Structural insights into the catalytic mechanism of aldehyde-deformylating oxygenases
- First Online:
- 1.9k Downloads
The fatty alk(a/e)ne biosynthesis pathway found in cyanobacteria gained tremendous attention in recent years as a promising alternative approach for biofuel production. Cyanobacterial aldehyde-deformylating oxygenase (cADO), which catalyzes the conversion of Cn fatty aldehyde to its corresponding Cn-1 alk(a/e)ne, is a key enzyme in that pathway. Due to its low activity, alk(a/e)ne production by cADO is an inefficient process. Previous biochemical and structural investigations of cADO have provided some information on its catalytic reaction. However, the details of its catalytic processes remain unclear. Here we report five crystal structures of cADO from the Synechococcus elongates strain PCC7942 in both its iron-free and iron-bound forms, representing different states during its catalytic process. Structural comparisons and functional enzyme assays indicate that Glu144, one of the iron-coordinating residues, plays a vital role in the catalytic reaction of cADO. Moreover, the helix where Glu144 resides exhibits two distinct conformations that correlates with the different binding states of the di-iron center in cADO structures. Therefore, our results provide a structural explanation for the highly labile feature of cADO di-iron center, which we proposed to be related to its low enzymatic activity. On the basis of our structural and biochemical data, a possible catalytic process of cADO was proposed, which could aid the design of cADO with improved activity.
Keywordsaldehyde-deformylating oxygenase di-iron center crystal structure catalytic mechanism alk(a/e)ne production
ADO from cyanobacteria
ferredoxin, ferredoxin reductase, and NADPH
inductively coupled plasma optical emission spectrometer
phenazine methosulfate and NADH
Prochlorococcus marinus MIT9313
the R2 protein of ribonucleotide reductase
Synechococcus elongates PCC7942
Fatty alk(a/e)nes are major components of fuel oil and one of the ideal alternatives for fossil-based biofuels. Recently, Schirmer et al. identified two genes from cyanobacteria that encode an acyl-ACP reductase (AAR) and an aldehyde-deformylating oxygenase (ADO), both of which are responsible for alkane production in cyanobacteria (Li et al., 2012; Schirmer et al., 2010). This pathway has drawn considerable attention, since this route provides a promising approach for photosynthetic production of alka(e)ne biofuels (Krebs et al., 2011; Wang et al., 2013). In this two-step pathway, AAR reduces fatty acyl-ACPs or -CoAs into their corresponding aldehydes, and subsequently the Cn fatty aldehydes are converted into their corresponding Cn-1 alk(a/e)nes by ADO (Schirmer et al., 2010).
Cyanobacterial aldehyde-deformylating oxygenase (cADO) belongs to the superfamily of ferritin-like di-iron proteins (Krebs et al., 2011) and contains a di-iron center (Das et al., 2011). However, metal analysis indicated that less than one fifth of the purified protein in a typical preparation contains iron (Das et al., 2011), assuming two iron atoms per enzyme molecule, the results indicated that the iron atoms may be loosely bound to cADO. The in vitro reaction catalyzed by cADO requires both the dioxygen as co-substrate and the presence of a reducing system, which provides four electrons per turnover (Warui et al., 2011) and can either be biological (ferredoxin, ferredoxin reductase, and NADPH; Fd/FR/N) (Schirmer et al., 2010; Warui et al., 2011) or chemical (phenazine methosulfate and NADH; P/N) (Das et al., 2011). It was proposed that the incorporation of O2 into the reduced cofactor generates an iron-peroxo species that attacks the substrate aldehyde to form a hemiacetal and followed by the scission of its C1–C2 bond (Li et al., 2012; Li et al., 2011). The ensuing product of the reaction is a Cn-1 alkane (Li et al., 2011), and the C1-derived co-product was demonstrated to be formate (Warui et al., 2011). Isotope-tracer assay revealed that the oxygen atom in the product formate originates from the co-substrate O2 (Li et al., 2012). Recent reports suggested that cADO also catalyzes the incorporation of an oxygen atom from O2 into its alkane product for C9–10 aldehyde substrates, which yields both Cn-1 alcohol and aldehyde products, implying a function for cADO in oxygenation, in addition to deformylation (Aukema et al., 2013; Das et al., 2014). An intriguing feature of cADO is its low in vitro activity (Das et al., 2011; Eser et al., 2012; Li et al., 2011), with only 3–5 turnovers (Andre et al., 2013; Warui et al., 2011). Andre et al. reported that cADO is reversibly inhibited by the side product H2O2, and its inhibition can be relieved by adding catalase in enzyme assays (Andre et al., 2013). Subsequently, it was demonstrated that the output of H2O2 can be reduced by over 30% in the presence of a cognate biological reducing system Fd/FR/N (Zhang et al., 2013). In addition, when cADO was fused with cognate FR and Fd in a specific order, it displayed a 3-fold increase in activity relative to native cADO (Wang et al., 2014).
