Probing the specificity of CYP112 in bacterial gibberellin biosynthesis

Abstract

Biosynthesis of the gibberellin A (GA) plant hormones evolved independently in plant-associated fungi and bacteria. While the relevant enzymes have distinct evolutionary origins, the pathways proceed via highly similar reactions. One particularly complex transformation involves combined demethylation and γ-lactone ring formation, catalyzed in bacteria by the cytochrome P450 CYP112 in three individual steps, which involves large structural changes in the transition from substrate to product, with further divergence in the recently demonstrated use of two separate mechanistic routes. Here the substrate specificity of the isozyme from Erwinia tracheiphila, EtCYP112, was probed via UV-Vis spectral binding studies and activity assays with alternate substrates from the GA biosynthetic pathway. EtCYP112 tightly binds its native substrate GA₁₂ and the intermediates GA₁₅ and GA₂₄, as well as the methylated derivatives of GA₁₂ and GA₁₅. It however only poorly binds methylated GA₂₄, its GA₉ final product and the C-20 carboxylate side-product GA₂₅. These distinct affinities are consistent with the known reactivity of EtCYP112. However, while it binds to the immediately preceding pathway metabolite GA₁₂-aldehyde and even earlier oxygenated ent-kaurene precursors, EtCYP112 only reacts with GA₁₂-aldehyde, and not the earlier ent-kaurene derived metabolites. But even with GA₁₂-aldehyde conversion is limited to the first two steps and the final combined demethylation and γ-lactone ring forming reaction is not catalyzed. Thus, CYP112 has evolved specificity at the catalytic rather than substrate binding level to enable its role in GA biosynthesis.

Introduction

Gibberellins (GAs) are essential hormones in plants, and certain plant-associated fungi and bacteria also have acquired the ability to produce GAs in order to manipulate their host plants (1). The bacterial gibberellin biosynthetic pathway was elucidated very recently (Figure 1)(2–4). The steps of the individual pathways are quite similar between plants, fungi and bacteria. Nevertheless, at the enzyme level it is evident from the low sequence identity that GA biosynthesis evolved separately in plants, fungi and bacteria (1–4). One hallmark transformation in GA biosynthesis is the conversion of 20-carbon GAs to the 19-carbon GAs, involving the combined loss of a methyl group and the formation of an intramolecular γ-lactone bridge. Plants mediate this complex multi-step reaction with 2-oxoglutarate dependent dioxygenases (2ODDs) termed GA 20-oxidase (GA20ox) and, although this transformation is mediated by cytochromes P450 (CYPs) in both fungi (CYP68B) and bacteria (CYP112), these two CYP sub-families share less than 15% sequence identity (1, 4). Moreover, while the relevant fungal CYP68B is membrane-bound, the bacterial CYP112 is soluble. Nonetheless, the chemical transformations catalyzed by the three individual enzymes appear to be identical, involving stepwise oxidation of C-20. Specifically hydroxylation of GA₁₂ at this position to form GA₁₅, followed by further oxidation to the aldehyde equivalent GA₂₄ and, finally, loss of C-20 as CO₂ with formation of the intramolecular γ-lactone bridge that yields GA₉ (5–9)(Figure 1). While the fungal enzyme does not react with the intermediates GA₁₅ or GA₂₄ and will only release them under sub-optimal conditions, the bacterial and plant enzymes release their intermediates more readily, and can use GA₁₅ and GA₂₄ as substrates to produce GA₉ (1, 4, 7, 9). Although all of these oxidases produce the C-20 carboxylate GA₂₅, this seems to be a side-product rather than intermediate, as it cannot be transformed to GA₉ by any of them.

Figure 1.

Bacterial gibberellin operon (A) Schematic representation of the GA biosynthetic operon. Arrows indicate the direction of translation, abbreviations for the genes are CYP, cytochrome P450; Fd, ferredoxin; SDR, short-chain alcohol dehydrogenase/reductase; IDS, isoprenyl diphosphate synthase; CPS, copalyl diphosphate synthase; KS, ent-kaurene synthase; and IDI, isopentenyl diphosphate isomerase. (B) Catalyzed reactions, with emphasize on the sequential oxidation and elimination of C-20 from GA₁₂, to produce GA₉, by CYP112.

