Purification and Characterization of 3-Ketoacyl-Acyl Carrier Protein Synthase I11 from Spinach A CONDENSING ENZYME UTILIZING ACETYL-COENZYME A TO INITIATE FATTY ACID SYNTHESIS*

The 3-ketoacyl-acyl carrier protein (ACP) synthase I11 from spinach was purified to homogeneity by an eight-step procedure that included an ACP-affinity column. The size of the native enzyme was M, = 63,000 based on gel filtration, and its subunit size was M, = 40,500 based on sodium dodecyl sulfate-polyacryl-amide gel electrophoresis, suggesting that 3-ketoacyl- ACP synthase I11 may be a homodimer. The purified enzyme was highly specific for acetyl-coA and malonyl-ACP. The K,,, for acetyl-coA was 5 NM when assayed in the presence of 10 p~ malonyl-CoA. Acetyl-, butyryl-, and hexanoyl-ACP would not substitute for acetyl-coA as substrates. The specificity for acetyl- CoA suggested that the physiological function of 3- ketoacyl-ACP synthase is to catalyze the initial condensation reaction in fatty acid biosynthesis. The homogeneous 3-ketoacyl-ACP synthase was capable of catalyzing acetyl-CoA:ACP transacylation but at a rate about 90-fold slower than the condensation reaction with malonyl-ACP. The 3-ketoacyl-ACP synthase was inhibited 100% by 5 mM N-ethylmaleimide or 20 mM and malonyl- CoAACP transacylase assays were all based on the conversion of radiolabeled acyl-CoA substrates which were soluble in 10% trichlo- roacetic acid to radiolabeled acyl-ACP products which precipitated in 10% trichloroacetic acid. The standard KAS-111 assay, in a volume of 25 pl, contained 100 mM Tris, pH 8.0, 800 pM dithiothreitol, 10 pM spinach ACP, 10 p~ malonyl-CoA, 0.06 nmol/min malonyl-CoAACP transacylase, 0.002 nmol/min KAS-111, and 10 p~ [l-'4C]acetyl-CoA.

The reactions resulting in the biosynthesis of fatty acids are common to all organisms and are catalyzed by acetyl-coA carboxylase and the fatty acid synthase (Bloch and Vance, 1977;Wakil et al., 1983;Stumpf, 1987). Fatty acid synthase is responsible for the sequential two-carbon elongation process and consists of six distinct enzyme activities. In addition, ACP,' a small (9-kDa) protein, is the acyl carrier during each of these reactions. The plant and prokaryotic fatty acid synthases are classified as type I1 synthases in which each reaction is catalyzed by a discrete and dissociable enzyme (Bloch and Vance, 1977). This is in contrast to the animal and yeast synthases which are a type I fatty acid synthase and consist of two large multifunctional subunits (Wakil et al., 1983). The type I1 fatty acid synthase from Escherichia coli (Volpe and Vagelos, 1973;Bloch and Vance, 1977), safflower, and spinach (Stumpf, 1987) have been substantially characterized. Com-* This work was supported by National Science Foundation Grant DCB-8904395 and the Committee for Faculty Research at Miami University. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore he hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
7l To whom correspondence should be addressed. Tel.: 513-529-2094;Fax: 513-529-4221. The abbreviations used are: ACP, acyl carrier protein; HIC, hydrophobic interaction chromatography; KAS, 3-ketoacyl-ACP synthase; PAGE, polyacrylamide gel electrophoresis; NEM, N-ethylmaleimide; HPLC, high performance liquid chromatography; MES, 4morpholineethanesulfonic acid. parison of these enzymes from the plant and bacterial synthases with respect to their sizes and common physical properties reveal that they are quite similar.
The 3-ketoacyl-ACP synthase (KAS), commonly referred to as the condensing enzyme, catalyzes the condensation of an acyl-ACP with malonyl-ACP to produce the corresponding 3-ketoacyl-ACP, elongated by two carbons.
This enzyme was purified from E. coli sufficiently to yield a size determination of 80 kDa (Garwin et al., 1980). KAS was also capable of catalyzing partial reactions (Alberts et al., 1972), which included an acyl-CoAACP transacylase. 0 R A S-ACP + COA Ft R tt S-COA + ACP Multiple forms of E. coli KAS, KAS-I and KAS-11, were subsequently discovered (D'Agnolo et al., 1975;Garwin et al., 1980), with KAS-I1 being distinguished from KAS-I by size (85 kDa) and its ability to catalyze preferentially the final condensation in cis-vaccenic acid biosynthesis. Similar KASs were also found in plants (Shimakata and Stumpf, 1982a, 1982b, 1982c, 1983aMacKintosh et al., 1989). Plant KAS-I also has the substrate specificities expected for the enzyme which catalyzes the initial condensation reactions of fatty acid synthase resulting in palmitoyl-ACP synthesis, whereas KAS-I1 is specific for long chain acyl-ACPs (14:O and 16:O).