Crystal structures of ADO from cyanobacterium Prochlorococcus marinus (Pm) MIT9313 (wild-type and two single-point mutants) were reported (Khara et al., 2013). Structures showed that PmADO adopted an α-helical folding, with two iron atoms coordinated by histidines and carboxylate ligands (Khara et al., 2013; Schirmer et al., 2010). A long-chain ligand, which was subsequently identified as a mixture of fatty acid molecules (Khara et al., 2013), was observed in the vicinity of the di-iron center in these structures (Khara et al., 2013). This pocket accommodating fatty acid molecules were suggested to be the substrate channel. However, due to the structural difference between fatty acid and the real substrate fatty aldehyde, the interaction between ADO and the substrate remains elusive. In addition, their structural resemblance indicated that they may represent one similar state of the reaction process, thus structures of cADO at different dynamic stages during its reaction cycle will be of great help for understanding its detailed catalytic process.
To gain insights into the reaction catalyzed by cADO, we determined structures of wild type ADO from Synechococcus elongates PCC7942 (Se) with different treatment (WT0, no treatment; WT1, co-crystallized with iron; WT-HP, soaked in H2O2), and solved the structures of two SeADO mutants (Y122F and F86YF87Y). Comparison of all our structures revealed that their overall conformations are closely resembling each other, while the conformation of their active sites are largely different. Analysis of all these structures indicated that they may represent different states of the enzyme in the reaction cycle. Except WT-HP structure, which is similar to that of the PmADO, the other four structures exhibit different conformations in their active sites that were not described previously.
Results and Discussion
Overall structure of SeADO
Statistics of data processing, structure refinement and summary of different structures of SeADO
Cell parameters (Å)
a = 61.7
a = 64.73
a = 61.7
a = 62.0
a = 61.7
b = 61.7
b = 64.71
b = 61.8
b = 62.1
b = 61.9
c = 110.5
c = 101.4
c = 124.9
c = 125.0
c = 124.8
No. of unique refls
Rmerge (%overall/outmost shell)
I/σ(I) (overall/outmost shell)
Completeness (% overall/outmost shell)
Bond lengths (Å)
Bond angles (º)
Mean B value (Å2)
Iron incorporation before crystallization
Treatment of crystals
No. of molecules in a.u.
No. of di-iron center in a.u.
1 + 0.5
Location of di-iron center
Molecules A and B
Molecules A and B
Lred (Fatty alcohol)
Lred (Fatty alcohol)
Lred (Fatty alcohol)
Lox (Fatty acid)
Either with or without di-iron cluster, the overall structures of all these molecules are similar, with the exception of one specific helix conformation. In the structures of Y122F and molecule A of other threes (WT1, F86YF87Y and WT-HP), the ADO molecule adopts an all-helical folding, comprising of eight α-helices (H1–H8) that form a compact structure (Fig. 1A and 1B). Two iron atoms (Fe1 and Fe2) are surrounded by a four-helix bundle, which is composed of H1 (S16–M45), H2 (R50–L74), H4 (V103–I126) and H5 (A131–D160).
