Purification and characterization of recombinant spinach acyl carrier protein I expressed in Escherichia coli.

Expression of plant acyl carrier protein (ACP) in Escherichia coli at levels above that of constitutive E. coli ACP does not appear to substantially alter bacterial growth or fatty acid metabolism. The plant ACP expressed in E. coli contains pantetheine and approximately 50% is present in vivo as acyl-ACP. We have purified and characterized the recombinant spinach ACP-I. NH2-terminal amino acid sequencing indicated identity to authentic spinach ACP-I, and there was no evidence for terminal methionine or formylmethionine. Recombinant ACP-I was found to completely cross-react immunologically with polyclonal antibody raised to spinach ACP-I. Recombinant ACP-I was a poor substrate for E. coli fatty acid synthesis. In contrast, Brassica napus fatty acid synthetase gave similar reaction rates with both recombinant and E. coli ACP. Similarly, malonyl-coenzyme A:acyl carrier protein transacylase isolated from E. coli was only poorly able to utilize the recombinant ACP-I while the same enzyme from B. napus reacted equally well with either E. coli ACP or recombinant ACP-I. E. coli acyl-ACP synthetase showed a higher reaction rate for recombinant ACP-I than for E. coli ACP. Expression of spinach ACP-I in E. coli provides, for the first time, plant ACP in large quantities and should aid in both structural analysis of this protein and in investigations of the many ACP-dependent reactions of plant lipid metabolism.

All de nouo plant fatty acid synthesis (FAS)' and certain reactions of end product transfer and lipid synthesis require acyl carrier protein (ACP) (1). ACP provides these reactions with thioester free energy potential via a phosphopantetheine group derived from coenzyme A (2). Previously, the purification of ACP from spinach and barley leaf tissue resulted in the separation of two isoforms (3,4). ACP-I is the major isoform in leaf tissue whereas ACP-I1 is expressed as a minor * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
To whom correspondence should be addressed: Dept. of  ' The abbreviations used are: FAS, fatty acid synthesis; ACP, acyl carrier protein; IPTG, isopropyl-0-D-thiogalactopyranoside; TCA, trichloroacetic acid; MCT, malonyl-CoAACP transacylase); GLC, gasliquid chromatography; HPLC, high pressure liquid chromat~graphy; DTT, dithiothreitol; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; FAME, fatty acid methyl esters; MES, 2-(N-morpho1ino)ethanesulfonic acid; PVP, polyvinylpyrrolidone; RIA, radioimmunoassay. component (3, 5). The relationship between ACP structure and expression may be important to the molecular basis of' FAS tissue specificity. In this regard, analyses in uitro have indicated that leaf ACP-I and ACP-I1 react differently in the reaction of oleoyl-ACP thioesterase and glycerol 3-phosphate acyl transferase (6). These enzymes provide a branch point for oleic acid utilization and may act to control the proposition of acyl groups retained in or exported from chloroplasts.
In the past, careful kinetic analyses using plant ACP have been limited due to the relative difficulty in obtaining plant ACP (6, 7). Recently, Beremand et al. (8) synthesized a spinach ACP-I gene and obtained expression in a bacterial system. We report here the purification and characterization of the synthetic ACP-I gene product. Acid precipitation, ion exchange chromatography, and gel permeation high pressure liquid chromatography (HPLC) were used to purify the recombinant ACP-I protein to homogeneity.
Eucaryotic proteins synthesized from plasmids in Escherichia coli often possess the amino-terminal methionine residue. This host modification in vivo may cause alterations in the reactivity and immunogenicity of recombinant proteins (9).
Evidence suggests that the nearest neighbor residue to the amino terminus plays a role in determining if the methionine is retained (10). To address this transgenic phenomenon and to verify that the purified ACP-I was from the pIasmid-born synthetic gene, the first 30 amino acid residues were sequenced.
To determine the reactivity of synthetic ACP-I, a series of in uitro FAS reactions was conducted. Immunological crossreactivity of synthetic ACP-I in two independent analyses was also performed with the homogeneous protein.
ACP possesses @-alanine in the phosphopantetheine prosthetic group. An investigation of holo-ACP synthesis and acylation was conducted in the E. coli @-alanine auxotroph, SJ16. Cells were labeled with [3H]@-alanine and radioactive proteins were analyzed by electrophoresis ~u o r o~a p h y (11).
Finally, gas-liquid chromatography analyses were conducted on fatty acid methyl ester profiles of transformed E. coli JMlOl cells grown to midlog phase. These results and certain aspects of recombinant ACP-I structure and function relationships will be discussed.

