Fusion of farnesyldiphosphate synthase and epi -aristolochene synthase, a sesquiterpene cyclase involved in capsidiol biosynthesis in Nicotiana tabacum Maria Brodelius 1 , Anneli Lundgren 1 , Per Mercke 2,† and Peter E. Brodelius 1 1 Department of Chemistry and Biomedical Sciences, University of Kalmar, Sweden; 2 Department of Plant Biochemistry, Lund University, Sweden A clone encoding farnesyl diphosphate synthase (FPPS) was obtained by PCR from a cDNA library made from young leaves of Artemisia annua. A cDNA clone encoding the tobacco epi-aristolochene synthase (eAS) was kindly supplied by J. Chappell (University of Kentucky, Lex- ington, KY, USA). Two fusions were constructed, i.e. FPPS/ eAS and eAS/FPPS. The stop codon of the N-terminal en- zyme was removed and replaced by a short peptide (Gly-Ser- Gly) to introduce a linker between the two ORFs. These two fusions and the two single cDNA clones were separately introduced into a bacterial expression vector (pET32). Escherichia coli was transformed with the expression vectors and enzymatically active soluble proteins were obtained after induction with isopropyl thio-b- D -thiogalactoside. The recombinant enzymes were purified using immobilized metal affinity chromatography on Co 2+ columns. The fusion enzymes produced epi-aristolochene from isopentenyl diphosphate through a coupled reaction. The K m values of FPPS and eAS for isopentenyl diphosphate and farnesyl diphosphate, respectively, were essentially the same for the single and fused enzymes. The bifunctional enzymes showed a more efficient conversion of isopentenyl diphosphate to epi-aristolochene than the corresponding amount of single enzymes. Keywords: bifunctional enzyme; epi-aristolochene synthase; farnesyl diphosphate synthase; gene fusion; recombinant expression. The enzymatic machinery of a living cell is very complex. Thousands of enzymes are present and the flow of metabolites has to be tightly regulated. Consequently, enzymes are localized to different organelles and within a specific organelle the enzymes are organized in different ways. They may be found as soluble, membrane-associated or membrane-integrated enzymes. In order to make meta- bolism more efficient, enzymes catalysing sequential reac- tions are often found in close proximity to each other. They may form aggregates, be immobilized close to each other by adsorption to cellular structures or they may be organized in bi- or multifunctional enzymes within one single polypep- tide chain. Such enzymes exhibit substrate channelling which is a process by which two or more sequential enzymes of a pathway interact to transfer a metabolite directly from one active site to the next without allowing free diffusion of the intermediate [1,2]. Channelling is believed to play an important role in metabolic regulation and cellular control of enzymatic activities. The three-dimensional structures of bifunctional enzymes indicate that channelling can be achieved in different manners. In tryptophan synthase from Salmonella typhimurium, a hydrophobic 25 A ˚ tunnel, which matches the dimensions of the intermediate indole, connects the two active sites [2]. In the bifunctional enzyme thymidine synthase/dihydrofolate reductase from the protozoan Leish- mania major, the dihydrofolate intermediate is channelled on the basis of electrostatic interactions at the protein surface [3]. In abietadiene synthase from grand fir, two distinct active sites within a structural domain catalyse two sequential, mechanistically different cyclizations to form the tricyclic perhydrophenanthrene-type structure of abietadi- ene from the universal diterpene precursor geranylgeranyl diphosphate [4]. The copalyl diphosphate intermediate diffuses between the two active sites in this monomeric enzyme. Artificial bi- or multi-functional enzymes may be obtained by fusion of two or more structural genes [5]. The translational 3¢ terminus of the first gene is deleted along with any prosequence at the 5¢ terminus of the second gene and the genes are ligated in-frame. A small linker sequence coding for a few amino acids is often introduced between the two structural genes. This linker separates the two proteins in space by a small distance allowing each of them to fold properly without constrains from the other protein molecule. Linkers of different length have been used but it has been shown that if a too long linker is used the proximity effect is abolished [6]. Direct fusion of enzymes Correspondence to P. E. Brodelius, Department of Chemistry and Biomedical Sciences, University of Kalmar, S-39182 Kalmar, Sweden. Fax: + 46 480 446262, Tel.: + 46 480 447358, E-mail: peter.brodelius@hik.se Abbreviations: ADS, amorpha-4,11-diene synthase; eAS, epi-aristo- lochene synthase; FPP, farnesyl diphosphate; FPPS, farnesyl diphos- phate synthase; GPP, geranyl diphosphate; IPP, isopentenyl diphosphate; IMAC, immobilized metal affinity chromatography; IPTG, isopropyl thio-b- D -thiogalactoside. Enzymes: farnesyl diphosphate synthase (EC 2.5.1.10); epi-aristolo- chene synthase (EC 4.1.99.7). Present address: Plant Research International, Business Unit Cell Cybernetics, PO Box16, 6700 AA Wageningen, the Netherlands. (Received 18 February 2002, revised 13 May 2002, accepted 13 June 2002) Eur. J. Biochem. 