Kinetic and biochemical analyses on the reaction mechanism of a bacterial ATP-citrate lyase Tadayoshi Kanao, Toshiaki Fukui, Haruyuki Atomi and Tadayuki Imanaka Department of Synthetic Chemistry and Biological Chemistry, Graduate School of Engineering, Kyoto University, Japan The prokaryotic ATP-citrate lyase is considered to be a key enzyme of the carbon dioxide-fixing reductive tricarboxylic acid (RTCA) cycle. Kinetic examination of the ATP-citrate lyase from the green sulfur bacterium Chlorobium limicola (Cl-ACL), an a 4 b 4 heteromeric enzyme, revealed that the enzyme displayed typical Michaelis-Menten kinetics toward ATP with an apparent K m value of 0.21 ± 0.04 m M . However, strong negative cooperativity was observed with respect to citrate binding, with a Hill coefficient (n H ) of 0.45. Although the dissociation constant of the first citrate mole- cule was 0.057 ± 0.008 m M , binding of the first citrate molecule to the enzyme drastically decreased the affinity of the enzyme for the second molecule by a factor of 23. ADP was a competitive inhibitor of ATP with a K i value of 0.037 ± 0.006 m M . Together with previous findings that the enzyme catalyzed the reaction only in the direction of citrate cleavage, these kinetic features indicated that Cl-ACL can regulate both the direction and carbon flux of the RTCA cycle in C. limicola. Furthermore, in order to gain insight on the reaction mechanism, we performed biochemical analyses of Cl-ACL. His273 of the a subunit was indicated to be the phosphorylated residue in the catalytic center, as both cat- alytic activity and phosphorylation of the enzyme by ATP were abolished in an H273A mutant enzyme. We found that phosphorylation of the subunit was reversible. Nucleotide preference for activity was in good accordance with the preference for phosphorylation of the enzyme. Although residues interacting with nucleotides in the succinyl-CoA synthetase from Escherichia coli were conserved in AclB, AclA alone could be phoshorylated with the same nucleotide specificity observed in the holoenzyme. However, AclB was necessary for enzyme activity and contributed to enhance phosphorylation and stabilization of AclA. Keywords: ATP-citrate lyase; reductive tricarboxylic acid cycle; Chlorobium limicola. ATP-citrate lyase (ACL) (EC 4.1.3.8) catalyzes one of the most complex enzyme reactions in which acetyl-CoA and oxaloacetate are produced from citrate and CoA with the hydrolysis of ATP to ADP and phosphate. The enzyme has received much attention in mammalian cells, as it is presumed to play a vital role in providing acetyl-CoA and oxaloacetate in the cytosol as starting materials for a variety of biosynthetic pathways. Rat and human ACLs from various organs and tissues have been extensively studied in terms of biochemical and genetic analyses [1–3], as well as transcriptional regulation [4] and post-translational phos- phorylation [5]. Subsequently, ACL has been investigated in many eukaryotic cells, including fungus [6], yeast [7], and plant cells [8]. It has been proposed that the eukaryotic ACL reaction consists of the following three steps: Enzyme þ Mg 2þ - ATP () Enzyme-PO 2À 3 þ Mg 2þ -ADP ð1Þ Enzyme-PO 2À 3 þ citrate () Enzyme-citryl-PO 2À 3 ð2Þ Enzyme-citryl-PO 2À 3 þ CoA-SH () oxaloacetate þ acetyl-CoA+Enzyme ð3Þ In contrast to the eukaryotic ACLs, little is known about the prokaryotic ACL. The prokaryotic ACLs have been identified only from autotrophic bacteria and archaea that utilize the reductive tricarboxylic acid (RTCA) cycle as a carbon dioxide (CO 2 ) assimilation pathway [9–11]. The prokaryotic ACLs have been purified and characterized from a few bacteria [12–14]. Studies with the purified proteins have raised the possibilities that ACL plays a key role in controlling the flux in the RTCA cycle. We have previously isolated the green sulfur bacterium Chlorobium limicola strain M1, and found that the strain carries out carbon dioxide fixation via the RTCA cycle. We have identified and performed initial characterization of ACL along with NADP-dependent isocitrate dehydroge- nase from this strain [15,16]. We were able to clone for the first time the prokaryotic ACL gene, and heterologous gene expression and characterization of the recombinant protein revealed that the enzyme (Cl-ACL), unlike its mammalian counterpart, was comprised of two distinct gene products, AclA (a subunit) and AclB (b subunit). By comparing the primary structures of AclA and AclB with that of the mammalian enzyme, we found that AclA and AclB Correspondence to T. Imanaka, Department of Synthetic Chemistry and Biological Chemistry, Graduate School of Engineering, Kyoto University, Yoshida-Honmachi, Sakyo-ku, Kyoto 606–8501, Japan. Fax: + 81 75 7534703, Tel.: + 81 75 7535568, E-mail: imanaka@sbchem.kyoto-u.ac.jp Abbreviations: RTCA cycle, reductive tricarboxylic acid cycle; ACL, ATP-citrate lyase; Cl-ACL, ATP-citrate lyase from Chlorobium limi- cola;AclA,a subunit of ATP-citrate lyase; AclB, b subunitofATP- citrate lyase. Enzyme: ATP-citrate lyase (EC 4.1.3.8). (Received 4 February 2002, revised 13 May 2002, accepted 23 May 2002) Eur. J. Biochem. 269, 3409–3416 (2002) Ó FEBS 2002 doi:10.1046/j.1432-1033.2002.03016.x corresponded to the C-terminal (33–39% identical) and N-terminal (27–34% identical) regions of the single peptide mammalian ACL, respectively. Cl-ACL did not catalyze the reverse reaction, the citrate synthase reaction, indicating that ACL could control the direction of carbon flux in the RTCA cycle in C. limicola. Furthermore, we found that Cl-ACL activity was inhibited under the presence of higher ADP/ATP ratios. This result suggests that the enzyme may also contribute in regulating the amount of carbon flux in the cycle depending on the levels of intracellular energy available from light. Here, we report a biochemical and kinetic examination of the bacterial heteromeric ACL from C. limicola,mainly focusing on the enzyme reaction mechanism. Interesting kinetic features were observed with the enzyme in terms of citrate binding, as well as inhibition by ADP. In addition, our results indicate the steps that govern the nucleotide dependency of the enzyme and the inhibition observed with ADP. MATERIALS AND METHODS Purification of the recombinant Cl -ACL Construction of the expression vector pETACL harboring the aclBA genes from C. limicola strain M1, and the expression procedure of the genes in E. coli BL21(DE3) have been previously described [15]. The recombinant enzyme was purified by using A ¨ KTA explorer 10S appar- atus (Amersham Pharmacia Biotech, Uppsala, Sweden) at 4 °C in all steps. The cell-free extract after ultracentrifuga- tion (110 000 g) was applied onto HiTrapQ HR5/5 anion exchange column (Amersham Pharmacia Biotech) equili- brated with 20 m M potassium phosphate buffer (KPB) (pH 7.4), and the enzyme was eluted with a linear gradient of KCl (0–0.5 M ). The active fraction was concentrated by ultrafiltration treatment and was applied onto TSKgel G4000SW (Tosoh, Tokyo, Japan) gel-filtration column equilibrated with 20 m M KPB containing 0.1 M KCl. The active fraction was then applied onto HiTrap Blue HR5/5 affinity column (Amersham Pharmacia Biotech) equilibrat- ed with 20 m M KPB (pH 7.4), and purified ACL was eluted with a linear gradient of KCl (0–1.0 M ). The homogeneity of active fractions after each step was confirmed by SDS/ PAGE analysis. In order to dissociate the AclAB complex, the purified enzyme was applied on Bio-Scale CHT5-I hydroxyapatite column (Bio-Rad Laboratories, Hercules, CA, USA) equil- ibrated with 10 m M KPB (pH 7.4). AclB was eluted with a linear gradient of KPB (10–100 m M ). After the active fraction (AclAB complex) was eluted with 100 m M KPB (pH 7.4), AclA was then eluted with 400 m M KPB (pH 7.4). The dissociation of the subunits was confirmed by SDS/ PAGE analysis. Assay of enzyme activity ACL activity was assayed by the coupled malate dehydrog- enase (MDH) method [17]. The reaction mixture contained 10 m M MgCl 2 ,10m M dithiothreitol or 2-mercaptoethanol, 0.2 m M NADH, 50 l M CoA, 2 m M sodium citrate, 1 m M ATP, 3.3 U MDH, and recombinant ACL solution in 0.1 M Tris/HCl buffer (pH 8.4) in a total volume of 1 mL. All measurements were performed at 30 °C.Oneunitofactivity wasdefinedas1lmol of NADH oxidized per min. Kinetic analyses were calculated with SigmaPlot (SPSS Science, Chicago, IL, USA). Site-directed mutagenesis The point mutation was introduced using QuikChange TM Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA, USA). The two complementary oligonucleotide primers were primer 1; 5¢-GGTATGAAGTTCGGCGCC GCCGG TGCCAAGGAAGG-3¢ and primer 2; 5¢-CCTTCCTTGG CACC GGCGGCGCCGAACTTCATACC-3¢. The codon for His273 (CAC) was replaced by an alanine codon (GCC; underlined). Experiments were carried out as advised by the manufacturer. Introduction of the point mutation, and absence of unintended mutations were confirmed by DNA sequencing. The expression and purification of the mutant enzyme was performed by the same procedure as that of the wild-type enzyme as described above. Phosphorylation of ACL Phosphorylation of Cl-ACL was measured as follows; Cl- ACL was incubated with 0.2 lCi [c- 32 P]ATP for 15 min at 30 °C. The 50-lL reaction mixture contained 50 m M Tris/ HCl buffer (pH 8.4), 5 m M MgCl 2 ,and2 m M dithiothreitol. The reaction was terminated by addition of 25 lLSDS/ PAGE loading buffer (3 · concentrated) into the mixture or by taking a 10-lL aliquot of the mixture and mixing it with 5 lL loading buffer. SDS/PAGE loading buffer contained 10% (v/v) 2-mercaptoethanol, 10% (v/v) gly- cerol, 5% (w/v) SDS, 60 l M bromophenol blue (BPB), and 0.1 M Tris/HCl buffer (pH 6.8). Additional procedures of these experiments are mentioned in each figure legend. The samples were applied on SDS/PAGE without heat treat- ment. After the electrophoresis, the gel was dried by RAPIDRY gel-dry system (ATTO, Tokyo, Japan) for 90 min at 80 °C and used for autoradiography. ATP and ADP were separated and detected by thin layer chromatography (TLC) with a previously described method [18]. In order to generate [a- 32 P]ADP, [a- 32 P]ATP was treated with D -glucose and hexokinase (Sigma, St Louis, MO, USA). [a- 32 P]ATP (1 lCi) was added into a reaction mixture containing 20 m M Tris/HCl buffer (pH 7.5), 10 m MD -glucose, 10 m M MgCl 2 , and 0.5 U of hexokinase. The reaction mixture was incubated at 30 °Cfor2hand then treated at 80 °C for 15 min in order to denature hexokinase. The purified ACL incubated with nonlabeled ATP at 30 °C for 15 min was mixed together with the [a- 32 P]ADP, and the reaction mixture was spotted directly onto Polygram CEL300PEI TLC plates (Macherey-Nagel, GmbH & Co., Duren, Germany). The substrate and product of the reaction were separated by one dimension chromatography using 1 M LiCl. RESULTS Purification of Cl -ACL from recombinant E. coli Cl-ACL consists of two distinct subunits a (AclA) and b (AclB), with molecular masses of 65 535 Da and 43 657 Da, respectively. The two subunits were supposed 3410 T. Kanao et al. (Eur. J. Biochem. 