Structural Insight Into Aeromonas Hydrophila AHL Synthase AhyI Driving Acyl-ACP Selective Recognition

Background: The gram-negative bacterium Aeromonas hydrophila as the major causative agent of the sh disease motile aeromonad septicemia, uses N-acyl-homoserine lactone quorum sensing signals to coordinate biolm formation, motility and virulence gene expression in pathogens. Thus, AHL signaling pathway is considered as a therapeutic target against pathogenic A. hydrophila infection. AHL autoinducers biosynthesis in A. hydrophila are specically catalyzed by an ACP-dependent AHL synthase AhyI using SAM and acyl-ACP as the precursors. Our previously reported AhyI protein heterologously expressed in E. coli strain showed the production characteristics of medium-long chain AHLs, although AhyI was only considered as a short-chain C 4 /C 6 -HSL synthase during the past two decades. Results: In this study, we carried out the in vitro biosynthetic assays of six AHL molecules and kinetic studies of recombinant AhyI with a panel of four linear acyl-ACPs. These resulting data all indicate that C 4 /C 6 -ACP are the native acyl substrates for AhyI against acyl-ACPs with longer linear chains as the non-native acyl donor. In an effort to further understand AhyI acyl-donor substrates preferences, we performed a structural comparison of three ACP-dependent LuxI homologs (TofI, BmaI1 and AhyI), and identied three key hydrophobic residues (I67, F125 and L157) as part of the acyl-chain binding pocket that confer AhyI to selectively recognize native C 4 /C 6 -ACP substrates. The predictions were further supported by computational Ala mutation assay. Conclusions: Our current studies redened AhyI protein that is a multiple short- to long-chain AHL molecules synthase with longer acyl-ACPs (C8~C14) as the non-native substrates, and we also theorized that with knowledge of the key residues in AHL signal synthase AhyI to drive acyl-ACP selective recognition. -ACP, CoA, coenzyme acyl-coenzyme 14 N-tetradecanoyl-homoserine lactone; QS, quorum sensing; PAGE, polyacrylamide gel electrophoresis; UPLC-MS/MS, ultraperformance liquid chromatography-tandem mass spectrometry.


Background
The opportunistic pathogen Aeromonas hydrophila is a ubiquitous inhabitant of various aquatic environments worldwide and infects sh, reptiles, amphibians, and mammals, including humans [1,2]. In particular, motile aeromonad septicemia (MAS) caused by A. hydrophila has become the most important bacterial disease in sh species, and frequent outbreaks lead to huge economic losses periodically per year [3,4]. To date, antibiotics usually are the rst prevention and treatment for A. hydrophila infections, but the extensive use of antibiotics leads to the development of multidrug resistance [5].
Many Gram-negative bacteria use autoinducers as signal molecules to alter speci c genes expression for enabling population density control termed quorum sensing (QS) [6,7]. N-acyl-homoserine lactones (AHLs), the best characterized QS signals, are wide distributed in most gram negative bacteria [8]. AHL molecules possess conservative homoserine lactone ring (HSL), but vary in acyl chain length from C 4 to C 18 , in backbone branch or unsaturation and decoration (i.e., 3-oxo or 3-OH substitution at the β-carbon) [9]. AHL quorum sensing has been implicated as an important factor to affect the virulence in some bacterial pathogens [10]. For example, AHL-based QS enhances the bio lm maturation, modulates the exoenzymes and hemolysin production and is involved in the regulation of type III and type secretion system in the zoonotic agent A. hydrophila [11][12][13][14]. Importantly, targeting AHL signaling circuit asserts less selective pressure for developing drug resistances, and is therefore considered as an alternative strategy of antibiotic usage to provide protection against A. hydrophila that depends on QS to initiate pathogenic expression (Fig. 1).
The crystal structures of three LuxI members known as ACP-dependent AHL synthases have been resolved, including Pseudomonas aeruginosa LasI (PDB Code: 1RO5) [24], Pantoea stewartii EsaI (PDB Code: 1KZF) [21] and Burkholderia glumae TofI (PDB Code: 3P2F) [25]. In addition, several cocrystal structures of complexes with various ligands and a CoA-dependent AHL synthase BjaI from Bradyrhizobium japonicum were also identi ed recently [19]. These LuxI-type AHL synthases exhibited a similar α-β-α fold with a V-shaped cleft and a prominent active-site cavity, reveal a detectable structural similarity to the GCN5-related N-acetyltransferases (GNATs) family [26]. The site-speci c variants LuxI suggest that some identi ed crucial residues play integral roles in catalysis, and establish structural basis for substrate speci city. AHL synthase speci city is tight but not absolute, likely affected from cognate acyl-ACP pool supply [27]. For example, a C 8 -HSL synthase Burkholderia mallei BmaI1 can utilize nonnative acyl-ACP substrates from the E. coli acyl-ACP pool to synthesize nonspeci c AHLs, although the catalytic e ciencies are lower than with the native octanoyl-ACP (C 8 -ACP) [28]. Currently, the general features of molecular structure that determine substrate selectivity in partial LuxI-type AHL synthase are clear, but many details for other well-recognized AHL synthases (e.g. AhyI) remain to be de ned.
AHL molecules in A. hydrophila are typically produced by the LuxI homolog AhyI and perceived by the AhyR receptors [29]. For a long time past, A. hydrophila AhyI was only identi ed as a short-chain AHLs (C 4 -HSL and C 6 -HSL) producer [30]. However, our recent work showed that the recombinant AhyI protein expressed in E. coli can synthesize six types of AHLs, namely, C 4 -HSL, C 6 -HSL, C 8 -HSL, C 10 -HSL, C 12 -HSL, and C 14 -HSL [31]. Due to a lack of in vitro biosynthetic assays of medium-and long-chain AHLs by AhyI, there is not any substantial evidence yet to support whether or not the longer-chain AHLs observed in the heterologous expression strain was an artifact. In this paper, we presented kinetic studies of AhyI with multiple acyl-ACP substrates derived from recombinant E. coli to verify this hypothesis as longer acyl-ACPs might be the acyl substrates for AhyI. Moreover, we propose new insights into the acyl-donor substrates preferences and the structural determinants of substrate speci city for AhyI.

