Crystal Structures of L-DOPA Dioxygenase from Streptomyces sclerotialus

Abstract

Extradiol dioxygenases are essential biocatalysts for breaking down catechols. The vicinal oxygen chelate (VOC) superfamily contains a large number of extradiol dioxygenases, most of which are found as part of catabolic pathways degrading a variety of natural and human-made aromatic rings. The l-3,4-dihydroxyphenylalanine (L-DOPA) extradiol dioxygenases compose a multitude of pathways that produce various antibacterial or antitumor natural products. The structural features of these dioxygenases are anticipated to be distinct from those of other VOC extradiol dioxygenases. Herein, we identified a new L-DOPA dioxygenase from the thermophilic bacterium Streptomyces sclerotialus (SsDDO) through a sequence and genome context analysis. The activity of SsDDO was kinetically characterized with L-DOPA using an ultraviolet-visible spectrophotometer and an oxygen electrode. The optimal temperature of the assay was 55°C, at which the K_m and k_cat of SsDDO were 110 ± 10 μM and 2.0 ± 0.1 s⁻¹, respectively. We determined the de novo crystal structures of SsDDO in the ligand-free form and as a substrate-bound complex, refined to 1.99 and 2.31 Å resolution, respectively. These structures reveal that SsDDO possesses a form IV arrangement of βαβββ modules, the first characterization of this assembly from among the VOC/type I extradiol dioxygenase protein family. Electron paramagnetic resonance spectra of Fe–NO adducts for the resting and substrate-bound enzyme were obtained. This work contributes to our understanding of a growing class of topologically distinct VOC dioxygenases, and the obtained structural features will improve our understanding of the extradiol cleavage reaction within the VOC superfamily.

Graphical Abstract

Dioxygenase chemistry is essential for catechol breakdown. The largest natural source of catechols, or 1,2-dihydroxybenzenes, is the plant woody-tissue polymer, lignin. Lignin is a polymer of catecholic monomers that strengthens the cell walls of a plant but impedes access to the cellulosic sugars used in biofuel production.^1,2 It is also a relatively untapped source of carbon that could be repurposed and valorized into feedstocks and natural products.³ Ever since the first example of a “pyrocatechase” (catechol-1,2-dioxygenase) was discovered 70 years ago,⁴ numerous catechol dioxygenases from all domains of life and in all types of metabolic pathways have been reported, including those that use molecular oxygen to break the aromatic ring adjacent to the diol, generating an aldehyde and a carboxylic acid at either end of the six-carbon chain. These enzymes were the inaugural members of two functional classes of oxygenases: the intradiol and the extradiol dioxygenases.⁵ When found in degradation pathways, the extradiol dioxygenase reactions linearize the catecholic substrates such that downstream products can enter central metabolism.⁵ In some cases, the linearized products undergo cyclization and further derivatization to yield a natural product.^6–8

The extradiol cleavage reaction has been adapted to the structural scaffolds of three distinct superfamilies: vicinal oxygen chelate (VOC or type I),⁹ LigAB (type II),¹⁰ and cupin (type III).¹¹ VOC/type I dioxygenases are the best characterized group of extradiol dioxygenases, and they are mostly found as part of catabolic pathways degrading a variety of natural and synthetic aromatic compounds. All VOC superfamily members use a metal center to coordinate an organic substrate, intermediate, or transition state through both of its vicinal oxygen atoms;⁹ for example, the neighboring hydroxyl groups of catechol are typically chelated by an active site Fe²⁺ in the VOC/type I extradiol dioxygenases.¹² Despite significant knowledge of VOC structure, domain architecture, and enzymatic function, the relationships between VOC domain architecture and function remain complex and poorly understood.¹³ While the chemistry throughout the VOC superfamily varies, each superfamily member is structurally defined as an assembly of βαβββ modules.⁹ The connectivity of these βαβββ modules is permuted across the superfamily into six different topologies [forms I–VI (Figure S1)].^9,14 The functional class of nonredox VOC enzymes contains representative domain architectures of every possible assembly, with the exception of monomeric form II. However, among the six possible domain architectures, nearly every characterized VOC/type I extradiol dioxygenase is of form V architecture, which is a four-module monomer of two back-to-back stacked domains.⁹

The first L-DOPA extradiol dioxygenase, LmbB1, was identified almost 25 years ago from the biosynthetic gene cluster of lincomycin from Streptomyces lincolnensis.^15–17 Later, several of its homologues from other bacterial sources were characterized.¹⁸ These dioxygenases are involved in the biosynthesis of antibacterial or antitumor natural products, such as HrmF of hormaomycin,¹⁹ SibV of anthramycin,^20,21 Por13 of porothramycin,²² SibV of sibiromycin,²³ and TomH of tomaymycin.²⁴ L-DOPA dioxygenases employ a ferrous iron to catalyze the insertion of dioxygen into the aromatic ring of L-DOPA, linearizing it to the chromophoric 5-alanyl-2-hydroxymuconate 6-semialdehyde (AHMS), which undergoes a prompt, non-enzymatic cyclization to 3-carboxy-3-hydroxyallylidene-3,4-dihydropyrrole-2-carboxylic acid (CHAPCA) (Scheme 1).²⁵ L-DOPA dioxygenases were labeled as one-domain VOC enzymes based only on the length of sequence and homology to VOC family enzymes.⁵ These one-domain VOC dioxygenases were in contrast to the larger (>40 kDa), well-studied, two-domain/form V enzymes, such as catechol 2,3-dioxygenase (C23O), 2,3-dihydroxybiphenyl 1,2-dioxygenase (DHBD), and homoprotocatechuate 3,4-dioxygenase (HPCD).⁵ While the smaller VOC dioxygenases, typified by L-DOPA dioxygenases, have been studied biochemically, no detailed structural information has been available to date.

