Cyclic AMP Inhibits the Activity and Promotes the Acetylation of Acetyl-CoA Synthetase through Competitive Binding to the ATP / AMP Pocket *

Xiaobiao Han (韩小彪), Liqiang Shen (申立强), Qijun Wang (王启军) , Xufeng Cen (岑旭峰), Jin Wang (王金), Meng Wu (吴萌)**, Peng Li (李鹏), Wei Zhao (赵维), Yu Zhang (张余), and Guoping Zhao (赵国屏) From the Key Laboratory of Synthetic Biology, Institute of Plant Physiology and Ecology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai 200032, China, the Key Laboratory of Synthetic Biology, Chinese Academy of Sciences Center for Excellence in Molecular Plant Sciences, Institute of Plant Physiology and Ecology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai 200032, China, the Department of Medical Microbiology and Parasitology, Shanghai Jiao Tong University School of Medicine, Shanghai 200025, China, the **Shanghai Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai 200031, China, the State Key Lab of Genetic Engineering & Institutes of Biomedical Sciences, Department of Microbiology and Microbial Engineering, School of Life Sciences, Fudan University, Shanghai 200433, China, the Shanghai-MOST Key Laboratory of Disease and Health Genomics, Chinese National Human Genome Center at Shanghai, Shanghai 201203, China, the Department of Microbiology and Li Ka Shing Institute of Health Sciences, The Chinese University of Hong Kong, Prince of Wales Hospital, Shatin, New Territories, Hong Kong SAR, China, and the University of Chinese Academy of Sciences, Beijing 100049, China

Acetyl-coenzyme A (acetyl-CoA) is a central metabolic intermediate that modulates the balance of anabolism and catabolism by functioning as the fuel for energy generation or as the precursor of lipogenesis for energy storage in many organisms (1). Acetyl-CoA is also deemed a second messenger for physiological regulation by supplying an acetyl group for protein acetylation (1,2), an important mechanism of post-translational modification (PTM). Acetyl-coenzyme A synthetase (Acs) 5 (EC 6.2.1.1) belongs to the acyl-or aryl-CoA synthetase family and catalyzes the biosynthesis of acetyl-CoA from acetate, coenzyme A (CoA), and ATP, and is well conserved from bacteria to mammals. The Acs of Escherichia coli scavenges environmental acetate with high affinity (Յ10 mM in general) and converts it to acetyl-CoA during glucose limitation (3). In humans, Acs is implicated in disposing of alcohol in normal hepatocytes (1,4) and maintaining cell growth under metabolic stress conditions in certain tumors (5,6). Acs is also reported to be involved in chromatin regulation (7) and bacterial chemotaxis (8,9) by supplying acetyl-CoA as the acetylation donor.
Acs is strictly regulated at both transcriptional and posttranslational levels in enteric bacteria. Under carbon limitation, cAMP-CRP (cAMP receptor protein) binds upstream of the acs proximal promoter (acsP2) and initiates transcription of acs (10). However, this cAMP-CRP-dependent activation may be inhibited by the nucleoid proteins IHF and FIS by competitive binding to the CRP-binding sites or other unknown steric hindrances (11). The activity of Acs is regulated by acetylation catalyzed by the protein acetyltransferase Pat and deacetylation catalyzed by the NAD ϩ -dependent deacetylase CobB. Pat transfers an acetyl group from acetyl-CoA onto the Salmonella enterica Acs (SeAcs) Lys 609 and blocks the first half of the catalytic reaction, acetyl-AMP formation (12), whereas CobB restores activity by specifically removing this acetyl group (13).
cAMP is a master second messenger that is involved in rapid response to environmental nutritional changes in bacteria (14,15). The acetate scavenging by Acs is an alternative pathway for acetyl-CoA production under a low-glucose condition, and cAMP regulates Acs in multiple ways. Besides the direct transcription activation of the acs gene by cAMP-CRP as described above (10), cAMP indirectly modulates Acs activity by activating transcription of the pat gene in the presence of non-phosphotransferase system carbon sources in enteric bacteria (16), or by activating Pat activity via binding to its allosteric site in mycobacteria (17,18).