EXPERIMENTAL PROCEDURES' RESULTS AND DISCUSSION
In Vitro Characterization of Recombinant Spinach ACP-I-All characterizations in vitro were conducted with homoge-Portions of this paper (including "Experimental Procedures," part of "Results," Tables 1-111, and  are presented in miniprint at the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are included in the microfilm edition of the Journal that is available from Waverly Press. neous preparations of recombinant ACP-I. The HPLC-purified recombinant ACP-I UV spectra was analyzed and compared with purified E. coli ACP (Fig. 2). Only the absorbance range between 245 and 300 nm is displayed because lower UV wavelength absorbances are essentially identical. This type of analysis is similar to that described by Hoj and Svendsen (22). It is clear that recombinant ACP-I does not possess the same absorption maxima as E. coli ACP and that the spectral properties of the two ACP structures are quite dissimilar. The difference in aromatic amino acid residues is probably the basis for spectral divergence in the UV absorbance between E. coli ACP and recombinant ACP-I.
A radioimmunoassay (17) was used to assess the structural relationship of recombinant ACP-I and authentic spinach ACP-I. Only minor differences in cross-reactivity to polyclonal antibody were evident in the comparisons shown in Fig. 3. Spinach ACP-I (isolated from spinach leaves) and recombinant ACP-I competed with approximately equal effectiveness for binding to polyclonal antibodies raised to spinach ACP-I. Further evidence for recombinant ACP-I reactivity to ACP-I specific antibody is shown by the decrease of acyl-ACP synthetase activity by adding antibody to the reaction mixtures. Less than 20% maximal activity was observed at the highest antibody concentration (Fig. 3, inset). Therefore, expression of ACP-I in E. coli did not significantly alter its immunological properties.
Using a cell-free fraction from E. coli JMlOl cells, an in vitro fatty acid synthetase assay was performed. Table I1 shows that recombinant ACP-I was a relatively poor substrate for the E. coli FAS. With the ratio of ACP-I/[14CJmalonyl-CoA kept at a constant (<1), increasing the amount of recombinant ACP-I from 449 to 898 pmol only slightly increased FAS activity (0.28 and 0.34 pmol of fatty acid produced per h). Raising the amount of E. coli ACP within the same set of conditions also increased FAS activity (1.20 versus 1.76 pmol/ h). It was, however, clear that recombinant ACP-I was at least 4 times less reactive than E. coli ACP in this in vitro E.
To further examine the relationship of ACP structure to E.
coli FAS an experiment was conducted using altered ratios of ACP to ['*C]malonyl-CoA, Fig. 4 shows the result of raising the amount of ACP relative to the amount of m~onyl-CoA from a ratio of 1 to 5. A consistent difference in FAS activity between recombinant ACP-I and E. coli ACP is evident. In each increase of the ACP/['*C]malonyl-CoA molar ratio a significant drop in FAS was observed. In all cases tested, E.
coli ACP was several times more reactive than recombinant ACP-I. A time course analysis of the reaction (Fig. 4, inset) further demonstrated that recombinant ACP was a very poor substrate for E. coli FAS in vitro.
One of the first committed reactions in both plant and bacterial FAS is the thioester transfer of the malonyl group from coenzyme A to ACP (2). To examine the effect of ACP structure on an individual reaction of E. coli FAS, the ACP preparations were tested for activity with malonyl-CoA:ACP transacylase. With increasing molar ratios of ACP/['"C]malonyl-CoA the malonyl-CoA:ACP transacylase reaction rate increases up to a substrate ratio of 3 (Fig. 5). Further increases in substrate ratio result in an immediate decline in reactivity. Recombinant ACP-I always showed half the activity as cosubstrate in the E. coli malonyl-CoA:ACP transacylase assay, but the shape of the activity curve versus ACP/['4C]malonyl-CoA was identical to that of E. coli ACP. These data are consistent with the reaction kinetics for E. coli malonyl-CoA:ACP transacylase which has been studied in detail elsewhere (23, 24). It is apparent that recombinant ACP-I was only 50% efficient in the reaction. This result may be one of the factors causing very low levels of E. coli FAS in vitro when using the recombinant ACP-I.
Previously we found that two chromato~aphically distinct malonyl-CoA:ACP transacylase isozymes of Ricinus cornmunis showed no selectivity in ACP structure when reactions were conducted with E. coli ACP, spinach ACP-I, and spinach ACP-I1 (6,16). In this study, malonyl-CoA:ACP transacylase from 3. napus was equally reactive with both recombinant ACP-I and E. coli (Fig. 5, inset). These results verify the lack of plant malonyl-CoAACP transacylase specificity with ACP structure.