Unlike other structural elements in SeADO, helix H5 is unique in representing two distinct conformations in different structures. In the structures mentioned above, H5 exists as a long helix (Fig. 1C), where two residues (Glu144 and His147) in the second EX2H motif are coordinated or close to iron atoms. By contrast, the helix H5 is unwound in the middle in the structures of WT0 and molecule B of other three structures (WT1, F86YF87Y and WT-HP), forming two short helices (H5A and H5B) that are connected by a loop L5 (Fig. 1D and 1E). The helix to loop transition results in a different conformation for a number of amino acids (from residue 144 to 150), thus the two iron-coordinating residues (Glu144 and His147) move away from the di-iron site (Fig. 1E). As a result, most of our SeADO structures with the L5 conformation lose their metal atoms. The only exception is molecule B in the F86YF87Y structure, which contains the di-metal cluster while exhibiting the L5 conformation. However, the relatively weak electron density indicates a low occupancy of the two iron atoms in this molecule.
Iron content of wild type and mutants of SeADO measured by ICP-OES assay
Number of iron atom per ADO molecule
Percentage of iron content compared with wild type 3
LB medium + Fe
Wild type 1
0.28 ± 0.004
17.7% ± 0.25%
Wild type 2
0.67 ± 0.02
42.4% ± 1.27%
Wild type 3
1.58 ± 0.09
100% ± 5.7%
1.58 ± 0.06
99.9% ± 3.81%
1.69 ± 0.12
107.38% ± 7.7%
1.30 ± 0.26
82.29% ± 16.78%
1.60 ± 0.29
101.32% ± 18.17%
1.58 ± 0.03
100.25% ± 1.63%
1.48 ± 0.01
93.96% ± 0.76%
1.53 ± 0.08
96.64% ± 5.17%
1.35 ± 0.23
85.50% ± 14.71%
Substrate channel in SeADO
To identify the precise nature of these ligands, the ligands extracted from SeADO protein samples were analyzed by GC-QqQ-MS/MS. The results showed that the majority of the ligands examined were a mixture of long-chain fatty acids, and the dominant of the remaining components was identified as a C18 fatty alcohol (Fig. 2D). Following the analysis of the extracts together with the observed electron densities, we built a long-chain fatty acid (Lox) into the structure of WT-HP, and a long-chain fatty alcohol (Lred) into three other structures. On the basis of these observations, we propose that the Lred ligand may represent the real substrate, namely fatty aldehyde, because of the similarity in their chemical structures.
SeADO structures that represent the different states during catalytic reaction
The substrate-free structure of Y122F represents the initial state of the reaction
Coordinating ligands and distances of two iron atoms in SeADO structures
Water2 (only in F86YF87Y)
Water2 (only in F86YF87Y)
The WT1 structure represents the state of ADO bound with substrate
In the WT1 structure, a Lred ligand, which was proposed to mimic the substrate aldehyde, was modeled in the substrate channel according to the electron density (Fig. 2B). The ligand approaches the di-iron cluster from the opposite side of His147, with its head-group binding directly to the Fe2 atom. This coordination mode suggests that Fe2 is the preferred iron for substrate binding. The water molecule, located near the di-iron site in the Y122F structure, is displaced by the head group of the Lred ligand. The residue Glu144, which is located in helix H5, changes its rotamer and is not coordinated with either iron atoms. In addition, E115 alters its mode of Fe2 coordination, from bidentate to monodentate, while other four iron-coordinating residues stay unchanged, resulting in a distorted tetrahedral coordination of both Fe1 and Fe2 (Fig. 5B and Table 3). The structure of WT1 is likely to represent the state where the enzyme is bound with substrate. The structural comparison between WT1 and Y122F implied that the swing of Glu144 might be induced by ligand binding, thus enabled us to visualize the conformational change coupled with substrate binding in cADO active site for the first time.