In this study, we discovered that cAMP inhibits the activity of SeAcs by directly binding to the enzyme and competing against one of its substrates, ATP. The binding of cAMP increases SeAcs Lys 609 acetylation, which further inhibits SeAcs activity. The acetylation is apparently caused by concomitant promotion of Pat-dependent acetylation and inhibition of CobB-dependent deacetylation. We further determined a crystal structure of SeAcs complexed with cAMP and CoA, and the crystal structure not only confirms that cAMP binds to the conserved ATP/AMP binding pocket of SeAcs, but also suggests cAMP, in addition to CoA, restrains SeAcs in an open conformation. This structure provides possible scenarios for the mechanism that promotes SeAcs Lys 609 acetylation. In addition, we found cAMP also inhibited S. enterica propionyl-CoA synthetase (SePrpE) and long-chain acyl-CoA synthetase (SeFadD), which are from the same protein family of acyl-or aryl-CoA synthetases. As the residues of cAMP-binding site are generally conserved in this family, we infer that cAMP possibly inhibits other acyl-or aryl-CoA synthetases in S. enterica.

Results
cAMP Binds to SeAcs Directly-We measured the thermodynamic parameters of the proposed interaction between cAMP and non-acetylated SeAcs isolated from a pat-null S. enterica strain using the isothermal titration calorimetry (ITC) method to test whether cAMP binds directly to Acs. The result shows that cAMP directly interacted with SeAcs, with one cAMPbinding site for each SeAcs (blue in Fig. 1A). The dissociation constant (K d ) was estimated to be 164 Ϯ 11 M, which is very close to the physiological concentration of cAMP in live bacterial cells under starvation conditions (19,20), implicating a physiological relevance. We next measured the interaction between cAMP and acetylated SeAcs isolated from a cobB-null S. enterica strain, and the K d was estimated to be 297 Ϯ 15 M (red in Fig. 1A).
Given that the 0.8-fold increase in K d of cAMP binding to acetylated SeAcs compared with the non-acetylated enzyme, we investigated if acetylation of SeAcs also affects the binding of its natural substrate, ATP. The results indicate a similar 2.4fold increase in K d of ATP binding to the acetylated SeAcs (1.09 Ϯ 0.079 mM; red in Fig. 1B) compared with the non-acetylated SeAcs (322 Ϯ 19 M; blue in Fig. 1B). Therefore, both cAMP and ATP preferably bind to the non-acetylated SeAcs. We next measured the acetylation level of Lys 609 , the primary site on SeAcs under acetylation regulation. There is a clear difference in the Lys 609 acetylation level of SeAcs isolated from pat-null or cobB-null S. enterica cells (Fig. 2, A and B). As Lys 609 participates in the catalytic reaction of SeAcs, it is not surprising that acetylation at Lys 609 affects the binding of ATP to SeAcs.
The results also suggest that cAMP might bind to a location similar to that of ATP in SeAcs.
cAMP Inhibits SeAcs by Competing with the Substrate ATP-We measured SeAcs activity (isolated from pat-null S. enterica cells) in the presence or absence of cAMP to test whether binding of cAMP inhibits the activity of SeAcs, SeAcs catalyzes the formation of acetyl-CoA using CoA, acetate, and ATP as substrates. The potential inhibitory effect of cAMP on Acs was tested using a subsaturated concentration for each of the substrates. cAMP substantially inhibited SeAcs activity only in an enzymatic reaction system with a low concentration of ATP (Fig. 1C), suggesting cAMP inhibits SeAcs via an ATP-competitive mechanism.
To further confirm that cAMP inhibits SeAcs by competing against the substrate ATP, we measured the apparent K m constants of ATP under various cAMP concentrations. The Lineweaver-Burk plot for cAMP shows a characteristic signature of a competitive inhibitor (Fig. 1D), in agreement with the above observation that cAMP barely inhibits SeAcs when ATP is saturated (Fig. 1C).
The inhibition constant (K i ) for cAMP-SeAcs was estimated as 185 Ϯ 25 M by plotting the apparent K m (toward substrate ATP) against cAMP concentration (inset of Fig. 1D), which is similar to the ITC estimated K d (164 Ϯ 11 M; blue in Fig. 1A).
cAMP Promotes Lys 609 Acetylation of SeAcs-Acetylation of the catalytic lysine residue (Lys 609 in SeAcs) is an efficient way to modulate Acs activity (12,13). cAMP has been reported to enhance Acs acetylation by activating the lysine acetyltransferase Pat in mycobacteria (17,18) via binding to the allosteric site in the cyclic nucleotide (cNMP)-binding domain of Pat. This domain is unique to mycobacteria species (17), and we speculate that the mechanism by which cAMP affects the Acs in enteric bacteria, such as SeAcs, must be different.
To determine whether direct binding of cAMP to SeAcs affects its Lys 609 acetylation level, we performed in vitro enzymatic acetylation assays using purified non-acetylated SeAcs (from a pat-null S. enterica strain) and acetyl-CoA as the substrates. The acetylation assays were catalyzed by purified S. enterica protein acetyltransferase (SePat), with or without cAMP in the reaction system. The level of SeAcs acetylation was determined by Western blotting using an antibody that specifically recognizes acetylated Lys 609 ( Fig. 2A). As shown in . The addition of cAMP to the acetylation assay system further decreases SeAcs activity, suggesting SeAcs acetylation promoted by cAMP also contributes to the inhibitory effect of cAMP on SeAcs. Because a cNMP-binding domain is absent in SePat (17), we infer that cAMP promotes SeAcs acetylation through direct binding to its ATP/AMP pocket as well.
Lysine acetyltransferase Pat and deacetylase CobB exist normally as a pair to modulate cellular protein acetylation levels in most bacteria (21). To test if cAMP also affects deacetylation of SeAcs by SeCobB, we performed in vitro enzymatic deacetylation assay using purified acetylated SeAcs from a cobB-null S. enterica strain as the substrate, NAD ϩ as a coenzyme, and purified S. enterica Sir2-like lysine deacetylase (SeCobB) as the catalytic enzyme. The result in Fig. 2D confirmed that SeCobB was able to remove the acetyl group on Lys 609 of SeAcs (lane 4), and rescued SeAcs activity. However, decrease in the acetyla-tion level as well as the recovery of SeAcs activity by CobB treatment was inhibited remarkably in the presence of cAMP (lane 7). Because cAMP does not bind to SeCobB (data not shown), we infer that cAMP inhibits CobB-dependent deacetylation by directly binding to SeAcs.
cAMP Binds to the ATP/AMP Binding Pocket of SeAcs-To understand the structural basis of SeAcs inhibition by cAMP, we determined the crystal structure of SeAcs complexed with cAMP, CoA, and acetate ( Fig. 3, Table 1) at 1.65-Å resolution. The structure defined the binding site of cAMP on SeAcs, and the effects of cAMP and CoA binding on SeAcs conformation. The simulated annealing omit F o Ϫ F c electron density map shows unambiguous electron density of cAMP within the ATP/AMP binding pocket of SeAcs (Fig. 3, A and E), in agreement with our observation that cAMP inhibits SeAcs by competing with the substrate ATP ( Fig. 1, C and D). The map also shows unambiguous electron density for adenosine and diphosphate moieties of CoA within the adjacent CoA binding pocket (right in Fig. 3A). However, no electron density for acetate was observed within the acetate-binding site of SeAcs, probably due to effect of steric hindrance by cAMP.
Our crystal structure revealed that in the presence of cAMP and CoA, the C-terminal domain of SeAcs adopted an open conformation (red in Fig. 3B), exposing Lys 609 to the solvent. Lys 609 plays an important role in the first half of the reaction catalyzed by SeAcs and is subject to negative regulation by acetylation (13). We We also compared our open conformation with a previously determined "closed" conformation of a crystal structure of yeast Acs in complex with AMP. We observed a large rotation of ϳ100°in the Acs C-terminal domain, which includes Lys 675 (Lys 609 in SeAcs), in the closed conformation (Fig. 3D), as a result Lys 675 is buried into the active center. The exposed conformation of Lys 609 in our current structure would allow Lys 609 to be more accessible for acetylation, consistent with the observation that the acetylation of SeAcs was remarkably increased in the presence of cAMP (Fig. 2C).
Contacts between cAMP and residues of SeAcs are shown in Fig. 4, A and B. The adenine moiety of cAMP makes the same interactions with SeAcs as the adenine moiety of AMP makes in the crystal structure of SeAcs in complex with AMP, CoA, and acetate (PDB 2P2F; Fig. 3E) (22). These interactions are also the same for the adenine moiety of 5Ј-propyl-AMP in the crystal structure of SeAcs in complex with 5Ј-propyl-AMP and CoA (PDB 1PG4; Fig. 3E) (23). These interactions include: stacking effects of the adenine base to side chain atoms of Trp 413 and Ile 512 from one side and main chain atoms of Val 386 , Glu 388 , and Pro 389 from the other side; one H-bond through the N6 atom of FIGURE 2. cAMP promotes acetylation of SeAcs Lys 609 in vitro. A, the antibody used is specific to acetylated SeAcs Lys 609 . The same amounts of SeAcs isolated from cobB-null or pat-null S. enterica strains were separated by SDS-PAGE, transferred to the nitrocellulose membrane, then blotted with the antibody specific to the acetylated