Scheme 1. Biosynthesis Pathway of Pyrroline Carboxylic Acid, in which SsDDO Catalyzes the Second Step, L-DOPA Dioxygenation^a.

^aThe functionalized pyrroline carboxylic acid is further modified and incorporated into the natural product scaffold.

Here, we identify a new L-DOPA extradiol dioxygenase from a thermophilic bacterium Streptomyces sclerotialus (SsDDO) by sequence and genome context analysis. Moreover, the first three-dimensional structural study of an L-DOPA dioxygenase was completed by X-ray crystallography. A temperature-dependent activity assay was performed by monitoring the product chromophore to investigate the optimal temperature for catalysis. The iron center was also characterized by electron paramagnetic resonance (EPR) spectroscopy. These results demonstrate that SsDDO is the inaugural member of a new class of topologically distinct VOC dioxygenases, whose structure and mechanism will improve our understanding of the extradiol cleavage reaction.

MATERIALS AND METHODS

Protein Overexpression and Purification.

The synthetic gene of the codon-optimized SsDDO in the pET28a expression vector was purchased from GenScript. The Escherichia coli BL21 (DE3) cells (Merck) containing the expression plasmid for N-terminally His₆-tagged SsDDO were cultured in Luria-Bertani medium with kanamycin (50 μg/mL) at 37°C. When the OD₆₀₀ reached 0.6 AU, the temperature was decreased to 20°C and isopropyl β-thiogalactopyranoside and ferrous ammonium sulfate were added to final concentrations of 600 and 50 μM, respectively. The cells were harvested after 18 h and resuspended in buffer A [50 mM Tris-HCl and 200 mM NaCl (pH 8.0)]. An LM20 cell disruptor (Microfluidics) was used to lyse cells, and the supernatant was recovered after centrifugation (25000×g for 40 min at 4°C).

The protein was then purified by affinity chromatography using a HisTrap column (GE Healthcare) with a gradient of buffer B [50 mM Tris-HCl, 200 mM NaCl, and 500 mM imidazole (pH 8.0)]. The eluted protein was concentrated using an Amicon centrifugal filter with a 3 kDa cutoff (Millipore) and desalted into 50 mM Tris-HCl and 50 mM NaCl (pH 8.0) for further use.

For structural studies, seleno-l-methionine (SeMet)-substituted protein was cultured in M9 minimal medium with the addition of 0.25 mM SeMet according to the published method.²⁶ The purification procedure was the same as the aforementioned wild-type SsDDO. The eluted SsDDO was then treated with a Thrombin CleanCleave Kit (Sigma) to remove the His tag. The untagged protein was further purified again with a HisTrap column and followed by a Superdex-75 column (GE Healthcare) equilibrated with 50 mM Tris-HCl (pH 7.4).

Temperature-Dependent L-DOPA Dioxygenase Activity Assay.

SsDDO was freshly prepared by anaerobic iron reconstitution prior to an activity assay using a published method.²⁷ Apo-SsDDO was prepared by overnight incubation with EDTA (10 mM, pH 8) at 4°C, followed by desalting to remove EDTA. Fe(II)-SsDDO was prepared by incubating degassed apo-SsDDO with 10 equiv of O₂-free ammonium ferrous sulfate in a glovebox for 30 min, and finally, iron ions were removed by desalting using an argon-saturated buffer. After iron reconstitution, the iron occupancy in SsDDO was around 95%, as assessed by a ferrozine assay.²⁸ The steady-state kinetic parameters were evaluated on the basis of Fe(II)-SsDDO concentrations. Michaelis–Menten steady-state constants were obtained by reacting SsDDO (0.2–0.5 μM) with L-DOPA (25 μM to 1 mM) in reaction buffer [50 mM potassium phosphate (pH 8.0)]. An Aligent 8453 ultraviolet–visible (UV–vis) spectrophotometer with a temperature-regulated cuvette holder was used to detect the resulting cyclized product of AHMS, CHAPCA, at 414 nm with an extinction coefficient of 47500 M⁻¹ cm⁻¹.¹⁶ The temperature was adjusted to 25, 40, 50, 55, 60, and 65°C to determine the optimal temperature of the assay. An oxygen electrode (Oxygraph, Hansatech Instruments) was used to measure oxygen consumption under the same conditions. Initial rates were measured and plotted versus L-DOPA concentration and fit with the Michaelis–Menten equation using standard nonlinear regression (Origin 8.5, Microcal).

Crystallization.

The untagged protein was concentrated to 10–15 mg/mL and mixed at a 1:1 volume ratio with a crystallization buffer of 0.1 M Tris-HCl (pH 8.5), 0.2 M MgCl₂, and 16% (w/v) PEG 8000 using the hanging drop, vapor-diffusion method at 295 K. Microcrystals formed the next day and grew to an optimal size suitable for X-ray diffraction after 3–4 days. Crystals were cryoprotected with crystallization buffer containing an additional 20% (v/v) glycerol and then flash-cooled in liquid nitrogen. For substrate-bound crystals, crystallization trays were set up inside a glovebox with an argon-saturated protein solution and crystallization buffer. After anaerobic crystals were formed, the ES complex was prepared by soaking the crystals with 5 mM L-DOPA under a strictly anaerobic condition for half an hour before they were rapidly cooled in liquid nitrogen.

Data Collection and Structure Determination.

All data were collected at 100 K and processed using HKL2000.²⁹ Single-wavelength anomalous diffraction data for the SeMet-substituted SsDDO were collected from beamline SSRL 9–2, and a 2.40 Å resolution data set was obtained at a wavelength of 0.97913 Å, corresponding to an absorption peak of selenium. PHENIX.AutoSol³⁰ and PHENIX.AutoBuild³¹ were used for phasing and model building. A SsDDO structure was also collected at beamline SSRL 9–2 with a resolution of1.99 Å, which was later termed an initial model for molecular model building and refinement for other structures using PHENIX³² software packages. The ES complex (SsDDO in complex with L-DOPA) structure of 2.31 Å resolution was collected at SBC 19BM. The E only and ES complex SsDDO original data sets have a strong anisotropic problem; therefore, original data were truncated using an online server.³³ All of the data collection and refinement statistics are summarized in Table 1.

Table 1.

X-Ray Data Collection and Refinement Statistics

	SeMet SsDDO	SsDDO (E only)	SsDDO (ES complex)
	SAD	single wavelength	single wavelength
Protein Data Bank entry		6ON1	6ON3
Data Collectiona
beamline station	SSRL 9–2	SSRL 9–2	SBC 19BM
wavelength (Å)	0.97913	0.97946	0.97933
space group cell dimensions	P21	P21	P21
a, b, c (Å)	100.1, 46.2, 130.8	100.0, 45.2, 131.2	99.8, 40.7, 128.4
α, β, γ (deg)	90, 106, 90	90, 105, 90	90, 105, 90
resolution (Å)	50–2.40 (2.44–2.40)	50–1.99 (2.02–1.99)	50–2.31 (2.35–2.31)
total no. of reflections	498715	510762	193367
no. of unique reflections	41737	76745	42332
redundancy	11.9 (8.2)	6.7 (6.2)	4.6 (3.8)
Rmergeb (%)	18.7 (92.8)	9.1 (36.5)	12.4 (55.2)
I/σI	15.80 (1.04)	20.93 (3.86)	12.0 (1.17)
completeness (%)	91.0 (66.9)	96.9 (91.1)	94.5 (82.0)
CC1/2, highest-resolution shell	0.74	0.97	0.83
FOMc	0.345
Refinement
resolution (Å)		30.33–1.99	44.01–2.31
no. of reflections		76545	32257
Rworkd/Rfreee		20.7/25.6	23.5/29.4
no. of atoms
protein		7190	7246
iron		6	6
ligand		N/A	84
solvent		376	342
B-factor (Å2)
protein		46.86	32.04
iron		34.98	33.14
ligand		N/A	32.7
solvent		47.41	31.07
root-mean-square deviation
bond lengths (Å)		0.006	0.002
bond angles (deg)		0.769	0.498
Ramachandran analysis (%)
favored		98.46	96.05
allowed		1.18	3.37
outlierf		0.35	0.58

^aNumbers in parentheses refer to data in the highest-resolution shell.

^b$R_{\text{ merge }}=\sum \left|I_{\text{h}}-\langle I_{\text{h}}\rangle \right|/\sum I_{\text{h}}$, where I_h is the observed intensity and ⟨I_h⟩ is the average intensity.

^cFigure of merit after RESOLVE.^15,16

^d$R_{\text{work}}=\sum \Vert F_0\left|-k\right|F_{\text{c}}\Vert /\sum \left|F_0\right|$.

^eR_free is the same as R_obs for a selected subset (10%) of the reflections that was not included in prior refinement calculations.

^fOutliers are from the surface flexible region, which are irrelevant to substrate binding. Total of three outliers in the E only structure: Ala85 in chain B and Ala79 and Asp81 in chain C. Total of five outliers in the ES structure: Pro159 in chain B, chain D, and chain F, Leu27 in chain E, and Ala14 in chain F.

Preparation and EPR Measurement of Nitrosyl Complexes.

The preparation of nitrosyl samples and the X-band EPR setup were described previously.³⁴ Argon-saturated SsDDO (500 μM) was reduced by 5 mM sodium ascorbate for 30 min anaerobically. A slight excess of NONOate (Cayman) was added to reduced SsDDO forming the E-NO^• complex. L-DOPA (5 mM) was then added to generate the ES-NO^• complex. Spectra for high-spin signals were collected at 6 K and 0.796 mW with a 6 G modulation amplitude with a 100 kHz modulation frequency, while spectra for the low-spin dinitrosyl complex were collected at 50 K and 0.05 mW. All spectra were the average of four scans.

RESULTS AND DISCUSSION

Identification of SsDDO as a Putative L-DOPA Dioxygenase.

Previous work on L-DOPA dioxygenases, LmbB1, Orf12, and SibV, reported the enzyme existed in purified form as a dimer of identical subunits,^17,21 implying a topology distinct from that of existing, structurally characterized, two-domain VOC/type I dioxygenases. However, further structural study has been impeded due to the failures in the crystallization of LmbB1 and reported homologues; therefore, sequence and genome context were analyzed from a variety of organisms to identify homologues of L-DOPA dioxygenase that would be more amenable to crystallographic study.

Genome context analysis was made possible because the extradiol cleavage of L-DOPA was first proposed within the context of a biosynthetic pathway to C₂- or C₃-proline units of several natural products.³⁵ In fact, the pathways to lincomycin, hormaomycin, and the pyrrolobenzodiazepines of anthramycin, sibiromycin, porothramycin, and tomaymycin have at least three genes in common.¹⁸ l-Tyrosine hydroxylases, L-DOPA dioxygenases, and γ-glutamyltransferase-like N-terminal hydrolases that catalyze C–C bond cleavage are present in all six gene clusters that produce 4-alkyl-l-(dehydro)proline synthons for incorporation into natural products (Scheme 1).³⁶ When found co-located on the bacterial chromosome, these genes are indicative of a minipathway to a 4-alkyl-l-(dehydro)proline synthon. We identified such a pathway in the organism S. sclerotialus ISP-5269, a species first described from an Indian hot spring³⁷ and later sequenced as part of a genome mining project.³⁸ The L-DOPA dioxygenase homologue from S. sclerotialus is 54% identical to LmbB1 from S. lincolnensis, and it clusters on the chromosome with homologues of l-tyrosine hydroxylase (S. lincolnensis, LmbB2), the γ-glutamyltransferase (S. lincolnensis, LmbA), an O-methyltransferase (S. lincolnensis, LmbW), and a protein of unknown function (S. lincolnensis, LmbX) (Figure 1).

Figure 1.

Identification of SsDDO as a putative L-DOPA dioxygenase through a genome context analysis of the biosynthetic gene cluster of pyrroline carboxylic acid in S. lincolnensis and S. sclerotialus. l-Tyrosine hydroxylases (blue): lmbB2 and 2337. L-DOPA dioxygenases (red): lmbB1 and 2352. γ-Glutamyltransferase-like C–C bond hydrolases (yellow): lmbA and 3003. O-Methyltransferase (purple): lmbW and 3013. PhzF phenazine biosynthetic protein/unknown function (green): lmbX and 2336.

A bioinformatics utility EFI-EST (https://efi.igb.illinois.edu/efi-est/) was employed to generate the enzyme sequence similarity network (SSN)³⁹ of L-DOPA dioxygenases (Figure S3), and 65 proteins (as of February 2019) were found to be similar with L-DOPA dioxygenases in the UniProt database. At an e value of 10⁻⁵⁵, this new class of VOC dioxygenases, including LmbB1, Ant12, HrmF, Orf12, Por13, SibV, and TomH, is spread over several smaller clusters. The biosynthetic function of the L-DOPA dioxygenases might be the reason why this homologous group of extradiol dioxygenases is quite distinct from other proteins in primary sequence, or this result could be further evidence of the difficulty in aligning the primary sequence within a superfamily where the essential βαβββ modules can be coupled in frame or inverted with respect to their primary sequence.¹³ Nevertheless, all of the resulting proteins are similar in size, and most are annotated as dioxygenases, such as catechol extradiol dioxygenases. Furthermore, the SSN also includes annotations of glyoxalase I-type enzymes, another member of the vicinal oxygen chelate superfamily. The genomic context and the sequence similarity of the SsDDO validated its identity, and the thermophilic nature of the source organism indicated this homologue was likely more amenable to crystallographic study. Consequently, the gene encoding SsDDO was codon-optimized, synthesized, cloned, and expressed in E. coli a an ~20 kDa protein that cleaved L-DOPA.

Catalytic Activity of SsDDO.

The optimal temperature of the Fe(II)-SsDDO activity assay was determined by monitoring the formation of its cyclized product, CHAPCA, whose extinction coefficient at 414 nm is 47500 M⁻¹ cm⁻¹ at pH 8.¹⁶ The transient, linearized product AHMS (λ_max = 378 nm) immediately cyclizes to form CHAPCA (Scheme 1 and Figure 2A). The cyclization rate of AHMS is related to the substrate concentration, reaction temperature, pH, and other factors. Therefore, consistent with previous observations of LmbB1,²⁵ both enzyme-catalyzed cleavage and product cyclization affect the observed activity of SsDDO, and the rate of cyclized product formation is employed to reflect the enzymatic activity. Herein, K_M′ and k_cat′ indicate the observed activity of forming CHAPCA, which are distinct from the steady-state kinetic parameters. Via measurement of the specific activity at various temperatures, the enzymatic activity was observed to increase from 25 to 55°C, with a decrease in activity when the temperature exceeded 55°C (Figure S2). Alternatively, steady-state kinetic parameters can also be evaluated by oxygen consumption. Therefore, the oxygen electrode method was employed to measure K_M and k_cat at both 25 and 55°C (Figure 2C and Table 2). The results suggest the K_M obtained with an oxygen electrode is comparable with the K_M′ obtained by the UV–vis spectrophotometer while the k_cat is slightly higher than k_cat′. This is an expected result considering product cyclization was involved in measuring K_M′ and k_cat′, and the non-enzymatic cyclization is not the rate-limiting step or competitive with respect to the enzymatic reaction. Approximately 70% of the rate of dioxygenation activity at 55°C (both k_cat′/K_M′ and k_cat/K_M) was lost at room temperature (Figure 2B and Table 2).

Figure 2.

(A) UV–vis spectra of the SsDDO-catalyzed reaction. The transient linearized product (AHMS; λ_max = 378 nm) immediately cyclizes to CHAPCA with a λ_max of 414 nm. The reaction was performed with 0.2 μM enzyme with 0.5 mM L-DOPA at 55°C. Initial rate k_obs = 1.15 s⁻¹ (inset). Steady-state kinetic assay of SsDDO with L-DOPA at 25 and 55°C, obtained using UV–vis spectrometer (B) and oxygen electrode (C), respectively. The data were fit to the Michaelis–Menten equation shown with red lines.

Table 2.

Steady-State Kinetic Parameters for the Reaction of SsDDO with L-DOPA at 25 and 55°C^a

temperature (°C)	KM′ (μM)	kcat′ (s−1)	kcat′/KM′ (×103 s−1 M−1)
25	130 ± 20	0.48 ± 0.04	3.8 ± 0.6
55	130 ± 20	1.5 ± 0.1	12 ± 2
temperature (°C)	KM (μM)	kcat (s−1)	kcat/KM (×103 s−1 M−1)
25	99 ± 6	0.59 ± 0.01	6.0 ± 0.4
55	110 ± 10	2.0 ± 0.1	18 ± 2

^aThe results were obtained using a UV–vis spectrophotometer (top) and an oxygen electrode (bottom).

A kinetic study was previously reported for a SsDDO homologue, LmbB1, the L-DOPA dioxygenase from S. lincolnensis. Although the reported K_M′ for LmbB1 with L-DOPA (30 μM) was lower than that of SsDDO, the k_cat′/K_M′ value of 2.3 × 10³ M⁻¹ s⁻¹ around room temperature is comparable with the measured values for SsDDO.²⁵ The temperature-dependent activity of a keratinase from the same thermophilic strain, S. sclerotialus, was previously studied, and the corresponding keratinase had a temperature-dependent activity tendency similar to that of SsDDO with a maximum activity that was also at 55°C.³⁷

Global Structure of SsDDO.

In solution, LmbB1, Orf12, and SibV have been reported to exist as dimers,^17,21 and their small size (~20 kDa) was termed “one-domain” with respect to other known VOC extradiol dioxygenases, such as the ~40 kDa, two-domain catechol 2,3-dioxygenase.⁵ These differences implied a distinct three-dimensional structure for L-DOPA dioxygenase; however, no additional structural investigations were available to understand how the topological arrangement of the protein is accomplished with a one-domain sequence. To provide the first three-dimensional view of this group of dioxygenases, we determined the resting-state crystal structure of SsDDO by single-wavelength anomalous diffraction (SAD) X-ray crystallography and refined it to 1.99 Å resolution. The diffraction data solution was achieved in space group P2₁, and three dimers were found in an asymmetric unit. The monomeric SsDDO is composed of three α-helices and eight β-strands, which fold into two structural domains (residues 13–78 and residues 92–160) connected by an intervening segment (residues 79–91) (Figure 3A), and the helices and β-strands are organized into two functional modules with a repeated βαβββ pattern (β1–α1–β2–β3–β4 and β5–α3–β6–β7–β8). The two domains are not identical, and one of tham (denoted the first domain hereafter) has an additional short α-helix (α2) inserted between β2 and β3. In particular, each domain forms a catalytic center with an associated domain from a neighboring protomer. The β-strands (β1–β4) of the first domain associate with the β-strands (β5*–β8*) of the second domain from a neighboring subunit in an antiparallel manner, rendering a concave curvature of an eight-stranded β-sheet surrounding the iron center. Another curved eight-stranded β-sheet is formed vice versa, yielding two catalytic iron centers in two β-sheets. α2 is situated on the opposite side of the β-sheet, and it completes the cup-shaped active pocket, while two long α-elices (α1 and α3) closely interact with each other to associate the two domains (Figure 3B). Multiple-sequence alignment of SsDDO with its homologues using Clustal Omega⁴⁰ indicates that the secondary structural elements are well conserved (Figure 3C). This arrangement of β-sheets with the formation of a metal binding center from two separate monomers is known for the non-oxidative glyoxalase enzymes but has not been observed in a VOC/type I extradiol dioxygenase.

Figure 3.

Structural view of the SsDDO monomer and dimer and amino acid sequence alignment of L-DOPA dioxygenases. (A) Monomer structure with labeled secondary structure elements. (B) Two monomers in a dimer colored pink and blue. The active site is formed by dimerization with an iron (orange). (C) Amino acid sequence alignment of SsDDO and its homologues from other Streptomyces bacteria. The included sequences are SsDDO from S. sclerotialus (WP_051872352), LmbB1 from S. lincolnensis (CAA55747), Por13 from Streptomyces albus (AEA29636), Orf12 from Streptomyces refuineus (ABW71843), SibV from Streptosporangium sibiricum (ACN39745), HrmF from Streptomyces griseoflavus (AEH41784), and TomH from Streptomyces achromogenes (ACN39021). The TomH sequence is truncated relative to those of the other homologues. Highly conserved residues are colored red. Strictly conserved residues are colored white on a red background. The secondary structure elements are defined from the SsDDO structure. Blue stars and circles represent residues involved in the iron coordination and L-DOPA binding site, respectively.

Nearly every characterized VOC/type I extradiol dioxygenase is of form V architecture, which is a four-module monomer of two back-to-back stacked domains (Figure S1).⁹ The well-characterized examples are C23O,^41,42 DHBD,^43,44 and HPCD.⁴⁵ To date, the only exceptional scaffold of a deposited VOC dioxygenase is 2,6-dichlorohydroquinone dioxygenase (DCHQD)⁴⁶ as a form VI, which is a four-module monomer of two edge-to-edge stacked subunits. Some have speculated that the different topology between DCHQD and the many, characterized, form V VOC dioxygenases might explain how the same VOC scaffold is adapted to oxidatively cleave different classes of hydroxylated aromatic substrates,¹³ but the lack of structural data for other scaffolds has limited progress in this area. SsDDO fills a long-standing gap in our understanding by demonstrating that the VOC L-DOPA dioxygenases are of the form IV module arrangement, with an edge-to-edge dimer of back-to-back fused subunits, the first example of its kind from within the VOC/type I extradiol dioxygenase superfamily. These data reveal a previously unrecognized structural class of VOC/type I extradiol dioxygenases, whose structure and mechanism will improve our understanding of the extradiol cleavage reaction.

Metal Coordination.

The de novo structure of SsDDO shows that the catalytic iron ion is octahedrally coordinated by two histidine residues (His19 and His95*, where an asterisk indicates a residue from a different subunit), a glutamate residue (Glu154*), and three water molecules (Figure 4A). This coordination environment is typical for extradiol dioxygenases across different superfamilies⁴⁷ and was first described in the structure of the two-domain VOC enzyme DHBD.⁴⁸ The coordinated water molecules interact with second-coordination sphere residues, including His74 and Tyr144*, and other surrounding water molecules. Overall, the catalytic center is located at the interface of two protein molecules in a dimer, and the active site pocket is well-organized with a hydrophobic character, which provides a good cavity for substrate L-DOPA binding. The residues involved in metal binding are strictly conserved among this type of L-DOPA dioxygenases (Figure 3C).

Figure 4.

Active site of (A) enzyme only and (B) substrate-bound SsDDO structures. The green F_o – F_c omit map is contoured at 3σ, and the light blue 2F_o – F_c electron density map after refinement is contoured at 1σ. Oxygen, nitrogen, and iron atoms are colored red, blue, and magenta, respectively. Carbon atoms from subunit A, subunit B, and L-DOPA are colored white, dark gray, and yellow, respectively. Residues and secondary structural elements with asterisks are from subunit A. Distances between atoms are shown as gray dashed lines.

Permutations in sequence caused by gene duplication and domain swapping events somewhere in the evolutionary history of VOC enzymes have made it challenging to identify metal binding residues for this class of one-domain extradiol dioxygenases from sequence alone and resulted in misidentification of Fe binding residues in the past.⁷ However, with the increase in the availability of sequences homologous to that of L-DOPA dioxygenase, it is readily apparent that the Fe binding residues identified by the SsDDO structure are conserved across L-DOPA dioxygenases from various organisms. In fact, when SsDDO is superposed with a more distantly related, VOC family protein with a similar topology, such as glyoxalase I (PDB entry 1FA7), the active site and three of the four glyoxalase I metal binding residues are conserved (Figure S4A). Further, the overall structure of SsDDO is quite similar with that of glyoxalase I, which has a Z score of 15.8 and a root-mean-square deviation (rmsd) of 1.8 from the result of the DALI search.⁴⁹ Glyoxalases are prototypical form IV, VOC family enzymes that use four amino acid ligands (two His and two Glu residues) spread over adjoining βαβββ modules to chelate Ni(II), Zn(II), or Cd(II) and catalyze isomerizations,^50,51 highlighting the spatial conservation of residues within the βαβββ motif of VOC metal binding enzymes. Nevertheless, molecular replacement using glyoxalase structures was not successful in solving the SsDDO phasing problem during structural determination. Molecular replacement was attempted using the structure of glyoxalase I (PDB entry 1FA7), but phasing was unsuccessful due to weak conservation of sequence between SsDDO and glyoxalase, only12.5% identity and 24.5% similarity (Figure S4B). While the amino acid sequence level conservation between L-DOPA dioxygenases and other VOC/type I extradiol dioxygenases is low, the evolutionary relationship is apparently preserved at the structural level, as previously found in a similar case between aerobic and anaerobic ribonucleotide reductases.⁵²

Crystal Structure of SsDDO in Complex with Its Native Substrate, L-DOPA.

The SsDDO and L-DOPA (ES) binary complex crystals were obtained by crystallizing SsDDO under anaerobic conditions and then soaking them with a 5 mM L-DOPA solution. A 2.31 Å resolution data set is shown in Figure 4B. Without fitting any ligand, strong additional density for L-DOPA was observed in the active site. Iterative structural refinements suggest that L-DOPA binds to the iron in a slightly asymmetric, bidentate fashion. An L-DOPA molecule is found coordinated to the iron atom of each of the six subunits with iron–oxygen distances of 2.51 ± 0.03 Å for O1 and 2.45 ± 0.03 Å for O2 Å (subunit E was excluded because its distances failed the Q-test with 90% confidence⁵³). These distances are longer and more symmetric than the values of 2.06 and 2.18 Å,⁴⁴ 2.0 and 2.4 Å,⁵⁴ 1.98 and 2.15 Å,⁵⁵ and 2.0 and 2.3 Å⁴⁵ reported for other form V and VOC/type I enzymes, respectively. The asymmetric binding of catecholic substrates to extradiol dioxygenase catalytic Fe(II) centers has been interpreted as evidence of ionization at one of the −OH groups of the substrate; that is, the substrate is bound as a monoanion.⁵⁴ Previous interpretations of VOC dioxygenase structure assert that the hydroxyl adjacent to the molecular oxygen insertion site (C2−C3 in the case of SsDDO) is deprotonated and has a shorter distance to iron compared to the other oxygen. However, in the case of SsDDO, the opposite is true; the Fe−O1 distance is slightly longer than the Fe−O2 distance in all six subunits (Table S1). To verify our predictions about L-DOPA binding, a different model with a flipped L-DOPA aromatic ring was refined; however, the resulting map missed partial electron density on C5 even though the Fe−O1 distance (2.29 Å) was shorter than the Fe−O2 distance (2.42 Å) (Figure S5). At the current resolution, the model presented in Figure 4B provides a better fit of the observed density, despite the difference in bidentate chelation relative to other characterized VOC extradiol dioxygenases. The different substrate binding mode may be a result of the unique assembly of SsDDO as a form IV VOC dioxygenase.

The stabilization of the substrate oxygen atoms by active site residues has also been well-documented in other VOC/type I enzymes, and in all cases, there is a histidine on one side and a tyrosine on the other, and the aromatic bond adjacent to the tyrosine is cleaved.^54,56 In SsDDO, O1 of the substrate is stabilized by Tyr144* while O2 is stabilized by His74 with distances of 2.7 and 3.4 Å, respectively. In the resting state, these two second-coordination sphere residues initially interact with water ligands (Figure 4A), which are presumably displaced upon substrate binding. The positioning of second-coordination sphere residues once again supports the idea that our current model with C2−C3 facing toward Tyr144* is more appropriate.

In the studies of other extradiol dioxygenases, the same second coordination sphere is also present to assist the asymmetric binding of the substrate and subsequent dioxygen activation. Additionally, it was proposed that hydrogen bonding from this tyrosine promotes the iron-bound superoxide intermediate to activate the organic substrate and, thus, forms the alkylperoxo intermediate.^12,57 Structural alignment of ES complexes of SsDDO and another well-characterized, form V/two-domain VOC/type I extradiol dioxygenase HPCD indicates that His74 and Tyr144* of SsDDO align well with His200 and Tyr257 of HPCD (Figure S6), respectively. Moreover, the residues known to play critical roles in the extradiol dioxygenase pathway of HPCD and other form V/two-domain VOC/type I extradiol dioxygenases^{44,54,56–61} are also found in similar positions.

Notably absent from the SsDDO active site is the putative base that interacts with the active site tyrosine and is proposed to assist in deprotonation of the substrate catechol;¹² this is the role proposed for His248 of HPCD,⁵⁷ His246 of C23O,⁵⁹ and His241 of DHBD.⁵⁴ In SsDDO, there is no such histidine; rather, an arginine (Arg56) occupies the equivalent location. Arg56 is located in short α-helix α2 (Asp51–Arg56) that recognizes the amino acid main chain of the substrate, L-DOPA. The hydroxyl groups of Ser50 and Thr53, as well as the backbone amide of Ser50, also form hydrogen bonding contacts with the carboxylate of L-DOPA (Figure 4B). However, the amino of L-DOPA is surrounded by hydrophobic residues and shows no direct polar interactions with any active site residues. The lack of specific, polar binding determinants for the amino group of L-DOPA could explain why molecules lacking the amino group such as 3,4-dihydroxyphenylacetic acid are substrates of the enzyme.⁷

The overall structure of the ES complex is largely unchanged when compared to the E only structure with an rmsd value of0.344 Å for 147 C_α atoms; one exception is helix α2, which slightly moves away from the active site to accommodate substrate binding (Figure S7). In particular, Arg56 exhibits a large side chain movement upon substrate binding and is supported by Glu152*. The guanidinium group of Arg56 points away from the active site in the holoenzyme structure (Figure 4A), while it sits over the aromatic ring of Tyr144* in the ES complex, forming a parallel cation–π interaction (Figure 4B). This evidence again supports the hypothesis that Arg56 replaces His248 in HPCD. Moreover, the stabilization of Tyr144* by Arg56 upon substrate binding suggests that the positioning of Tyr144* is critical for substrate recognition and catalysis.

It has been reported that an extradiol dioxygenase employs loop dynamics to accommodate substrates with disparate polarities for rapid enrichment of O₂ at the enzyme active site.⁶² The binding of the primary substrate causes conformational changes and results in an increased hydrophobicity in the active site, which subsequently facilitates O₂ enrichment for rapid binding to the iron center. A well-known example is 3-hydroxyanthrinilate 3,4-dioxygenase (HAO), which efficiently catalyzes dioxygenation with a turnover rate of 25 s⁻¹ and a k_cat/K_M of 1.12 × 10⁶ s⁻¹ M⁻¹.⁶² Interestingly, this phenomenon was not observed in SsDDO by comparing structures with and without L-DOPA, presumably because of a much lower turnover rate of thermophilic SsDDO and a nearly 200-fold smaller k_cat/K_M value in the solution. Therefore, the dynamic loop movement is unnecessary for SsDDO to enrich oxygen rapidly.

EPR Characterization of the Non-Heme Ferrous Center.