Recently, a third 3-ketoacyl-ACP synthase, KAS-111, was discovered in E. coli Jackowski et al., 1989) and plants Walsh et al., 1990). KAS-I11 analyzed in crude cell homogenates had a substrate specificity that was consistent with it catalyzing the condensation step during the initial cycles of fatty acid synthase. In addition, KAS-I11 differed from all previously characterized condensing enzymes with respect to its ability to utilize acetyl-coA directly as a primer rather than an acyl-ACP (Jackowski et al., 1989;Jaworski et al., 1989). Thus, this enzyme appears to provide a means to bypass the need for acetyl-ACP. Since KAS-I11 may obviate the requirement for acetyl-ACP in fatty acid synthesis and since previously studied condensing enzymes have transacylase activity associated with them as a partial reaction (Alberts et al., 1972), it was suggested that previously characterized acetyltransacylase (Williamson and Wakil, 1966;Alberts et al., 1969;Shimakata and Stumpf, 1983b) was actually a condensing enzyme (Jackowski et al., 1989), which is contrary to the widely accepted pathway for fatty acid synthesis.
The KAS-I11 gene from E. coli has been isolated and its expression product purified (Tsay et al., 1992). This gene directed the synthesis of a 34,800-dalton protein, and after purification, it catalyzed both the condensation of acetyl-coA with malonyl-ACP, as well as the acetyl-CoAACP transacylase reaction. Although the purified KAS-I11 catalyzed both reactions, the specific activity of the condensing enzyme reaction was 200-fold higher than the transacylase reaction, and it was not determined if the transacylase activity of KAS-I11 could account for all the acetyltransacylase in E. coli. Nonetheless, the properties of the purified expression product of the KAS-I11 gene were consistent with the role of KAS-I11 catalyzing the initial reaction of fatty acid synthesis.
In this paper, the purification and characterization of KAS-I11 from spinach leaf are described. To begin to address the in vivo role of KAS-I11 in fatty acid biosynthesis, an analysis of the substrate specificity of the purified enzyme was carried out. In addition, the ability of the purified enzyme to catalyze acetyl-CoA:ACP transacylation was analyzed.

Isolation of Spinach ACP
Recombinant spinach ACP-I was isolated from E. coli E103S (pPB104), a gift of J. B. Ohlrogge. All volumes in this procedure refer to the starting volume of cell paste, usually 50-200 ml. Cells suspended in 3 volumes of 50 mM Tris, pH 8.0, 10 mM MgCl, were broken with a French pressure cell and the 105,000 X g supernatant collected. The supernatant was diluted 5-fold and adjusted to 2.5% (w/v) trichloroacetic acid. After 1 h at 4 "C, the precipitate was collected by centrifugation at 5,000 X g for 10 min. The pellet was suspended in 1 volume of water with a blender or Polytron, the pH adjusted to approximately 8.0 with 1.0 M Tris base, and 2-mercaptoethanol was added to a final concentration of 100 mM. Finally, the p H was carefully raised to 9.0 with 1 M NaOH, and the suspension was stirred slowly for 1 h at room temperature to hydrolyze all acyl-ACPs and completely reduce ACPs (Lakin- Thomas and Brody, 1986). Undissolved material was removed by centrifugation (5,000 X g for 10 min), and the supernatant was adjusted to 5% (w/v) trichloroacetic acid. After 30 min at 4 "C, the precipitate was collected by centrifugation. The pellet was redissolved in 1 volume of 50 mM Tris, and the pH was adjusted to 7-8 with 1.0 M NaOH and the pellet dialyzed overnight against 20 volumes of 50 mM Tris, pH 7.5. A large precipitate containing about 30% of the total ACP formed and was removed by centrifugation. The supernatant was adjusted stepwise to 40, 60, and finally 70% saturated ammonium sulfate with the supernatant saved each time. The final supernatant was adjusted to 10% (w/v) trichloroacetic acid; the precipitate was collected by centrifugation, redissolved in 50 mM Tris, and dialyzed against 50 mM Tris, pH 7.5, overnight.
The ACP solution containing both spinach ACP-I and E. coli ACP was further purified on a DEAE-cellulose column. The column was equilibrated with 10 mM Tris, pH 7.5, and ACPs were separated with 20 column volumes of a linear 0-0.5 M LiCl gradient. Peaks of ACP activity were at conductivities of 1.2 X lo4 mhos (spinach ACP-I) and 1.6 X lo4 mhos (E. coli ACP). The spinach ACP-I was adjusted to 70% saturated ammonium sulfate, the precipitate removed by centrifugation, and the ACP-I collected by precipitation with 10% (w/v) trichloroacetic acid. This preparation of ACP-I was essentially pure as judged by SDS-PAGE with Coomassie staining. The E. coli ACP was contaminated with as much as 10-15% spinach ACP and could be purified further with a shallower gradient on DEAE or on a Mono Q column.

KAS-III Purification
All steps were carried out at 4 "C. Between indicated purification steps, the active fractions were routinely concentrated with an Amicon PM-10 filter.
Step 1. Ammonium Sulfate Precipitate-A crude extract was prepared by homogenizing batchwise 10 kg of spinach leaf in a blender containing 2 volumes (w/v) 20 mM potassium phosphate, pH 7.6,0.3 mM EDTA, and 0.5 mM dithiothreitol. The slurry was filtered by gravity through six layers of cheesecloth, then centrifuged at 13,000 X g for 10 min. The supernatant was collected and adjusted to 40% saturation with solid ammonium sulfate and to pH 7.6 with KOH, stirred 1 h on ice, and then centrifuged. The supernatant was adjusted to 80% saturation with solid ammonium sulfate, and to pH 7.6, stirred 1 h, and centrifuged. The pellet was resuspended into 1.0 M ammonium sulfate, 20 mM KDG buffer (potassium phosphate, pH 7.6, 1 mM dithiothreitol, 20% (v/v) glycerol).
Step 2. Hydrophobic Interaction Chromatography-A Baker WP HI-Propyl (40 pm, 275 A particle size) HIC column was preequilibrated with 1.8 M ammonium sulfate, 25 mM KDG, pH 7.0. Batches of the ammonium sulfate precipitate containing 15 g of protein were separated on a 250-ml column eluted with a 1.8-0 M ammonium sulfate, 25 mM KDG, pH 7.0 gradient (2,400 ml) at a flow rate of 10 ml/min. The KAS-I11 eluted at 1.0 M ammonium sulfate. The KAS-111 fractions were pooled, concentrated, and dialyzed against 20 mM KDG.
Step 4. Affi-Gel Blue Separation-Two Affi-Gel blue (Bio-Rad) columns were used sequentially. DEAE-purified KAS-I11 was loaded onto a 200-ml column equilibrated with 20 mM KDG. The column was washed with 1.0 M NaC1, 0.2 M KDG (1,000 ml), rinsed with 0.2 M KDG (300 ml), and the KAS-111 was eluted with 1.0 M KSCN in 0.2 M KDG buffer. All of the KAS-111 activity appeared in the initial fractions (50 ml total) containing the 1.0 M KSCN. Because KAS-I11 was unstable and could not be assayed in KSCN, these fractions were dialyzed as they eluted from the column by passing the eluant through a 200-filament Spectra/Por hollow fiber bundle suspended in 2.0 liters of rapidly stirring 20 mM potassium phosphate, pH 7.6, with several changes of buffer occurring during the elution. Active fractions were pooled and extensively dialyzed against 20 mM KDG. After dialysis, KAS-I11 was loaded onto a second Affi-Gel blue column (15 ml) and eluted with a 75-ml gradient of 0-1.0 M KSCN in 0.2 M KDG buffer at a flow rate of 0.6 ml/min. The eluant was also dialyzed through the 200-filament Spectra/Por hollow fiber bundle against 20 mM potassium phosphate, pH 7.6. The KAS-111 eluted at 0.25 M KSCN, and the active fractions were pooled and dialyzed against 20 mM KDG.
Step 5. Hydroxylapatite Separation-The KAS-111 was loaded onto a 30-ml hydroxylapatite column (Bio-Gel HTP, Bio-Rad) equilibrated with 20 mM KDG and eluted with a 20 mM to 0.2 M KDG gradient (300 ml) at a flow rate of 2 ml/min. The KAS-I11 eluted at 0.04 M KDG, and the fractions were pooled, concentrated, and adjusted to 1.8 M ammonium sulfate, 25 mM KDG, pH 7.
Step 6. HIC-HPLC Separation-The KAS-I11 was loaded onto a Baker 4.6 X 50 mm HI-Propyl HIC-HPLC column and eluted with a 1.8 to 0 M ammonium sulfate, 25 mM KDG, pH 7 gradient (15 ml), at a flow rate of 0.5 ml/min. KAS-I11 eluted at 0.24 M ammonium sulfate.
Step 7. Mono Q Separation-KAS-I11 was loaded onto a Mono-Q HR 5 / 5 column equilibrated with 20 mM KDG (10% glycerol), pH 7.6. The enzyme was eluted with a 0-0.2 M KC1 gradient (25 ml) at a flow rate of 1 ml/min. It should be noted that the performance of this column was inexplicably quite variable from one purification series to another, and often the KAS-I11 did not stick to the column. The KAS-111 fractions were pooled, concentrated, and dialyzed against 20 mM Tris, pH 7.2, 0.5 mM dithiothreitol, 20% glycerol, 0.01% Triton X-100 buffer.
Step 8. ACP-affinity Purification-For the final purification step, the enzyme was purified twice on a 600-pl spinach ACP-affinity column. The ACP-affinity medium was prepared by mixing ACP with activated Affi-Gel 15 (Bio-Rad) and rotating continuously in a 1.5ml microcentrifuge tube at room temperature for 1 h. Typically, 7 mg of ACP was mixed with a 0.6-ml bed volume of Affi-Gel 15 in a total volume of 1-1.2 ml of 20 mM MES, pH 6.1.
Before use, the ACP-affinity column was incubated for 15 min with 5 mM dithiothreitol, 20 mM Tris, pH 7.2, and then equilibrated with 7 ml of 20 mM Tris, pH 7.2,0.5 mM dithiothreitol, 20% glycerol, 0.01% Triton X-100 buffer. One column volume of enzyme was loaded and allowed to incubate for 10-20 min before loading more enzyme or proceeding with elution of the KAS-111. After loading, the column was sequentially eluted with equilibration buffer (2 ml), with 0.2 M NaCl in buffer (3 ml) to remove nonspecifically bound proteins, and finally with 1.0 M NaCl in buffer (3 ml) to elute the KAS-111. Usually, about 50% of the loaded enzyme was eluted with the 1.0 M NaCl. After dialysis against the equilibration buffer, the enzyme was again loaded onto the reequilibrated ACP-affinity column and eluted by the same procedure. Some ACP was released from the affinity matrix during the process, but this ACP was easily removed from the KAS-111 by batch elution from a 100-111 DEAE-cellulose column, to which ACP binds much more tightly than KAS-111. The purified enzyme was stored at -80 "C in KDG buffer with 0.01% Triton X-100.

Enzyme Assays
The KAS-111, acetyl-CoAACP transacylase, and malonyl-CoAACP transacylase assays were all based on the conversion of radiolabeled acyl-CoA substrates which were soluble in 10% trichloroacetic acid to radiolabeled acyl-ACP products which precipitated in 10% trichloroacetic acid. The standard KAS-111 assay, in a volume of 25 pl, contained 100 mM Tris, pH 8.0, 800 pM dithiothreitol, 10 p M spinach ACP, 10 p~ malonyl-CoA, 0.06 nmol/min malonyl-CoAACP transacylase, 0.002 nmol/min KAS-111, and 10 p~ [l-'4C]acetyl-CoA. In addition, 100 p~ cerulenin was present in all assays to inhibit completely any KAS-I activity that may be present. The Tris, dithiothreitol, and spinach ACP were incubated for 15 min at 32 "C before the remainder of the reaction mix was added. The reaction was initiated by the addition of [1-"Clacetyl-CoA, incubated at 32 "C, and stopped by the addition of 2.8 pl of 100% trichloroacetic acid (w/ v). The mixture incubated on ice for 10 min, was diluted with 400 pl of 5% trichloroacetic acid, centrifuged for 5 min at 15,000 X g, and the supernatant removed by aspiration. The pellet was washed with an additional 400 pl of 5% trichloroacetic acid and then redissolved into 40 pl of 0.1 M NaOH and analyzed by liquid scintillation spectrometry. When a progress curve for an assay was obtained, the reaction was stopped by removal of an aliquot of reaction mix at a designated time, and the aliquot was immediately added it to 0.11 volume of 100% trichloroacetic acid (w/v). The activity was dependent on malonyl-CoA and ACP except for the small amount of acetyl-CoA-ACP transacylase activity (less than 5%), which was not subtracted.
When KAS-111 was analyzed to determine substrate specificity, the 3-ketoacyl-ACP products in the 10% trichloroacetic acid precipitate were subsequently reduced enzymatically to the saturated acyl-ACP before being separated by native PAGE and analyzed by autoradiography . This type of analysis was necessary because the substrates examined were unlabeled acetyl-coA or acyl-ACPs with [2-14C]malonyl-CoA and ACP. Since the reactions also contained [2-"C]malonyl-ACP which precipitated in 10% trichloroacetic acid, the KAS-111 assay described above was inappropriate. Briefly, the reductases and dehydrase of fatty acid synthase were obtained from the starting 40-80% saturated ammonium sulfate precipitate which was treated with 10 mM NEM to remove KAS activity. After the standard KAS assay, the precipitated 3-ketoacyl-ACPs were redissolved and mixed with the NEM-treated fatty acid synthase, 1 mM NADH, 2 mM NADPH, and 0.1 M Tris, pH 7.5, to yield the saturated acyl-ACPs without any further extension of the products. Incubations were for 10 min at room temperature before again precipitating the acyl-ACPs with 10% trichloroacetic acid and preparing samples for SDS-PAGE.
Acetyl-CoAACP transacylase was assayed by the procedure described for the KAS-111 except the malonyl-CoA and malonyl-CoAACP transacylase were omitted.
Malonyl-CoAACP transacylase was assayed by the procedure described for the KAS-111 except the cerulenin and [1-"C]acetyl-CoA are omitted, and the final concentrations of spinach ACP and [2-"C] malonyl-CoA (10 pCi/pmol) were each 50 p~. The reaction was initiated by the addition of malonyl-CoA.

Inhibition Studies
Inhibition studies were carried out with NEM and sodium arsenite, covalent inhibitors of thiols. Because the KAS-111 assay includes ACP and malonyl-CoAACP transacylase which also contain active thiols, the KAS-111 was preincubated at the stated concentration of inhibitor, and the excess inhibitor was titrated away with dithiothreitol. The KAS-I11 was preincubated with 0.5 mM dithiothreitol in 20 mM potassium phosphate, pH 7.6, buffer for 10 min at room temperature in a volume of 3 pl. An equal volume of inhibitor was then added and incubated an additional 10 min at room temperature. The reported concentration of the inhibitor was its concentration during this incubation. To remove unreacted inhibitor, dithiothreitol was added a t a concentration equimolar to the inhibitor, and the mix was allowed to incubate for 10 min at room temperature. These enzyme preparations were then used immediately to assay for KAS-111 and acetyltransacylase, as described above.

Moleculnr Mass
The molecular mass of native KAS-I11 was determined by gel filtration using a Superose 12 HR 10/30 column (Pharmacia LKB Biotechnology Inc.). The column was equilibrated with 50 mM potassium phosphate, pH 7.0, 150 mM NaCl, and 0.5 mM dithiothreitol. Elution volumes were calculated using the leading side of solute peaks extrapolated to base line.
Electrophoresis SDS-PAGE was performed essentially by the method of Laemmli (1970), fixed with 6% perchloric acid, and stained with ammoniacal silver nitrate solution (Hochstrasser et al., 1988).

RESULTS
Analysis of the substrate specificity of KAS-I11 in spinach leaf homogenates suggested that KAS-I11 catalyzed the initial condensation reaction of fatty acid synthesis. Because the substrate specificity of KAS-I11 was central to our understanding of this enzymes role in fatty acid biosynthesis, the spinach KAS-I11 was purified to homogeneity to determine its substrate specificity in the absence of any potentially interfering reactions.
Purification of US-IZI-KAS-I11 was purified to homogeneity from spinach leaf, as judged by SDS-PAGE with protein detection by silver staining (Fig. 1) and as summarized in Table I  pure sample purified on the ACP-affinity column resulted in a nearly homogeneous KAS-I11 (Fig. 1, compare lane 1 with  lane 4 ) , and repurification of this KAS-I11 on the same affinity column removed the remaining impurities (Fig. 1, lane 5). KAS-I11 binding to the ACP affinity matrix was relatively weak. Consequently, a concentrated KAS-I11 sample was required to obtain significant binding of the KAS-I11 to the column. The extensive purification obtained from the earlier columns allowed the KAS-111 to be concentrated sufficiently without creating too viscous a solution due to high protein levels. Even under these conditions, only about 50% of the KAS-I11 was obtained in the 1.0 M NaCl elution. The remainder of the enzyme either did not stick or was present in the 0.2 M NaCl elution. During the final purification steps, the enzyme was most stable at -80 "C in 20 mM KDG buffer (or 20 mM Tris, 1 mM dithiothreitol, 20% glycerol buffer) with 0.01% purified Triton X-100. The Triton X-100 was necessary to prevent loss of the KAS-I11 caused by nonspecific binding, and no specific loss of activity was observed because of its addition. Under these conditions, the KAS-I11 slowly lost activity over several weeks. For the final steps of the purification, the level of protein was too low to measure without sacrificing a significant portion of the enzyme. The protein level in the pure enzyme was estimated from the silver stained SDS-polyacrylamide gel, and included in this estimation was the assumption that the KAS-I11 was stained with approximately as the same intensity as the molecular weight standards. As the staining of proteins is quite variable, this estimate could be in substantial error. Nonetheless, the approximately 3,500-fold purification and a final specific activity of 0 7,000 nmol/min/mg protein were in reasonable agreement with the same data reported for the KAS-I purification from Brassica napw (MacKintosh et al., 1989). The abundance of KAS-I and KAS-I11 in plants would be expected to be similar.

3-Ketoacyl-Acyl Carrier Protein Synthase
Substrate Specificity-The role of different isozymes of KAS in fatty acid biosynthesis is determined primarily by their substrate specificity with respect to the acyl group. KAS-111 was distinct from other condensing enzymes in that it used acetyl-coA instead of acyl-ACPs . We began to address the question the acyl group specificity for KAS-I11 by determining its activity in response to varying acetyl-coA concentrations. KAS-I11 followed normal Michaelis-Menten kinetics with respect to acetyl-coA when assayed in the presence of near saturating levels (10 p~) of malonyl-CoA (Fig. 2). Under these conditions, the K,,, for acetyl-coA was determined to be 5 PM based on linear regression analysis of the double-reciprocal plot.
Initial investigation of the substrate range of KAS-I11 using cell homogenates indicated that acyl chains could be extended at least to c8 in the presence of 100 p M cerulenin. Cerulenin is an antibiotic inhibitor of KAS-I which covalently binds to the active site. Complete loss of KAS-I activity is observed with cerulenin at concentrations of less than 10 pM (Shimakata and Stumpf, 1982b). Thus, the synthesis of 80-ACP suggested that 6:0-ACP was a substrate for KAS-I11 . To determine the substrate specificity of the purified KAS-111, unlabeled acyl-ACPs were prepared and the reactions carried out with 10 p~ ['4C]malonyl-ACP. Because both the labeled substrate and the expected products were precipitated with trichloroacetic acid, the reaction was monitored by separation of the reaction components by native PAGE followed by transfer of the acyl-ACPs to nitrocellulose and autoradiography. Initial analysis of the reaction products indicated that the 3-ketoacyl-ACPs were unstable to PAGE, requiring modification of the standard KAS assay to include enzymatic reduction and dehydration of the 3-ketoacyl-ACPs to the saturated acyl-ACP.
The substrate specificity of the KAS-I11 was determined using acetyl-coA and acyl-ACPs at concentrations of 1, 3, and 10 pM (Fig. 3A). Only acetyl-coA was an effective substrate at the concentrations analyzed. Even when the autoradiogram was overexposed, there were no visible products in the lanes for the other substrates. Since earlier analysis with crude enzyme preparations had suggested a broader substrate specificity for KAS-I11 , we repeated this substrate analysis with crude enzyme and partially purified enzyme (Fig. 3B). The crude enzyme preparation was able to extend acetyl-coA as well as 2:O-4:O-and 6O-ACP, as expected based on previous results. The KAS activity with 4:0-ACP was slightly less than that with acetyl-coA, whereas the activity with 6:0-ACP was much less. When 20-ACP was the substrate, the KAS activity was barely detectable, indicating that if KAS-I11 catalyzes the initial condensation of fatty acid biosynthesis, it does so almost exclusively with acetyl-coA. KAS-I11 that had been purified through the Affi-Gel blue step had the same substrate specificity as the purified KAS-111, indicating loss of the cofactor or enzyme occurred early in the purification that resulted in extended substrate specificity for condensing activity. This KAS activity with the short chain acyl-ACPs was not caused by KAS-I, since these reactions were carried out with 100 p M cerulenin. We confirmed the earlier report (Shimakata and Stumpf, 1982b) that this concentration of cerulenin was 3-10-fold greater than was necessary to inhibit completely KAS-I activity and fatty acid synthesis (data not shown). Furthermore, this preparation was completely inactive with 80-and 100-ACP (data not shown), which are normal substrates for KAS-I. Acetyltransacylase Analysis-It has been suggested that acetyltransacylation in E. coli may be catalyzed by KAS-I11 rather than a separate acetyltransacylase, i.e. that acetyltransacylase and KAS-I11 may be the same enzyme (Jackowski et al., 1989). This suggested that KAS-I11 was similar to other condensing enzymes, and its transacylase activity was one of the partial reactions observed in vitro for the other previously studied condensing enzymes. Therefore, acetyltransacylase activity was analyzed during the purification of KAS-I11 to determine if the acetyltransacylase activity could be separated from KAS-I11 activity. Acetyltransacylase activity was compared directly with KAS-I11 activity in fractions from two of the initial KAS-I11 purification steps and in the purified KAS-I11 (Table 11). During the purification, the acetyltransacylase activity relative to KAS-I11 decreased approximately 2-fold, changing from a 51-fold difference after HIC purification to a 92-fold difference in the purified KAS-111.
Significantly, acetyltransacylase activity was detected in the all of KAS-I11 fractions that were analyzed, including the KAS I11 and acetyl transacylase assay conditions were as described under "Experimental Procedures," except, as indicated below, some of the assays were carried out using E. coli ACP. homogeneous enzyme. Furthermore, after DEAE separation, through five additional purification steps, the ratio of KAS-I11 to acetyltransacylase activity was essentially unchanged. The constancy of the ratio indicated that the acetyltransacylase was not simply a minor contaminant of the KAS-I11 fractions. This clearly demonstrated that acetyltransacylation can be catalyzed by KAS-111, although the acetyltransacylase activity of KAS-I11 was at least 50 times less than the KAS-I11 activity. Furthermore, we were not able to determine if KAS-I11 could account for all the acetyltransacylase activity in crude extracts.

Assay Purification step KAS IIP A C F (KAS III,ACT)
Interestingly, when E. coli ACP was used in place of spinach ACP-I, a 3-fold increase in the acetyltransacylase activity was observed (Table 11), and this %fold increase was consistent at each step of the purification. In contrast, the ACP source made no difference to the KAS-I11 activity. Both ACPs were purified to homogeneity and contained no detectable KAS-I11 or acetyltransacylase activity.
KAS-111 Inhibition-To examine further if acetyltransacylase and KAS-I11 reactions are catalyzed by the same enzyme, KAS-I11 inhibition was studied using two different covalent thiol inhibitors, arsenite and NEM. It has been demonstrated previously that NEM is an inhibitor of both condensing enzymes and acetyltransacylase from spinach Stumpf, 1983a, 1983b). In contrast, sodium arsenite was a potent and somewhat specific inhibitor of acetyltransacylase. However, sodium arsenite was a weak inhibitor of KAS-I1 and had no effect on the other enzymes of fatty acid synthase (Shimakata and Stumpf, 1983b). Both NEM and arsenite were potent inhibitors of KAS-I11 activity (Fig. 4). At 5 and 20 mM, arsenite inhibition of KAS-I11 was 45 and 98%, respectively. NEM was 100% inhibitory of KAS-111 at 5 mM. Thus, KAS-I11 inhibition by sodium arsenite and NEM was similar to that observed previously for acetyltransacylase.
Molecular Characterization of KAS-111-Chromatography of KAS-I11 by gel filtration on a Superose 12 HR column indicated an M, = 63,000 for the native protein (Fig. 5).
However, when subunit size was analyzed by SDS-PAGE, M, = 40,500 was obtained (Fig. 1). Considering that KAS-I and I1 from E. coli (Gamin et al., 1980) andB. mpus (MacKintosh et al., 1989) are all known to be a homodimer these data suggest that spinach KAS-I11 may also be homodimer.

DISCUSSION
The 3-ketoacyl-ACP synthases catalyze the condensation reactions of fatty acid synthase. Prior to the discovery of KAS-111, the accepted pathway of fatty acid synthesis in-

KAS-111.
FIG. 6. Scheme for fatty acid biosynthesis incorporating cluded initiation of the process with acetyltransacylase catalyzing the synthesis acetyl-ACP, the presumed substrate for the initial condensation. However, KAS-I11 uses acetyl-coA directly at a rate that is at least 20-fold faster than acetyltransacylase in vitro. It therefore seems likely that, as illustrated in Fig. 6, initial reaction of fatty acid synthesis is a condensation reaction catalyzed by KAS-I11 and that the transacylase reaction to form acetyl-ACP is not essential.
To study the extent to which KAS-I11 may participate in the initial reactions of fatty acid biosynthesis, its relative activity with acetyl-coA, acetyl-ACP, and other short chain acyl-ACPs was determined. A clear difference between KAS-111 and the other condensing enzymes studied in plants (Shimakata and Stumpf, 1982a, 1982b, 1982c, 1983a was found in its substrate specificity. The purified KAS-I11 used exclusively acetyl-coA, whereas other condensing enzymes use acyl-ACPs of varying chain length. The inability of KAS-I11 to use acetyl-ACP suggests that if acetyl-ACP is an intermediate in fatty acid biosynthesis in vivo, it is used by KAS-I or an unidentified KAS (see Fig. 6). Moreover, the strong preference of KAS-111 for acetyl-coA suggests that catalysis of the initial condensation reaction is the primary role of KAS-I11 in fatty acid biosynthesis.
A puzzling observation was the extended substrate specificity of the 40-80% saturated ammonium sulfate precipitate in the presence of 100 p~ cerulenin (Fig. 3B). Since cerulenin inhibits KAS-I by alkylation of the active site (Kauppinen et al., 1988) and we demonstrated that this cerulenin-treated preparation was devoid of condensing enzyme activity for longer chain acyl-ACPs, the extended substrate specificity presumably was not caused by any residual KAS-I. It does not seem likely that the change in substrate specificity during purification is a result of an alteration in the KAS-I11 itself, since the K,,, for acetyl-coA is essentially unchanged during the purification. Since the ability to use 4:0-ACP is lost at an early step in purification, presumably either another unknown KAS or an unknown factor that influences KAS-111 specificity was removed. These results are analogous to the in vivo observations made for E. coli in the presence of high cerulenin concentrations . Under conditions that completely inhibit KAS-I activity, short chain acyl-ACPs are synthesized. In contrast to plants, KAS-I11 in crude homogenates of E. coli is only capable of using acetyl-coA.
A fundamental question resulting from the study of KAS-I11 concerns the role of the acetyltransacylase reaction in fatty acid biosynthesis. There are two major aspects to this problem that relate to our basic understanding of this primary pathway. First, is the acetyltransacylase reaction catalyzed by a separate transacylase or by a condensing enzyme, i.e. KAS-III? Second, is acetyl-ACP normally used in vivo as an intermediate in fatty acid biosynthesis?
The suggestion that the acetyltransacylase reaction may be catalyzed by KAS-I11 resulted from the study of a spontaneous mutant of E. coli, CDM5, that had acquired resistance to the antibiotic inhibitor of KAS-111, thiolactomycin (Jackowski et al., 1989). In the parent strain, both acetyltransacylase and KAS-111 activities were inhibited by thiolactomycin, but both activities from CDM5 were much less inhibited by thiolactomycin. Consistent with that suggestion is our observation that the purified KAS-I11 retained acetyltransacylase activity, and it was strongly inhibited by arsenite. It was shown previously that acetyltransacylase was the only fatty acid synthase enzyme from spinach that was inhibited by this dithiol inhibitor (Shimakata and Stumpf, 1983b). Finally, the gene for KAS-I11 in E. coli has recently been cloned (Tsay et al., 1992), and its purified expression product also contained both KAS-I11 and acetyltransacylase activity, although the KAS-I11 activity was 200-fold greater than the acetyltransacylase activity.
While the suggestion that KAS-I11 catalyzes the acetyltransacylase reaction is reasonable, there are some data that suggest there may be a separate acetyltransacylase. Enzymes that catalyze the acetyltransacylase reaction have been purified from E. coli (Lowe and Rhodes, 1988) and from spinach (Shimakata and Stumpf, 1983b). The enzyme from E. coli was purified more than 30,000-fold to homogeneity. This protein consisted of two apparently identical 29-kDa subunits and was estimated by gel filtration to be 61 kDa. The acetyl-transacylase from spinach has been partially purified (180fold), and its size was reported to be 48 kDa, also based on elution from a gel filtration column, and its subunit size was also 29 kDa (Shimakata and Stumpf, 198313). In both cases, the purified enzymes had subunits that were considerably smaller than KAS-I11 from these organisms. Furthermore, all previously characterized condensing enzymes have an active site cysteine, whereas transacylases have active site serines (Wakil et al., 1983). Our analysis of acetyltransacylase activity relative to KAS-I11 during purification (Table 11) does not preclude the presence of a separate acetyltransacylase. Our earlier report of KAS-I11 in crude extracts of spinach leaf demonstrated levels of acetyltransacylase that were as high as 20% of KAS-I11 . Since these were much higher levels than seen in partially purified fractions of KAS-111, it suggests that there are additional enzymes that can catalyze the acetyltransacylase reaction. Similarly, it is unclear if in E. coli the acetyltransacylase activity of the KAS-I11 can account for all the of the transacylase activity of these cells (Tsay et al., 1992). In addition, in both E. coli and plants the acetyltransacylase reaction could be catalyzed by another condensing enzyme instead of a specific acetyltransacylase. It should be noted that the enhanced transacylase activity observed when E. coli ACP was used in place of spinach ACP (Table 11) was unexpected. Since earlier studies of plant fatty acid synthase enzymes have consistently used the E. coli ACP, this result emphasizes that some caution must be exercised at least when interpreting kinetic data from plant enzymes assayed with E. coli ACP.
In plants, acetyltransacylase has the lowest activity of the fatty acid synthase enzymes (Shimakata and Stumpf, 1983b). If this activity is actually the transacylase reaction of KAS-111, then is there any evidence that acetyltransacylase activity is functioning in uiuo? Until recently, the presence of acetyl-ACP in plant cells had not been demonstrated. However, it is now known that acetyl-ACP is present in vivo, and its pool size is responsive to changes in the rate of fatty acid biosynthesis (Post-Beittenmiller et al., 1991). In fact, in dark spinach leaf, acetyl-ACP is the most abundant acyl-ACP. Thus, whether acetyltransacylase is a separate enzyme or a partial reaction of a condensing enzyme, it appears to be active in viuo. Its contribution to overall fatty acid biosynthesis appears to be minor.' The primary objective for purifying and'characterizing the KAS-I11 was to better understand its role in fatty acid biosynthesis in plants. Additional objectives were to obtain a preliminary understanding of its relationship to other KASs, espe-* J. G. Jaworski, D. Post-Beittenmiller, and J. B. Ohlrogge, submitted for publication. cially with respect to mechanistic similarities and substrate specificity, and to compare the plant KAS-I11 with the E. coli . The transacylase activity found associated with KAS-I11 is common to all condensing enzymes studied thus far. Furthermore, its subunit size is similar to other condensing enzymes. Finally, the subunit mass and substrate specificity are essentially the same as the E. coli  In summary, we have presented the initial report on the purification to homogeneity and characterization of a KAS-I11 from plants. The data are consistent with positioning this enzyme in metabolism at the beginning of fatty acid biosynthesis, catalyzing the initial condensation reaction, similar to the role of KAS-I11 in E. coli. Along with acetyl-coA carboxylase, KAS-III's position at the beginning of the pathway suggests the potential for a regulatory role for this enzyme. Our current studies are aimed at exploring this possibility.