SeAcs Lys 609 in the presence or absence of unmodified peptide (KTRSGKIMRRI), or acetylated peptide (KTRSG(K ac )IMRRI), and detected using an ImageQuant LAS 4000 (GE Healthcare). Coomassie Brilliant Blue staining was used for loading controls. B, semiquantification of the acetylation level of SeAcs Lys 609 purified from CobB-null or pat-null S. enterica cells. C, cAMP activates SePat-dependent acetylation and enhances the activity inhibition upon SeAcs. Recombinant SeAcs purified from the pat-null S. enterica strain is treated by acetylation in the presence or absence of cAMP followed by activity measurement and Western blotting detection. The SeAcs activity in the absence of cAMP, acetyl-CoA, and SePat (Lane 1) was treated as 100%. Coomassie Brilliant Blue-stained proteins on SDS-PAGE gels are shown below as a loading control. D, cAMP inhibits SeCob-dependent deacetylation and weakens the activity recovery of SeAcs. Recombinant SeAcs purified from a cobB-null S. enterica strain was treated by deacetylation in the presence or absence of cAMP followed by activity measurement and Western blotting detection. The SeAcs activity (lane 4) was treated as 100%. Each experiment was repeated in triplicate, and the data are presented as mean Ϯ S.D., ***, p Ͻ 0.001; **, p Ͻ 0.01. the adenine with Asp 411 ; and one water-mediated H-bond through the N3 atom of the adenine with Asp 500 (Fig. 4, A and  B). The ribose moiety of cAMP makes three H-bonds through its 2Ј OH group with Asp 500 , Gln 415 , and Arg 515 , respectively, and one H-bond through its 5Ј O atom with Asn 521 . The phosphate group of cAMP contacts SeAcs by one H-bond with Thr 416 , one salt-bridge bond with Arg 526 , and Van der Waals interactions with Thr 311 and Glu 417 . cAMP binds to the highly conserved ATP/AMP binding pocket of Acs and shares most contact residues with AMP (Figs. 3E and 4, A and B). Therefore it is not surprising that most contact residues of cAMP are conserved among bacterial species, yeast, and humans (Fig. 4C). We thereby propose that the regulatory effect of cAMP on Acs is possibly an important mechanism that has been preserved during evolution.
Trp 413 , Gln 415 , Asp 500 , and Arg 526 in SeAcs Are Determinants for cAMP Binding and Promotion of SeAcs Acetylation-Although the SeAcs crystal structure shows 14 total residues interact with cAMP either by H-bonds or Van der Waals forces (Fig. 4), it is still unknown which site(s) is critical. To further define the determinants of cAMP binding, we performed ITC experiments analyzing the binding affinity of cAMP to SeAcs derivatives bearing alanine substitution of residues that are involved in side chain interactions. As shown in Table 2, the derivatives of SeAcs W413A, Q415A, D500A, and R526A lost their ability to bind cAMP; whereas D411A, R515A, and N521A displayed decreased affinity; and T416A displayed a similar affinity to wild-type SeAcs.
To verify that cAMP promotes SeAcs Lys 609 acetylation through direct binding of cAMP to the ATP/AMP pocket in SeAcs, we performed similar in vitro Pat-dependent acetylation and CobB-dependent deacetylation assays using SeAcs derivatives W413A, Q415A, and D500A, which lost their cAMP bind-ing ability. The W413A, Q415A, and D500A derivatives of SeAcs isolated from pat-null S. enterica cells were acetylated in vitro by SePat to a similar level to wild-type SeAcs (  (Table 2), the above results provide further support that the modulation of SeAcs acetylation by cAMP is dependent on direct cAMP binding.
cAMP Generally Inhibits Other Acyl-or Aryl-CoA Synthetases-Acs belongs to a large acyl-or aryl-CoA synthetase protein family (EC 6.2.1-), which all contain a two-domain architecture and harbors a conserved ATP/AMP binding pocket (24). To explore whether cAMP might inhibit other members of the acyl-or aryl-CoA synthetase family in S. enterica, we tested the effect of cAMP against two other acyl-CoA synthetases from S. enterica, SePrpE and SeFadD. The potential inhibitory activities were measured in the presence of subsaturation ATP concentration. The activities of SePrpE and SeFadD were inhibited by cAMP by 80 and 70%, respectively (Fig. 6A). These results support our hypothesis that cAMP possibly regulates other members of the acyl-or aryl-CoA synthetase family in S. enterica.
We aligned the protein sequences of all available acyl-or aryl-CoA synthetases (1190 records) from the UniPortKB data- base, and found that 13 of the 14 cAMP contact residues are generally conserved (Fig. 6B). We speculate that the regulation by cAMP might be a more widespread occurrence in this protein family in organisms apart from S. enterica.

Discussion
This study establishes that cAMP modulates Acs activity through a novel mechanism. We show that cAMP occupies the ATP/AMP binding pocket of SeAcs, inhibiting SeAcs activity by competitive binding against its natural substrate ATP. We found that cAMP cooperates with Pat/CobB to regulate SeAcs acetylation and activity. Finally, we observed a similar inhibitory effect of cAMP against SePrpE and SeFadD, both harboring the ATP/AMP-binding domain and forming AMP during catalysis, and thus, this novel regulatory mechanism is proposed to be extendable to acyl-or aryl-CoA synthetase family enzymes, subject to future confirmation. cAMP adjusts cellular anabolism and catabolism in some bacteria to allow adaptation to an environment with a nonpreference carbon source. cAMP activates/represses the transcription of about 200 genes by directly binding to and allosterically activating CRP (25). One of them is the gene that encodes Acs, which converts acetate to acetyl-CoA to fuel the life when glucose is absent and the environmental concentration of acetate is low. This reaction is costly and is deemed to be regulated by cAMP (10). Acs consumes one ATP for activating one acetate, releasing one AMP and one PP i . Two more ATPs in total are consumed by adenylate kinase to catalyze the recycling of AMP to ADP (the ATP precursor), and by pyrophosphate phosphohydrolase to catalyzes PP i hydrolysis (26). Uncontrolled overexpression of Acs has been reported to cause a low-energy physiology state and arrest growth of S. enterica cells (26). Therefore, either the acs gene expression level or Acs activity has to be tightly regulated.
Previous studies have revealed activity-based negative regulation of cAMP upon Acs through PTM. cAMP coordinates expression of yfiQ genes in E. coli, which encodes the lysine acetyltransferase Pat (16). Pat completely abolishes the activity of Acs by adding an acetyl group to the catalytic lysine residue in Acs active center (13). cAMP also allosterically activates the activity of the protein acetyltransferase Pat in Mycobacterium tuberculosis and Mycobacterium smegmatis by targeting the Pat cNMP-binding domain, hence inhibiting the activity of Acs (17,18). The new cAMP regulation mechanism identified in this study provides an alternative means for directly sensing the cellular ATP/cAMP levels, and also a rapid way to rescue energy deficiency by inhibiting Acs activity. We found binding of cAMP could influence the acetylation potential of Acs by promoting Pat-dependent acetylation and inhibiting CobB-dependent deacetylation concomitantly, thus enhancing the PTM inhibition of Acs activities.
SeAcs enzymatic activity is subject to post-translational regulation by acetylation of the catalytic Lys 609 residue (13). Lys 609 is located at the C-terminal domain of SeAcs, which adopts open and closed conformations by rotating ϳ100°relative to the N-terminal domain (22,23,27). The C-terminal domain of SeAcs is proposed to fold into a closed conformation during the first half-reaction (converting ATP and acetate into acetyl-AMP), in which Lys 609 inserts into the active center and makes critical contacts with the substrates (27). The C-terminal domain adopts an open conformation in the second half-reaction (converting acetyl-AMP and CoA into acetyl-CoA and AMP), in which Lys 609 moves out from the active center and is exposed to solvent (27).
The two conformations have been captured in the crystal structures of SeAcs complexed with CoA, acetate, and AMP (open conformation; PDB code 2P2F) (22), and the yeast Acs in complex with AMP (closed conformation; PDB code 1RY2) (27), respectively. It is proposed that binding CoA triggers the conformation change (22). In the current crystal structure of SeAcs in complex with cAMP and CoA, the C-terminal domain of SeAcs is held to an open conformation, which exposes Lys 609 to the protein surface. A recently reported crystal structure of SeAcs in complex with SePat (28) revealed that SeAcs adopts an open conformation during Lys 609 acetylation exactly as that of our crystal structure, which may account for enhanced acetylation by SePat via cAMP binding. However, it is puzzling why cAMP also inhibits the SeCobB catalyzed deacetylation of SeAcs. Because no crystal structures of either bona fide acetylated SeAcs or SeAcs complexed with SeCobB are available, it is unclear how Lys 609 acetylation affects SeAcs conformation and what conformation of SeAcs would adopt during deacetylation. We speculate that SeCobB might favor a slightly different conformation of SeAcs from the open one observed in our structure, and the presence of cAMP hinders the conformation flexibility.
This study also broadens our knowledge of the complexity of cAMP signaling. This second messenger was initially discovered as a regulator for carbohydrate metabolism in bacteria (29,30), and then increasingly recognized as an important factor involved in other biological processes such as cell division, cell motility, and microbial virulence (31-33). cAMP has been cAMP Directly Inhibits Acs Activity JANUARY 27, 2017 • VOLUME 292 • NUMBER 4 implicated into multiple events in humans, ranging from metabolism (34) to memory formation (35) and innate immunity (36). The major outcome of cAMP signaling is regulating gene expression, either by directly modulating transcription through cAMP-CRP in bacteria, or by indirectly controlling gene expression through protein kinase A as an intermediate in eukaryotes (37). Besides gene expression, cAMP also controls gating of some ion channels (38) and activates guanine exchanging factors (39). For all the target proteins discussed above, cAMP binds to a conserved cNMP-binding domain (40) shared by all of them, characterized by a ␤ barrel flanked by ␣ helixes folded from ϳ120 continuous residues. The cAMP binding pocket of Acs discovered in our crystal structure exhibits a typical Rossmann-fold (Fig. 3B, six ␤ sheets and three ␣ helixes form the cAMP binding pocket), which is radically different from the fold of the conserved cNMP-binding domain. Moreover, this new cAMP binding pocket is well conserved based on our sequence alignments of Acs from bacteria, yeast, and humans (Fig. 4C). Therefore, we propose that the regulation mechanism of cAMP on Acs is evolutionally conserved from bacteria to mammals.
The newly identified cAMP binding pocket is also generally conserved in other acyl-or aryl-CoA synthetases (Fig. 6B), as implicated by the sequence alignment of 1190 acyl-or aryl-CoA synthetase family members across multiple species. We tested cAMP inhibitions of two other family members from S. enterica experimentally, and the inhibition effects upon SePrpE and SeFadD were similar to that for SeAcs. We infer that the regulatory mechanism of cAMP on Acs might be extendable to other family members of acyl-or aryl-CoA synthetases in  . cAMP has no effect on acetylation promotion of mutant SeAcs. A, cAMP activates SePat-dependent acetylation of Lys 609 on wild-type SeAcs but has no effect on mutant SeAcs in vitro. B, cAMP inhibits SeCobB-dependent deacetylation of Lys 609 on wild-type SeAcs but has no effect on mutant SeAcs in vitro. Top band, acetylation detection by Western blotting using anti-SeAcs Lys 609ac antibody; bottom band, Coomassie Brilliant Blue-stained proteins on SDS-PAGE gels are shown as a loading control. Each experiment was independently repeated three times.

cAMP Directly Inhibits Acs Activity
S. enterica by competitively binding the substrate pockets of these enzymes. We speculate that the regulatory mechanism by cAMP on this protein family might also exist in other organisms. The regulation of Acs by cAMP binding is yet to be characterized in vivo. However, the K d of cAMP measured in this study is comparable with the physiological concentration found in bacterial cells under glucose depletion conditions (19,20), which implicates the physiological relevance of cAMP. It is intriguing that Acs is regulated by cAMP in such a complex network involving several different layers of transcriptional activation, PTM, and direct inhibition. The concentration of cAMP is ever changing in cells corresponding to its growth phase and the growth environment (Table 3). In particular, cAMP is low when glucose is abundant in the culture medium, the concentration increases as glucose is consumed, and it reaches a maximum when glucose is depleted. It was reported that cAMP activates the transcription of Acs at a micromolar level but represses this transcription when cAMP is present a millimolar level (41,42). We therefore speculate that one dominant regulatory means is preferred at a certain level of cAMP concentration for Acs activities in vivo. Once the concentration of cAMP reaches to more than 100 M, a hostile environment may be represented and cells need to curtail most energy costs including inhibiting acetate scavenging by directly binding cAMP to the Acs protein.
Our data present a new means of Acs regulation allowing direct sensing and rapid response by bacterial to cellular energy level. Our crystal structure also uncovered a new cAMP binding pocket, adding another layer to the complexity of cAMP signaling.

Experimental Procedures
Materials-Bacteria strains and plasmids used in this study are described and listed in supplemental Tables S1 and S2. Antibody of anti-acetylated SeAcs Lys 609 was kindly gifted from Prof. Chen Yang. Acetate, acetyl-CoA, CoA, NAD ϩ , cAMP, ATP, myokinase, pyruvate kinase, lactate dehydrogenase, and other chemicals were purchased as high purity from Sigma-Aldrich. Modified sequencing grade trypsin was purchased from Promega (Madison, WI).
Construction of cobB-null and pat-null Strains-The completely knock-out mutants of ⌬cobB and ⌬pat were constructed in S. enterica G2466 by adapting a method described by Datsenko and Wanner (43). Genetic information of strains and primer sequences are listed in supplemental Tables S1 and S3. Obtained deletion mutations were verified by PCR.
Protein (SeAcs, SePat, SeCobB, SePrpE, and SeFadD) Expression and Purification-The acs (STM4275), pat (STM2651), cobB (STM1221), prpE (STM0371), and fadD (STM1818) genes were PCR amplified from S. enterica G2466 genome and cloned into the pET22b-tac or pET28b-tac vector (supplemental Table  S2). Recombinants were then introduced into S. enterica G2466 wild-type or the indicated mutant cells, and successful transformants were selected by plating the cells on LB plates containing 100 g/ml of ampicillin. Single colonies were first grown in 5 ml LB overnight at 37°C, and then inoculated to 250 ml LB. Protein expressions were induced with isopropyl ␤-D-thiogalactoside to a final concentration of 0.5 mM overnight at 18°C. Cells were harvested and resuspended in lysis buffer containing 50 mM Tris (pH 7.5), 300 mM KCl, 10% (v/v) glycerol, and 1 mM PMSF, and disrupted using an EmulsiFlex-C5 cell disruptor (AVESTIN, Inc., Ottawa, Canada). Cellular debris was removed by centrifuging at 13,000 ϫ g for 30 min at 4°C. The recombinant proteins were purified with affinity chromatography using a nickel-nitrilotriacetic acid Superflow column (Qiagen) and dialyzed into a buffer containing 50 mM Tris (pH 7.5), 50 mM NaCl, 1 mM DTT, and 10% (v/v) glycerol. After purification, the target proteins with sufficient purity (above 95%) were quantified and kept frozen at Ϫ80°C. The variants of SeAcs were constructed via PCR site-directed mutagenesis (Agilent, Inc.) and then expressed in S. enterica G2466 and purified in the same way as that of wild-type SeAcs. Specially, SeAcs used in Pat-dependent acetylation assay and enzymatic inhibition assay was produced from the pat-null S. enterica G2466 strain, whereas SeAcs used in CobB-dependent deacetylation assay was produced from cobB-null S. enterica G2466 strain.
SeAcs for Crystallization-A single colony of BL21(DE3) transformed with plasmid Ph 3 (pET22b-tac-SeAcs) was used to inoculate 20-ml LB broth containing 100 g/ml of ampicillin, and cultures were incubated overnight at 37°C with shaking. LB broth (1 liter) was inoculated with a 1:100 ratio, incubated at
In Vitro SeAcs Enzymatic Assay-SeAcs enzymatic activity was measured by a spectrophotometric coupled assay at 37°C as described in Ref. 22. During the assay, SeAcs first converts acetate, CoA, and ATP to acetyl-CoA and AMP. Then, myokinase converts AMP to ADP. Pyruvate kinase converts ADP and phosphoenolpyruvate to pyruvate and ATP. Finally, lactate dehydrogenase reduces pyruvate and oxidizes NADH to NAD ϩ . The change of NADH level is detected spectrometrically at 340 nm. The enzymatic reaction mixtures (90 l) contain 50 mM HEPES (pH 7.5), 5 mM MgCl 2 , 1 mM TCEP, 0.02 M SeAcs, 5 units of myokinase, 1 unit of pyruvate kinase, 1.5 units of lactate dehydrogenase, 3 mM phosphoenolpyruvate, and 0.1 mM NADH. Reaction mixtures were incubated at 37°C for 2 min, and then the assay was initiated by adding 10 l of substrate mixtures (final concentration: 1 mM CoA, 1 mM ATP, and 20 mM acetate).
To determine the ratio of cAMP inhibition, the potential inhibitory effect of cAMP (1 mM) on SeAcs was tested using subsaturated concentrations for one substrate (i.e. 20 M for CoA, 20 M for ATP, or 20 M for acetate), and saturated concentrations for other substrates (i.e. 1 mM for CoA, 1 mM for ATP, or 20 mM for acetate). To determine the dissociate constant K i of cAMP toward SeAcs, different concentrations of cAMP (0, 0.15, 0.3, 0.45, 0.6 mM) was added to the reaction mixtures. For each cAMP concentration, the apparent K m value of SeAcs toward the substrate of ATP was measured by adding 10 l of substrate mixture (final concentration: 20 mM acetate, 1 mM CoA, and varying ATP: 20, 40, 80, 120, 200, 300, 400, 600, 1000, 1500, and 2000 M). The reactions were continually monitored by measuring absorbance at A 340 every 2 s for 10 min using a microplate reader (Thermo Multiskan FC, Thermo-Fisher Inc.). The initial velocities were analyzed using Prism 5 (GraphPad Software Inc.), and the apparent K m values in the presence of inhibitors were estimated by non-linear regression using the Michaelis-Menten equation. The mode of cAMP inhibition was determined by a Lineweaver-Burk plot and the substrate binding affinity K i was estimated (intercept on x axis) in a Dixon plot with an apparent K m toward ATP against the concentration of cAMP as described in Ref. 44. Each point was measured in triplicate.
In Vitro SeAcs Acetylation and Deacetylation Assay-Pat-dependent SeAcs acetylation assays were carried out as follows: reaction mixtures (100 l) containing 50 mM Tris (pH 7. CobB-dependent SeAcs deacetylation assays were carried out as follows: reaction mixtures (100 l) containing 50 mM Tris (pH 7.5), 200 M TCEP, 1 mM NAD ϩ , 0.3 M SeCobB, and 1 M SeAcs and in the presence or absence of 1 mM cAMP, were incubated at 37°C for 60 min. The acetylation level of SeAcs Lys 609 was measured by Western blotting. Meanwhile, the activity of treated SeAcs was detected by adding 10 l of deacetylation reaction mixtures (containing 2 pmol of SeAcs) to the enzymatic reaction mixtures as described above.
Western Blotting-Standard Western blotting procedures were followed for protein acetylation analysis. Proteins were separated by SDS-PAGE, transferred onto NC membrane (Millipore), subjected to immunoblotting with the homemade antibody specific to the acetylated SeAcs Lys 609 , and detected by ImageQuant LAS 4000 (GE Healthcare). Coomassie Blue staining (SDS-PAGE) was used for loading controls.
In Vitro SePrpE, SeFadD Enzymatic Assay-The SePrpE and SeFadD enzymatic activities were measured by a similar spectrophotometric-coupled assay as described above. Reaction mixtures (90 l) contain 50 mM HEPES (pH 7.5), 5 mM MgCl 2 , 1 mM TCEP, 10 nM SePrpE or 800 nM SeFadD, 5 units of myokinase, 1 unit of M ATP for SeFadD). The reactions were continually monitored by measuring absorbance at A 340 every 2 s for 10 min using a microplate reader (Thermo Multiskan FC, ThermoFisher Inc.). Isothermal Titration Calorimetry-ITC experiments were carried out using a Microcal iTC200 microcalorimeter (GE Healthcare). SeAcs and SeAcs derivatives were dialyzed against a buffer containing 20 mM HEPES (pH 7.5), 100 mM NaCl, and 1 mM TCEP and concentrated to 100 M by ultrafiltration. cAMP or ATP was dissolved in the same buffer to the concentration of 4 mM. During the experiments, 200 l of 100 M SeAcs or SeAcs derivatives was titrated with 2 l of 4 mM cAMP for 20 steps. The resulting titration curves were fitted with MicroCal ORIGIN software. The dissociation constant K d was estimated by non-linear regression with ORIGIN 7.0 software (OriginLab Inc.).
SeAcs Crystallization-SeAcs (180 M) was incubated with cAMP (2 mM), coenzyme A (1 mM), and acetate (1 mM), and incubated 1 h at room temperature. Crystal growth conditions of the complex were screened with commercial solutions (Emerald Biosystems, Inc. and Hampton Research, Inc.) using a sitting-drop vapor-diffusion technique (drop: 0.2 l of protein complex plus 0.2 l of reservoir solution; reservoir: 60 l of commercial screening solution; 22°C). Small shell-like crystals appeared within 3 days during screening, and were able to reach a dimension of 0.2 ϫ 0.4 ϫ 0.1 mm when growing in drops of larger size (2 l; 1 l of protein complex plus 1 l of reservoir) using hanging-drop vapor diffusion with reservoir (0.2 M KH 2 PO 4 , pH 4.8, 20% PEG3350). Crystals were harvested by transferring into reservoir solution containing 12.5% (v/v) (2R,3R)-(Ϫ)-2,3-butanediol (Sigma), and then flash-cooled in liquid nitrogen.
Structure Determination-Diffraction data were collected from cryo-cooled crystals at SSRF beamline BL17U1 (51). Data were processed using HKL2000 (45). The structure was solved by molecular replacement with Molrep (46) using the crystal structure of SeAcs (PDB code 1PG4) (23) as the search model, and refined with Phenix (47) and Coot (47). cAMP and coenzyme A were built into the model during the late stage of refinement. The final SeAcs-cAMP-CoA model, has been refined to R work and R free of 0.169 and 0.207, respectively.
Author Contributions-X. H. conceived the study, performed most in vitro biochemical experiments and data analysis; L. S. purified and crystallized SeAcs-cAMP-CoA, and determined the crystal structure; P. L., Q. W., X. C., J. W., and M. W. assisted in data analysis; W. Z., Y. Z., and G. Z. planned experiments, and wrote the manuscript. All the authors reviewed the results and approved the final version of the manuscript.