The F86YF87Y structure represents the state of ADO where the oxygen path is formed
The WT-HP structure represents an intermediate state of ADO where oxygen reacts with the substrate
To explore the effects of H2O2 on ADO activity, a crystal of wild type SeADO was soaked in H2O2, and its structure (WT-HP) was solved. No electron density was visible for the H2O2, probably due to its unstable nature. A Lox ligand closely resembling a fatty acid molecule was built according to the electron density observed within this structure (Fig. 2A), which is nearly identical to the previously solved structure of PmADO (PDB code 2OC5). In the WT-HP structure, the helix conformation of H5 was restored, and the two coordinating residues His63 and His147 approach the active site and ligate the Fe1 and Fe2 atoms, respectively. The water molecule that bridges the two iron atoms in the F86YF87Y structure moves towards Fe1 and is not coordinated with Fe2 in the WT-HP structure. The two iron ions are bridged by one carboxylate ligand E60, the Lox ligand and an oxo group. Both iron atoms show saturated coordination with an octahedral geometry (Fig. 5D and Table 3). The higher coordination numbers of di-iron center, together with the fact that the crystal was obtained after being treated with strong oxidizer H2O2, lead to the speculation that the WT-HP structure represents an oxidized form of ADO. Furthermore, the coordination mode of two iron atoms in WT-HP structure shows a similar pattern to that observed for the R2 protein of ribonucleotide reductase (RNR-R2), in which the oxo bridge and the coordinated water exist at the di-iron site in the oxidized structure, while both are absent in the reduced structure (Logan et al., 1996). The superimposition of the F86YF87Y and WT-HP structures revealed that the hole at the protein surface, which is caused by the distortion of helix H5, exposes the di-iron sites and the oxo bridge (Fig. 6C and 6D). This observation supports our hypothesis that the identified hole serves as dioxygen channel. We proposed that the Lox ligand is likely to be the analog of the possible intermediate product hemiacetal, thus the WT-HP structure may represent the intermediate state in reaction of cADO, following the entering of dioxygen and its reaction with the substrate.
The WT0 structure represents the inactive state of ADO
The WT0 structure is characterized by losing the di-iron cluster and by exhibiting a distorted conformation of helix H5. This structure is likely to represent the inactive state of SeADO, as it has lost its cofactor iron. We assume that if the loop conformation of helix H5, as revealed in the F86YF87Y structure, was not restored into a helical structure at the appropriate time, the solvent-accessible di-iron center may become unstable. Whereas the instability of the di-iron cluster further promotes the loop conformation, and thus results in a complete loss of the di-iron cluster, as shown in the WT0 structure (Fig. 1D and 1E). The structural flexibility of SeADO provides an explanation for its low iron occupancy, and may be in part responsible for the low enzymatic activity observed for cADO. To the best of our knowledge, the observation that the conformational change of one specific helix towards the loop correlates with the loss of the di-iron cluster is the first case described for this superfamily.
The proposed process of the catalytic reaction by cADO
Among all the iron-coordinating residues, Glu144 remarkably alters its conformation among structures. A similar case was observed in other members of the di-iron protein family, such as RNR-R2 and the hydroxylase component of soluble methane monooxygenase. Both possess an essential iron-coordinating glutamic acid, which exhibits distinct conformations during their respective reactions (Kolberg et al., 2004; Sazinsky and Lippard, 2006; Whittington and Lippard, 2001). Apart from the structural information, our biochemical analysis showed that the E144A mutant has only 5% activity remaining, with similar iron content as measured for the wild type (Fig. 4 and Table 2). We therefore suggest that Glu144 plays an important role in catalysis.
In summary, we determined five crystal structures of SeADO and revealed novel structural features around their active site. Snapshots of these consecutive states allow us to visualize the morphing of the active site during the reaction. Analysis of our structural and biochemical data highlights a number of important structural features that can influence the catalytic process and activity of cADO, including the conformational switch of the central part of helix H5, and the flexibility of residue Glu144. Together, these results provide new structural insights into the catalytic mechanism and allow us to propose a possible catalytic process of cADO, thus provides crucial information required for developing new strategies to improve its enzymatic activity, with the ultimate goal of producing fuel-grade alk(a/e)nes in a renewable and sustainable manner.
Materials and Methods
Protein expression and purification
The codon-optimized gene encoding cADO from Synechococcus elongates PCC7942 was synthetized and cloned into the expression vector pET-28a(+) (Novagen). The constructs for mutant of SeADO were generated using the QuikChange site-directed mutagenesis kit (Stratagene). The constructs were confirmed by DNA sequencing and transformed into E. coli BL21(DE3). Protein expression was induced by adding isopropyl β-D-thiogalactoside (IPTG) to a final concentration of 1 mmol/L. After being shaken at 37°C for approximately additional 4 h, cultures were harvested by centrifugation at 6000 ×g at 4°C for 15 min. To obtain the iron-bound cADO proteins, 2 mmol/L (NH4)2SO4FeSO4·6H2O was added in medium.
The cell pellet was homogenized in buffer containing 20 mmol/L Tris-HCl pH 7.8 and 300 mmol/L NaCl (buffer A), and sonicated. Cell debris was removed by centrifugation at 40,000 ×g for 30 min. The supernatant was collected and loaded onto Ni-IDA resin (Chelating sepharose FF, GE Healthcare) and rinsed with buffer A 20 mmol/L imidazole. The protein was eluted from the affinity resin with buffer A containing 250 mmol/L imidazole. The eluted fraction was concentrated and further purified by gel filtration on Superdex 75 (GE Healthcare) with elution buffer 50 mmol/L Hepes pH 7.2, 150 mmol/L NaCl. The purified protein was concentrated to 10 mg/mL for crystallization.
Ferredoxin reductase and ferredoxin derived from Synechococcus elongates PCC7942 were constructed and purified as previously described (Zhang et al., 2013).
Crystallization and X-ray data collection
Crystallization trials were carried out at 18°C by mixing equal volume of protein and reservoir solution using the sitting-drop vapor diffusion method. Crystals of WT0 structure was harvested in condition of reservoir solution of 0.2 mol/L Magnesium chloride hexahydrate, 0.1 mol/L Tris hydrochloride pH 8.5, 30% (w/v) PEG 4,000. To obtain the iron-bound crystals, 4 mmol/L ferrous ammonium sulfate was added to SeADO protein solutions right before crystallization. The crystals of WT1 structure were harvested in reservoir solution containing 0.2 mol/L L-proline, 0.1 mol/L Hepes pH 7.1, 25% (w/v) PEG1500. The crystals of Y122F and F86YF87Y were grown in the same solution with 0.02 mol/L adenosine-5’-triphosphate disodium salt hydrate added. The crystal of WT1 was soaked in 10 mmol/L H2O2 for 30 min before being flash frozen to obtain the crystal of WT-HP.
Crystals were flash-cooled in a nitrogen-gas stream at 100 K for data collection. Diffraction data of Y122F and F86YF87Y were collected on beamline BL17U at Shanghai Synchrotron Radiation Facility. Diffraction data of WT0 and WT1 were collected utilizing Rigaku RAXIS IV image plate detector at Institute of Biophysics (Chinese Academy of Sciences). Diffraction data of WT-HP were collected on BL17A at Photo Factory, Japan. Diffraction data were processed and scaled with HKL-2000 package (Otwinowski and Minor, 1997).
Structure determination and refinement
The WT0 model, which subsequently make a searching model for molecular replacement of other four structures, was solved by molecular replacement with Phaser_MR (Mccoy et al., 2007) using the PmADO structure (PDB code 2OC5) as a searching model. The model were rebuilded by AutoBuild in PHENIX package (Adams et al., 2010), following subjected to refinement by Phenix.refine (Adams et al., 2010) and COOT (Emsley et al., 2010). Figures of the structures were prepared by Pymol (DeLano Scientific, LLC). A summary of data collection and structure refinement statistics is given in Table 1.
Metal content determination and enzyme activity assay
The metal contents of the wild-type enzyme and mutants were determined by inductively coupled plasma optical emission spectrometer (ICP-OES), PerkinElmer 5300DV.
N-heptanal was selected as the substrate to measure the activity of cADO proteins. Assays were performed in 1.5 mL gastight vials with a total volume of 500 μL, and reactions were conducted in 100 mmol/L HEPES buffer pH 7.2 containing 100 mmol/L KCl and 10% glycerol, with 15 μmol/L protein samples, 2 mmol/L of n-heptanal in final 4% DMSO, 30 μg/mL ferredoxin, 0.04 U/mL ferredoxin reductase, 800 μmol/L NADPH, 60 μmol/L ferrous ammonium sulphate. After being enrolled of all the components, reactions were shaken at 220 rpm at 37°C for 30 min. To determine the amount of n-hexane produced, a sample of the headspace was collected using a gastight sample lock Hamilton syringe and analysed by Shimadzu GC-2010 with DB-5 column. The amount of n-hexane produced was quantified by a standard curve of known concentrations of n-hexane.