269, 3570–3577 (2002) Ó FEBS 2002 doi:10.1046/j.1432-1033.2002.03044.x without a linker may also result in an active bifunctional enzyme [7]. Fused genes may be expressed in a suitable host (e.g. Escherichia coli) and the recombinant bi- or multi- functional enzyme used to study the effects of fusion on enzyme kinetics and stability. In a number of studies, it has been shown that recombinant fused enzymes exhibit a higher catalytic efficiency than the correspond- ing mixture of single enzymes. Some examples are D -hydantoinase/N-carbamylase [7], b-galactosidase/galac- tokinase [8], citrate synthase/malate dehydrogenase [9], aminocyclopropane-carboxylic acid synthase/aminocyclo- propane-carboxylic acid oxidase [10], and trehalose-6- phosphate synthetase/trehalose-6-phosphate phosphatase [11]. However, it has recently been argued, based on kinetic studies, that these bifunctional enzymes do not exhibit substrate channelling [12,13]. The higher catalytic efficiency, observed for these bifunctional enzymes, is entirely due to proximity effects. The use of fused enzymes for metabolic engineering at a branching point of a biosynthetic pathway is of great potential. However, so far this approach has been used to a relatively limited extent. The lactose utilization [6] and the osmotolerance of E. coli [14] have been influenced by introduction of the bifunctional enzymes b-galactosidase/galactokinase and c-glutamyl kinase/ c-glutamyl phosphate reductase, respectively. The starch degrading bifunctional enzyme a-amylase/glucose isom- erase was expressed in potato tubers and upon heating (65 °C for 45 min) of the crushed fresh tubers, glucose and fructose was produced from the starch present in the tubers [15]. The a-amylase/glucose isomerase fusion is not active at ambient temperatures and therefore it did not have any adverse effects on plant development and metabolism. We are involved in studies on the biosynthesis of sesquiterpenoids in plants with the aim of improving by metabolic engineering the amount of sesquiterpenes pro- duced. The common precursor for sesquiterpenoids is farnesyl diphosphate (FPP), which is also a substrate for the biosynthesis of other terpenoid metabolites such as sterols (Fig. 1). Sesquiterpene and sterol biosynthesis occurs in the cytosol while the biosynthesis of mono- and diterpenes take place in plastids. From studies with cell cultures of tobacco, it is well established that the sesquiterpene synthase, epi-aristolo- chenesynthase(eAS), involved in biosynthesis of the phytoalexin capsidiol, is induced upon treatment of the culture with a fungal elicitor [16,17]. A coordinated reduction in activity of squalene synthase (SS), an enzyme catalysing the formation of squalene from two molecules of FPP, is observed. Obviously, a shift in flow of metabolites, i.e. FPP, is achieved by these changes of enzyme activities and the conversion of FPP to epi-aristolochene is a regulatory step in sesquiterpene biosynthesis. FPP is produced from isopentenyl diphosphate/dimethylallyl diphosphate by farnesyl diphosphate synthase (FPPS). Thus, fusion of the two enzymes, FPPS and eAS, would give a bifunctional enzyme that catalyses the conversion of the C 5 -substrate isopentenyl diphosphate (IPP) to the complex C 15 -product epi-aristolochene (Fig. 1). The expression of this bifunctional enzyme in plant cells may result in an increased metabolic flow into sesquit- erpene biosynthesis. This phenomenon may be even more pronounced as the enzymes are located on each side of an important branching point of terpene metabolism. Transformation of tobacco with a gene construct enco- ding the FPPS/eAS will lead to increased formation of epi-aristolochene and possibly the phytoalexin capsidiol with a simultaneous decrease in biosynthesis of sterols. We have constructed, expressed in E. coli and partially characterized fusions of FPPS and eAS as a step in our efforts to increase the yield of sesquiterpenes in plants by metabolic engineering. MATERIALS AND METHODS Reagents Restriction enzymes, [1- 14 C]IPP (55 mCiÆmmol )1 )and [1- 3 H]FPP (16 CiÆmmol )1 ) were from Amersham-Pharma- cia Biotech. Isopropyl thio-b- D -thiogalactopyranoside (IPTG), IPP, geranyl diphosphate (GPP) and FPP were from Sigma. The tobacco eAS cDNA clone (TEAS) was kindly provided by J. Chappell, University of Kentucky, Lexington. PCR cloning of FPPS from Artemisia annua A cDNA library, previously constructed from poly(A + ) RNA extracted from young leaves of A. annua,wasusedto amplify a fragment encoding FPPS by PCR [18]. Primers for the PCR reaction were designed according to a published sequence