Plant FAS from spinach chloroplasts was previously shown to react about equally with E. coli ACP, ACP-I, and ACP-11. In this study, a 3. napus chloroplast FAS preparation was used to evaluate the activity of recombinant ACP-I. With an increasing molar ratio of ACP/(14C]malonyl-CoA the B. napus chloroplast FAS activity decreased for both E.

napus
coli ACP and recombinant ACP-I (Fig. 6). This relationship is identical with that found using the E. coli FAS system (Fig.  4). The major difference between E. coli FAS and B. napus FAS was that recombinant ACP-I and E. coli ACP cosubstrate reactivity did not differ in the plant in vitro system.
In 1967, Simoni et al. (7) isolated ACPs from spinach and avocado and tested their activity in FAS reactions of E. coli and spinach. They observed that E. coli ACP was 4-10-fold higher in activity with the E. coli enzymes and 2-3-fold higher with the spinach enzymes. A partial explanation of their data might be that the plant ACPs required much greater purification and may have lost activity during this process. Guerra et al. (6) examined spinach versus E. coli ACP with spinach FAS enzymes and found similar activity. In this paper we have re-examined the activity of plant and bacterial ACP using spinach ACP that was produced in E. coli and, therefore, underwent purification very similar to that of E. coli ACP. Our data in general support the observations of Simoni et al. (7) that plant ACPs are relatively inactive in bacterial FAS. In contrast to Simoni et al. (7) and in agreement with our previous observations (6) we find that E. coli and plant ACP have very similar activity when assayed with plant FAS enzymes.
The recombinant ACP-I was further characterized with E. coli acyl-ACP synthetase (Fig. 7). A standard assay for determination of ACP in plant and bacterial extracts utilizes this reaction (14). It is important to determine if recombinant ACP-I can be accurately quantified with the standard assay procedure.
Both E. coli ACP and recombinant ACP-I present at 2.2 gmol (44 p M ) caused the reaction to reach saturation within 5 rnin (Fig. 7 A ) . Increasing the amount of ACP beyond 2.2 pmol in the reaction mixture had no additional effect on enzyme activity. Recombinant ACP-I (2.2 pmol added) was 1.5 times more reactive in the acyl-ACP synthetase reaction than E. coli ACP (0.13 pmol of 16:O ACP/5 min/gl versus 0.09 pmol of 16:0-ACP/5 min/pl). Kuo and Ohlrogge (14) also observed slightly higher activity of spinach ACP-I in the synthetase reaction when compared to E. coli ACP. ACP-I structure is apparently more reactive for the acyl-ACP synthetase reaction than E. coli ACP when 160 is the acyl cosubstrate. Time course analyses revealed that the reaction was essentially complete after 5 min with recombinant ACP-I (Fig. 7, B and C). Acyl-ACP synthetase reactions generally reached saturation at lower ACP concentrations and in shorter time when recombinant ACP-I was used in the assay. Analysis of the reaction products by PAGE indicated that under conditions of limiting ACP and excess enzyme all recombinant ACP-I was converted from ACP-SH to acyl-ACP (data not shown). Thus, our data indicate that the E. coli acyl-ACP synthetase reaction can be used for determination of recombinant ACP-I (or authentic spinach ACP-I) as long as the reaction is carried to completion. We emphasize that while recombinant ACP-I was a very good substrate for the acyl-ACP synthetase (Fig. 6) it was a poor substrate for E. . These data suggest that E. coli fatty acid metabolism in uiuo may be selectively altered by recombinant ACP-I expression.
Physiological Response of E. coli to Expression of Plant ACP-E. coli strains harboring plasmid pPB104 and expressing spinach ACP-I grew at rates indistinguishable to nontransformed cells. Thus, expression of plant ACP in E. coli appears to have no major influence on cell physiology. We examined the fatty acid composition of E. coli cells expressing plant ACP. As shown in Table I11 there were no major differences in either total fatty acid/cell weight or in the proportions of individual fatty acids. These data may in part reflect our in uitro findings which indicated the low activity of recombinant ACP-I with E. coli FAS enzymes (Fig. 4).