The recently discovered bacterial CYP112 offers distinct advantages for study as, in contrast to the 2ODDs utilized by plants, CYPs change their spectral properties upon substrate binding. Unbound CYPs have a characteristic absorbance maximum around 420 nm, called the Soret-band, which shifts to around 390 nm upon binding of their substrates, and this spectral shift can be used to measure affinity (10). While the fungal CYP68B presumably undergoes a similar spectral shift, it is membrane bound, which significantly complicates expression and purification. Moreover, the fungal CYP68B does not release its reaction intermediates or exhibit reactivity with these, at least in the context of incubation with cultures of Gibberella fujikuroi to which these studies have been limited (11, 12). In contrast, the CYP112 from Erwinia tracheiphila (EtCYP112) can be readily recombinantly expressed in Escherichia coli, purified and assayed in vitro, with characterization of its reaction mechanism via ¹⁸O labeling experiments recently reported (9). Specifically, EtCYP112 variably employs two diverging reaction mechanisms, with some significant difference in intermediates – i.e., either the open or closed (δ-lactone and lactol, respectively) forms of the intermediates GA₁₅ and GA₂₄ – that reconverge at a transient cyclic anhydride C-20 geminal-diol intermediate (Figure 1B). Thus, EtCYP112 is able to react with all potential intermediates, including those not utilized by either of the functionally equivalent enzymes from fungi or plants. In particular, as the fungal CYP68B has not been shown to react with any intermediate (11), while the plant GA20ox does not react with the closed lactone of GA₁₅ (7). This sparked interest as to the substrate specificity of EtCYP112, which was investigated here by UV-Vis monitored binding studies and in vitro activity assays with alternate substrates. The results demonstrate that, although EtCYP112 is able to bind multiple upstream metabolites, it only reacts with GA₁₂-aldehyde. However, the resulting transformation is limited to production of GA₁₅-aldehyde and GA₂₄-aldehyde.

Experimental

Cloning, Expression and Purification of EtCYP112 and EtFdR

Cloning of EtCYP112 and EtFdR was described previously (9). Plasmids containing either EtCYP112 or EtFdR were transformed into E. coli BL21-Star (Invitrogen) for expression. Starter cultures of 25 ml liquid NZY medium (10 g/L NaCl, 10 g /L casein, 5 g/L yeast extract, 1 g/L MgSO4 (anhydrous), pH 7.0), including 50 μg/mL carbenicillin, were inoculated with 3 individual colonies and incubated at 18 °C for 2 days with constant shaking at 200 rpm. Theses starter culture were used to inoculate 1 L of NZY with 50 μg/mL carbenicillin, which was kept at 18 °C with constant shaking at 200 rpm until an OD600 of 0.8 was reached. At this point the culture was induced with 1 mM IPTG and, for cultures expressing EtCYP112, 1 mM aminolevulinic acid, 1 mM riboflavin and 0.1 mM FeCl₃ were added. After 36 hours, the cells were harvested by centrifugation at 5000 x g for 20 min. The EtCYP112 cell pellet was re-suspended in 10 mL 25 mM 3-(N-morpholino)-2-hydroxypropanesulfonic acid (MOPSO), pH 7.2, 10% (v/v) glycerol), while the EtFdR cell pellet was re-suspended in 10 mL 50 mM phosphate, pH 7.5, 10% glycerol. Both cell suspensions were homogenized with an EmulsiFlex C-3 (Avestin, Canada). The lysate was centrifuged at 17,000 x g for 30 min. The cleared lysate was then added to 1 mL of Ni-NTA Agarose (Qiagen), with the corresponding buffer containing 1 M imidazole added to reach a final concentration of 20 mM imidazole. Columns were incubated with gentle shaking at 4 °C for 1 h. The column was washed with 5 mL buffer containing 20 mM imidazole and then 5 mL with 60 mM imidazole, before being eluted with the corresponding buffer with 250 mM imidazole. The eluted protein was then transferred to dialysis tubing with a molecular weight cutoff of 15 kDa and dialyzed 3 times against 1L of 10 mM MOPSO, pH 7.2, 5% (v/v) glycerol in the case of EtCYP112 when used for spectral binding studies or 25 mM phosphate, pH 7.5, 10% glycerol in the case of EtFdR, as well as EtCYP112 when used for enzymatic activity assays. The resulting enzymes were either used fresh or stored at −80 °C after flash freezing by brief immersion in liquid nitrogen of 0.5 mL aliquots in 1.5 mL Eppendorf tubes. Only fractions with ≥95% purity, as assessed by SDS-PAGE analysis, were used. The concentration of EtCYP112 was measured by optical absorbance at 280 nm, using the calculated extinction coefficient of 43,555 M⁻¹ cm⁻¹.