of FPPS from A. annua [19]. Primers P1 (forward) and P2 (reverse) contained an NcoIandaXhoI restriction site, respectively (Table 1). PCR was carried out in a total volume of 50 lLwiththe following reagents: 1 · cloned Pfu polymerase buffer (Stratagene), 0.2 m M dNTPs (Pharmacia), 20 pmol of each primer and 1 U Turbo Pfu polymerase (Stratagene). PCR cycling was: two cycles of 94 °C(0.5min),50°C(1.0min), 72 °C (1.5 min); 29 cycles of 94 °C (0.5 min), 56 °C(1 min), 72 °C (1.5 min); 72 °C (5 min). The amplified fragment was resolved on a 1% agarose gel and visualized by staining with ethidium bromide. The FPPS-wild type fragment was digested with NcoIand XhoI. The bacterial expression vector pET32c (Novagen) Fig. 1. Biosynthetic pathway from IPP to the sesquiterpene epi-aristo- lochene and squalene. The reaction carried out by the bifunctional enzymes described here is depicted within the shaded area. Ó FEBS 2002 Fusions of FPPS and epi-aristolochene synthase (Eur. J. Biochem. 269) 3571 was cleaved with the same enzymes and treated with alkaline phosphatase. Vector and the fragment were isolated from agarose gel bands using the JETQUICK gel extraction spin kit (Genomed). Ligation of the fragment in frame with a multifunctional tag (including a hexa-His) in the vector was carried out according to standard procedures using T4 DNA ligase (Boehringer). The plasmid obtained, pET32FPPS, was transformed into E. coli NovaBlue (Novagen). Colonies were analysed by PCR using primers P1 and P2 to confirm the presence of the FPPS gene. Cells were grown in Luria– Bertani medium containing 50 lgÆmL )1 ampicillin and the plasmid was purified using the JETQUICK plasmid puri- fication spin kit (Genomed) and used as template for PCR amplification as described below. PCR amplification of FPPS and e AS For fusions of FPPS and eAS a number of DNA fragments were prepared by PCR amplification using the same conditions as above. Two sets of primers (P6/P2 and P1/ P5) were used for amplification of FPPS using the pET32FPPS as template (Table 1). In a similar manner three sets of primers (P3/P4, P3/P7 and P8/P4) were used to amplify the eAS gene using the TEAS cDNA as template (Table 1). Construction of expression vectors for production of single and fused enzymes Wild type eAS was digested and cloned into pET32c as described for FPPS above. The resulting plasmid pET32eAS was transformed into E. coli NovaBlue. Ligation of FPPS and eAS was achieved by sequential cloning into the bacterial expression vector pET32c. Two fused enzymes, i.e. FPPS/eAS and eAS/FPPS, were con- structed. First, fragments FPPS-nsc and eAS-nsc were digested with NcoIandBamHI and separately cloned into pET32c as described above yielding the plasmids pET32FPPS-nsc and pET32eAS-nsc. These two plasmids were transformed into E. coli NovaBlue for production of the plasmids. The plasmids were purified using the JETQUICK plasmid purification spin kit. Subsequently the fragments FPPS-L and eAS-L were digested with BamHI and XhoI and cloned into the plasmid pET32eAS-nsc and pET32FPPS-nsc, respectively, as des- cribed above yielding plasmids pET32FPPS/eAS and pET32eAS/FPPS. These two plasmids were transformed into E. coli NovaBlue. DNA sequencing DNA sequencing of cloned PCR fragments was performed using a DNA BigDye TM Terminator Cycle Sequencing Kit (Perkin Elmer) for the labelling of the sequencing reactions. Analyses were then carried out on an ABI PRISM TM 310 Genetic Analyzer. Oligonucleotides (15-mers) were synthes- ized according to sequence information and were used as primers for sequencing. Expression of the recombinant proteins and preparation of bacterial extracts NovaBlue cells carrying the plasmids pET32FPPS, pET32eAS, pET32FPPS/eAS and pET32eAS/FPPS were grown overnight in Luria–Bertani medium containing ampicillin (50 lgÆmL )1 )at37°C. The plasmids were purified using the JETQUICK plasmid purification spin kit and transferred into E. coli strain BL21(DE3) pLysS. The BL21 cells were grown at 37 °Cin20mLLuria– Bertani medium containing ampicillin (50 lgÆmL )1 )toan D 660 of  0.6. IPTG was then added to the final concen- tration of 1 m M . The cells were harvested after 4 h of cultivation at 30 °C by centrifugation at 200 g for 10 min at room temperature and the pellet was resuspended in 2 mL extraction buffer (50 m M Tris/HCl pH 8.0, containing 15 m M MgCl 2 and 20% glycerol). The cells were disrupted by sonication (Braun-Sonic 2000 microprobe at maximum power for 3 · 20 s bursts with 0.5 min chilling period on ice between bursts). The extract was centrifuged at 10 000 g for 15 min at 4 °C. The supernatant was collected and analysed. Optimization of expression Optimization of expression was carried out for the pET32eAS/FPPS construct. The parameters investigated were IPTG concentration (0, 0.2, 0.4, 0.6, 0.8 and 1.0 m M ) used for induction, induction temperature (12, 22 and 30 °C) and time of harvest after induction (90, 135, 210, 295 and 330 min). BL21 cells carrying the pET32eAS/FPPS plasmid were grown at 37 °C in 20 mL Luria–Bertani medium containing ampicillin (50 lgÆmL )1 )toanD 660 of  0.6. Induction under various conditions was subsequently carried out. The cells were harvested by centrifugation at 200 g for 10 min at room temperature and the pellet was resuspended in 2 mL extraction buffer. The cells were disrupted by sonication. The extract was centrifuged at Table 1. Primers used in cloning. Restriction sites are underlined. Primer Sequence Template P1 5¢-TAGAT CCATGGGTAGTACCGATCTG-3¢ FPPS N-terminal P2 5¢-CTA CTCGAGCTACTTTTGCCTCTTGTA-3 FPPS C-terminal P3 5¢-TAGAG CCATGGCCTCAGCAGCAGTT¢-3¢ eAS N-terminal P4 5¢-CTACTCGAGTCAAATTTTGATGGAGTC-3¢ eAS C-terminal P5 5¢-TA GGATCCCTTTTGCCTCTTGTAAAT-3¢ FPPS C-terminal P6 5¢-AT GGATCCGGAATGAGTAGTACCGATCTG-3¢ FPPS N-terminal P7 5¢-TA GGATCCAATTTTGATGGAGTCCAC-3 eAS C-terminal P8 5¢-AT GGATCCGGAATGGCCTCAGCAGCAGTT-3¢ eAS N-terminal 3572 M. Brodelius et al. (Eur. J. Biochem. 269) Ó FEBS 2002 10 000 g for 15 min at 4 °C. The supernatant was collected and enzyme activities determined. Production and purification of recombinant proteins BL21 cells were grown and induced (0.4 m M IPTG) as described above in 600 mL Luria–Bertani medium contain- ing ampicillin. Proteins were extracted with 40 mL buffer. Purification of recombinant proteins was carried out in a single step using immobilized metal affinity chromato- graphy (IMAC). The supernatant was applied to a 5-mL HiTrap Chelating HP column (Amersham Pharmacia Biotech) loaded with Co 2+ . Nonbound proteins were removed by washing with buffer. Elution of adsorbed recombinant proteins was achieved with extraction buffer containing 500 m M imidazole. Fractions of 0.5 mL were collected and fractions showing enzyme activity were pooled and frozen in liquid nitrogen in aliquots (0.5 mL) and stored at )80 °C until used. Electrophoresis and Western blotting SDS/PAGE electrophoresis was carried out on 4–20% Tris/ glycine gels or on NuPAGE 4–12% Bis/Tris gels from Invitrogen according to instructions supplied by the manu- facturer. After electrophoresis, the gels were either stained with Coomassie blue or the proteins transferred to nitro- cellulose membranes. Blotted proteins were detected with the Stag Alkaline Phosphatase Western blot kit (Novagen) according to the instructions supplied by the manufacturer. Enzyme assays FPPS. An aliquot of bacterial extract (10–46 lL) or purified enzyme (2–10 lL)wasassayedin50m M Tris/ HCl pH 8.0, containing 15 m M MgCl 2 ,10m M 2-mercap- toethanol, 20% glycerol, 55 l M GPP and 50 l M [1- 14 C]IPP (1.8 lCiÆmmol )1 ) in a total volume of 50 lL. The samples were incubated for 10 min at 30 °C and subse- quently an aliquot of 6 M HCl (5 lL)wasaddedtostopthe reaction and incubation was continued for 30 min to hydrolyse FPP formed to extractable farnesol (the substrate IPP is stable under these conditions). Neutralization was performed with 6 M NaOH (7.5 lL). The mixture was first extracted with hexane (400 lL) and subsequently the hexane phase (350 lL) was removed and extracted with water (300 lL). An aliquot (200 lL) of the hexane phase was taken into a scintillation vial and measured in a scintillation counter. eAS. An aliquot of bacterial extract (10–48 lL) or purified enzyme (2–10 lL) was assayed in 50 m M Tris/HCl pH 8.0, containing 15 m M MgCl 2 ,10m M 2-mercaptoethanol, 20% glycerol and 20 l M [1- 3 H]FPP (0.1 CiÆmmol )1 )inatotal volume of 50 lL. After a 10-min incubation at 30 °C, the reactions were stopped by addition of an equal volume of 0.2 M KOH containing 0.1 M EDTA. Subsequently, the reaction mixture was extracted with hexane (2 · 0.4 mL) and the hexane extracts were passed over a small column filled with 200–250 mg silica (Merck; size: 0.2–0.5 lm) and the column was rinsed with additional hexane (1.0 mL). The hexane extract was examined for radioactivity by scintilla- tion counting. Coupled enzyme assay. The coupled enzyme reaction was analysed for the bifunctional enzymes. An aliquot of bacterial extract (10–46 lL) or purified enzyme (2–10 lL) wasassayedin50m M Tris/HCl pH 8.0, containing 15 m M MgCl 2 ,10m M 2-mercaptoethanol, 20% glycerol, 55 l M GPP and 50 l M [1- 14 C]IPP (1.8 lCiÆmmol )1 ) in a total volume of 50 lL. The samples were incubated for 10 min at 30 °C. Two sets of samples were made. In one set, the reaction was stopped with 0.2 M KOH and the final product was analysed according to the eAS assay description. To the other set, HCl was added and the evaluation of the FPPS activity was performed as described for the FPPS assay. Protein concentrations Protein concentrations of extracts and partly purified recombinant enzymes were determined according to Bradford [20] using BSA as standard. RESULTS AND DISCUSSION Expression of recombinant enzymes For production of single and fused enzymes the bacterial expression vector pET32c was used. This expression vector was selected because the