269) Ó FEBS 2002 to comprise an (ab) 4 structure, resembling the homotetra- meric quaternary structure of mammalian ACL. Recom- binant Cl-ACL was purified from E. coli cells harboring the aclBA genes from C. limicola strain M1. The purification procedure was slightly modified from a previous report [15], and mentioned in Materials and Methods. The homogen- eity of the recombinant protein was analyzed by SDS/ PAGE (see below). Kinetic analysis of Cl -ACL Kinetic examination of Cl-ACL with citrate was performed. The velocity plot and Lineweaver–Burk plot are shown in Fig. 1A,B, respectively. The 1/v value displayed a steep decrease at concentrations above 2 m M citrate (1/s < 0.5) (Fig. 1B). This phenomenon suggested negative coopera- tivity [19] with respect to citrate binding. We calculated the ratio of substrate concentrations that support V max /2 and V max /4, S 0.5 and S 0.25 , respectively. Indeed in the substrate- velocity plot, S 0.5 /S 0.25 ratio was 9.2, which was well above the expected ratio of 3 in the case of a nonallosteric enzyme (Fig. 1A). A Hill plot was constructed, and a slope with an extremely low value of 0.45 was observed at concentrations between 0.05 and 2.0 m M (Fig. 1C). At concentrations above 2 m M , the slope value was 1.0, indicating that no allosteric effects were present under these concentrations. For the calculation of our kinetic results, the velocity data for citrate were fitted by nonlinear regression analysis with the following equation [19]. m V max ¼ ð½S=K S þ 3½S 2 =aK 2 S þ 3½S 3 =a 2 bK 3 S þ½S 4 =a 3 b 2 cK 4 S Þ ð1 þ 4½S=K S þ 6½S 2 =aK 2 S þ 4½S 3 =a 2 bK 3 S þ½S 4 =a 3 b 2 cK 4 S Þ where v is the initial velocity of the reaction, V max is the maximum velocity, [S] is the concentration of citrate, K s is the dissociation constant of the enzyme-citrate (ES) com- plex, and a, b, and c are the interaction factors of the first, second and third substrate molecule(s) toward vacant substrate binding sites, respectively. In this case, the dissociation constant of the first citrate molecule K s1 can be represented by K s1 ¼ K s /4, while K s2 ¼ a2K s /3, K s3 ¼ ab3K s /2, and K s4 ¼ abc4K s [19]. Our data fitted very well as represented in Fig. 1A with an R 2 value of 0.999, and we obtained a V max value of 3.7 ± 0.1 U mg )1 .The dissociation constants were K s1 ¼ 0.057 ± 0.008 m M , K s2 ¼ 1.3 ± 0.4 m M , K s3 ¼ 18 ± 20 m M ,andK s4 ¼ 1.6 ± 2.0 m M . These facts clearly indicated that Cl-ACL was an allosteric enzyme displaying strong negative coop- erativity in citrate binding. Consequently, the S 0.5 was high for citrate, with a value of 2.5 m M . InthecaseofATP,Cl-ACL exhibited typical Michaelis- Menten kinetics with an apparent Michaelis constant (K m value) of 0.21 ± 0.04 m M (Fig. 2A). We previously found the activity of Cl-ACL was strongly inhibited with a higher ratio of ADP to ATP [15], with 50% inhibition observed at an equimolar ratio. Lineweaver-Burk plots with or without ADP against ATP are shown in Fig. 2B. The plots indicated that ADP was a competitive inhibitor of ATP. The inhibition constant, or K i value, for ADP was determined to be 0.037 ± 0.006 m M . Nucleotide dependency The effect of different nucleotides (ATP, GTP, CTP, UTP, and dATP) on ACL activity was re-examined by using a malate dehydrogenase (MDH)-linked assay, a much more sensitive and accurate assay than the hydroxamate method used in our previous report [15]. Each nucleotide was supplied at a concentration of 1 m M into the reaction mixture. The results are shown in Table 1. The enzyme exhibited maximum activity in the presence of ATP, and in the presence of dATP showed 40% of this activity. In the presence of CTP, activity was slightly observed (less than 0.2%), whereas no activity could be detected with GTP and UTP. This nucleotide dependency showed the same ten- dency as the previous result, although the relative activities with CTP and dATP were determined to be lower. A single mutation H273A in AclA In the human ACL, the His765 residue has been identified as the catalytic site which is autophosphorylated by ATP- Mg 2+ in the first step of the reaction [3]. It was also suggested that phosphohistidine was generated by ATP- Mg 2+ in the enzyme from the green sulfur bacterium C. tepidum [14]. Comparison of primary structure predicted Fig. 1. Kinetic analysis of recombinant Cl-ACL with citrate. (A) Velocity plot with various concentrations of citrate. The solid curve was fitted by nonlinear regression analysis of the experimental data as described in the text. The dotted line represents a calculated curve of activity if the enzyme were to follow typical Michaelis–Menten kinetics. The small panel is focused on lower citrate concentrations (0–4 m M ). (B) A Lineweaver–Burk plot of the kinetic data. (C) A Hill plot of the kinetic data. Hill coefficients are indicated as n H values. The n H value of 0.45 is given in the lower citrate concentrations (0.05–2 m M ). Ó FEBS 2002 ATP-citrate lyase from Chlorobium limicola (Eur. J. Biochem. 269) 3411 His273 on AclA as the phosphorylated catalytic site. A mutant gene was constructed in which His273 residue was replaced with alanine, and subsequently coexpressed in E. coli BL21(DE3) cells together with aclB. The H273A mutant protein was purified as a heteromeric enzyme with identical elution profiles to those of the wild type ACL in the purification steps (data not shown). The results indicate that the mutation did not affect the subunit assembly of the enzyme. However, no ACL activity was observed in the mutant protein, indicating that His273 played an essential role in the activity of the enzyme, most likely as the residue phosphorylated by ATP. Phosphorylation of Cl -ACL In order to investigate the reaction mechanism of prokary- otic ACL, we examined whether the enzyme was phos- phorylated with the c-phosphate group of ATP. When incubated with [c- 32 P]ATP, the 65 kDa subunit corres- ponding to AclA was phosphorylated, but the 40 kDa subunit (AclB) was not (Fig. 3A, lane 1). As substrates other than ATP were not added in the reaction mixture, this indicated that phosphorylation of the protein can occur prior to the interactions with other substrates such as citrate or CoA. Furthermore, the H273A mutant enzyme was not phosphorylated under these conditions (Fig. 3A, lane 2), indicating that His273 in AclA is phosphorylated by ATP, and that this is essential for enzyme activity. Moreover, we have found that the phosphorylated protein was dephos- phorylated in the presence of citrate (Fig. 3A, lane 4). Examination with thin layer chromatography also displayed the specific release of the phosphate group with the addition of citrate, suggesting the transfer of the labeled phosphate from the enzyme to citrate (data not shown). The results support an ordered mechanism of the enzyme reaction where the phosphate group of ATP is first covalently bound to the His273 catalytic residue, and subsequently transferred to the citrate molecule to form citryl phosphate. Enzyme activity of Cl-ACL displayed a high preference for ATP (100%), followed by dATP (40%) and CTP (0.2%). Taking into account the results of the above experiments, this preference can be assumed to reflect the phosphorylation efficiencies by various nucleotides. We have clarified this by examining the phosphorylation of the AclA with various [c- 32 P]NTPs. Results with c- 32 P-labeled ATP (0.2 lCi), GTP (5 lCi), CTP (5 lCi), and dATP (0.5 lCi) are shown in Fig. 4. Radioactive signals corres- ponding to phosphorylated AclA were observed in the reaction mixtures when ATP, CTP, and dATP were used as a phosphate donor. The intensities of the signals were in good accordance with the activity levels observed for each nucleotide. Efficient dephosphorylation of the phosphoryl- ated proteins was observed upon addition of citrate, regardless of the nucleotide that provided the phosphate group. Subunit dissociation of Cl -ACL We have previously described that the two subunits of Cl-ACL (AclA and AclB) could be dissociated from each other by hydroxyapatite column chromatography [15]. In order to clarify the role of each subunit, AclA and AclB were dissociated and subjected to further individual exam- ination. The homogeneity of the purified wild type and H273A mutant ACL were examined by SDS/PAGE (Fig. 5A, lanes 1 and 2). Efficient dissociation of the individual subunits of the wild type ACL was confirmed in lanes 3 and 4. The individual subunits of the H273A mutant Fig. 2. Kinetic analysis of recombinant Cl-ACL with ATP and its inhibition by ADP. (A) Lineweaver–Burk plots for various con- centrations of ATP. (B) Double reciprocal plots for various concentrations of ATP with or without ADP. ADP concentrations were 0m M (circles), 0.1 m M (squares), and 0.3 m M (triangles). Table 1. ACL activities for each nucleotide. NA, no activity. Nucleotides (1 m M ) ATP dATP CTP GTP UTP Activity (UÆmg )1 ) 1.30 0.52 Trace NA NA Activity (%) 100 40 < 0.2 0 0 Fig. 3. Phosphorylation and dephosphorylation of Cl-ACL and the H273A mutant protein. Samples were subjected to SDS/PAGE (12.5%) and autoradiography. Lane 1, purified wild type enzyme was incubated with [c- 32 P]ATP (0.2 lCi) for 15 min at 30 °C; lane 2, purified H273A mutant protein incubated with [c- 32 P]ATP (0.2 lCi) for 15 min at 30 °C; lane 3, same as in lane 1 before addition of 2 m M citrate; lane 4, enzyme after addition of 2 m M citrate and incubation for 15 min at 30 °C with the sample in lane 3. Molecular masses (kDa) are indicated on the left of the panel. 3412 T. Kanao et al. (Eur. J. Biochem. 269) Ó FEBS 2002 ACL were also observed in lanes 5 and 6. No activity could be detected in the fractions containing individual subunits (Fig. 5A, lane 3 and 4). Interestingly, up to 70% of activity was recovered when the individual subunit fractions (AclA and AclB) were mixed together and incubated at 25 °Cfor 5min. Phosphorylation of AclA with or without AclB The catalytic residue that is the target for phosphorylation (His273) is located on the AclA subunit, whereas a sequence comparison with bacterial succinyl-CoA synthetase dis- played that residues interacting with ATP were conserved in the AclB subunit (K50, R58, E97, E104, N190, and D205) [15]. The holoenzyme and individual subunits of both wild type and H273A mutant ACL were incubated with [c- 32 P]ATP. Although no ACL activity was detected in the AclA fraction without AclB, we found that in the case of the wild type enzyme, AclA alone was phosphorylated by [c- 32 P]ATP to a similar extent as the AclAB complex after 90 min (Fig. 5B, lane 2). We then presumed that AclB might be responsible for nucleotide specificity, and not for phosphorylation itself. However, further examinations proved otherwise. No significant difference was observed in the nucleotide specificity of AclA phosphorylation between the AclAB complex (holoenzyme) and the AclA subunit alone (Fig. 6). We then followed the phosphoryla- tion levels of the AclAB complex and AclA subunit at various time intervals. The phosphorylation by [c- 32 P]ATP was performed at 30 °C, and aliquot samples after desired periods were applied to SDS/PAGE. Although a prompt phosphorylation was observed in the holoenzyme (Fig. 7A), the AclA fraction was phosphorylated at a much slower rate (Fig. 7B). Furthermore, a labeled degradation product of the AclA subunit was observed in the absence of the AclB subunit (Figs 7B and 5B). However, the actual amount of the degradation product was very low, as it could not be detected in Coomassie Brilliant Blue (CBB) stained gels before (Fig. 5A, lanes 3 and 5) or after (data not shown) incubation with ATP. The relatively high intensity of the degradation product in the autoradiograph compared to the CBB stained gels indicated that a high percentage of the degradation product was phosphorylated. The phosphory- lation of the AclA protein may lead to a slight decrease in stability of the protein. As no signals were detected in the case of the H273A mutant protein (Fig. 5B, lanes 4 and 5), the phosphorylation of the AclA protein and its degradation product is most likely to occur specifically at the His273 Fig. 4. Phosphorylation of Cl-ACL with various [c- 32 P]NTPs. The nucleotides which were used in each reaction are indicated above the lanes. The enzyme was incubated with the indicated nucleotide at 30 °C for 15 min. Samples were subjected to SDS/PAGE (12.5%) and autoradiography. The mixtures before addition of 2 m M citrate are indicated with (–), while those after addition of 2 m M citrate are indicated with (+). Nucleotides were added at concentrations of 0.2 lCi for [c- 32 P]ATP, 5 lCi for [c- 32 P]GTP, 5 lCi for [c- 32 P]CTP, and 0.5 lCi for [c- 32 P]dATP. Molecular masses (kDa) are indicated on theleftofthepanel. Fig. 5. Dissociation of AclAB. (A) SDS/PAGE of wild type and H273A mutant enzymes and their individual subunits. Samples were applied to a 12.5% gel, and then stained with Coomassie Brilliant Blue R250. Purified wild type and H273A mutant enzymes were applied onto lanes 1 and 2, respectively. The subunits AclA and AclB of wild type and H273A Cl-ACL were dissociated by hydroxyapatite column chromatography. AclB eluted with 10–100 m M KPB (lanes 4 and 6), while AclA eluted with 400 m M KPB (lanes 3 and 5). The dissociated subunits are derived from the enzymes which are indicated above the lanes (WT; wild-type Cl-ACL, H273A; H273A mutant protein). Lane M, molecular marker. (B) Autoradiograph of wild type (lanes 1–3) and H273A mutant (lanes 4–6) enzymes and their individual subunits after incubation with 0.2 lCi of [c- 32 P]ATP for 90 min. Each reaction mixture contained 1 lg protein. Purified enzymes before subunit dis- sociation (AB) are shown in lanes 1 and 4, the individual AclA subunits (a) are shown in lanes 2 and 5, and AclB subunits (b) in lanes 3 and 6. Molecular masses (kDa) are indicated on the side of each panel. The asterisk indicates the degradation product of AclA described in the text. Fig. 6. Nucleotide specificities of AclA phosphorylation with or without AclB. Nucleotides used for each reaction are indicated above the lanes. Each reaction mixture contained 1 lg protein and was incubated with the indicated nucleotide at 30 °C for 15 min. Samples were subjected to SDS/PAGE (12.5%) and autoradiography. Nucleotides were added at concentrations of 0.2 lCi for [c- 32 P]ATP, 5 lCi for [c- 32 P]GTP, 5 lCi for [c- 32 P]CTP, and 0.5 lCi for [c- 32 P]dATP. Molecular masses (kDa) are indicated on the right of the panel. Samples using purified enzyme before subunit dissociation and those with the AclA subunit alone are indicated with ÔABÕ and ÔaÕ, respectively. The asterisk indi- cates the degradation product of AclA described in the text. Ó FEBS 2002 ATP-citrate lyase from Chlorobium limicola (Eur. J. Biochem. 269) 3413 residue. Our results suggest that the AclB subunit contri- butes in enhancing the efficiency of phosphorylation of the AclA subunit, as well as stabilizing its structure. Examination of the inhibitory effect of ADP As an increased ratio of ADP towards ATP significantly inhibits Cl-ACL activity, we investigated the effect of ADP on the phosphorylation of AclA. Addition of 10–100 l M ADP to the phosphorylated Cl-ACL resulted in dephos- phorylation of the enzyme (Fig. 8A, lanes 2–5). This tendency increased with higher concentrations of ADP (data not shown). An apparent inhibitory effect on AclA phosphorylation was observed with an increase in ADP concentration (Fig. 8A, lanes 6–10), indicating a competi- tion of the labeled phosphate group between enzyme and ADP. In order to elucidate the fate of the phosphate group after the enzyme is dephosphorylated by ADP, the follow- ing experiment was carried out. [a- 32 P]ADP was produced by treating [a- 32 P]ATP with hexokinase and glucose (Fig. 8B, lane 2). Purified enzyme was phosphorylated using unlabeled ATP. The phosphorylated enzyme and [a- 32 P]ADP were incubated together, and the reaction product was applied to TLC (Fig. 8B, lane 3). We clearly detected the generation of ATP in the lane. These results leave no doubt that the first step of the ACL reaction, phosphorylation by ATP, is reversible. DISCUSSION We have examined the kinetic and biochemical features of a prokaryotic ATP-citrate lyase. The results provide the first insight on the reaction mechanism of an ATP-citrate lyase with a heteromeric structure, and display some interesting features of the enzyme. The His273 residue, located on AclA, is phosphorylated by c-phosphate of ATP. This phosphorylation can take place in the absence of other substrates, and was found to be reversible. These results coincided with the previously reportedfeaturesofmammalianACL[20]andsuccinyl- CoA synthetase from E. coli (Ec-SCS) displaying similarit- ies to ACLs in terms of structure and catalytic mechanism [21,22]. In addition to these results, we demonstrated that the AclA subunit alone could be phosphorylated when incubated with ATP, and AclB markedly accelerated the phosphorylation rate of AclA. Similar results were also reported in the case of Ec-SCS [21], although the phos- phorylation of the a-subunit alone (80% of the subunit after 24 h) was much slower than AclA alone (90 min). A major contribution of AclB, as well as the b subunit of Ec-SCS, in the catalytic process seems to be assistance of the nucleotide binding to and phosphorylation of the a subunit. Another function of AclB was found to be stabilization of the enzyme, as AclB prevented the degradation of AclA that was otherwise observed in the absence of AclB. After the phosphorylation, we detected that the addition of citrate subsequently removed the phosphate from the enzyme, presumably forming citryl phosphate. This occurred in the absence of CoA. Taking into account the reaction mechanism of mammalian ACL [23], the final step of the reaction can be assumed to be the nucleophilic attack of CoA to the phosphorylated carbonyl carbon of citryl phosphate, and the cleavage of the resulting citryl-CoA to acetyl-CoA and oxaloacetate. In our previous report, we could not detect citrate synthase activity in Cl-ACL [15]. It has been described that the reaction of mammalian ACL was reversible, although it is much stronger in the cleavage direction [20]. Since the phosphorylation of ACLs was reversible, the unidirectional characteristics of the enzymes were likely to be due to the low efficiencies in the condensation between acetyl-CoA and oxaloacetate. We also revealed that the nucleotide specificity of the phosphorylation of AclA alone displayed the same tendency with the overall enzyme reaction. This finding suggests that the nucleotide is discriminated by AclA, and the phos- phorylation step governs the overall nucleotide specificity of the holoenzyme. It has been reported that a histidine residue in ACL from rat liver was autophosphorylated by GTP even though GTP did not support the overall reaction [24]. This was not the case in Cl-ACL, as GTP did not lead to phosphorylation of the enzyme, nor support activity. Fig. 7. Effect of AclB on the phosphorylation of AclA. AclA with (A) or without AclB (B) was incubated with 0.2 lCi of [c- 32 P]ATP at 30 °C. The incubation times (min) are indicated above each lane. Samples were subjected to SDS/PAGE (12.5%) and autoradiography. Equal amounts of AclA (0.11 pmol) were used in (A) (purified enzyme before subunit dissociation) and (B) (AclA alone). Molecular masses (kDa) are indicated on the left of the panel. The asterisk indicates the deg- radation product of AclA described in the text. Fig. 8. Effects of ADP on the phosphorylation of Cl-ACL by ATP. (A) After the following reactions, each sample was subjected to SDS/ PAGE (12.5%) and autoradiography. Purified Cl-ACL after incuba- tion with 0.2 lCi of [c- 32 P]ATP at 30 °C for 15 min (lane 1). Samples of lane 1 with addition of 100 l M (lane 2), 50 l M (lane 3), 20 l M (lane 4) and 10 l M (lane 5) ADP were further incubated at 30 °Cfor15 min. Purified Cl-ACL was incubated with 0.2 lCi of [c- 32 P]ATP in the presence of 100 l M (lane 6), 50 l M (lane 7), 20 l M (lane 8), 10 l M (lane 9) and absence (lane 10) of ADP at 30 °C for 15 min. (B) Autoradi- ograph of TLC. Lane 1, [a- 32 P]ATP; lane 2, [a- 32 P]ATP treated with hexokinase and glucose; lane 3; sample from lane 2 after incubation with phosphorylated Cl-ACL (AB-p) for 15 min at 30 °C. 3414 T. Kanao et al. (Eur. J. Biochem. 269) Ó FEBS 2002 Although the nucleotide preference in the phosphorylation of Ec-SCS has not been examined, a recent crystal structure of Ec-SCS revealed that the residues interacting with nucleotides were mostly located in the b-subunit [25]. In addition, ATP- and GTP-specific SCS isozymes from pigeon breast and liver are composed of the same asubunit but different b subunits, indicating the importance of the b subunit in the nucleotide preference [26]. As alignment of Cl-ACL with Ec-SCS identified the corresponding residues for the interaction to nucleotides in AclB, AclB may also participate in the binding and recognition of nucleotides. With respect to Cl-ACL, some nucleotide binding residues, sufficient for nucleotide discrimination, should at least be present in AclA. In our kinetic studies, the most striking feature of Cl-ACL was the strong negative cooperativity observed towards citrate binding. As K s1 <<K s2 <<K s3 , binding of the first and second molecules of citrate drastically decrease the affinity of the enzyme for the second and third molecules, respectively. The dissociation constant of the first citrate molecule (K s1 ) was 0.057 ± 0.008 m M , lower than the K m values of ACLs from other sources [14]. However, the K s2 value was 23-fold higher, with a value of 1.3 ± 0.4 m M ,and the K s3 value was even higher at 18 ± 20 m M . Owing to this strong negative cooperativity, it can be presumed that under physiological conditions, the majority of Cl-ACL binds to only one molecule of citrate. In the presence of sufficient levels of ATP, the low K s1 value would contribute to the efficient conversion of citrate at low concentrations. If citrate were to accumulate in the cells, the high K s2 and K s3 values due to strong negative cooperativity would limit the reaction rate in comparison with a nonallosteric enzyme. This feature of Cl-ACL would serve as a valve that limits the flux of the RTCA cycle in C. limicola. In the literature, we found that while the ACL from C. tepidum displayed typical Michaelis–Menten kinetics [14], the unphosphorylated human ACL showed similar negative cooperativity toward citrate [5]. The Hill coefficient was 0.65, indicating a weaker negative cooperativity than that of Cl-ACL. This negative cooperativity was abolished when the enzyme was phosphorylated either by cAMP- dependent protein kinase alone or in combination with glycogen synthase kinase-3 [27–30]. These kinases phos- phorylate the Thr446, Ser450, and Ser454 residues of human ACL. As corresponding residues are not present in Cl-ACL [15], it is likely that this sort of absolving mechanism does not exist in Cl-ACL. Another feature of Cl-ACL was the inhibition observed with ADP. It has been reported that ADP inhibited the activities of both mammalian and bacterial ACLs [12,13,31]. The double reciprocal plots with or without ADP showed that ADP was a competitive inhibitor of ATP with a K i value of 0.037 ± 0.006 m M . This would result in a decrease in Cl-ACL activity when intracellular energy is at an insufficient level. Together with the negative cooperativity and the unidirectional features of the enzyme, Cl-ACL is likely to regulate both the direction and carbon flux of the RTCA cycle in C. limicola. REFERENCES 1. Houston, B. & Nimmo, H.G. (1984) Purification and some kinetic properties of rat liver ATP citrate lyase. Biochem. J. 224, 437–443. 2. Elshourbagy, N.A., Near, J.C., Kmetz, P.J., Sathe, G.M., Southan, C., Strickler, J.E., Gross, M., Young, J.F., Wells, T.N.C. & Groot, P.H.E. (1990) Rat ATP citrate-lyase. Molecular cloning and sequence analysis of a full-length cDNA and mRNA abun- dance as a function of diet, organ, and age. J. Biol. Chem. 265, 1430–1435. 3. Elshourbagy, N.A., Near, J.C., Kmetz, P.J., Wells, T.N.C., Groot, P.H.E., Saxty, B.A., Hughes, S.A., Franklin, M. & Gloger, I.S. (1992) Cloning and expression of a human ATP-citrate lyase cDNA. Eur. J. Biochem. 204, 491–499. 4. Sato, R., Okamoto, A., Inoue, J., Miyamoto, W., Sakai, Y., Emoto, N., Shimano, H. & Maeda, M. (2000) Transcriptional regulation of the ATP citrate-lyase gene by sterol regulatory ele- ment-binding proteins. J. Biol. Chem. 275, 12497–12502. 5. Potapova, I.A., El-Maghrabi, M.R., Doronin, S.V. & Benjamin, W.B. (2000) Phosphorylation of recombinant human ATP: citrate lyase by cAMP- dependent protein kinase abolishes homotropic allosteric regulation of the enzyme by citrate and increases the enzyme activity. Allosteric activation of ATP: citrate lyase by phosphorylated sugars. Biochemistry 39, 1169–1179. 6. Pfitzner, A., Kubicek, C.P. & Ro ¨ hr, M. (1987) Presence and reg- ulation of ATP: citrate lyase from the citric acid producing fungus Aspergillus niger. Arch. Microbiol. 147, 88–91. 7. Shashi, K., Bachhawat, A.K. & Joseph, R. (1990) ATP: citrate lyase of Rhodotorula gracilis: purification and properties. Biochim. Biophys. Acta 1033, 23–30. 8. Rangasamy, D. & Ratledge, C. (2000) Compartmentation of ATP: citrate lyase in plants. Plant Physiol. 122, 1225–1230. 9. Shiba, H., Kawasumi, T., Igarashi, Y., Kodama, T. & Minoda, Y. (1985) The CO 2 assimilation via the reductive tricarboxylic acid cycle in an obligately autotrophic, aerobic hydrogen-oxidizing bacterium Hydrogenobacter thermophilus. Arch. Microbiol. 141, 198–203. 10. Schauder, R., Widdel, F. & Fuchs, G. (1987) Carbon assimilation pathways in sulfate-reducing bacteria II. Enzymes of a reductive citric acid cycle in the autotrophic Desulfobacter hydrogenophilus. Arch. Microbiol. 148, 218–225. 11. Beh, M., Strauss, G., Huber, R., Stetter, K O. & Fuchs, G. (1993) Enzymes of the reductive citric acid cycle in the autotrophic eubacterium Aquifex pyrophilus andinthearchaebacterium Thermoproteus neutrophilus. Arch. Microbiol. 160, 306–311. 12. Antranikian, G., Herzberg, C. & Gottschalk, G. (1982) Char- acterization of ATP citrate lyase from Chlorobium limicola. J. Bacteriol. 152, 1284–1287. 13. Ishii, M., Igarashi, Y. & Kodama, T. (1989) Purification and characterization of ATP: citrate lyase from Hydrogenobacter thermophilus TK-6. J. Bacteriol. 171, 1788–1792. 14. Wahlund, T.M. & Tabita, F.R. (1997) The reductive tricarboxylic acid cycle of carbon dioxide assimilation: initial studies and pur- ification of ATP-citrate lyase from the green sulfur bacterium Chlorobium tepidum. J. Bacteriol. 179, 4859–4867. 15. Kanao, T., Fukui, T., Atomi, H. & Imanaka, T. (2001) ATP- citrate lyase from the green sulfur bacterium Chlorobium limicola is a heteromeric enzyme composed of two distinct gene products. Eur. J. Biochem. 268, 1670–1678. 16. Kanao,T.,Kawamura,M.,Fukui,T.,Atomi,H.&Imanaka,T. (2002) Characterization of isocitrate dehydrogenase from the green sulfur bacterium Chlorobium limicola. A carbon dioxide- fixing enzyme in the reductive tricarboxylic acid cycle. Eur. J. Biochem. 269, 1926–1931. 17. Takeda, Y., Suzuki, F. & Inoue, H. (1969) ATP citrate lyase (citrate-cleavage enzyme). Methods Enzymol. 13, 153–160. 18. Rashid, N., Morikawa, M., Nagahisa, K., Kanaya, S. & Imanaka, T. (1997) Characterization of a RecA/RAD51 homologue from the hyperthermophilic archaeon Pyrococcus sp. KOD1. Nucleic Acids Res. 25, 719–726. Ó FEBS 2002 ATP-citrate lyase from Chlorobium limicola (Eur. J. Biochem. 269) 3415 19. Segel, I.H. (1993) Enzyme Kinetics: Behavior and Analysis of Rapid Equilibrium and Steady-State Enzyme Systems. Wiley Classics Library Edition, New York. 20. Inoue, H., Suzuki, F., Tanioka, H. & Takeda, Y. (1968) Studies on ATP citrate lyase of rat liver. III. The reaction mechanism. J. Biochem. 63, 89–100. 21. Pearson, P.H. & Bridger, W.A. (1975) Catalysis of a step of the overall reaction by the a subunit of Escherichia coli succinyl coenzyme A synthetase. J. Biol. Chem. 250, 8524–8529. 22. Majumdar, R., Guest, J.R. & Bridger, W.A. (1991) Functional consequences of substitution of the active site (phospho) histidine residue of Escherichia coli succinyl-CoA synthetase. Biochim. Biophys. Acta 1076, 86–90. 23. Wells, T.N.C. (1991) ATP-citrate lyase from rat liver. Character- isation of the cytryl-enzyme complexes. Eur. J. Biochem. 199, 163–168. 24. Tuha ´ c ˇ kova ´ ,Z.&Kr ˇ iva ´ nek, J. (1996) GTP, a nonsubstrate of ATP citrate lyase, is a phosphodonor for the enzyme histidine auto- phosphorylation. Biochem. Biophys. Res. Commun. 218, 61–66. 25. Joyce, M.A., Fraser, M.E., James, M.N.G., Bridger, W.A. & Wolodko, W.T. (2000) ADP-binding site of Escherichia coli suc- cinyl-CoA synthetase revealed by x-ray crystallography. Bio- chemistry 39, 17–25. 26. Johnson, J.D., Muhonen, W.W. & Lambeth, D.O. (1998) Char- acterization of the ATP- and GTP-specific succinyl-CoA synthe- tases in pigeon. The enzymes incorporate the same a-subunit. J. Biol. Chem. 273, 27573–27579. 27. Pierce, M.W., Palmer, J.L., Keutmann, H.T. & Avruch, J. (1981) ATP-citrate lyase. Structure of a tryptic peptide containing the phosphorylation site directed by glucagon and the cAMP-depen- dent protein kinase. J. Biol. Chem. 256, 8867–8870. 28. Ramakrishna, S., Pucci, D.L. & Benjamin, W.B. (1981) ATP- citrate lyase kinase and cyclic AMP-dependent protein kinase phosphorylate different sites on ATP-citrate lyase. J. Biol. Chem. 256, 10213–10216. 29. Ramakrishna, S., D’Angelo, G. & Benjamin, W.B. (1990) Sequence of sites on ATP-citrate lyase and phosphatase inhibitor 2-phosphorylated by multifunctional protein kinase (a glycogen synthase kinase 3 like kinase). Biochemistry 29, 7617–7624. 30. Hughes, K., Ramakrishna, S., Benjamin, W.B. & Woodgett, J.R. (1992) Identification of multifunctional ATP-citrate lyase kinase as the a-isoform of glycogen synthase kinase-3. Biochem. J. 288, 309–314. 31. Inoue, H., Suzuki, F., Fukunishi, K., Adachi, K. & Takeda, Y. (1966) Studies on ATP citrate lyase of rat liver. I. Purification and some properties. J. Biochem. 60, 543–553. 3416 T. Kanao et al. (Eur. J. Biochem. 269) Ó FEBS 2002 . Kinetic and biochemical analyses on the reaction mechanism of a bacterial ATP-citrate lyase Tadayoshi Kanao, Toshiaki Fukui, Haruyuki Atomi and Tadayuki. cycle; ACL, ATP-citrate lyase; Cl-ACL, ATP-citrate lyase from Chlorobium limi- cola;AclA ,a subunit of ATP-citrate lyase; AclB, b subunitofATP- citrate lyase. Enzyme: