There are three annotated C. acetobutylicium fabF homologues designated as CAC3573, CAC2008 and CAA0093 10. We will temporarily call these genes fabF1, fabF2 and fabF3, although our data indicate that only the first of these genes functions in fatty acid synthesis. To test the functions of these homologues, the three genes were inserted into the arabinose-inducible vector pBAD24. The resulting plasmids were then introduced into two E. coli fabB(Ts) fabF strains, CY244 and JWC275. At the non-permissive temperature these mutant strains lack both long chain 3-ketoacyl-ACP synthase activities and thus are unable to grow even when the medium is supplemented with the unsaturated fatty acid, oleate 1112. Derivatives of strains CY244 or JWC275 carrying pHW36 encoding fabF1 grew at 42°C in the presence of oleate whereas the strains carrying pHW37 and pHW38 (encoding fabF2 and fabF3, respectively) failed to grow (Fig. 2) (similar results were seen with plasmids of both low and high copy number vectors). Thus, only fabF1 complemented the E. coli fabF mutation showing that C. acetobutylicium FabF1, like E. coli FabF, is able to catalyze all of the elongation reactions required in the synthesis of saturated fatty acids. Furthermore, expression of FabF1 restored thermal control of fatty acid composition to a FabF null mutant strain (Table 1). An E. coli fabF strain in which C. acetobutylicium FabF1 was expressed from the lac promoter of a low copy number vector closely mimicked the changes in fatty acid composition seen in wild type E. coli strains upon changes in growth temperature 13. Expression of FabF1 restored cis-vaccenate synthesis at all temperatures, but was much more effective at 30°C than at 37°C or 42°C (Table 1). This effect seems likely to be due to the effects of temperature on FabF1 synthase activity since thermal regulation disappeared upon expression of FabF1 from a high copy number vector (Table 1) and the enzyme was thermolabile in vitro (see below). Apparently, at high growth temperatures low levels FabF1 elongation activity was overcome by high-level expression of the protein. We also found high levels of cis-vaccenate at the non-permissive temperature upon expression of fabF1 in an E. coli fabB fabF strain that carried the fabB gene of Haemophilus influenzae, a bacterium naturally defective in both cis-vaccenate synthesis and in regulation of fatty acid composition by temperature 14 (data not shown).

Although the presence of plasmid pHW36 (fabF1) allowed growth of the two E. coli fabB(Ts) fabF strains at the non-permissive temperature, growth of both strains required oleate. The lack of growth in the absence of oleate argued that either FabF1 lacked the ability to replace FabB or that FabF1 was unable to simultaneously perform the tasks of both FabB and FabF under these conditions. To decide between these alternatives we transformed pHW36 into strain K1060, a strain that carries an unconditional fabB allele, and into strain CY242 which carries the same fabB(Ts) allele as strain CY244. The complementation experiments showed that C. acetobutylicium fabF1 allowed strain K1060 to grow on RB medium lacking oleate at 37°C (Fig. 2). However, fabF1 failed to complement growth of the temperature sensitive fabB mutant strain, CY242 at 42°C (Fig. 2). If FabF1 possessed FabB activity at 37°C, unsaturated fatty acids should be synthesized. To test this hypothesis, we grew strain K1060 carrying pHW36 or pHW33 (both encoding fabF1) at different temperatures and the fatty acid compositions of these strains were determined by collision induced dissociation electrospray mass spectroscopy (CID ES-MS) (Table 2). The strains clearly synthesized unsaturated fatty acids when grown at all of the different temperatures. However, the level of unsaturated fatty acids synthesized was lower than that seen in K1060 carrying a plasmid (pCY9) that encoded E. coli fabB and the amount of cis-vaccenate decreased with increased growth temperature. Moreover, despite the differing copy numbers, the two plasmids that encoded C. acetobutylicium FabF1 gave similar levels of unsaturated fatty acids. These results provide an explanation for lack of complementation of the fabB(Ts) phenotype at 42°C by the fabF1-encoding plasmids. At 42°C the low activity of FabF1 did not allow enough unsaturated fatty acid synthesis to support growth. To test whether or not C. acetobutylicium FabF1 has FabB function at 42°C we assayed unsaturated fatty acid synthesis in strain CY242 carrying the fabF1 plasmid pHW36 (growth was supported by cyclopropane fatty acid supplementation) (Fig. 3). Under these conditions [¹⁴C] acetate labeling showed low levels of unsaturated fatty acids synthesis upon arabinose induction of FabF1 expression (Fig. 3). Therefore, FabF1 has the ability to replace FabB in E. coli unsaturated fatty acid synthesis but its expression allows growth only when the host FabF is present to perform the bulk of the chain elongation reactions.

The sole fabZ homologue in the C. acetobutylicium genome is located within a large cluster of putative fab genes 10. To test function of the encoded protein the fabZ gene was inserted into the arabinose-inducible pBAD24 expression vector to give plasmid pHW22. Since E. coli fabZ null strains are nonviable 1516, we first introduced pHW22 into strain DY330, a "recombineering" strain 17. We then expressed the C. acetobutylicium FabZ in this strain and used standard phage γ recombinase manipulations to delete the host fabZ gene. These manipulations gave strain HW7, which grew well in presence of arabinose but failed to grow in the presence of fucose, an anti-inducer of arabinose promoter expression (Fig. 4). The fatty acid composition of the complemented mutant strain grown in presence of arabinose was similar to that of the parental strain, DY330, indicating that C. acetobutylicium FabZ functionally replaced E. coli FabZ (Table 3). The lack of fabA and fabM homologues in C. acetobutylicium raised the possibility that the FabZ of this organism might function as both an isomerase and a dehydratase as does the E. faecalis FabZ-like protein, FabN 9. To test this possibility plasmid pHW22 was introduced into both the fabA(Ts) E. coli strain CY57 and the fabA null mutant strain MH121. Neither stain grew in the absence of unsaturated fatty acid supplementation (data not shown) indicating that C. acetobutylicium FabZ lacks isomerase function and thus was unable to functionally replace FabA. However, it remained possible that C. acetobutylicium FabZ catalyzed UFA synthesis, but that the levels of UFA produced were too low to support growth. This possibility was tested by [¹⁴C]-acetate labeling of the fatty acids synthesized by strain CY57 carrying pHW22 and analysis of the resulting radioactive fatty acids for traces of UFA (Fig. 5). No UFA synthesis was detected. Another possible explanation for the observed lack of UFA synthesis was that FabI, the enoyl-ACP reductase of E. coli, converted the intermediate trans-2-decenoyl-ACP to decanoyl-ACP before the putative isomerase activity of C. acetobutylicium FabZ could act. Thus, we repeated the labeling experiment in the presence of a low dose of triclosan, a specific E. coli FabI inhibitor 6, in order to give the putative isomerase a better opportunity to act on the trans-2-decenoyl-ACP intermediate. Again no synthesis of unsaturated fatty acid was observed (data not shown). These in vivo results argued strongly that that C. acetobutylicium FabZ was unable to isomerize trans-2-decenoyl-ACP.

To allow direct assay of C. acetobutylicium FabF1 and FabZ activities we expressed the proteins in E. coli to facilitate their purification. [³⁵S]Methionine labeling showed that strain BL21 (DE3) carrying plasmids encoding either C. acetobutylicium fabF1 or fabZ under control of a phage T7 promoter expressed proteins of the expected sizes (Fig. 6A). However, the expression level of the FabZ protein was so low that it was not detected upon staining the SDS gels (Fig. 6B). We attributed this poor expression to the fact that the C. acetobutylicium FabZ gene contains 24 codons that correspond to nonabundant (rare) tRNA species in E. coli. We therefore changed these codons to synonymous codons that correspond to abundant E. coli tRNA species thereby resulting in a modified gene we call fabZm. Plasmid pHW74m (which encoded the His-tagged fabZm under T7 promoter control) abundantly expressed a protein with an apparent mass of 17 kDa (Fig. 6B) in good agreement with the expected value for the His₆-tagged protein (17.5 kDa). The His₆-tagged FabZ protein was purified to essential homogeneity using nickel-chelate chromatography (Fig. 6B). We also purified the N-terminally His₆-tagged versions of C. acetobutylicium FabF1 and the E. coli fatty acid biosynthetic proteins FabD, FabG, FabA, FabZ, FabB and FabI plus the Vibrio harveyi AasS acyl-ACP synthetase 18 by nickel-chelate chromatography. AasS was used to synthesize the 3-hydroxydecanoyl-ACP substrate whereas the other enzymes were used to assemble a defined in vitro fatty acid synthesis system in which the activities of E. coli FabA and C. acetobutylicium FabZ or E. coli FabB and C. acetobutylicium FabF1 could be directly compared. In reactions containing FabA 3-hydroxydecanoyl-ACP was converted to a mixture of trans-2 and cis-3-decenoyl-ACPs as expected from prior work 1920. E. coli FabB is unable to elongate trans-2-decenoyl-ACP, but elongates the cis-3 species to 3-keto-cis-5-dodecenoyl-ACP in the presence of malonyl-ACP 20. This product is then reduced by FabG and dehydrated by FabA to form trans-2-cis-5-dodecadienoyl-ACP20. The trans-2-cis-5-dodecadienoyl-ACP product accumulates because the reaction mixtures lacked enoyl-ACP reductase which precluded further elongations 20. Using this system we showed (as has long been known) that FabA is a bifunctional enzyme that catalyzes both the dehydration of 3-hydroxydecanoyl-ACP and well as the reversible isomerization of trans-2-decenoyl-ACP to cis-3- decenoyl-ACP (Fig. 7). As expected E. coli FabZ converted 3-hydroxydecanoyl-ACP to trans-2-decenoyl-ACP. However, addition of E. coli FabB to this reaction failed to give the 12-carbon unsaturated elongation product seen with FabA (Fig. 7) in agreement with prior reports that E. coli FabZ acts solely as a dehydratase and that FabB is unable to elongate trans-2-decenoyl-ACP 20. If C. acetobutylicium FabZ was capable of the isomerization reaction, then upon addition of E. coli FabB the reaction would yield trans-2, cis-5-dodecadienoyl-ACP 20. However, the only product formed was trans-2-decenoyl-ACP, the product of E. coli FabZ (Fig. 7A). Hence, we conclude that C. acetobutylicium FabZ possesses only dehydratase activity and introduction of the cis double bond requires another enzyme that has yet to be discovered. In parallel experiments, we replaced E. coli FabB with C. acetobutylicium FabF1 in the E. coli FabA reaction mixture to test if C. acetobutylicium FabF1 could elongate cis-3-decenoyl-ACP (Fig. 7B). We found that addition of FabF1 gave a modest conversion of cis-3-decenoyl-ACP to trans-2-cis-5-dodecadienoyl-ACP and at 37°C the product yields were lower than those seen at 25°C and 30°C consistent with the low activity of FabF1 at high temperature seen in vivo (Fig 7B).

The E. coli strains and plasmids used in this study are listed in Additional file 1. Luria-Bertani medium was used as the rich medium for E. coli. The phenotypes of fab strains were assessed on rich broth (RB) medium 12. Oleate neutralized with KOH was added to RB medium at final concentration of 0.1% and solubilized by addition of Brij 58 detergent to final concentration of 0.1 to 0.2%. Antibiotics were used at the following concentrations (in mg/L) sodium ampicillin, 100; chloramphenicol, 30; kanamycin sulfate and rifampicin, 200. L-Arabinose and D-fucose were used at concentrations of 0.01%. Isopropyl-β-D-thiogalactoside (IPTG) was used at final concentration of 1 mM.

Restriction enzymes, T4 DNA ligase and Taq DNA polymerase were from Invitrogen or New England Biolabs unless indicated otherwise. All enzymatic reactions were carried out according to the manufacturer's specifications. Qiagen products were used to isolate plasmids, purify DNA fragments from agarose gels and purify PCR products. Plasmids were introduced into E. coli strains by CaCl₂-mediated transformation. C. acetobutylicium ATCC824 genomic DNA was extracted using the GNOME DNA kit (Bio 101). DNA sequencing and the synthesis of oligonucleotides were done at the University of Illinois Keck Genomics Center.

The C. acetobutylicium fabF homologues were amplified from genomic DNA using the primers fabF1, fabF2 and fabF3 (Additional file 1). The PCR products were cloned into vector pCR2.1TOPO to give plasmids pHW40 (fabF1), pHW41 (fabF2) and pHW42 (fabF3). Plasmids pHW40 and pHW42 were then digested with EcoRI, the appropriate fragments were isolated and these were ligated into pHSG576 28 digested with the same enzyme to give plasmids pHW33 and pHW35, respectively. The orientation of the C. acetobutylicium ORFs in these plasmids were such that the genes would be transcribed by the vector lac promoter. The HindIII-XhoI fragment of pHW41 was ligated into vector pSU20 29 digested with the same enzymes to give pHW43 which was then digested with HindIII plus SalI and the fabF2-containing fragment was inserted into the same sites of vector pHSG576 to give pHW34. Plasmids pHW16, pHW31 and pHW32 were constructed as follows. The upstream primers were primers12, 34 and 56 (Additional file 1) and the downstream primer was the M13 (-) forward primer. Plasmids pHW33, pHW34 and pHW35 were used as templates for PCR amplification. The products were cloned into vector pCR2.1 TOPO to yield pHW16, pHW31 and pHW32, respectively. The BspHI-PstI fragments of pHW16 and pHW32 were then ligated into NcoI and PstI sites of pBAD24 30 to give plasmids pHW36 and pHW38, respectively. Likewise, the BspHI-HindIII fragment of pHW31 was inserted into the NcoI and HindIII sites of pBAD24 to yield pHW37.

The fabZ homologue was amplified by PCR using C. acetobutylicium genomic DNA as template with primers Zprimer1 and Zprimer2 (Additional file 1). The PCR product was inserted into pCR2.1 TOPO vector to give pHW15. The BspLU11I-HindIII fragment of pHW15 was inserted into the sites of pBAD24 digested with NcoI and HindIII to give pHW22. The BspHI-EcoRI fragments of pHW15 and pHW16 was inserted into the NcoI and EcoRI sites of pET28b to give pHW39 and pHW28, respectively. By site-directed mutagenesis using the primers listed in Additional file 1, an NdeI site was introduced into pHW39 and pHW28. The NdeI-EcoRI fragment of this two new plasmids were inserted into the NdeI and EcoRI sites of pET28b to give pHW74 and pHW76. To increase FabZ expression, 24 codons that correspond to rare E. coli tRNA species were substituted with codons favored in E. coli by site-directed mutagenesis using the primers listed in Additional file 1 to give pHW74m. The NcoI-HindIII fragment of pHW74m was inserted into the NcoI and HindIII sites of pBAD24 to give pHW22m.

A linear DNA fragments carrying a kan cassette was amplified from pKD13 by PCR 931 using primers, HZ1 and HZ2 listed in Additional file 1. These primers were homologous at the 3' end for priming sequences in pKD13 and contained 45-nucleotide extensions at the 5' end homologous to the E. coli fabZ sequence. The 1.4 kb PCR product was purified, treated with DpnI, and then introduced into a pHW22-containing derivative of DY330 a strain lysogenic for a defective prophage that contains the recombination genes under control of temperature-sensitive cI-repressor 9. The transformed cells were spread on LB plates containing ampicillin, kanamycin and arabinose. The E. coli fabZ deletion strain, HW7, was verified by PCR using primers P1, P2 plus HZ1, and HZ2.

Cultures (5 ml) were grown aerobically at different temperatures in RB medium overnight. The cells were then harvested and the phospholipids extracted as described previously 14. The fatty acid compositions were analyzed by mass spectroscopy as described previously 914. For analysis of radioactive fatty acids, 100 μl of a culture grown overnight in LB medium was transferred into 5 ml of RB medium supplemented with 0.1% cis-9, 10-methylenehexadecanoic acid (a cyclopropane fatty acid) plus 0.01% L-arabinose. After incubation of these cultures for 1 h, 5 μCi of sodium [1-¹⁴C] acetate was added and the culture allowed continuing growth for 4 h. The phospholipids were then extracted as described above. The phospholipid acyl chains were converted to their methyl esters, which were separated by argentation thin-layer chromatography, and analyzed with autoradiography 12

To assay expression of the products of C. acetobutylicium fabF1 and fabZ, pHW28 and pHW39 were introduced into E. coli strain BL21 (DE3), which encodes T7 RNA polymerase under the control of the IPTG-inducible lacUV5 promoter. The products of the cloned gene were selectively labeled with [³⁵S]methionine as described 32. The proteins were separated on a sodium dodecyl sulfate-12% polyacrylamide gel (pH 8.8). The destained gels were dried, and the labeled proteins were visualized by autoradiography 32. The molecular mass markers (Bio-Rad, Richmond, Calif) were rabbit phosphorylase, bovine serum albumin, rabbit actin, bovine carbonic anhydrase, trypsin inhibitor and hen egg white lysozyme.

Plasmid pHW76 and pHW74m were introduced into strain BL21 (DE3), respectively, and the proteins were overexpressed and purified as described previously20. The enzymes were homogeneous as judged by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. The E. coli FabD, FabH, FabG, FabA, FabZ, FabB, FabI and Vibrio harveyi AasS proteins were purified by their hexahisitidine tags described previously 1820.

Fatty acid synthesis was reconstituted in vitro to assay FabF1 and FabZ activity using the purified enzymes that catalyze the fatty acid biosynthesis essentially. The assay utilized the AasS acyl-ACP synthetase from Vibrio harveyi 18 to generate 3-hydroxydecanoyl-ACP. The reaction mixtures to synthesize 3-hydroxydecanoyl-ACP contained 20 μM ACP, 10 mM ATP, 10 mM MgSO₄, 5 mM DTT, 0.1 M sodium phosphate buffer (pH 7.0), 100 μM 3-hydroxydecanoic acid (Sigma) and AasS (0.2 μg) in a final volume of 50 μl. The mixtures were incubated at 37°C for 1 h. To assay C. acetobutylicium FabF1, the following incubation 1 μg each of E. coli FabD, FabG and FabA, 100 μM NADPH, 100 μM NADH, 100 μM malonyl-CoA, and 1 μg of either E. coli FabB or C. acetobutylicium FabF1 was added. To assay C. acetobutylicium FabZ, the following incubation contained 1 μg each of E. coli FabD, FabG and FabB, 100 μM NADPH, 100 μM NADH, 100 μM malonyl-CoA, and 1 μg of E. coli FabA or C. acetobutylicium FabZ was added. The resulting mixture was incubated for an additional 1 h and the reaction products were analyzed by conformationally sensitive gel electrophoresis on 20% polyacrylamide gels containing 2.5 M urea 2024. The gel was stained with Coomassie Brilliant Blue R250.