target protein is expressed as a fusion with a tag containing the thioredoxin for increased protein solubility [21], a histidine tag sequence facilitating protein purification and a Stag sequence for sensitive protein quantification and detection. The sequence also contains an enterokinase cleavage sites for removal of the fusion tag. The four plasmids pET32FPPS, pET32eAS, pET32eAS/FPPS and pET32FPPS/eAS were isolated and transferred into the E. coli BL21(DE3)pLysS for production of the recombinant enzymes. The transformed bacteria were grown and induced with IPTG (1 m M ), and cell-free extracts were prepared and evaluated for FPPS and/or eAS activity. All extracts exhibited the expected activities. Recombinant proteins of the predicted molecu- larmassweredetectedinWesternblotsusingStag alkaline phosphatase staining. The recombinant bifunc- tional enzymes show both activities and obviously the linker (Gly-Ser-Gly) is sufficiently long to permit the two enzymes to fold properly. The three-dimensional structure of eAS has been reported [22] but so far no structure for a plant FPPS has been published. However, the three- dimensional structure of chicken FPPS, which shows high amino acid identity (46.6%) and similarity (67.2%) to the Artemisia enzyme, has been reported [23]. Assuming a conserved structure between the two FPPSs, these two structures may be used to model the fusion enzyme. It is evident from such models that the linker used is sufficiently long to permit proper folding of the fused enzymes. Linkers of different length have been used in constructs of bifunctional enzymes [6,24,25]. However, long linkers may be exposed and sensitive to proteolytic attack and a too long distance between the two active sites may lead to reduced or no channelling of substrate as has been reported for the bifunctional b-galactosidase/galacto- kinase [6]. The Gly-Ser-Gly linker is relatively short and is convenient to use as the corresponding nucleotide Ó FEBS 2002 Fusions of FPPS and epi-aristolochene synthase (Eur. J. Biochem. 269) 3573 sequence contains a BamHI site, which may be used for the fusion of two genes. The Gly-Ser-Gly linker was used in a functional fusion of citrate synthase and malate dehydrog- enase [9]. Optimization of expression was carried out for one of the constructs, i.e. pET32 eAS/FPPS. The amount of eAS activity was determined as a function of incubation time after addition of IPTG, concentration of IPTG and induction temperature. The FPPS activity showed the same pattern. Based on these results the following conditions were selected for large-scale induction of all four recombinant proteins: incubation time, 4 h; IPTG concentration, 0.4 m M ; induction temperature, 30 °C. These conditions are consistent with the conditions used for the production of a number of other recombinant proteins in our laboratory. It is interesting to note that the highest activity of the relatively large recombinant protein (119.9 kDa) is observed at 30 °C. We assume this to be due to the fact that the fused enzyme is expressed as a fusion with the thioredoxin protein, which is improving the solubility of the expressed protein [21]. In fact, no recombinant protein could be detected in the insoluble fraction by SDS/PAGE. For expression in E. coli of eAS without any tag a significant part of the recombinant protein was found in inclusion bodies [26]. However, a twofold increase in recombinant eAS activity was obtained when the induction temperature was lowered from 37 to 27 °C. Similarly, an increased amount of recombinant epi-cedrol synthase was obtained by lowering the induction temperature to 20 °C [18]. The high activity of the bifunctional enzyme in extracts from cells grown at 30 °C is reflected in a higher total protein content. The specific activity in extracts obtained at 30 °C is lower than for extracts prepared from cells induced at lower temperatures. However, for large-scale production of the recombinant proteins an induction temperature of 30 °C was used. Production and purification of recombinant proteins The four recombinant proteins were produced on a large scale (2 · 600 mL cultures) using the conditions established above. The proteins carrying a (His) 6 -tag were purified in a one-step procedure by chromatography on IMAC-columns. This convenient procedure is widely used to purify recom- binant proteins [27]. Columns charged with three different ions, i.e. Ni 2+ ,Zn 2+ and Co 2+ , were tested. Some unspecific binding of E. coli proteins appeared to occur on all three ions, which may be due to the presence of metal binding sites in some proteins. In fact, on SDS/PAGE the same contaminating proteins appeared to be present in all four purified proteins. The least unspecific binding was observed for columns charged with Co 2+ , which were used for the large-scale purifications. Analysis by SDS/PAGE of proteinpurifiedonaCo 2+ column showed that they were at least 95% pure (Fig. 2). No attempts were made to remove the impurities by other purification steps. During prolonged incubations with enterokinase at 30 °C for removal of the affinity tag, a significant loss of enzyme activity was observed for the recombinant proteins. There- fore, the characterization of the recombinant proteins was carried out on enzymes containing the thioredoxin-Stag- His-tag as an N-terminal fusion. Kinetic properties of recombinant enzymes The IMAC-purified enzymes were used to determine K m values according to standard techniques. The K m for IPP was determined to be 3.3 for recombinant Artemisia FPPS. No significant difference in K m values was observed for the fused and single enzymes. The K m values for IPP was calculated to be 3.8 and 4.0 l M for the two fusion enzymes eAS/FPPS and FPPS/eAS, respectively. These K m values are similar to those reported for FPPS from other sources [28–31]. The K m value for FPP with the recombinant tobacco eAS was estimated to be 1.7 l M , which is the same as the 2–5 l M reported for the purified wild-type tobacco eAS [32]. The K m values for FPP were calculated to be 1.6 and 2.6 l M for the two fusion enzymes eAS/FPPS and FPPS/eAS, respectively. The K m values for FPP of other wild-type and recombinant plant sesquiterpene synthases have been reported to be in the range 0.5–7 l M [18,33–37]. In conclusion, essentially the same K m values were obtained for the single FPPS and eAS as for the two fusion enzymes FPPS/eAS and eAS/FPPS and these K m were similar to those reported previously for FPPS and sesquiterpene synthases from other sources. Apparently, the fusion of the two enzymes does not affect the affinity for the substrates. Folding of recombinant single and bifunc- tional enzymes is appropriate. Furthermore, these results indicate that an N terminal tag does not affect the catalytic properties of the recombinant enzymes. Coupled activity The fused enzymes FPPS/eAS and eAS/FPPS convert IPP to epi-aristolochene via FPP (Fig. 1). With increasing recombinant enzyme amount an increased formation of epi-aristolochene from IPP is obtained (Fig. 3). It is evident from Fig. 3 that the amount of enzyme used in an assay must be carefully adjusted for linearity of the assay and that the incubation time should not be too long under the conditions used. Fig. 2. SDS/PAGE of recombinant enzymes produced in E. coli. Lane 1, molecular mass standards; lane 2, crude extract FPPS/eAS (14.0 lg protein); lane 3, purified FPPS/eAS (5.1 lg);lane4,crudeextracteAS/ FPPS (11.2 lg); lane 5, purified eAS/FPPS (4.4 lg); lane 6, crude extract FPPS (13.4 lg);lane7,purifiedFPPS(3.3lg); lane 8, crude extract eAS (13.0 lg); lane 9, purified eAS (4.2 lg). The calculated molecular weights of FPPS, eAS and the fusion enzymes are 57, 80 and 120 kDa, respectively. 3574 M. Brodelius et al. (Eur. J. Biochem. 269) Ó FEBS 2002 It is interesting to note that the relative amount of epi- aristolochene formed by the bifunctional enzymes increases at higher enzyme activities, i.e. a larger portion of the FPP produced by the FPPS part of the enzyme is converted to final product by eAS (Fig. 4). This may be expected as the building-up of the intermediate FPP is more rapid at higher amounts of FPPS and the subsequent eAS experiences a higher substrate concentration, i.e. a steady-state condition is approached. No difference can be observed between the two bifunctional enzymes. Obviously, the order in which the two enzymes are fused does not influence the activity of the enzymes. This is also reflected in the K m values determined for the two bifunctional enzymes. Finally, to further evaluate the performance of the bifunctional enzymes, the level of epi-aristolochene pro- duced from IPP by the eAS/FPPS was compared with that produced by the corresponding amounts of the two single enzymes. As shown in Fig. 5 the amount of FPP produced is essentially the same for the two systems. However, the amount of epi-aristolochene produced is considerably higher for the fusion enzyme than for the mixture of single enzymes. Apparently, a proximity effect or substrate channelling operates in the fusion enzyme and increases the overall catalytic activity of the reaction. The FPP produced by the first enzyme is transferred to the active site of eAS with limited diffusion into the surrounding solution. Similar substrate channelling has been observed for a number of artificial bifunctional enzymes [7–11]. CONCLUDING REMARKS The fused enzymes described above are fully active when expressed in E. coli. Next these gene constructs will be transferred to a plant transformation vector. Transgenic tobacco plants producing the bifunctional enzymes will be established and the effects on sesquiterpene and sterol biosynthesis investigated. We are involved in studies on the Fig. 3. Time course for formation of FPP and epi -aristolochene from IPP and GPP as function of the amount of purified recombinant bifunctional enzymes. (A) FPPS/eAS. (B) eAS/FPPS. Open symbols, FPP; solid symbols, epi-aristolochene. s, d,2lL purified enzyme; n, m,4lL purified enzyme; h, j,6lL purified enzyme. Each point is the mean of two determinations. The protein concentrations of the purified enzyme preparations were 1.9 and 3.0 mg proteinÆmL )1 for FPPS/eAS and eAS/FPPS, respectively. Fig. 4. Amount of epi-aristolochene formed as function of FPP produced by the purified bifunctional enzymes FPPS/eAS and eAS/FPPS using IPP and GPP as substrates. Each point corresponds to a separate enzymatic assay containing either different amount of enzyme or being incubation for different times. d, FPPS/eAS; j, eAS/FPPS. Fig. 5. Time course for formation of FPP and epi-aristolochene from IPP and GPP by the two purified recombinant single enzymes (open symbols) or the purified recombinant bifunctional eAS/FPPS (solid symbols) as a function of incubation time. The activities of the single enzymes in the assay were carefully adjusted to the corresponding activities of the bifunctional enzyme. h, j, FPP; s, d, epi-aristo- lochene. Each point is the mean of two determinations. Ó FEBS 2002 Fusions of FPPS and epi-aristolochene synthase (Eur. J. Biochem. 269) 3575 biosynthesis of the antimalarial sesquiterpene artemisinin in A. annua. The sesquiterpene cyclase, amorpha-4,11-diene synthase (ADS), converting FPP to the first intermediate of artemisinin biosynthesis was recently cloned in our labor- atory [33]. We will make fusions of FPPS and ADS and introduce the bifunctional enzyme into plants of A. annua. We expect to obtain an increased biosynthesis of artemisinin in transgenic plants of A. annua expressing the FPPS/ADS fusion. ACKNOWLEDGEMENTS The financial support to P.E.B. from the Swedish Research Council for Engineering Sciences and the Swedish Council for Forestry and Agricultural Research. During a part of this work M.B. received a Marie Curie Scholarship from the European Union. We thank professor J. Chappell for the kind gift of the TEAS cDNA clone. REFERENCES 1. Meek, T.D., Garvey, E.P. & Santi, D.V. (1985) Purification and characterization of the bifunctional thymidylate synthetas-dihy- drofolate reductase from methotrexate-resistent Leishmania tro- pica. Biochemistry 24, 678–686. 2. Hyde, C.C., Ahmed, S.A., Padlan, E.A., Miles, E.W. & Davies, D.R. (1988) Three-dimensional structure of the tryptophan syn- thase alpha 2 beta 2 multienzyme complex from Salmonella typhimurium. J. Biol. Chem. 263, 17857–17871. 3. Knighton, D.R., Kan, C C., Howland, E., Janson, C.A., Hostomska, Z., Welsh, K.M. & Matthews, D.A. (1994) Structure of and kinetic channelling in bifunctional dihydrofolate reductase- thymidylate synthase. Nat. Struct. Biol. 1, 186–194. 4. Peters, R.J., Ravn, M.M., Coates, R.M. & Croteau, R.B. (2001) Bifunctional abietadiene synthase: Free diffusive transfer of the (+)-copalyl diphosphate intermediate between two distinct active sites. J. Am. Chem. Soc. 123, 8974–8978. 5. Bu ¨ low, L. & Mosbach, K. (1991) Multienzyme systems obtained by gene fusion. Trends Biotechnol. 9, 226–231. 6. Carlsson, H., Ljungcrantz, P., Bu ¨ low, L. & Mosbach, K. (1992) Engineering of lactose metabolism in E. coli by introducing b-galactosidase/galactokinase fusion enzymes. Biotechnol. Lett. 14, 439–444. 7. Kim, G J., Lee, D E. & Kim, H S. (2000) Construction and evaluation of a novel bifunctional N-carbamylase- D -hydantoinase fusion enzyme. Appl. Environ. Microbiol. 66, 2133–2138. 8. Bu ¨ low, L. (1987) Characterization of an artificial bifunctional enzyme, b-galactosidase/galactokinase, prepared by gene fusion. Eur. J. Biochem. 163, 443–448. 9. Lindbladh, C., Rault, M., Hagglund, C., Small, W.C., Mosbach, K., Bu ¨ low, L., Evans, C. & Srere, P.A. (1994) Preparation and kinetic characterization of a fusion protein of yeast mitochondrial citrate synthase and malate dehydrogenase. Biochemistry 33, 11692–11698. 10. Li, N., Jiang, X.N., Cai, G.P. & Yang, S.F. (1996) A novel bifunctional fusion enzyme catalyzing ethylene synthesis via 1-aminocyclopropane-1-carboxylic acid. J. Biol. Chem. 271, 25738–25741. 11. Seo, H.S., Koo, Y.J., Lim, J.Y., Song, J.T., Kim, C.H., Kim, J.K., Lee, J.S. & Choi, Y.D. (2000) Characterization of a bifunctional enzyme fusion of trehalose-6-phosphate synthase and trehalose- 6-phosphate phosphatase of Escherichis coli. Appl. Environ. Microbiol. 66, 2484–2490. 12. Pettersson, H., Olsson, P., Bulow, L. & Pettersson, G. (2000) Kinetics of the coupled reaction catalysed by a fusion protein of yeast mitochondrial malate dehydrogenase and citrate synthase. Eur. J. Biochem. 267, 5041–5046. 13. Pettersson, H. & Pettersson, G. (2001) Kinetics of the coupled reaction catalysed by a fusion protein of b-galactosidase and galactose dehydrogenase. Biochim. Biophys. Acta 1549, 155–160. 14. Meijer, P L., Lilius, G., Holmberg, N. & Bu ¨ low, L. (1996) An artificial bifunctional enzyme, c-glutamyl kinase/c-glutamyl phosphate reductase, improves NaCl tolerance when expressed in E. coli. Biotechnol. Lett. 18, 1133–1138. 15. Beaujean, A., Ducrocq-Assaf, C., Sangwan, R.S., Lilius, G., Bu ¨ low, L. & Sangwan-Norreel, B.S. (2000) Engineering direct fructose production in processed potato tubers by expressing a bifunctional alpha-amylase/glucose isomerase gene complex. Biotechnol. Bioeng. 70, 9–16. 16. Vo ¨ geli, U. & Chappell, J. (1988) Induction of sesquiterpene cyclase and suppression of squalene synthetase activities in plant cell cultures treated with fungal elicitor. Plant Physiol. 88, 1291–1296. 17. Threlfall, D.R. & Whitehead, I.M. (1988) Coordinated inhibition of squalene synthetase and induction of enzymes of sesquiterpe- noid phytoalexin biosynthesis in cultures of Nicotiana tabacum. Phytochemistry 27, 2567–2580. 18.Mercke,P.,Crock,J.,Croteau,R.&Brodelius,P.E.(1999) Cloning, expression and characterization of epi-cedrol synthase, a sesquiterpene cyclase from Artemisia annua L. Arch. Biochem. Biophys. 369, 213–222. 19. Matsushita, Y., Kang, W. & Charlwood, B.V. (1996) Cloning and analysis of a cDNA encoding farnesyl diphosphate synthase from Artemisia annua. Gene 172, 207–209. 20. Bradford, M.M. (1976) A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein dye binding. Anal. Biochem. 72, 248–254. 21. LaVallie, E.R., DiBlasio, E.A., Kovacic, S., Grant, K.L., Schen- del, P.F. & McCoy, J.M. (1993) A thioredoxin gene fusion expression system that circumvents inclusion body formation in the E. coli cytoplasm. Bio/Technology 11, 187–193. 22. Starks, C.M., Back, K., Chappell, J. & Noel, J.P. (1997) Structural basis for cyclic terpene biosynthesis by tobacco 5-epi-aristolochene synthase. Science 277, 1815–1820. 23. Tarshis,L.C.,Yan,M.,Poulter,C.D.&Sacchettini,J.C.(1994) Crystal structure of recombinant farnesyl diphosphate synthase at 2.6-A resolution. Biochemistry 33, 10871–10877. 24. Rock, F., Everett, M. & Klein, M. (1992) Overexpression and structure-function analysis of a bioengineered IL-2/IL-6 chimeric lymphokine. Protein Eng. 5, 583–591. 25. Carlsson, H., Ljung, S. & Bu ¨ low, L. (1996) Physical and kinetic effects on induction of various linker regions in beta-galactosidase/ galactose dehydrogenase fusion enzymes. Biochim. Biophys. Acta. 1293, 154–160. 26. Back, K., Yin, S. & Chappell, J. (1994) Expression of a plant sesquiterpene cyclase gene in Escherichia coli. Arch. Biochem. Biophys. 315, 527–532. 27. Nilsson, J., Sta ˚ hl, S., Lundeberg, J., Uhle ´ n, M. & Nygren, P A ˚ . (1997) Affinity fusion strategies for detection, purification, and immobilization of recombinant proteins. ProteinExpressionPurif. 11, 1–16. 28. Eberhardt, N.L. & Rilling, H.C. (1975) Prenyltransferase from Saccharomyces cerevisiae. Purification to homogeneity and mole- cular properties. J. Biol. Chem. 250, 863–866. 29. Reed, B.C. & Rilling, H.C. (1976) Substrate Binding of avian liver prenyltransferase. Biochemistry 15, 3739–3745. 30. Barnard, G.F. & Popjak, G. (1981) Human liver prenyltransferase and its characterization. Biochim. Biophys. Acta 661, 87–99. 31. Ohnuma, S., Hirooka, K., Ohto, C. & Nishino, T. (1997) Con- version from archaeal geranylgeranyl diphosphate synthase to farnesyl diphosphate synthase. J. Biol. Chem. 272, 5192–5198. 32. Vo ¨ geli, U., Freeman, J.W. & Chappell, J. (1990) Purification and characterization of an inducible sesquiterpene cyclase from elicitor-treated tobacco cell suspension cultures. Plant Physiol. 93, 182–187. 3576 M. Brodelius et al. (Eur. J. Biochem. 269) Ó FEBS 2002 33. Mercke, P., Bengtsson, M., Bouwmeester, H.J., Posthumus, M.A. & Brodelius, P.E. (2000) Cloning, expression and characterization of amorpha-4,11-diene synthase, a key enzyme of artemisinin biosynthesis in Artemisia annua L. Arch. Biochem. Biophys. 381, 173–180. 34. Zook, M.N., Chappell, J. & Kuc, J.A. (1992) Characterization of elicitor-induction of sesquiterpene cyclase activity in potato tuber tissue. Phytochemsitry 31, 3441–3445. 35. Crock,J.,Wildung,M.&Croteau,R.(1997)Isolationandbac- terial expression of a sesquiterpene synthase cDNA clone from peppermint (Mentha x piperita L.) that produce the aphid alarm pheromone (E)-b-farnesene. Proc. Natl Acad. Sci. USA 94, 12833–12838. 36. Steele,C.L.,Crock,J.,Bohlmann,J.&Croteau,R.(1998)Ses- quiterpene synthases from grand fir (Abies grandis). J. Biol. Chem. 273, 2078–2089. 37. Schmidt, C.O., Bouwmeester, H.J., Bu ¨ low, N. & Ko ¨ nig, W.A. (1999) Isolation, characterization and mechanistic studies of (–)-a-gurjunene synthase from Solidago canadensis. Arch. Bio- chem. Biophys. 364, 167–177. Ó FEBS 2002 Fusions of FPPS and epi-aristolochene synthase (Eur. J. Biochem. 269) 3577 . Fusion of farnesyldiphosphate synthase and epi -aristolochene synthase, a sesquiterpene cyclase involved in capsidiol biosynthesis in Nicotiana tabacum Maria. are D -hydantoinase/N-carbamylase [7], b-galactosidase/galac- tokinase [8], citrate synthase/ malate dehydrogenase [9], aminocyclopropane-carboxylic acid synthase/ aminocyclo- propane-carboxylic