For GC analysis the flow rate of the nitrogen carrier gas was 1.1 mL/min and the inlet temperature was maintained at 220°C. Injections were made in split mode with a split ratio of 2:1 and a total flow of 2 mL/min. The oven temperature was held at 40°C for 3 min and then increased to 120°C at 10°C/min, and finally maintained at 120°C for 2 min. The FID detector was at 250°C with a continuous flow of H2 at 40 mL/min and air at 400 mL/min. Chromatographic data were analyzed using the associated software.
Determination of ligand(s) using GC-QqQ-MS/MS.
2 g of purified SeADO protein was acidified using 1 mol/L HCl to pH 3.0 and extracted with ethyl acetate. The organic layer was collected and dried by passing through MgSO4. The solvent was evaporated by rotary evaporator and nitrogen (N-EVAP) to 100 μL. The sample was trimethylsilylated for analysis using BSTFA + 1% TMCS (Sigma). All spectra were recorded on an Agilent 7890A GC system connected to an Agilent 7000B triple quadrupole MSD with electron impact ionization mode. A 1-μL portion of the derivatized extract was injected in splitless mode onto the column. The column used was a DB-5ms (30 m × 250 μm × 0.25 μm film thickness, Agilent J&W ScientiWc, USA) fused silica capillary column. Injector temperature was 280°C and the oven program was as follows: oven temperature was held at 60°C for 2 min and then increased to 240°C at 10°C/min and then increased to 300°C at 20°C/min finally maintained at 300°C for 5 min. Helium was used as the carrier gas for GC at a flow rate of 1.0 mL/min, and the inlet temperature was maintained at 280°C with splitless. Chromatographic data were acquired and processed using MassHunter Workstation Quantitative Software.
We would like to thank Yi Han and Shengquan Liu at the Institute of Biophysics, CAS and the staffs at Shanghai Synchrotron Radiation Facility and Photo Factory, Japan, respectively, for help during X-ray data collection. We also thank Wei Shao at Beijing Center for Physical and Chemical Analysis for help with the GC analysis, Zhen Xue at Institute of Botany, CAS for providing help for the GC-QqQ-MS/MS analysis, and Hongzhi Zhang at Institute of Geographic Sciences and Natural Resources Research, CAS for technical support with the metal content analysis by ICP-OES. This work was supported by the National Basic Research Program (973 Program) (Nos. 2011CBA00902 and 2011CBA00907), Strategic Priority Research Program of the Chinese Academy of Sciences (XDB08020302), National Natural Science Foundation of China (Grant Nos. 31021062 and 31170765), and CAS Cross and Cooperation Team for Scientific Innovation (Y31102110A).
COMPLIANCE WITH ETHICS GUIDELINES
Chenjun Jia, Mei Li, Jianjun Li, Jingjing Zhang, Hongmei Zhang, Peng Cao, Xiaowei Pan, Xuefeng Lu, and Wenrui Chang declare that they have no conflict of interest.
This article does not contain any studies with human or animal subjects performed by the any of the authors.
- Lindqvist Y, Huang W, Schneider G, Shanklin J (1996) Crystal structure of delta9 stearoyl-acyl carrier protein desaturase from castor seed and its relationship to other di-iron proteins. EMBO J 15:4081–4092Google Scholar
- Logan DT, deMare F, Persson BO, Slaby A, Sjoberg BM, Nordlund P (1998) Crystal structures of two self-hydroxylating ribonucleotide reductase protein R2 mutants: structural basis for the oxygen-insertion step of hydroxylation reactions catalyzed by diiron proteins. Biochemistry 37:10798–10807CrossRefGoogle Scholar
- Pandelia ME, Li N, Norgaard H, Warui DM, Rajakovich LJ, Chang WC, Booker SJ, Krebs C, Bollinger JM (2013) Substrate-triggered addition of dioxygen to the diferrous cofactor of aldehyde-deformylating oxygenase to form a diferric-peroxide intermediate. J Am Chem Soc 135(42):15801–15812CrossRefGoogle Scholar
- Yang YS, Baldwin J, Ley BA, Bollinger JM, Solomon EI (2000) Spectroscopic and electronic structure description of the reduced binuclear non-heme iron active site in ribonucleotide reductase from E. coli: comparison to reduced delta(9) desaturase and electronic structure contributions to differences in O-2 reactivity. J Am Chem Soc 122:8495–8510CrossRefGoogle Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited.