for producing both octanoyl and long chain length acyl-ACPs is supported by mutational disruption of the mitochondrial FAS system of fungi. For example, disruption of the nuclear-encoded gene for mitochondrial ACP in Neurospora crassa results in an accumulation of lysophospholipids in mitochondrial membranes and an accompanying respiratory-deficient phenotype. Disruption of the nuclear-encoded gene for either the mitochondrial ␤-ketoacyl synthase or ACP in S. cerevisiae also produces a respiratory-deficient phenotype and, in the case of the ACP-defective strain, it has been established that cellular lipoic acid concentration is reduced to Ͻ10% of that of the wild-type strain (4,33). Furthermore, addition of lipoic acid to the growth medium could not compensate for the block in endogenous lipoic acid synthesis, suggesting that in this species lipoic acid may not readily be taken up into mitochondria (4). Our finding that the human mitochondrial ␤-ketoacyl synthase is able to restore normal growth to the cem1 mutant strain implies that the substrate specificity and kinetic properties of the human enzyme are entirely compatible with those of the endogenous S. cerevisiae enzyme and allow the synthesis and release of octanoyl-ACP from the elongation cycle for conversion to lipoyl-ACP. An interesting feature of the kinetic properties of the human mitochondrial ␤-ketoacyl synthase is the higher K m for octanoyl-ACP (11 M), compared with the values for hexanoyl-and decanoyl-ACP (ϳ2 M). Possibly, this property may result in a higher pool size for the C-8 intermediate that could facilitate diversion into the lipoate pathway. For many years it has been generally assumed that in mammals the requirement for lipoic acid is met by dietary intake. In this pathway, free lipoate, taken up via vitamin transporters (34,35) is activated to lipoyl-AMP by a lipoate-activating enzyme, and then the lipoyl moiety is transferred to a lysine residue in the acceptor protein by a lipoyl-AMP:N⑀-lysine lipoyltransferase; both enzymes have been cloned and characterized from mammalian sources (36). However, recent studies have revealed that mammalian mitochondria also contain a lipoic acid synthase, which converts octanoyl-ACP to lipoyl-ACP (37). Although a lipoyl-ACP-dependent transferase required for the lipoylation of mitochondrial proteins (performed by the lipB and LIP2 gene products of prokaryotes and plant mitochondria, respectively (38,39)) has yet to be identified in mammalian mitochondria, this finding raised the possibility that endogenously synthesized lipoic acid may be utilized for the lipoylation of mitochondrial proteins, as it is in fungal and plant mitochondria. In mammals, all of the known lipoate-containing proteins, pyruvate dehydrogenase, ␣-ketoglutarate dehydrogenase, and glycine cleavage enzyme, are located in the mitochondria (40) and the presence of a lipoate-producing system in this organelle may ensure an adequate supply of this cofactor, essential for mitochondrial function, independent of dietary availability.
Collectively then, the complementation of the CEM1 knockout by the human mitochondrial ␤-ketoacyl synthase, the similar substrate specificities, and close evolutionary relationship of type II human mitochondrial and prokaryotic ␤-ketoacyl synthases suggest that the type II enzymes fulfill similar roles in these different environments, namely the supply of octanoyl-ACP for lipoic acid production and the production of long chain length fatty acyl moieties for membrane phospholipid biosynthesis. Nevertheless, until unequivocal experimental evidence is obtained validating this hypothesis, the possibility that the FAS pathway might generate other products important for mitochondrial function, such as myristate for protein modification, should not be discounted. In this regard, the finding that ACP isolated from N. crassa mitochondria appeared to have a ␤-hydroxymyristoyl moiety attached to the phosphopantetheine (41) remains something of an enigma. Evaluation of the effects of down-regulation of expression of mitochondrial FAS components on mitochondrial function may resolve these questions.
In recent years the type II fatty acid-synthesizing systems of microorganisms have been identified as promising targets for the development of new therapeutic agents for the treatment of a variety of infectious diseases, based primarily on the premise that such agents likely would have little or no effect on the type I FAS system present in the cytosol of the animal host cells (31,42,43). Demonstration of the presence of a type II FAS system in animal mitochondria indicates that caution should be exercised in the screening of such compounds, and assurance should be sought that they have no serious side effects on mitochondrial function.