Enzyme assays

Enzyme assays under regular atmosphere or ¹⁸O₂ were performed as described previously (9). Briefly, 0.01 mL each of purified EtCYP112, spinach ferredoxin (Sigma-Aldrich), and EtFdR (all at 1 mg/mL in 25 mM phosphate, pH 7.5, 10% glycerol) were diluted with 1.5 mL of 25 mM phosphate, pH 7.5, 10% glycerol in 4 mL clear glass vials, and 20 μM of the specified substrate and 60 μM NADPH were added. The vial was sealed with a Teflon-septum cap and incubated for 12 h at 25 °C. For ¹⁸O₂ labeling experiments these vials were first flushed with nitrogen for 5 min, then 0.05 mL each of purified EtCYP112, spinach ferredoxin (Sigma-Aldrich), and EtFdR (all at 1 mg/mL in 25 mM phosphate, pH 7.5, 10% glycerol), and 1.7 mL of 25 mM phosphate, pH 7.5, 10% glycerol were added using a syringe. ¹⁸O₂ (Sigma-Aldrich, 97% labeled; max loaded pressure less than 2.4 bar) was added from the gas cylinder via a needle through the septum until the pressure equilibrated between the tank and the vial. The reaction was started by the addition of substrate to a final concentration of 40 μM. Assays from ¹⁸O₂ labeling experiments were analyzed with an Agilent 6540 GC coupled to a Waters GCT Premier mass spectrometer and a 3900 Saturn GC coupled to a 2100T ion trap mass spectrometer (Varian, Palo Alto, CA, USA), enzyme assays under regular atmosphere were only analyzed with the 3900 Saturn GC coupled to a 2100T ion trap mass spectrometer. Separation was achieved over a DB-5MS column (30 m, 250 μm, 0.25 μm) with a previously described temperature gradient (3). For further conversion of the GA₁₂-aldehyde products, two-thirds of the organic solvent extract from the primary in vitro enzyme assay were split off and, after evaporation, dissolved in 300 μL of methanol:DMSO (1:1, v/v) without methylation. This material fed to the GA operon SDR from E. tracheiphila (EtSDR), much as previously described (2). Briefly, this was done via whole-cell feeding for three days to a recombinant culture of E. coli expressing EtSDR, with subsequent extraction and analysis by GC-MS.

UV-Vis spectroscopy

Purified EtCYP112 in 10 mM MOPSO, pH 7.2, 5% (v/v) glycerol was diluted in the same buffer so that absorbance at 419 nm was 0.6 – 0.7 and spectra from 300 to 700 nm were collected on a Cary 50 UV-VIS spectrometer (Varian) at a scan rate of 60 nm per minute. Substrates were added stepwise from a stock solution of 0.1 or 1 mg/mL dissolved in methanol. Methanol was added to EtCYP112 as a control for spectral shifts originating from the solvent alone, although no changes in the UV-Vis spectrum were observed with up to 8% methanol (v/v). K_d for individual binding curves were calculated using plots of the added substrate concentration versus the induced spectral shifts, the curve was then fitted using the model for tight binding ligands with the following formula.

$$A_{\mathit{obs}}=\left(\frac{A_{\mathit{max}}}{2\ast E_t}\right)\ast (\mathrm{S}+E_t+K_d)-\sqrt{{(S+E_t+K_d)}^2-(4\ast S\ast E_t)}$$

The mean and standard error were calculated from three individual experiments and reported in Table 1.

Table 1.

Et CYP112 affinity for GA metabolites and methylated derivatives.

Substrate	Kd [μM]	High Spin State	α-band Absorbtion
GA12	2.6 ± 0.3	> 90%	decreased
GA15 (open lactone)	0.11 ± 0.02	> 90%	decreased
GA15 (closed lactone)	0.6 ± 0.1	> 90%	decreased
GA24	0.32 ± 0.07	80 – 90%	decreased
GA9	n.a.	< 5%	unchanged
GA25	n.a.	10 – 20%	unchanged
GA12-aldehyde	2.0 ± 0.3	> 90%	decreased
MeGA12	14 ± 3	50 – 60%	decreased
MeGA15	0.11 ± 0.03	80 – 90%	decreased
MeGA24	n.a.	5 – 10%	unchanged
ent-kaurenoic acid	2.7 ± 0.6	50 – 60%	decreased
methyl-ent-kaurenoate	0.5 ± 0.1	10 – 15%	decreased
ent-7α-hydroxy kaurenoic acid	3.4 ± 0.6	50 – 60%	decreased
ent-kaurenal	18 ± 2	5 – 10%	decreased
ent-kaurenol	27 ± 7	20 – 30%	decreased
ent-kaurene	n.a.	< 5%	unchanged

The K_d and the standard error were calculated from 3 independently performed titration experiments.

Results

The previously described N-terminal His-tagged EtCYP112 construct was found to be expressed to about 15 mg/ml in E. coli and easily purified by nickel affinity chromatography. UV-Vis spectral analysis found that the eluted preparation exhibited a Soret-band peak at 424 nm. Following dialysis, the substrate-free enzyme showed a more characteristic Soret-band maximum at 419 nm. Upon reduction and binding of carbon monoxide, purified EtCYP112 also exhibited the eponymous absorption peak at 450 nm (Supplemental Figure S1).

Upon addition of substrate or intermediates to purified EtCYP112, typical Type I binding spectra were observed. These are characterized by an almost complete shift of the Soret-band peak to 391 nm, as well as decreased α-band absorption at 570 nm. Given the evident high-affinity of EtCYP112 for these, their K_d values were calculated from Soret-band shift data using the equation that applies to such tight-binding ligands (13). K_d values for GA₁₂, GA₁₅ (open lactone, added from a stock with 1 M KOH), GA₁₅ (closed δ-lactone) and GA₂₄ were calculated to be 2.6, 0.11, 0.6 and 0.3 μM respectively (Figures 2, S2 and Table 1). By contrast, the GA₂₅ side-product and the GA₉ final product only shift the Soret-band from 419 to 391 nm about 10–20% or 5% respectively, without any decrease of the α-band.

Figure 2.

Spectral titration of EtCYP112 with substrates and products. UV-Vis spectra of EtCYP112 in the absence and with increasing concentrations of (A) GA₁₂, (B) GA₁₅ (closed lactone), (C) GA₁₅ (open lactone), (D) GA₂₄, (E) GA₂₅ and (F) GA₉.

Previously, to elucidate the mechanism of CYP112 the carboxylates at C-7 and C-19 were methylated and the resulting compounds tested as substrates. While MeGA₁₂ and MeGA₁₅ (δ-lactone form) are converted by EtCYP112 to GA₉, MeGA₂₄ was not (9). Binding of these methylated GAs also was investigated here by UV-Vis spectral analysis (Table 1, Figures 3 and S3). Methylation of GA₁₂ (MeGA₁₂) reduced its affinity for EtCYP112, with a 5-fold higher K_d observed than that measured for free GA₁₂. By contrast, MeGA₁₅ (closed δ-lactone) is bound about 6-fold more tightly by EtCYP112 than the analogous form of free GA₁₅, and its K_d is more similar to the open form of GA₁₅. MeGA₂₄ on the other hand only induces a partial shift of the Soret-band, the binding curve could not be fit to the model for tight binding ligands, and there was no decrease in the absorption of the α-band.

Figure 3.

Spectral titration of EtCYP112 with methylated substrates, as well as immediate precursor GA₁₂-aldehyde. UV-Vis spectra of EtCYP112 in the absence and with increasing concentrations of (A) MeGA₁₂, (B) MeGA₁₅ (closed lactone), (C) MeGA₂₄ and (D) GA₁₂-aldehyde.

Given the ability of EtCYP112 to act upon C-7 methyl ester derivatives, it seemed possible that it further might exhibit activity with the immediate precursor to GA₁₂ – GA₁₂-aldehyde – in which C-7 is an aldehyde rather than a carboxylate. Indeed, GA₁₂-aldehyde was found to be tightly bound by EtCYP112 in UV-Vis spectral studies, with a K_d of 2.0 μM (Table 1 and Figures 3 and S3). When fed to the previously reported reconstituted enzymatic system, GA₁₂-aldehyde is transformed into two products, although conversion was not complete even in the presence of a 10-fold excess of NADPH (Figure 4). The two products exhibited molecular ions with 30 mass units less than those of one of the corresponding C-7 carboxylate containing GAs, consistent with the expected mass difference between an aldehyde and a (methylated) acid function (note that all of the compounds examined here were methylated for analysis by GC-MS)(Figure 5). Based on the ability of the short-chain alcohol dehydrogenase/reductase (SDR) found in the bacterial GA biosynthetic operon to convert GA₁₂-aldehyde to GA₁₂ (2–4), the observed products were incubated with E. coli expressing this SDR from E. tracheiphila (EtSDR). This led to conversion of one product to GA₂₄, indicating that this was GA₂₄-aldehyde. These had very similar retention times (Figure 4). Another product exhibited a similarly close retention time to GA₁₅, and has a very similar mass fragmentation pattern to GA₁₅ as well, suggesting that it is GA₁₅-aldehyde. This tentative assignment was further supported by ¹⁸O labeling experiments, as the mass of the putative GA₁₅-aldehyde was increased by 2 mass units, while the mass of GA₂₄-aldehyde was increased by 2 mass units (about 30%) or 4 mass units (about 66%)(Figure 5). This labeling pattern is similar to that previously reported for the production of GA₂₄ from GA₁₂ (9, 13). The inability of EtSDR to convert the putative GA₁₅-aldehyde to the corresponding C-7 carboxylate (i.e., GA₁₅) is presumably due to the presence of the δ-lactone ring. However, no peak exhibiting the expected molecular mass and/or retention time for GA₉-aldehyde was observed.

Figure 4.

Conversion of GA₁₂-aldehyde by EtCYP112. GC-MS-chromatograms of in vitro assays with EtCYP112 and GA₁₂-aldehyde, incubation of the products with EtSDR and GA₁₅, GA₂₄ and GA₉ standards.

Figure 5.

Mass spectra of GA₁₅- and GA₂₄-aldehyde. MS spectra of standard compounds and the unknown products of enzyme assays with EtCYP112 and GA₁₂-aldehyde, either under regular atmosphere or ¹⁸O₂. Presented mass spectra were acquired with a Varian 2100T ion trap mass spectrometer.

GA biosynthetic intermediates prior to GA₁₂-aldehyde do not have the characteristic gibberellane 6-5-6-5 ring structure, but instead a 6-6-6-5 ent-kaurane ring structure. While addition of the olefin precursor ent-kaurene to EtCYP112 only induces a small shift of the Soret-band, with no decrease in α-band absorption, surprisingly, the oxygenated derivatives induced typical type I binding spectra. In particular, ent-kaurenoic acid, its methylated derivative (methyl ent-kaurenoate), ent-7α-hydroxy kaurenoic acid, ent-kaurenal and ent-kaurenol, for which the K_d values were calculated to be 2.7, 0.5, 3.4, 18 and 27 μM, respectively (Table 1 and Figures 6 and S4). However, no products were observed from assays with these compounds (data not shown).

Figure 6.

Spectral titration of EtCYP112 with kaurane precursors. UV-Vis spectra of EtCYP112 in the absence and with increasing concentrations of (A) ent-kaurenoic acid, (B) methyl ent-kaurenoate, (C) ent-7α-hydroxy-kaurenoic acid, (D) ent-kaurenal, (E) ent-kaurenol, and (F) ent-kaurene.

Discussion

The substrate specificity of the E. tracheiphila CYP112 isozyme was tested here. This was enabled by its ready over-expression in E. coli, perhaps reflecting their relatively close phylogenetic relationship (i.e., both species fall within the Enterobacteriaceae family). However, unlike EtCYP117, which catalyzes the first oxidative transformation in the GA biosynthetic pathway (Figure 1) and also is readily over-expressed in E. coli (14), EtCYP112 can be purified with retention of catalytic activity and exhibits the characteristic Soret-band at about 420 nm, as well as the characteristic Soret peak at 450 nm after reduction and CO binding.

Based on the ability to measure binding with straightforward UV-Vis spectral studies it was possible to examine the affinity of EtCYP112 for a number of GA biosynthetic intermediates, as well as selected methylated derivatives (Table 1). In contrast to at least a GA20ox from wheat (Triticum aestivum) that exhibited increased K_M values for the catalyzed reaction intermediates GA₁₅ (open) and GA₂₄ relative to GA₁₂ (15), EtCYP12 exhibits decreased K_d (i.e., higher affinity) for these. This is consistent with their observed retention in the active site during the course of the catalyzed series of reactions (9). While the K_M values for GA20ox and K_d values for EtCYP112 cannot be directly compared, they are both in the upper nM to lower μM range (no binding affinities have been reported for the fungal equivalent CYP68B). Given the highly bioactive nature of GAs low production levels are sufficient to induce an effect, consistent with the tight binding observed here. In addition to GA₁₂, GA₁₅ and GA₂₄, EtCYP112 was found to bind, at least to some degree, essentially all precursors in the bacterial GA biosynthetic pathway, as well as the methylated derivatives of its substrates and reaction intermediates. Surprisingly, this included precursors with the 6-6-6-5 ent-kaurane rather than 6-5-6-5 gibberellane ring structure. Although ent-kaurene and its derivatives did not serve as substrates, some spectral change was observed in Soret-band absorption upon their addition to EtCYP112. Accordingly, these precursors can bind and displace the water molecule that coordinates the heme-Fe(III) as its sixth ligand, leading to some conversion to the spectrally observed high-spin state. However, unlike the native reactants, these precursors do not fully convert EtCYP112 to the high-spin state (i.e., this is generally below <60%). This promiscuous binding may be linked to the complex reactions catalyzed by EtCYP112, as these involve relatively large changes in the structure of the reactant. Nonetheless, EtCYP112 exhibits higher fidelity in terms of activity, since only the natural substrate and intermediates are efficiently reacted upon. Indeed, even the immediate precursor GA₁₂-aldehyde, which particularly in its geminal-diol form is a close structural analog of GA₁₂, is not efficiently acted upon by EtCYP112. Accordingly, the different ring structure of ent-kaurene and its derivatives must alter their positioning in the EtCYP112 active site such that they are not correctly oriented for catalysis (i.e., relative to the thiolate heme-iron), despite their ability to (partially) displace the water molecule otherwise serving as the 6^th ligand.

It was previously shown that EtCYP112 does not react with either the C-20 carboxylate GA₂₅ or the 7,19-dimethyl ester derivative of MeGA₂₄ (9). Here this was found to be at least partially due to the fact that neither compound binds to EtCYP112 with a significant shift of the Soret-band and decrease of the α-band. Together these findings support the previously indicated use of a cyclic anhydride C-20 gem-diol intermediate, as methylation of the C-19 carboxylate in MeGA₂₄ prevents cyclization and GA_25, while theoretically able to cyclize to the anhydride, is energetically unlikely do so in an aqueous environment.

In conclusion, the studies reported here provide insight into the impact of the complex transformation of GA₁₂ to GA₉ on the substrate structure-function relationship of CYP112, which must accommodate significant changes in its substrate/reactants (see Figure 1B). This is shown to be reflected, in promiscuous binding of upstream precursors. Nevertheless, although these preceding metabolites can bind to CYP112, they only do so with a higher K_d or a reduced ability to convert the thiolate heme-iron to its high-spin state. Accordingly, CYP112 reactivity seems to be quite specific. Of the various biosynthetic precursors bound by CYP112, only GA₁₂-aldehyde is reacted upon, and even this is only slowly and partially transformed, with no catalysis of the final combined demethylation and γ-lactone ring forming reaction. This presumably is a function of the complexity of this final transformation (i.e., from GA₂₄ to GA₉), and corresponding requirement for precise positioning of the substrate/reactant within the active site. Regardless, these results indicate that, despite the need to accommodate significant alteration of its substrate/reactants during transformation of GA₁₂ to GA₉, leading to promiscuous binding of upstream metabolites, CYP112 only readily reacts with its natural substrate and reaction intermediates. Accordingly, the role of CYP112 in bacterial GA biosynthesis is specified at the catalytic rather than substrate binding level.

Abbreviations List

GA

Gibberellin

CYP

cytochrome P450

2ODD

2-oxoglutarate dependent dioxygenases

GA20ox

GA 20-oxidase

EtCYP112

Erwinia tracheiphila CYP112

EtFdR

E. tracheiphila ferredoxin reductase

EtSDR

E. tracheiphila short-chain alcohol dehydrogenase/reductase

GC-MS

gas chromatography mass spectrometry