It may be noted, however, that there was a relative trend for 16:l (9c) increase and 18:l (llc) decrease in pPB104transformed cells (e.g. 60 Whether this difference relates to ACP isoform effects on chain elongation from 16:l-ACP to 181-ACP awaits a more detailed analysis using fractionated enzyme preparations. E. coli strain SJ16 is a @-alanine auxotroph which allows convenient radiolabeling of ACP in the phosphopantetheine moiety (11). We used plasmid pPB104 to transform SJ16 and grew cultures in the presence of [3-3H]P-alanine. Fig. 8 shows the radiolabeled proteins as analyzed by sodium dodecyl sulfate-PAGE. We observed a major difference in the pattern of radioactive proteins in cells expressing recombinant ACP-I. During early log phase only a small proportion of E. coli ACP is acylated whereas recombinant ACP-I is approximately 50% acylated (Fig. 8, lanes 3 and 4 ) . This distribution of E. coli ACP forms agrees with results of Jackowski and Rock (11) who have determined that &12% of the ACP in log stage is in the acyl-ACP form. As shown in lane 4, in cells containing pPB104, two additional bands of approximately the same intensity are apparent. The upper band co-migrates with spinach ACP-I and the lower band with spinach acyl-ACP-I. Thus, in contrast to E. coli ACP, which is approximately 10% acylated, plant ACP expressed in E. coli is approximately 50% in the acyl form during early log phase. With isopropyl-@-Dthiogalactopyranoside induction both recombinant holo ACP-I and recombinant acyl ACP-I increase to the same relative level (compare lane 4 to lane 6). Simoni et al. (7) observed that the major products of incubations of E. coli FAS with plant ACP were hydroxy fatty acids and from this suggested that the plant hydroxyacyl-ACP derivatives were poor sub-strates for the E. coli dehydrase reaction. Extrapolation of their data to our in uiuo observations of an accumulation of spinach acyl-ACP in E. coli (Fig. 8) suggests that the acyl groups on spinach ACP may be hydroxy fatty acids. Although PAGE analysis of cell pellets immediately boiled in sample buffer revealed 50% acyl-ACP-I, we did not detect substantial acyl-ACP in our preparations after purification (Fig. 1). Thus, these acyl groups were evidently removed during storage of the cells or isolation of the protein.
Conclusions-Research on plant enzymes which utilize ACP has been hampered by the difficulty in obtaining sufficient quantities of plant ACP. Results presented in this study demonstrate a relatively facile method for isolation of much greater quantities of spinach ACP-I than can be obtained routinely from plant material. The yields are such that 5-10 mg of plant ACP can be obtained in several days starting from 2-3 liters of E. coli culture. We have shown that the plant ACP expressed in E. coli is fully in the holo-ACP form, is active with plant FAS enzymes, and is immunologically very closely related to authentic spinach ACP-I.
The in uitro characterization of recombinant ACP-I has revealed intriguing relationships between structure and function for plant and bacterial ACP. E. coli FAS and malonyl-CoA:ACP transacylase reactions work poorly with the plant ACP-I structure while the acyl-ACP synthetase appears to react more competently with this isoform. In contrast, plant FAS and malonyl-CoAACP transacylase reactions do not appear to discriminate between E. coli and ACP-I structures. These observations may partially reveal the evolutionary divergence of plant FAS from a remote prokaryotic ancestor. The structural form of E. coli ACP is recognized by both kingdoms (bacterial, plant) while the more "evolved" ACP-I has attained a more specific reactivity and has partially lost the structure necessary for ancestral FAS. An investigation of structure-function relationships for ACP isoform has been previously reported (6) from which it is clear that each set of reactions (bacterial, plant) must be carefully evaluated.
The effects of spinach ACP-I expression on in uiuo FAS in E. coli were in general agreement with the in uitro characterizations. No major changes in the amount or type of fatty acid synthesized were found, although acylation of ACP-I was prominent during recombinant ACP-I expression. There may be differences in 16-and 18-carbon acyl distributions in transformed JMlOl cells, but these trends must be investigated in greater detail.
In summary, our studies indicate that recombinant ACP-I is a fully functional clone of spinach ACP-I which can be expressed in E. coli in milligram/liter quantities. Recombinant ACP-I is immunologically cross-reactive with spinach ACP-I polyclonal antibodies, and amino acid sequence data confirms the homology between authentic and recombinant ACP-I. We also note that differences exist between ACP isoforms in E. coli in uitro FAS reactions and acyl-ACP synthetase. There are some indications that recombinant ACP-I and spinach ACP-I are slightly different (migration on polyacrylamide gels). These differences do not appear to affect the biological activity of recombinant ACP-I in the analyses reported here. This new source of plant ACP-I should prove to be an excellent molecular probe for plant fatty acid metabolism and may possibly serve as a model protein for in vivo structure-function relationships and molecular evolution.