Extradiol dioxygenases have an EPR-silent catalytic Fe(II) center. Therefore, binding of the organic substrate is very often studied by forming a nitrosyl complex with ^•NO, a spin probe of the EPR-invisible Fe(II) center and an oxygen surrogate.^43,63–65 EPR is sensitive to active site changes upon substrate binding in the solution state, which may be undetectable in the crystal form. By the addition of ^•NO to reduced SsDDO under anaerobic conditions, a light yellow enzyme–nitrosyl complex (E-NO^•) was generated. The corresponding EPR spectrum of E-NO^• is characteristic of $S=\raisebox{1ex}{$3$}\!\left/ \!\raisebox{-1ex}{$2$}\right.$ with resonances at g values of 4.266, 4.101, and 3.939 (Figure 5, trace A), indicative of unresolved signals from at least two E-NO^• species. The presence of two species is presumably due to slightly different NO^• binding modes in two active sites of one SsDDO dimer. The ES-NO^• complex was generated upon addition of the primary substrate L-DOPA, and the resulting EPR spectrum exhibited a predominant species with g values of 4.106 and 3.949 (Figure 5, trace B), suggesting a more homogeneous iron center was formed after binding the organic substrate. The ES-NO^• EPR signal was more intense than the E-NO^• signal, suggesting NO^•, an O₂ surrogate, has a higher affinity for the active iron center after chelation of the organic substrate.

Figure 5.

EPR spectra of SsDDO Fe^II–nitric oxide complexes. (A) In the absence of substrate, the EPR spectrum of E-NO^• is characteristic of at least two $S=\raisebox{1ex}{$3$}\!\left/ \!\raisebox{-1ex}{$2$}\right.$ species with resonances at g values of 4.266, 4.101, and 3.939. (B) In the presence of the organic substrate, the EPR spectrum of ES-NO^• exhibits a more homogeneous iron center with g values of 4.106 and 3.949. The ES-NO^• EPR signal is scaled to half-intensity for easy comparison with the E-NO^• spectrum. Spectra were collected at a microwave frequency of 9.4 GHz, a modulation frequency of 100 kHz, a modulation amplitude of 6 G, a temperature of 6 K, a microwave power of 0.8 mW, and with an average of four scans.

EPR experiments with extradiol dioxygenases are most often reported with an increased splitting of resonances in the g = 4 region, with signals becoming more rhombic after binding of the substrate. For instance, in studies of protocatechuate 2,3-dioxygenase, upon substrate binding, the resonances changed from g values of 4.11 and 3.95 to g values of 4.21 and 4.85,⁴³ for protocatechuate 4,5-dioxygenase, resonances changed from g values of 4.09 and 3.91 to g values of 4.21 and 3.77,⁶⁶ and in HPCD, resonances changed from g values of 4.04 and 3.96 to g values of 4.12 and 3.87.⁶⁷ When compared to these well-studied dioxygenases, the resonances observed for SsDDO did not show the same trend of increased rhombicity upon substrate binding, indicating a less perturbed electron environment. This observation is consistent with our crystallographic data, which revealed the substrate Fe–O distances were longer and more symmetric than for other prototypical extradiol dioxygenases and may indicate that DOPA binds to SsDDO as a diol rather than a monoanion. It should be noted that both ENO^• and ES-NO^• complexes exhibited an S = 1/2 dinitrosyl signal in the g = 2 region caused by excess ^•NO (Figure S8) which was unaffected by the addition of L-DOPA. The dinitrosyl signal has been observed in synthetic compounds and some non-heme iron proteins; however, it is rarely observed in extradiol dioxygenases.^68–71 The observed dinitrosyl species indicates that the solvent-derived ligands bind the iron center much more weakly than other extradiol dioxygenases and that the active site has more flexibility, allowing the diatomic gas to easily replace the water ligands.

CONCLUDING REMARKS

Cleavage of an aromatic ring with the controlled insertion of molecular oxygen is one of nature’s most elegant reactions. The VOC/type I extradiol dioxygenase superfamily is a collection of prototypical catechol-cleaving enzymes. This superfamily, first defined by catechol-2,3-dioxygenase, has grown to include the example discussed in this work, L-DOPA dioxygenase. L-DOPA dioxygenase proteins are present in biosynthetic pathways that yield antibacterial or antitumor natural products. A new L-DOPA dioxygenase, SsDDO, was identified, sequenced, and biochemically and structurally characterized. The de novo crystal structures of both the resting state and the ES complex were determined at high resolution. Together, these results represent the first comprehensive structural study of an L-DOPA dioxygenase and suggest that the substrate binding is significantly altered compared to that of other extradiol dioxygenases. The global structure indicates SsDDO is novel among VOC/type I extradiol dioxygenases, an edge-to-edge dimer of back-to-back fused subunits from one domain (form IV). When compared to other structurally characterized VOC/type I extradiol dioxygenases (form V), the active site second-coordination sphere residues playing essential catalytic roles are spatially well-conserved in SsDDO. The only exception is a histidine, often present to assist the deprotonation of the substrate catechol, is replaced in SsDDO with an arginine that undergoes significant movement upon substrate binding. However, in marked contrast to structurally characterized VOC/type I extradiol dioxygenases such as catechol 2,3-dioxygenase, SsDDO forms its active site by dimerization of individual monomers, a topology thus far unprecedented among extradiol dioxygenases of the VOC superfamily. These findings promise to provide new insights into the relationship between VOC domain architecture and enzymatic function.