Reagents and strains
The AHL standards purchased from Sigma-Aldrich Chemical Co. were dissolved in methanol to prepare 100 μM stock solutions. Chemicals for protein preparation and enzyme assays were purchased from Sangon Biotech (Shanghai) Co., Ltd., Bio-Rad Laboratories (Shanghai), Inc. or Sigma-Aldrich. The 6x His-Tag Protein Puri cation Kit and BCA Protein Assay Kit were from ProbeGene Inc. (Xuzhou, China).
Molecular biology reagents used for construction of cloning vectors were from Sangon Biotech. UPLC-MS/MS solvents and other conventional reagents were from Merck KGaA (Germany) or Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China). All primers and oligonucleotides for PCR were purchased from Sangon Biotech.
The pET-His plasmids carrying the genes for Escherichia coli MG1655 ACP, Vibrio harveyi B392 AasS, and A. hydrophila HX-3 AhyI were supplied by our previous work [31]. The pBAD plasmid (no His-tag) carrying E. coli ACPs was obtained as a kind gift from Prof. Haihong Wang at the South China Agricultural University, Guangzhou. E. coli BL21(DE3) for protein overexpression were grown on LB at 37 ℃.

Puri cation of holo-ACP
The plasmids pET28a-acpP and pBAD-acpS were transformed into E. coli BL21 (DE3) together, and positive clones were screened on LB medium containing kanamycin (50 μg/mL) and chloramphenicol (25 μg/mL). E. coli strain carrying plasmids pET28a-acpP and pBAD-acpS was cultured at 37 ℃ with shaking in 2 L of LB media supplemented with 0.1 mM L-arabinose and the same concentrations of antibiotics [37]. The culture was grown to an optical density of 0.8, and then induced by the addition of 0.5 mM isopropyl-β-D-thiogalactopyranoside (IPTG) followed by incubation for an additional 4 h. Cell pellets were harvested by centrifugation at 10000 rpm for 10 min at 4 °C and washed twice with equal volume of 20 mM Tris-HCl (pH 8.0). Cells were then resuspended in 10 mL ice-cold buffer A containing 20 mM Tris-HCl pH 8.0, 10 mM MgCl 2 and 5 mM DTT, and lysed by sonication. Cell debris was removed by centrifugation at 10000 rpm for 20 min at 4 °C, 1 mM CoA was added to the supernatant and incubated for 4 h at 37 °C.
An equal volume of ice-cold isopropanol was added to the extract and incubated with stirring at 4 °C for 1 h to remove most other proteins. Following centrifugation, the supernatant was concentrated under nitrogen to the half-volume and then dialyzed overnight against a buffer of 50 mM 2-(N-morpholino) ethanesulfonic acid (MES), pH 6.1. The dialyzed extract was cleared by centrifugation, and the supernatant was applied to a Ni-IDA column. The column was sequentially washed with 100 ml of buffer B (20 mM Tris-HCl pH 8.0, 2 M NaCl and 0.1% TritonX-100), 20 ml of buffer C (20 mM Tris-HCl pH 8.0, 50 mM NaCl and 0.1% TritonX-100) and 50 ml of 10 mM imidazole in buffer C. Holo-ACP with a C-terminal 6×His-tag was eluted from the column using appropriate volumes of buffer C with 250 mM imidazole.
The nickel ions and imidazole were e ciently removed from the ACP solution after dialysis twice against buffer D (20 mM Tris-HCl pH 8.0, 5 mM DTT). Puri ed holo-ACP proteins were concentrated to a volume of 4 mL and the nal concentration was quanti ed by Nanodrop UV-Vis analysis using a molar extinction coe cient of 1.8×10 3 at 280 nm [38]. The purity of the holo-ACP was monitored by using conformationally sensitive gel electrophoresis on a non-denaturing 17.5% polyacrylamide gel containing 2.5 M urea (urea-PAGE) based on the method described previously [33].

Preparation of acyl-ACP substrates
Vibrio harveyi AasS was utilized to synthesize the linear acyl-ACPs (C 4~C14 ) [39]. A reaction mixture contained 100 mM Tris-HCl (pH 7.8), 10 mM MgCl 2 , 5 mM DTT, 10 mM ATP, 100 μM fatty acid, 20 μM holo-ACP and 0.75 μM puri ed AasS and was incubated at 37 °C for 4 h. Notably, the additional reaction times (more than 12 hours) were required for the C 4 -ACP, C 12 -ACP and C 14 -ACP preparation. The reaction was stopped by addition of 50 % ice-cold isopropanol to remove AasS protein. This suspension resulting from centrifugation was treated with two volumes of acetone to precipitate acyl-ACP proteins and stayed at -20 °C overnight [40]. Following centrifugation and two washes with three volumes of acetone, the precipitates were air dried and resuspended in buffer D. Essentially complete conversion to acyl-ACPs were veri ed by urea-PAGE.

In vitro assay of AhyI activity
The acyl substrate recognition pro les of AhyI was analyzed using the reaction mixture (0.5 mL) containing 100 mM Tris-HCl (pH 7.8), 1 mM SAM and 100 μM acyl-ACP (C 4~C14 ). Reactions were initiated by the addition of 1 μM puri ed AhyI and then incubated at 37 °C for 60 min. The AHL products in the reaction mixtures were extracted twice with an equal volume of ethyl acetate containing 0.01% glacial acetic acid. The organic phase was dried with a nitrogen gun and residues were dissolved in 1.0 mL of methanol. The nal products were validated by ultraperformance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS) analysis according to the method described previously [31].
To determine the kinetic parameters for AhyI, the enzymatic reaction was monitored using a colorimetric assay that measured the decrease in 2,6-dichlorophenolindophenol (DCPIP) absorbance at 600 nm [23,41]. The typical reaction contained 30 μM DCPIP, 100 mM Tris-HCl (pH 7.8), 1 mM SAM, and 2-300 μM acyl-ACP. Following a 10 min incubation period, the reactions were then initiated by AhyI addition. AhyI was maintained at 0.75 μM for C 4 -ACP and C 6 -ACP, or 1 μM for C 8 -ACP and C 10 -ACP. The reduction of DCPIP by free holo-ACP released in AHL synthesis was monitored at 600 nm (Δε 600 = 21,000 M -1 cm -1 ) over 10 min and initial rates were calculated based on the progress curve. To estimate kinetic constants, the initial rate data were tted to Michaelis-Menten or substrate inhibition equation using the Graphpad Prism 8.0.

Molecular docking
The homology model of AhyI was generated using the program Modeler v9.19 with TofI structure (PDB code: 3P2F) as the molecular template. The detailed method for model construction and re nement was described in our previous publication [31]. The initial structures of mutated AhyI (I67A, F125A and L157A) were prepared using PyMOL 1.8, following by energy optimization to allow the mutant structure to nd the minimum energy conformation. The structures of acyl-4'-phosphopantetheine (C 4 -, C 6 -, C 8 -, C 10 -, C 12and C 14 -4'-PP) were processed with AutoDock Tools 1.5.6 by adding hydrogens, and further optimized using the PM3 Hamiltonian in MOPAC program. Molecular docking of acyl-4'-PP with AhyI was carried out by AutoDock 4.2.6 program [42]. The grid box was set up with 60×50×60 points in the XYZ axes at a grid spacing of 0.375 Å. The number of Genetic Algorithm (GA) run was set to 100, and the default settings were used for the rest parameters. Finally, the optimal 3D docking conformations with lowest energy scoring were selected for computational Ala mutation assay. The binding a nity values from three parallel dockings of each protein-ligand were collected for further statistical analysis.

Synthesis of acyl-ACP substrates
Overexpressed ACP in E. coli carrying plasmid pET28a-acpP was generally isolated primarily in the apoform. Apo-ACP must undergo 4'-phosphopantetheine (4'-PP) modi cation of the conserved Ser36 through a phosphodiester bond to form active holo-ACP, and then fatty acids can be bound in thioester linkage to the 4'-PP group thiol (Fig. S1) [32]. In the present study, E. coli strain carrying plasmid pET28a-acpP was additionally transformed with the plasmid pBAD-acpS expressing the E. coli AcpS which transfers 4'-PP from CoA to apo-ACP. The phosphopantetheinylation of apo-ACP was shown to be complete by urea-PAGE analysis (Fig. S2). Then, the puri ed Vibrio harveyi acyl-ACP synthetase (AasS) was used to catalyze reaction of holo-ACP and free fatty acids yielding the linear acyl chain of ACP substrates. The reaction products were also analyzed by urea-PAGE (Fig. S2). No single holo-ACP bands on polyacrylamide gel were obtained, indicating that each reaction of acyl-ACP biosynthesis was completed.
Although the molecular weight of the hexahistidine tagged ACPs are higher than native forms, the synthetic acyl-ACPs with C-terminal His-tags remain active but at somewhat less levels than the native protein. For subsequent enzymatic analysis, the activity of hexahistidine tagged acyl-ACPs will be su cient.

Analysis of acyl-ACP utilization pools
Prior UPLC-MS/MS analysis of metabolites from cultured supernatants of recombinant E. coli carrying pET30a-ahyI demonstrated production of six AHL signals. To further verify the in vitro enzymatic activity of AhyI, we tested a panel of linear acyl-ACP (C 4~C14 -ACP) and SAM as substrates for the formation of the corresponding AHL. Six typical characteristic peaks in total ion current (TIC) chromatograms were observed respectively by UPLC-MS/MS analysis, consistent with the retention times of the AHL standards ( Fig. 2). Moreover, all corresponding ion peaks for respective C 4 -HSL (m/z 172), C 6 -HSL (m/z 200), C 8 -HSL (m/z 228), C 10 -HSL (m/z 256), C 12 -HSL (m/z 284), and C 14 -HSL(m/z 312) along with the precursor ion peak (m/z 102) matched those of the AHL synthetic standards in our previous experiments [31], and mass data were shown in Fig. 3. These data suggest that AhyI can use these linear acyl substrates to synthesize the AHL products, including the short-chain C 4 /C 6 -HSLs and the medium-chain C 8 /C 10 -HSLs, as well as the long-chain C 12 /C 14 -HSLs. Thus, the in vitro experiments con rmed the conclusion that acyl-ACPs with longer linear chains than C6 were also the acyl-donor substrates for AhyI.

Kinetics of AHL synthesis by AhyI
The kinetic analyses for AhyI against four linear acyl-ACPs (C 4~C10 -ACP) were performed using a DCPIP colorimetric method. Under xed SAM conditions, the kinetic parameters for AhyI using each of the acyl-ACPs as an acyl-donor substrate are described in Table 1. The AhyI enzyme is clearly in favor of C 4 -ACP with the lowest K m (1.85 × 10 -6 M) and the highest k cat /K m (12.29 × 10 3 M −1 s −1 ) values. Catalytic e ciency was severely affected as the acyl chain length increased. For C 6 -ACP, the k cat /K m values decreased more than 8-fold compared to C 4 -ACP. However, AhyI shows 62-fold and 175-fold lower catalytic e ciencies in response to C 8 -ACP and C 10 -ACP, indicating enzyme activity could be signi cantly inhibited when the length of acyl chains is C8 or longer. These data are in agreement with in vivo observations that the C 4 -HSL with highest abundance and other AHL signals with much lower abundance were found in the recombinant E. coli and A. hydrophila strains [31]. It is worth mentioning that catalytic e ciencies were extremely low at less than 500 μM concentration of C 12 /C 14 -ACP substrates (data not shown). On the other hand, highly concentrated C 12 /C 14 -ACP proteins were prepared with great di culty due to solubility issues with long-chain fatty acids (or salt) during the AasS reaction. Finally, we failed to conduct the kinetic studies for these two acyl substrates with AhyI when the xed substrate was SAM.

Discussion
In contrast to many studies using E. coli DK574, DK574-pJT93 or DK574-pJT94 to prepare holo-ACP [19,28,33], the E. coli holo-ACP expression system in this study is easily conducted and the soluble holo-ACP with a hexahistidine tag could be routinely puri ed by Ni 2+ a nity chromatography in most laboratories. Normally the unmodi ed apo-ACP accumulation will strongly inhibit growth of E. coli [34]. However, our results indicated that overproduction of E. coli ACP therefore appeared to no directly impact strain growth, which was attributed to the overexpression of holo-ACP synthase AcpS, resulting in a rapid alteration of apo-to holo-ACP. For the acyl-ACP biosynthetic methods apart from AasS pathway, the phosphopantetheinyl transferase of Sfp from Bacillus subtilis to transfer the acyl-phosphopantetheine moiety of acyl-CoA to apo-ACP is also commonly used in the acyl-ACP synthesis [35]. However, compared to the two enzymatic methods, the acyl-ACP biosynthetic pathway in the present study could be more economical due to the high price and incomplete commercial supply of acyl-CoA products.
Notably, the substrate-velocity curves were hyperbolic for C 4 /C 6 -ACP and sigmoidal for C 8 /C 10 -ACP (Fig.   4). Interestingly, AhyI with C 4 -ACP utilization displayed a substrate inhibition property, which has also been observed for other LuxI type AHL synthases (e.g. BjaI and BmaI1) [19,28]. Prior kinetic studies on the BmaI1 established that hyperbolic behavior was appropriate for native acyl-ACP substrates with high reaction rates, while non-native acyl-ACPs reacting with BmaI1 showed sigmoidal response in rate curves [28]. Based on the above theory, our kinetic data suggest that C 4 /C 6 -ACP are the native acyl-donor substrates for AhyI and others are considered as non-native acyl-ACPs. Thus, it is reasonable to assume that the short-chain C 4 /C 6 -HSL are the speci c (native) AHL products for the AHL synthase AhyI, and nonspeci c medium-and long-chain AHLs with low synthesis rates could disrupt intercellular communication. However, the AHL-dependent regulation in A. hydrophila involving the medium-and longchain AHLs have not been reported, and quorum sensing mechanism in association with these nonspeci c AHL signaling molecules should be further evaluated to determine their actual impact.
Our previously reported AhyI model indicated the importance of a hydrophobic ligand pocket, hydrogen bonding interactions and several crucial residues with respect to AHL synthesis. However, the molecular basis enabling AhyI to selectively recognize native acyl-ACP substrates from the cellular acyl-ACP pool has yet to be de ned. As previously noted, nine hydrophobic residues (I67, L100, L103, F125, V144, I151, F152, L155, and L157) form an acyl-chain binding pocket of AhyI (Fig. S3). Indeed, similar residues have generally hydrophobic characteristics in other LuxI homologue proteins, but acyl chain size and length of native acyl-donor substrates varies. To understand the acyl-ACP substrate preference for AhyI, a structural comparison was performed for linear AHL synthases TofI (C 8 -HSL), BmaI1 (C 8 -HSL) and AhyI (C 4 -HSL) (Fig. 5). A notable difference between the respective acyl-chain binding pockets is the replacement of small aliphatic residues (A68, L126 and V158 in TofI and BmaI1) with larger hydrophobic residues (I67, F125 and L157, respectively in AhyI), which may constrain the binding pocket in AhyI to only accommodate the shorter acyl substrates. Of these changes, the replacement of A68 with a three carbons longer I67 that with L100 are located at the bottom of the acyl-chain pocket in AhyI would likely restrict acyl chain length to C4 or C6. Notably, two key residues (i.e. L103 and V144) locate adjacent to the pocket periphery, but TofI and BmaI1 contain two larger residues F105 and T145 at the equivalent position respectively, which could in uence in ligand acyl chain selection, in turn allowing AhyI to accommodate an expanded set of longer acyl group. Thus, variances at the positions provided a relatively reasonable explanation for how AhyI can recognize non-native acyl-ACP substrates (C8~C14), albeit the acyl-chain pocket volume limit. Recently, a similar tunnel prediction had been veri ed by the observation of the increase in C 4 -HSL production and decrease in C 12 -HSL after a corresponding residue T105Y mutation in The results obtained by Dong et al. [19,36] showed that the acyl-substrate tolerance of some CoA-based LuxI synthases is likely to depend on the volume of binding pocket, as a consequence of residues important for acyl group binding occuping the position of branched or linear alkyl-group of acyl-CoA substrates. Hence, in order to further test the relationship between the pocket size and the acyl-ACP substrate tolerance, we carried out computational alanine mutation using AutoDock program to compare the autodock-score values. Surface views of acyl-chain pocket of WT AhyI protein with six acyl-ACP substrates were shown in Fig S4. Auto dock results indicated that the relative binding a nities were increased upon introduction of mutations, excluding the replacing I67 with alanine reducing the binding a nity of C 8 -ACP to the protein by 0.44 kcal/mol (Fig. 6). Notably, the binding a nities have a more remarkable increase for non-native acyl-ACPs interacting with AhyI mutations, suggested that increasing the volume of binding pocket would signi cantly facilitate AhyI to recognize medium-long chain acyl substrate. However, the computational data of the ligand binding models presented in this study were insu ciency to prove the structure-function relationship, our future kinetic analysis of site-speci c variants will be performed to better understand the mechanism of acyl substrate selective recognition for AhyI.

Conclusions
In this study, six linear acyl-ACP proteins with C-terminal his-tags were synthesized by V. harveyi AasS using the fatty acids and active holo-ACP proteins from recombinant E. coli. Six types of AHL molecules were speci cally produced by the ACP-dependent AHL synthase AhyI through in vitro enzymatic reaction, indicating that AhyI can synthesize the short-, medium-and long-chain AHL signals using the SAM and corresponding linear acyl-ACP substrates. Kinetic studies of AhyI reacting with a panel of four linear acyl-ACPs showed a notable decrease in catalytic e ciency with increase of the acyl-chain length above C6. Hyperbolic or sigmoidal response in rate curves for varying acyl-donor substrates suggest that C 4 /C 6 -ACP are the native acyl-donor substrates for AhyI and others with longer linear chains than C6 are considered as non-native acyl-ACPs. Based on a structural comparison, three key hydrophobic residues (I67, F125 and L157) as part of the acyl-chain binding pocket were preliminarily proposed to be the structural determinants driving native acyl-ACP selective recognition for AhyI. The calculation data of molecular docking simulations further support this proposition by extremely increased binding a nities for nonnative acyl-ACPs interacting with a representative subset of AhyI mutations. Our structural data are expected to provide theoretical direction on the molecular basis for native acyl-ACP speci c recognition by AhyI.

Declarations
Ethics approval and consent to participate Not applicable Consent for publication Tables   Table 1. Kinetic constants for variable acyl-ACP substrates reacting with AhyI variable acyl-ACPs k cat (s −1 ) x 10 -3 K m (M) x 10 -6 k cat /K m (M −1 s −1 ) x 10 3 relative ratio a  Figure 1 Schematic diagram of biocontrol and prevention of MAS disease caused by A. hydrophila using AHL-QS target-speci c agents.