FAD-dependent C-glycoside–metabolizing enzymes in microorganisms: Screening, characterization, and crystal structure analysis

Significance

Microbial degradation of C-glycosides is only reported from human intestinal bacteria. This study isolated a C-glycoside–catabolizing bacteria, Microbacterium sp. 5-2b, from soil and identified a C-glycoside–metabolizing enzyme. Surprisingly, while the metabolic pathway of C-glycoside in the strain was similar to that of an intestinal one, the enzyme named CarA catalyzing the initial step of metabolism was completely different from an enzyme identified from the intestinal microorganism; CarA identified as C-glycoside 3-oxidase for C-glycoside metabolism exhibited amino acid sequence similarity to pyranose oxidase family proteins. This study clarified the structure, function, and distribution of the enzyme involved in C-glycoside catabolism and provides insight into the biogeochemical circulation of C-glycosides in nature.

Abstract

C-glycosides have a unique structure, in which an anomeric carbon of a sugar is directly bonded to the carbon of an aglycone skeleton. One of the natural C-glycosides, carminic acid, is utilized by the food, cosmetic, and pharmaceutical industries, for a total of more than 200 tons/y worldwide. However, a metabolic pathway of carminic acid has never been identified. In this study, we isolated the previously unknown carminic acid-catabolizing microorganism and discovered a flavoenzyme “C-glycoside 3-oxidase” named CarA that catalyzes oxidation of the sugar moiety of carminic acid. A Basic Local Alignment Search Tool (BLAST) search demonstrated that CarA homologs were distributed in soil microorganisms but not intestinal ones. In addition to CarA, two CarA homologs were cloned and heterologously expressed, and their biochemical properties were determined. Furthermore, a crystal structure of one homolog was determined. Together with the biochemical analysis, the crystal structure and a mutagenesis analysis of CarA revealed the mechanisms underlying their substrate specificity and catalytic reaction. Our study suggests that CarA and its homologs play a crucial role in the metabolism of C-glycosides in nature.

Various low–molecular mass plant-derived compounds, such as flavonoids, are glycosylated. Glycosides can be classified as O-, C-, N-, and S- by the manner of linkage between a sugar and aglycone, which is the nonsugar moiety of glycosides (Fig. 1A). Hundreds of C-glycosides have been isolated from various living organisms and show various bioactivities (1–4).

Fig. 1.

Schematic representation of O-glycosides and C-glycosides, the growth of Microbacterium sp. 5-2b, and HPLC analysis of reaction mixtures. (A) Structures of O-glycosides (Left) and C-glycoside (Right). The cleavage sites for deglycosylation are indicated by the dotted lines. (B, Left) Colonies of strain 5-2b on a culture plate containing carminic acid, the color of which is red. Strain 5-2b was cultured on this plate for 7 d. (B, Right) Corresponding to B, Left, the colonies of strain 5-2b are colored yellow, and the clear zone is indicated by a white dotted line so that it is easier to discern. (C) HPLC analysis of reaction mixtures; cell-free extracts of strain 5-2b were incubated with carminic acid for 0 min, 10 min, and 2 h. The numbers in this figure indicate carminic acid (1), compounds X₁ and X₂ (2, 3), and compound Y (kermesic acid, 4).

Humans ingest these glycosides in botanical foods and metabolize them in the intestine. In the small intestine, lactase-phlorizin hydrolase, which is present on the luminal side of the brush border, hydrolyzes glycosides (5) and, in the large intestine, glycosides are metabolized by intestinal microorganisms (6). The resulting aglycones are taken up from the intestine and show many beneficial bioactivities, such as antimicrobial, antiviral, and antioxidative ones (7). The metabolism of glycosides is closely related to the expression of their biological activities.

More than 100 glycoside hydrolase families in the CAZy database which hydrolyze glycosides to yield a sugar moiety and the corresponding aglycone have been identified from bacteria to mammals (8–11). However, C-glycosides are not deglycosylated by glycoside hydrolases, because the sugar moiety and the aglycone are linked by a carbon–carbon bond. In intestinal microorganisms, C-glycosides are deglycosylated through a two-step deglycosylation reaction consisting of oxidation of the sugar moiety and C–C bond cleavage (12–15). Although the enzymes catalyzing the two-step reaction have been identified (15), detailed biochemical characterization and crystal structure analysis have never been reported.

To clarify the metabolism of C-glycosides in nature, we started our study by screening carminic acid–catabolizing microorganisms from soil. Carminic acid, which is extracted from cochineal insects, is a C-glycoside of an anthraquinone derivative and is very important in the food, cosmetic, and pharmaceutical industries as a natural “red dye” all over the world (16).

In this study, we discovered a carminic acid–catabolizing microorganism and identified a C-glycoside–metabolizing enzyme, C-glycoside 3-oxidase (CarA), that catalyzes the first step of C-glycoside metabolism by oxidizing the C3 position of the sugar moiety. We also report the enzyme’s biochemical properties, crystal structure, and possible reaction mechanism.

Results

Screening of Carminic Acid–Catabolizing Bacteria.

The soil samples used for microbial screening were collected from around the University of Tsukuba. Using the enrichment culture method described in Materials and Methods, we isolated microorganisms that were able to grow on a medium containing carminic acid as the sole carbon source. One of them, strain 5-2b, was strongly suggested to degrade carminic acid on a plate because the red color of carminic acid disappeared around the colony of this strain (Fig. 1B). Regarding the 16S ribosomal RNA gene sequence, strain 5-2b showed 98% similarity to Microbacterium xylanilyticum S3-E(T). Additionally, the cell-free extract of strain 5-2b exhibited carminic acid–degrading activity (Fig. 1C). Strain 5-2b was identified as a so far unknown carminic acid-degrading as well as C-glycoside-metabolizing microorganism in soil.

In liquid media containing carminic acid, the carminic acid–degrading activity of the cell-free extract prepared from strain 5-2b increased after 20 h of cultivation, followed by a decrease in the amount of carminic acid in the medium (SI Appendix, Fig. S1). In contrast, carminic acid–converting activity was not observed in cell extracts prepared from cells cultured in the medium without carminic acid (SI Appendix, Fig. S1). These results indicated that carminic acid–degrading activity was induced by carminic acid in vivo.

Initial Metabolic Pathway of Carminic Acid in Strain 5-2b.

After incubation of strain 5-2b cells with carminic acid for 10 min, the reaction mixture was analyzed by high-performance liquid chromatography (HPLC), with three possible reaction products (named compounds X₁, X₂, and Y) being detected (Fig. 1C). When we incubated the reaction mixture for 2 h, compounds X₁ and X₂ were further converted to compound Y. Compound Y was purified by HPLC. According to ¹H, ¹³C, and heteronuclear multiple-bond connectivity (HMBC) NMR spectra, compound Y was found to be kermesic acid (17), which is the aglycone of carminic acid (SI Appendix, Fig. S2). The exact mass of deprotonated compound Y at m/z 329.0308 [M-H]⁻ corresponded exactly to that of deprotonated kermesic acid (C₁₆H₉O₈, 329.0303).

Compounds X₁ and X₂ showed the same mass at m/z 489.0687 [M-H]⁻, which was the best match with the exact mass of C₂₂H₁₇O₁₃, indicating that X₁ and X₂ were structural isomers and that two hydrogen atoms were removed from carminic acid. X₁ and X₂, which were purified by the method described in Materials and Methods, isomerized in an alkaline aqueous solution; however, isomerization was not observed at pH 4 (SI Appendix, Figs. S3 and S4). To determine the structures of X₁ and X₂, NMR spectral analysis of X₁ and X₂ was carried out in methanol-d₄. No isomerization was observed in methanol. ¹H NMR spectra of compound X₁ showed proton signals corresponding to the kermesic acid moiety and H1′, 2′, 4′, 5′, and 6′ of the sugar moiety. In the C3′ position, on the other hand, no proton signal was observed, and one of the ¹³C NMR signals was observed at 207 parts per million (ppm). Based on these findings and HMBC analysis, compound X₁ was identified as 3′-keto carminic acid (SI Appendix, Fig. S5). We also obtained ¹H and ¹³C NMR spectra of compound X₂. The proton signal of H1′ at 5.56 ppm was a broad singlet. Additionally, one of the ¹³C signals was observed at 211 ppm. These data and other NMR signals suggested that the structure of X₂ is 2′-keto carminic acid or 4′-keto carminic acid, or a mixture of them (SI Appendix, Fig. S6).

The metabolites obtained by the cell-free assay demonstrated that strain 5-2b initiated carminic acid metabolism through a two-step deglycosylation reaction; the sugar moiety of carminic acid was oxidized, followed by cleavage of the C–C bond between the sugar moiety and the aglycone. We then purified a metabolizing enzyme that catalyzes the first step of the deglycosylation reaction.

Purification and Identification of the Initial Enzyme for Carminic Acid Metabolism.

Proteins in cell-free extracts of strain 5-2b were precipitated with ammonium sulfate. The initial enzyme in carminic acid metabolism was further purified by column chromatography, as described in Materials and Methods (Table 1). The purification yield was 0.075%, and the overall increase in the specific activity was 163-fold. This enzyme (CarA) was observed as an ∼50-kDa protein band based on the sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) result (SI Appendix, Fig. S7). The N-terminal partial amino acid sequence of CarA was determined to be MSEAVDVLVVGSGPAGSSVA, and we identified an open reading frame (ORF) of 1,506 bp, of which 60 bp of the 5′ terminus matched the N-terminal sequence (Materials and Methods) from the draft genome sequence data for strain 5-2b. The deduced amino acid sequence of carA consisted of 501 amino acids with a theoretical molecular mass of 52.3 kDa, which was in good agreement with the protein size on SDS-PAGE (SI Appendix, Fig. S7). The molecular mass of the native enzyme was estimated to be 54.9 ± 0.3 kDa by size-exclusion chromatography–multiangle light scattering (SEC-MALS) analysis, indicating that this enzyme was a monomer.

Table 1.

Purification of CarA from Microbacterium sp. 5-2b

Step	Total protein, mg	Total activity, U	Specific activity, U/mg	Yield, %	Purification, -fold
Cell-free extract	380	1.75	0.00462	100	1
(NH4)2SO4	90.1	0.547	0.00607	31.0	1.31
HiPrep Phenyl	6.40	0.0933	0.0146	5.33	3.15
Resource Q	0.0198	0.0123	0.619	0.701	134
Superdex 200	0.00174	0.00131	0.756	0.0749	163

CarA was found to have a glucose–methanol–choline (GMC) oxidoreductase domain and showed 40 to 60% amino acid sequence similarity to the following: PeP2O (18), a fungal 2-pyranose oxidase (P2O) from Peniophora sp.; KaPOx (19) and AsP2Ox (20), bacterial P2Os from Kitasatospora aureofaciens and Arthrobacter siccitolerans, respectively; and flavin adenine dinucleotide (FAD)–dependent glycoside oxidoreductase (FAD-GO) (21), a fungal O-glycoside oxidoreductase from Rhizobium sp. (SI Appendix, Fig. S8). The 2-pyranose oxidase is an FAD-dependent oxidoreductase and catalyzes the oxidation of several aldopyranoses (including d-glucose) by O₂ at the C2 position to yield the corresponding 2-keto aldoses and hydrogen peroxide (H₂O₂). On the other hand, CarA did not share amino acid sequence similarity to the C-glycoside oxidoreductase reported from an intestinal microorganism (15).

Biochemical Characterization of CarA.

The 1.5-kb region of the CarA-coding gene was inserted into an expression vector, pET24a(+) (SI Appendix, Tables S1 and S2). The recombinant CarA was expressed as a His-tagged protein in Escherichia coli Rosetta2 (DE3) and purified by Ni-NTA column chromatography. The specific activity of the recombinant CarA was 6.7 (μmol⋅min⁻¹⋅mg⁻¹), which was higher than that of CarA purified from strain 5-2b. We carried out further experiments by using the recombinant CarA.

The stoichiometry of the CarA reaction was examined. We determined the amounts of carminic acid, keto carminic acid (X₁, X₂), O₂, and H₂O₂. After a 20-min incubation, the amounts of keto carminic acid (X₁, X₂) and H₂O₂ increased to 74 and 94 μM, respectively. In the same period, the amounts of carminic acid and O₂ coincidentally decreased by 81 and 75 μM, respectively (SI Appendix, Fig. S9). These results demonstrated that keto carminic acid (X₁, X₂) and H₂O₂ were formed stoichiometrically with the consumption of carminic acid and O₂ during the enzymatic reaction. The kinetic parameters of CarA were K_m = 0.019 ± 0.001 mM, V_max = 4.9 ± 0.1 (μmol⋅min⁻¹⋅mg⁻¹), and k_cat = 4.3 ± 0.1 s⁻¹. At high concentrations of carminic acid, substrate inhibition was observed with a K_i value of 0.70 ± 0.05 mM (SI Appendix, Fig. S10).

We examined the effects of temperature and pH on CarA activity. The optimal reaction temperature and pH were 30 to 40 °C and 7.6 to 8.4, respectively (SI Appendix, Fig. S11). CarA was stable at 30 °C and in the pH range of 4.2 to 11.4 (SI Appendix, Fig. S12).

As described above, X₁ and X₂ did not isomerize at pH 4. When the CarA reaction was carried out at pH 4, the main product was X₁ (SI Appendix, Fig. S13). This result indicated that CarA catalyzed the oxidation of the hydroxyl group at C3′ of carminic acid to form X₁.

The purified CarA gave an absorption spectrum typical of a flavoprotein, exhibiting bands with maxima near 360 and 460 nm (SI Appendix, Fig. S14). The flavin cofactor binding to CarA was identified as FAD by HPLC in comparison with the authentic compound (SI Appendix, Fig. S15). The amount of FAD obtained from 1.0 mM CarA was estimated to be 0.77 mM, suggesting that one FAD would bind one CarA monomer.

Effects of Small Molecules and Metals on the Enzyme Activity of CarA.

The effects of various compounds on the enzyme activity of CarA were investigated (SI Appendix, Tables S3 and S4). Li⁺, Na⁺, Mg²⁺, Ba²⁺, Mn²⁺, Cd²⁺, Ni²⁺, Rb²⁺, Cs²⁺, and chelating reagents (α,α′-dipyridyl, o-phenanthroline, 8-hydroxyquinoline, ethylenediaminetetraacetate [EDTA], diethyldithiocarbamate, NaN₃, and KCN) had no significant effect on the activity of CarA. On the other hand, the activity of CarA was inhibited by Ca²⁺, Zn²⁺, Co²⁺, Al²⁺, Pb²⁺, Hg²⁺, Cu²⁺, Fe²⁺, Fe³⁺, and Sr²⁺. When each of CaCl₂, CoCl₂, CuCl₂, FeSO₄, FeCl₃, and SrCl₂ was added to the reaction mixture, CarA was precipitated. Among the thiol reagents, p-chloromercuribenzoate (PCMB) inhibited the activity of CarA, while 5,5′-dithio-bis-2-nitrobenzoate, N-ethylmaleimide, and iodoacetate had no significant effect. On the other hand, some pyranose oxidases with amino acid sequence similarity to CarA are inhibited by PCMB (22, 23), although the inhibition mechanism has not been identified.

C-glycoside–Metabolizing Gene Cluster in Soil Bacteria.

Upstream of the carA gene there were two adjacent ORFs designated as carB and carC (Fig. 2A). The amino acid sequences of CarB and CarC were similar to those of DgpC and DgpB, respectively, which catalyze the second step of C-glycoside deglycosylation through the C–C bond-cleavage reaction in intestinal bacteria (15). This finding suggested that carA belonged to a gene cluster with the C–C bond-cleaving enzyme genes for C-glycoside metabolism. A Basic Local Alignment Search Tool (BLAST) search using the protein sequence of CarA as a query was performed against the protein database. CarA homologs were found in various soil bacteria. Each of these genes also forms a gene cluster with C–C bond-cleaving enzyme genes. However, a C-glycoside–metabolizing gene cluster containing a CarA homologous gene was not found in intestinal bacteria (Fig. 2A).

Fig. 2.

Gene clusters involved in the C-glycoside metabolism and phylogenetic analysis of CarA. (A) C-glycoside–metabolizing enzyme gene clusters in soil and intestinal bacteria. The genes homologous to carA, carB, and carC are colored green, blue, and black, respectively. CarB and CarC were similar to each subunit of C–C bond-cleaving enzyme homologs. The numbers under each arrow indicate the protein sequence identity/similarity of the corresponding homologs to CarA, CarB, and CarC. Purple arrows represent dgpA and homologous genes. (B) Phylogenetic tree including CarA, pyranose oxidases, other enzymes in the GMC family, and DgpA. Cholesterol oxidase (ChOx) and cellobiose oxidase (CDH) are other enzymes in the GMC family, which includes pyranose oxidases. DgpA identified from intestinal bacteria is a Gfo/Idh/MocA family oxidoreductase, which catalyzes C-glycoside oxidation. Closely related sequences were collapsed into triangles to reduce the complexity of the cladogram.

Activities of CarA and Homologous Enzymes toward Natural C- and O-Glycosides and Glucose.

Among the CarA homologous proteins found in the BLAST search, in particular, AgCarA and MtCarA were cloned from Arthrobacter globiformis and Microbacterium trichothecenolyticum, respectively, and heterologously expressed in E. coli (Fig. 2A and SI Appendix, Tables S2 and S5). The molecular masses of AgCarA and MtCarA were estimated to be 57.3 ± 2.2 and 55.3 ± 1.8 kDa, respectively, by SEC-MALS analysis for their purified enzymes. These proteins were monomers in solution, because the calculated masses of AgCarA and MtCarA were 55.23 and 54.75 kDa, respectively.

The enzymatic activities of CarA and its homologs toward various glycosides (such as C- and O-glycosides and glucose) were investigated (Fig. 3). CarA, AgCarA, and MtCarA showed oxidase activity toward C-glycosides such as carminic acid, mangiferin, and C6-glycosylated flavonoids; on the other hand, C8-glycosylated flavonoids and glucose were inert substrates (Fig. 3A). CarA and homologs showed enzymatic activity toward glycosides that had two sugar groups (i.e., rutin, which has glucose and rhamnose-sugar groups) as well (Fig. 3B). Although they also showed broad substrate specificities toward O-glycosides, the specific activities were 100 to 1,000 times lower than those of C-glycosides.

Fig. 3.

Substrate specificities of CarA and homologous enzymes. Specific activities of CarA, AgCarA, and MtCarA toward various (A) C-glycosides and (B) O-glycosides. Puerarin, orientin, aloesin, naringin, and aloenin were inert substrates for all enzymes. The sugar moiety of each compound is shown in blue. The dehydrated reaction products derived from O-glycosides were nonenzymatically cleaved, and the aglycone of each substrate was detected. N.D., no product could be detected.

Overall Crystal Structure of MtCarA.

To investigate the structure–function relationship of CarA, we initiated crystal structure analysis of this enzyme. While CarA could not be crystallized, MtCarA was crystallized by the hanging-drop vapor-diffusion method. The yellow color of the obtained crystals suggested that FAD in the MtCarA crystal was in the oxidized form. We solved the crystal structure of the substrate-free form of MtCarA at 2.4-Å resolution by the molecular replacement [MR]-native single-wavelength anomalous dispersion (SAD) method (Fig. 4A and SI Appendix, Table S6). While the asymmetric unit contained two MtCarA molecules and chains A and B, chain B was heavily disordered; only 299 out of the 515 residues could be modeled in chain B. Therefore, all subsequent structural analyses were performed with chain A. A structural homology search using the DALI server (24) revealed that the structure of MtCarA shared similarity with that of fungal pyranose 2-oxidase (PeP2O) (Protein Data Bank [PDB] ID code 1TZL). The resultant Z score and rmsd values were 44.2 and 2.5 Å (for 454 Cα atoms), respectively (28% amino acid sequence identity). Superposition of the structures showed that the crystal structure of MtCarA lacked an extra domain (Lys377 to Asp422) and a part of a long loop (Pro117 to Asp147) found in PeP2O (Fig. 4B). Amino acid sequence alignment revealed that the extra domain was unique to PeP2O among homologous proteins (SI Appendix, Fig. S8). On the other hand, the long loop in MtCarA adopted a different structure from that of PeP2O and was partly disordered. Since the long loop of PeP2O is involved in the tetramerization of PeP2O, the structural difference seems to contribute to the monomeric form of MtCarA. Another difference was found in FAD: While the isoalloxazine ring of FAD in PeP2O forms a covalent bond with His167, there was no corresponding covalent bond in MtCarA (SI Appendix, Fig. S16).

Fig. 4.

Crystal structure of MtCarA. (A) Cartoon representations of MtCarA. FAD and substrate-binding domains are shown in green and pink, respectively. FAD is shown as a sphere model with carbon atoms in yellow. An orange arrow indicates the substrate entrance reaching the active site of MtCarA. (B) Superposition of MtCarA and PeP2O. The two structures were superposed using the SSM superposition routine in Coot (39). MtCarA and PeP2O are shown in green and white, respectively. The extra domain and long loop in PeP2O are shown in red. (C) SSM superposition of the FAD domain of MtCarA and BphA4, which are shown in green and yellow, respectively. (D) Superposition of the active-site residues of MtCarA and PeP2O, of which carbon atoms are shown in green and yellow, respectively. The modeled carminate is shown in white, and the slow substrate (labeled as “Sub”) of PeP2O is in orange. The loop that collides with the aglycone part of the carminate is shown in purple and indicated by a purple arrow. (E) An orange arrow indicates the substrate entrance reaching the active site of MtCarA. FAD and substrate-binding domains are shown in green and pink, respectively. (F) The active-site structure of MtCarA. Carbon atoms in FAD are shown in yellow.

The overall structure of MtCarA can be divided into two domains, the FAD and substrate-binding domains (Fig. 4A). The FAD domain, which interacted with an FAD molecule, was found to adopt a Rossmann fold–like structure similar to those of glutathione reductase and BphA4 (25, 26) (Fig. 4C). Residues (Val46 to Pro227 and Asp289 to Ser459) which were not included in the FAD domain formed the substrate-binding domain.

Active-Site Structure of MtCarA.

The crystal structure of PeP2O in a complex with its reaction product (PDB ID code 2F5V) (18) suggests that the active site of MtCarA is located at the interface of the two domains. On the other hand, the substrate-binding crystal structure of P2O from Phanerochaete chrysosporium (hereafter PhP2O) was found in the PDB (PDB ID code 4MIG) (27). Least-squares fitting of the PeP2O and PhP2O structures using Cα atoms around the active site suggested that the substrate-binding sites are nearly the same as one another. Based on this observation, we prepared a carminate-bound model through superposition of the pyranose ring of carminate on the pyranose ring of the reaction product in PeP2O. This docking study revealed that MtCarA had enough space for the sugar-moiety binding at the re side (the right side in Fig. 4D) of the FAD isoalloxazine ring. However, the aglycone portion comes into contact with a loop region (residues 346 to 366), suggesting a conformational change is necessary to accommodate the substrate (Fig. 4D). The docking model suggested that carminate accessed the active site through a funnel-shaped hole in MtCarA (Fig. 4 A and E).

The active-site structure of MtCarA is similar to that of PeP2O. His127, Thr129, Asn344, His444, and Asn488 in the active site are conserved between the two structures (Fig. 4F). The docked carminate suggested that the hydroxyl groups on the C2′, C4′, and C5′ atoms of the sugar moiety are recognized by Arg94, Gln344, and Asp488, respectively (Fig. 4F). A hydroxyl group on the C3′ atom of the sugar moiety (hereafter the OH3′ group), which provides a hydride ion for FAD, is located in front of the N5 atom of the isoalloxazine ring. While the N5 atom forms a hydrogen bond with Thr129, the side chain of Thr129 may undergo a conformational change after accepting the hydride ion. The OH3′ group of the carminate seems to form a hydrogen bond with the Nε atom of His444. Substitution of His429 of CarA, which corresponds to His444 of MtCarA, with Ala resulted in inactivation of CarA (SI Appendix, Figs. S17 and S18 and Table S7).

Discussion

Natural C-glycosides are reported in over 300 species of plants, bacteria, and insects (1–3). The most consumed C-glycoside in the world would be carminic acid, which is synthesized by scale insects including American cochineals (Dactylopius coccus Costa) and is the main ingredient of carmine used for food coloring as a red pigment (16). Carmine is found in most sausages, sweets, red drinks, and syrups. The metabolism of carminic acid has never been identified in any living organisms, while the biosynthesis of carminic acid in insects has been reported (28, 29).

We initiated this study by screening carminic acid–catabolizing microorganisms. Strain 5-2b, which showed the highest carminic acid–degrading activity, was identified as Microbacterium sp. Microbacterium is a gram-positive bacterium found in diverse environments, including soil, plants, water, and human clinical specimens. We identified a carminic acid–metabolizing enzyme (CarA) after the several purification steps described in Table 1.

The results of the carminic acid–conversion assay using cell-free extracts of strain 5-2b indicated that strain 5-2b deglycosylated carminic acid through a two-step reaction (Fig. 1C): The first step is the oxidation of the sugar moiety in carminic acid by CarA and the second step is C–C bond cleavage of the reaction product of CarA. The two-step deglycosylation reaction for C-glycosides is also found in an intestinal microorganism involved in the metabolism of puerarin (12, 15), also known as C-glycosylated daidzein, which is a soy isoflavone showing estrogenic activity and is contained in the roots of Pueraria lobata, which is the main ingredient of “Kakkon-to,” a very famous “Kampo” medicine (traditional medicine in China and Japan) used for the treatment of colds. In the intestinal microorganism, the first and second steps are catalyzed by a C-glycoside oxidoreductase (DgpA) and C–C bond-cleaving enzyme (DgpB/C complex, which catalyzes a β-elimination–like reaction), respectively (15). The products of the neighboring genes of carA and carA homologs show amino acid sequence similarity to DgpB/C (Fig. 2A), suggesting soil and intestinal microorganisms have a similar C–C bond-cleaving enzyme (Fig. 5A). On the other hand, CarA homologs have never been identified from intestinal bacteria, which have DgpA homologs for C-glycoside metabolism (15). Although both CarA and DgpA catalyze the oxidation of a C-glycoside to form the corresponding 3′-keto C-glycoside, there are significant differences between them. First, the GMC family of CarA does not show amino acid sequence similarity to the Gfo/Idh/MocA family of DgpA (Fig. 2B). Second, CarA uses FAD as a cofactor, while DgpA uses NAD(H) and 3-oxo-glucose as cofactors. These differences in the initial enzymes for C-glycoside metabolism may arise from the adaptation to aerobic soil and anaerobic intestinal environments, respectively. While DgpA does not use oxygen for its catalytic reaction, CarA needs oxygen (Fig. 5A).

Fig. 5.

Proposed reaction mechanism of CarA and the C-glycoside metabolic pathway in soil and intestinal bacteria. (A) In C-glycoside metabolism, the C-glycoside is oxidized by an FAD-dependent enzyme (CarA) in soil bacteria or NAD-dependent enzyme (DgpA) in intestinal bacteria followed by a C–C bond-cleavage reaction through β-elimination. (B) CarA catalyzes FAD-dependent glycoside 3′ oxidation. 1, carminic acid; 2, 3′-keto carminic acid (compound X₁). The numbering of the amino acid residue in this figure is that for CarA.

CarA is a monomeric protein that catalyzes oxidization at C3′ of carminic acid and other C/O-glycosides to form the corresponding 3′-keto C-glycosides. The amino acid sequence of CarA exhibited similarity (∼60%) to that of pyranose oxidases, which are found in fungi and bacteria (Fig. 2B). Pyranose oxidase is an FAD-dependent protein belonging to the GMC oxidoreductase family, which includes choline dehydrogenase, methanol oxidase, and glucose oxidase.

Phylogenetic analysis suggested that there are four clades of pyranose oxidases (Fig. 2B): clade 1, fungal pyranose oxidases (18, 27); clade 2, a bacterial pyranose oxidase (19); clade 3, CarA and a bacterial pyranose oxidase (20); and clade 4, a fungal O-glycoside oxidoreductase (FAD-GO), which oxidizes the sugar moiety of O-glycosides at the C3 position (21). The enzymes in clades 3 and 4 are monomeric proteins, while those in clades 1 and 2 are multimeric (clade 1, tetramer; clade 2, dimer). Although clade 3 contains one bacterial pyranose oxidase, the results of our biochemical characterization of CarA and its homologs demonstrated that some bacterial putative pyranose oxidases in clade 3 were able to catalyze oxidation of C-glycosides but not glucose (Fig. 3). Therefore, we identified CarA and its homologs as unprecedented examples of FAD-dependent C-glycoside 3-oxidases. Although the pyranose oxidase from Coriolus sp. belonging to clade 1 catalyzes C3 oxidation of 1,6-anhydro-β-d-glucose (levoglucosan; LG) (30), the C3-oxidation reaction of CarA was unique in the pyranose oxidase family proteins which catalyze the oxidation at the C2 position. Other than pyranose oxidase, the glycoside hydrolases in the GH4 and GH109 families have been known to oxidize at C3 of a sugar (31, 32). In LG metabolism of Bacillus smithii, S-2701M, moreover, LG dehydrogenase (LGDH) catalyzes the C3 oxidation of LG, followed by a β-elimination reaction to cleave the C–O bond of LG (33). However, CarA does not show amino acid sequence similarity to those glycoside hydrolases and LGDH.

The K_m of CarA for carminic acid is low enough (0.019 ± 0.001 mM) to work in vivo, and the activity of CarA was induced in strain 5-2b by the addition of carminic acid to the culture medium, indicating that the C-glycoside–catabolic pathway is regulated by the substrate (SI Appendix, Fig. S1); strain 5-2b harboring CarA could physiologically catabolize C-glycosides in nature to play an essential role in its biogeochemical cycle.

While CarA was not crystallized, we succeeded in crystallizing MtCarA, a homolog of CarA. The crystal structure of MtCarA was determined at 2.4-Å resolution (Fig. 4A). The crystal structure of a C-glycoside 3-oxidase has never been reported except that of MtCarA here clarified by us. The crystal structure of MtCarA resembles those of pyranose oxidases, which belong to the GMC family (18, 27, 34). Our docking study revealed that conserved residues in the active site among these enzymes interacted with the substrate. The catalytic histidine (His429 in CarA and His444 in MtCarA), which is conserved in P2O and CarA homologs, is likely to interact with OH3′ of the substrate. Site-directed mutagenesis analysis using CarA revealed that this histidine was essential for the enzymatic activity of CarA (SI Appendix, Table S7). Based on our findings for CarA, we propose the following reaction mechanism for C-glycoside oxidation (Fig. 5B). 1) The catalytic histidine removes a hydrogen atom from the OH3′ of the sugar moiety, and a hydride is transferred to the N5 atom of the isoalloxazine ring of oxidized FAD to form 3′-keto carminic acid and reduced FAD. 2) The reduced FAD provides two electrons to an oxygen molecule and generates H₂O₂, resulting in the formation of oxidized FAD.

The substrate specificities of the CarA homologs revealed that the position of the sugar moiety in glycosides would be important for the enzymatic activity; C6-glycosylated compounds are more suitable than C8-glycosylated compounds. It was also analyzed using the model structure of MtCarA docked with the pyranose-superposed carminate. The docked carminate clearly showed that the aglycone part of the substrate collides with a loop region (residues 346 to 366) of MtCarA (Fig. 4D). A large conformational change of the loop must occur to bind the substrate, considering the above together with the finding that soaking experiments could not produce crystals of the MtCarA–carminate complex; the conformational change may be hampered in the crystal lattice. Structural comparison between PeP2O and MtCarA showed that the α-helix (residues 59 to 70) corresponding to α-helix-106 of PeP2O was significantly shifted in MtCarA. This shift of the α-helix seems to be needed for carminate binding. Our model suggested that the aglycone part of the carminate crashes into the α-helix if the α-helix is located at the same position as that of PeP2O.

While details of the substrate specificity cannot be discussed based on the docked structure of carminate, it is possible to argue the tendency of the substrate specificity roughly. Fig. 3 shows that CarA and its homologs could not catalyze the oxidation of orientin-type C-glycosides and C8-glycosylated flavonoids. A docking study of orientin suggested that the aglycone of the bound orientin collided with residues on the β-sheet above FAD. In the case of O-glycosides (Fig. 3B), aglycone seems to come into contact with residues around the substrate entrance, hampering their binding. Although some O-glycosides were oxidized by CarA and its homologs, the specific activities were about 1,000 times lower than that toward carminic acid. On the other hand, our docking study suggested that homoorientin-type C6-glycosylated flavonoids were able to bind the active site in the same manner as carmine. Indeed, they were oxidized by CarA and its homologs. For further analysis, a crystal structure containing a substrate is essential.

Carminic acid extracted from cochineal insects has been broadly used as a natural red dye in the food industry. The amount of natural compound in the earth has been maintained through metabolism, including synthesis and degradation. However, degradation pathways for C-glycosides have never been identified. In the present study, we are unique in our success in identifying C-glycoside–catabolizing microorganisms from soil and FAD-dependent C-glycoside 3-oxidases, which catalyze the initial step of C-glycoside catabolism. Moreover, we elucidated the crystal structure of C-glycoside 3-oxidase and proposed the reaction mechanism. Our findings provide an insight into the biogeochemical circulation of C-glycoside metabolism in nature.

Materials and Methods

Chemicals.

Carminic acid was purchased from Tokyo Chemical Industries. FAD, H₂O₂, formic acid, methanol, and acetonitrile were purchased from Nacalai Tesque. N-ethyl-N-(2-hydroxy-3-sulfopropyl)-3,5-dimethoxyaniline, sodium salt (DAOS) was purchased from Dojindo Laboratories. All chemicals used were of analytical grade.

Bacterial Strains, Plasmids, and Primers.

Please see SI Appendix, Tables S1 and S2.

HPLC and LC/MS Analyses.

A sample was applied to a Cosmosil πNAP column (4.6 × 150 mm; Nacalai Tesque). HPLC and LC/MS (mass spectrometry) analyses were carried out using a Prominence System with a photodiode array detector (SPD-M20A) and an LCMS-8040 (Shimadzu). The HPLC conditions were as follows: flow rate, 1 mL⋅min⁻¹; solvent A, 0.1% (volume [vol]/vol) HCOOH; and solvent B, methanol. After column equilibration with 50% solvent B, a linear gradient system of solvent B (50 to 100%) was applied over 13 min, followed by 100% solvent B for 2 min.

Structure Determination of Reaction Products.

The structures of reaction products were determined using high-resolution mass spectral (HRMS) data (SYNAPT G2; Waters) and NMR spectral data (AVANCE 600 MHz; Bruker).

Isolation of Carminic Acid–Metabolizing Microorganisms.

Using culture media containing carminic acid as the sole carbon source, carminic acid–metabolizing microorganisms were isolated from soil from the University of Tsukuba by the following enrichment method.

• Step 1: One gram of collected soil was added to 10 mL of culture media, which consisted of 0.1% (weight [wt]/vol) carminic acid, 1% (wt/vol) (NH₄)₂SO₄, 0.05% (wt/vol) KH₂PO₄, 0.05% (wt/vol) K₂HPO₄, 0.05% (wt/vol) MgSO₄·7H₂O, 0.0005% (wt/vol) FeSO₄·7H₂O, and 10% (vol/vol) tap water (adjusted to pH 7.0 with NaOH), followed by incubation at 28 °C for 1 wk.

• Step 2: Two percent (vol/vol) of the cultivated medium was added to fresh medium, followed by incubation at 28 °C for 1 wk. Step 2 was repeated two times.

After enrichment, the culture broth was spread on carminic acid sole-carbon agar plates, which contained 1.5% (wt/vol) agar in addition to the above carminic acid sole-carbon medium, and colonies that grew on these plates during a week’s incubation at 28 °C were isolated.

Assay of a Cell-Free Extract.

The isolated strain was inoculated into a test tube containing 10 mL of carminic acid sole-carbon media, followed by incubation at 28 °C for 2 d. Cells were harvested by centrifugation (4,000 × g, 10 min, 4 °C) and, after washing twice with 20 mM Tris⋅HCl (pH 8.0), were resuspended in 200 μL of the same buffer. Then, the cells were disrupted by sonication (INSONATOR 201M; Kubota), and the cell debris was removed by centrifugation (27,000 × g, 10 min, 4 °C) to prepare cell-free extracts. Two hundred microliters of the reaction mixture comprised 2 μL of 1 M Tris⋅HCl (pH 8.0), 10 μL of 10 mM carminic acid (in milliQ water), 100 μL of the cell-free extract, and milliQ water. After incubation at 28 °C for 16 h, the reaction was stopped by adding 100 μL of acetonitrile. The reaction samples were analyzed by HPLC and LC/MS.

Purification of the Carminic Acid–Metabolizing Enzyme from Microbacterium sp. 5-2b.

Microbacterium sp. 5-2b was cultured in 4 L of carminic acid 1/10 2×YT medium (0.025% [wt/vol] carminic acid, 0.1% [wt/vol] Bacto yeast extract [Difco Laboratories], 0.16% [wt/vol] Bacto tryptone [Difco], and 0.05% [wt/vol] NaCl) at 28 °C for 48 h. Purification of the carminic acid–metabolizing enzyme (CarA) was carried out through the steps described below. Each column chromatography was carried out using an AKTA purifier (GE Healthcare).

• Step 1: Cells were harvested and resuspended in 20 mM Tris⋅HCl buffer (pH 8.0) to disrupt them by sonication at 150 W for 30 min with an INSONATOR 201M (Kubota). The lysate was centrifuged at 27,000 × g at 4 °C for 20 min.

• Step 2: The cell-free extract was fractionated with ammonium sulfate (20 to 60% saturation), followed by resuspension of the precipitate in 20 mM Tris⋅HCl buffer (pH 8.0) containing 0.4 M ammonium sulfate.

• Step 3: The resulting supernatant was applied onto a HiPrep Phenyl FF (GE Healthcare) column equilibrated with 20 mM Tris⋅HCl buffer (pH 8.0) containing 0.5 M ammonium sulfate. The enzyme was eluted by decreasing the concentration of ammonium sulfate (0.5 to 0 M).

• Step 4: The active fractions were dialyzed against 20 mM Tris⋅HCl buffer (pH 8.0), and then placed on a Resource Q column (GE Healthcare) equilibrated with 20 mM Tris⋅HCl buffer (pH 8.0). The enzyme was eluted by increasing the concentration of NaCl (0 to 1 M).

• Step 5: The active fractions were dialyzed against 20 mM Tris⋅HCl buffer (pH 8.0), and then placed on a Superdex 200 fast protein liquid chromatography column (GE Healthcare). The conditions for the elution were as follows: flow rate, 1 mL/min; and buffer, 20 mM Tris⋅HCl (pH 8.0) containing 0.1 M NaCl.

Draft Genome Sequence of Microbacterium sp. 5-2b.

Total DNA from Microbacterium sp. 5-2b was prepared as follows: The strain was cultured at 28 °C for 48 h in 100 mL of 1/10 2×YT medium (0.1% [wt/vol] Bacto yeast extract [Difco], 0.16% [wt/vol] Bacto tryptone [Difco], and 0.05% [wt/vol] NaCl). Cells were harvested by centrifugation, washed with 10 mM Tris⋅HCl buffer (pH 8.0) containing 1 mM EDTA and 100 mM NaCl, and then suspended in 10 mL of 50 mM Tris⋅HCl buffer (pH 8.0) containing 10 mM EDTA and 15% (wt/vol) sucrose. The suspension was incubated with 7 mg/mL of lysozyme at 37 °C for 3 h, and then 2 mL of 0.5 M EDTA (pH 8.0), 2 mL of 10% SDS, and 2.7 mg of proteinase K were added to the solution, followed by incubation at 55 °C for 16 h. DNA was purified by extracting the lysate with phenol/chloroform/isoamyl alcohol (25/24/1; vol/vol/vol), followed by precipitation with isopropanol, treatment with RNase, and then reprecipitation with ethanol. Draft genome sequencing of strain 5-2b was performed using an Illumina HiSeq platform. We obtained 44.6 million reads of a 100-bp paired-end read. A total of 477 contigs comprising 189 to ∼1,692,717 bp were assembled. The genes in the draft genome sequence were annotated with DFAST (35) (https://dfast.ddbj.nig.ac.jp).

Cloning and Heterologous Expression of carA.

carA was amplified using the listed primers (SI Appendix, Table S2). The underlined letters represent the NdeI and EcoRI restriction sites, respectively. The PCR product was cloned into the linearized pET24a(+) vector by using In-Fusion (Clontech Laboratories). The resulting plasmid was designated as pET24a(+)-carA. Next, a His-tag sequence was added upstream of the carA sequence. The sequence of the His tag was amplified using the listed primers (SI Appendix, Table S2). The PCR product was cloned into the NdeI site of the pET24a(+)-carA vector by In-Fusion (Clontech Laboratories). The resulting plasmid was designated as pET24a(+)-His-carA. E. coli Rosetta2 (DE3) cells harboring plasmid pET24a(+)-His-carA were cultivated in 1 L of liquid 2×YT medium containing 50 μg/mL kanamycin and 30 μg/mL chloramphenicol, and grown at 37 °C to an OD₆₀₀ of 1.5. The temperature was lowered to 18 °C and isopropyl-β-d-thiogalactoside was added to a final concentration of 0.5 mM. The cells were cultured for a further 20 h and then harvested. Twenty milliliters of 20 mM Tris⋅HCl buffer (pH 8.0) was added to the pellet (20 g). The cells were disrupted with the sonicator described above. The lysate was centrifuged at 27,000 × g at 4 °C for 20 min. The recombinant CarA was purified by using a HisTrap HP column (GE Healthcare).

Enzyme Assay for CarA.

Measurement of enzyme activity was performed as follows. One hundred microliters of the reaction mixture (1 μL of 5.2 mg/mL CarA, 1 μL of 1 M Tris⋅HCl [pH 8.0], and 5 μL of 10 mM carminic acid [in milliQ water]) was used. One unit of carminic acid–metabolizing activity was defined as the amount of enzyme required to catalyze the formation of 1 μmol of the reaction product per minute. Specific activity is expressed as units per milligram of protein.

The reaction was initiated by adding the enzyme, followed by incubation at 28 °C for an appropriate time. After incubation, the reaction was stopped by adding 100 μL of acetonitrile.

For determination of the kinetic parameters, 200 μL of the reaction mixture consisted of 20 μL of 0.0044 mg/mL CarA, 2 μL of 1 M Tris⋅HCl (pH 8.0), 4 μL of 50 mM DAOS, 4 μL of 50 mM 4-aminoantipyrine, 2 μL of 5,000 U/mL peroxidase, and from 0.007 to 1.0 mM carminic acid. The reactions were initiated by the addition of CarA, followed by incubation at 28 °C. The produced H₂O₂ couples 4-aminoantipyrine and DAOS to yield a blue dye, which can be detected at 595 nm by spectrophotometry. The experiments were carried out in triplicate independently. k_cat values were calculated using an M_r of 52,318 for CarA.

Size-Exclusion Chromatography–Multiangle Static Light Scattering.

SEC-MALS analysis of CarA, AgCarA, and MtCarA was performed with a WTC-030S5 (Wyatt Technology) using a LaChrom Elite HPLC System (Hitachi). Light scattering and the refraction index were measured using a Dawn Heleos II detector (Wyatt Technology) and an RI-101 detector (Shodex), respectively. The column was equilibrated at 20 °C with 20 mM Tris⋅HCl buffer (pH 8.0) containing 100 mM NaCl. Samples (2 mg/mL) were injected at a buffer flow rate of 0.5 mL/min. The obtained data were recorded and processed using ASTRA 6.1 software (Wyatt Technology).

Determination of Kinetic Parameters of CarA.

Initial velocities of the CarA reaction were calculated from the amount of H₂O₂, which was determined by the following procedure. Two hundred microliters of reaction mixture comprising 158 μL of water, 20 μL of 0.92 μg/mL CarA, 2 μL of 1 M Tris⋅HCl (pH 8.0), 4 μL of 50 mM DAOS, 4 μL of 50 mM 4-aminoantipyrine, 2 μL of 5,000 U/mL peroxidase, and various concentrations of carminic acid (0.007, 0.01, 0.015, 0.03, 0.05, 0.06, 0.08, 0.1, 0.2, 0.5, 1.0 mM) were incubated at 28 °C for 0, 3, 6, and 9 min, and the absorbance was measured at 593 nm. Kinetic parameters were calculated by the enzyme kinetics module in SigmaPlot 12.0 (SYSTAT Software). The equation used for data fitting was uncompetitive substrate inhibition: v = V_max/(1 + K_m/S + S/K_i).

Temperature Stability and Dependency.

For thermal stability estimations, CarA was preincubated for 15 min at temperatures between 10 and 70 °C. The remaining activity was determined by incubating the enzyme at 28 °C for 10 min. Two hundred microliters of the reaction mixture consisted of 1 μL of 0.044 mg/mL CarA, 2 μL of 1 M Tris⋅HCl (pH 8.0), 2 μL of 10 mM carminic acid (in milliQ water), 4 μL of 50 mM DAOS, 4 μL of 50 mM 4-aminoantipyrine, and 10 U peroxidase. The amount of H₂O₂ was determined from the absorbance at 593 nm. For thermal dependency estimations, 100 μL of the reaction mixture consisted of 1 μL of 0.044 mg/mL CarA, 1 μL of 1 M Tris⋅HCl (pH 8.0), and 1 μL of 10 mM carminic acid (in milliQ water). The experiments were carried out in triplicate at 10, 20, 25, 30, 35, 40, 45, 50, 60, and 70 °C for 10 min. A Chill Heat CHT-101 (IWAKI Asahi Techno Glass) was used for incubation at 10 to 50 °C, and a Dry Thermo Unit DTU-1B (TAITEC) was used for incubation at 60 and 70 °C. The amounts of reaction products were determined by HPLC-PDA (photodiode array). One unit (μmol/min) is defined as the amount of enzyme that catalyzes the conversion of 1 μmol of substrate per minute.

pH Stability and Dependency.

For pH stability estimations, CarA was preincubated on ice for 30 min at each pH, and then the reaction was carried out at 28 °C in 200 μL of a reaction mixture containing 1 μL of 0.044 mg/mL CarA, 2 μL of 1 M Tris⋅HCl (pH 8.0), 2 μL of 10 mM carminic acid (in milliQ water), 4 μL of 50 mM DAOS, 4 μL of 50 mM 4-aminoantipyrine, and 10 U peroxidase. The amount of H₂O₂ was determined from the absorbance at 593 nm. For pH dependency estimations, 100 μL of the reaction mixture consisted of 1 μL of 0.044 mg/mL CarA, 25 μL of 0.4 M Britton–Robinson buffer (pH 4.0 to 10.0 [1.0 pH unit]), and 2 μL of 10 mM carminic acid (in milliQ water). The experiments were carried out for 10 min at 28 °C. The reaction was stopped by adding 50 μL of acetonitrile. The amount of reaction product was determined by HPLC-PDA. One unit (μmol/min) is defined as the amount of enzyme that catalyzes the conversion of 1 μmol of substrate per minute.

Identification of a Cofactor of CarA.

The cofactor was released from CarA by extraction with ethanol according to the following procedure. Eighty microliters of a 138 μM CarA solution was incubated with 120 μL of ethanol for 30 min at room temperature. Subsequently, the solution was centrifuged to remove aggregates. The supernatant was evaporated, and the obtained solids were dissolved in 100 μL of milliQ water. The resultant solution was loaded onto an HPLC column (πNAP, 4.6 × 150 mm; Nacalai Tesque). Separation was performed at 40 °C with a gradient from 0% (100% water with 0.1% formic acid) to 35% acetonitrile in 15 min at a flow rate of 1 mL/min. On HPLC (Shimadzu), the retention time and MS of the cofactor were compared with those of authentic FAD and flavin mononucleotide (FMN).

Determination of the Concentrations of CarA and FAD.

The concentration of CarA was determined from the absorption at 280 nm using the theoretical extinction coefficient of ε280 = 46,325 M⁻¹⋅cm⁻¹. FAD was separated from holoprotein by the method described in Identification of a cofactor of CarA. The concentration of FAD was calculated from the 450-nm absorbance using a molar extinction coefficient of ε450 = 11,300 M⁻¹⋅cm⁻¹. Absorption spectra were recorded using a UV-1700 spectrophotometer (Shimadzu).

Stoichiometry.

The amount of H₂O₂ was determined by an enzymatic photometric assay (36). Two hundred microliters of the reaction mixture consisted of 2 μL of 0.261 mg/mL CarA, 2 μL of 1 M Tris⋅HCl (pH 8.0), 10 μL of 10 mM carminic acid (in milliQ water), 4 μL of 50 mM DAOS, 4 μL of 50 mM 4-aminoantipyrine, and 10 U peroxidase. The consumption of O₂ was measured with an “oxygen electrode” (OXYT-1; Hansatech Instruments). One milliliter of the reaction mixture consisted of 10 μL of 0.261 mg/mL CarA, 10 μL of 1 M Tris⋅HCl (pH 8.0), and 5 μL of 100 mM carminic acid (in milliQ water). Then, the amounts of carminic acid and keto carminic acid (X₁, X₂) were determined by HPLC-PDA. One hundred microliters of the reaction mixture consisted of 1 μL of 0.261 mg/mL CarA, 1 μL of 1 M Tris⋅HCl (pH 8.0), and 5 μL of 10 mM carminic acid (in milliQ water).

Effects of Various Inhibitors and Metals.

For determination of the effects on CarA, 100 μL of reaction mixtures containing 1 μL of 0.044 mg/mL CarA, 1 μL of 1 M Tris⋅HCl (pH 8.0), 1 μL of 10 mM carminic acid (in milliQ water), and 1 mM each inhibitor (SI Appendix, Tables S2 and S3) were incubated at 28 °C for 10 min. The experiments were carried out in triplicate. The amounts of reaction products were determined by HPLC.

Substrate Specificity.

The following compounds were examined as to substrate specificity of CarA, AgCarA, and MtCarA at a final concentration of 0.1 mM: carminic acid, mangiferin, homoorientin, isovitexin, puerarin, emodin 8-glucoside, genistin, daidzin, rutin, naringin, and apigetrin for the CarA reaction. Instead of carminic acid, each of these compounds was added to the standard assay mixture. The production of the reaction product was detected by LC/MS. Reaction products of carminic acid were determined by MS and NMR spectroscopy. Other reaction products, which CarA synthesized by using alternate substrates, were identified by MS.

Time Courses of Cell Growth and Enzymatic Activity.

Microbacterium sp. 5-2b was cultured in 10 mL of 1/10 2×YT liquid medium for 48 h at 28 °C. The cells were harvested by centrifugation at 4,000 rpm and 4 °C for 10 min, and then washed twice with 10 mM potassium phosphate buffer (pH 7.0). The cells were suspended in 10 mL of potassium phosphate buffer (pH 7.0), and then inoculated 1% (vol/vol) into 100 mL of the following two types of media: 1) 1/10 2×YT liquid medium supplemented with 0.1% (wt/vol) carminic acid; and 2) 1/10 2×YT liquid medium. The cells were cultured at 28 °C. During cell growth, 1-mL samples were withdrawn every 3 h. The samples were used for measurement of OD₆₆₀, amount of carminic acid, and specific activity. The cell-free extracts were subjected to measurement of protein concentration and carminic acid degradation activity.

Circular Dichroism Analysis.

Circular dichroism (CD) measurements were carried out with a Jasco spectropolarimeter (model J-720W; Japan Spectroscopic) equipped with a thermal incubation system at 20 °C with a 1-mm-path-length cell. CD spectra were obtained at a protein concentration of 0.1 mg/mL in the far-ultraviolet region (200 to 240 nm).

Site-Directed Mutagenesis.

Site-directed mutagenesis was performed using a KOD-Plus Mutagenesis Kit (Toyobo), following the instructions of the manufacturer. The primers used for inverse PCR are given in SI Appendix, Table S2. A clone with the sequence for the desired mutation was chosen and transformed into E. coli BL21-Star (DE3). The recombinant cells were used for the overproduction and purification of the mutant enzymes.

Crystallization of MtCarA.

Crystallization conditions were initially screened using Crystal Screens 1 and 2 (Hampton Research), Crystal Screens Cryo 1 and 2 (Hampton Research), Wizard Screens I and II (Rigaku), PEGsII (Qiagen), Index (Hampton Research), PEGIon/PEGIon2 (Hampton Research), MembFac (Hampton Research), Footprint Screen (Molecular Dimensions), and Protein Complex Suite (Qiagen) with a Protein Crystallization System 2 (PXS2) at the Structural Biology Research Center, High Energy Accelerator Research Organization (37). Screening was performed by the sitting-drop vapor-diffusion method with crystallization drops consisting of 0.2 μL protein solution (10.2 mg/mL) and 0.2 μL screening solution at 293 K. Crystals of MtCarA were observed after 2 wk under condition 17 (30% polyethylene glycol [PEG] 4000, 0.2 M lithium sulfate, 0.1 M Tris⋅HCl, pH 8.5) of the Crystal Screen (Hampton Research). Before diffraction data collection, crystals of MtCarA were cryoprotected in a solution containing 30% ethylene glycol, 21% PEG 4000, 0.14 M lithium sulfate, and 70 mM Tris⋅HCl (pH 8.5) for 20 s.

X-ray Analysis and Structure Determination of MtCarA.

X-ray diffraction data were collected at 95 K, using an Eiger X 16M detector on BL-17A of the Photon Factory, KEK. Diffraction data were processed and scaled by XDS and XSCALE, respectively (38). The phases were determined by the MR-native SAD method (39). The coordinate of a pyranose 2-oxidase (PDB ID code 3PL8) was used as the initial model for MR calculation using BALBES (40), and the obtained initial phases were used for the MR-native SAD calculation by CRANK2 (41). Crystallographic refinement and model building were performed using phenix.refine (42) and Coot (43), respectively. All molecular graphics in this manuscript were prepared using PyMOL v2.4 (Schrödinger).

Manual Modeling of the MtCarA–Substrate Complex.

Initially, we superposed the crystal structure of pyranose 2-oxidase with a slow substrate (PDB ID code 2F5V; hereafter 2F5V) onto the crystal structure of MtCarA using the SSM routine of Coot (43). However, the positions of the FAD molecules significantly deviated between MtCarA and the superposed 2F5V structures. Therefore, the position of the slow substrate of the superposed 2F5V was inappropriate as a reference for preparing a substrate complex model of MtCarA. Then, we superposed the two structures using the residues around the active site. Residues 113 to 134, 270 to 280, 438 to 451, and 475 to 500 in MtCarA and their corresponding residues of 2F5V were used for least-squares fitting with LSQKAB (44), and the resultant rmsd was 0.473 Å. The two FAD molecules were well-superposed in this superposition. After the superposition, the pyranose ring of carminate or orientin was fitted onto the slow substrate’s pyranose ring in the superposed 2F5V using the pair-fitting routine of PyMOL v2.4 (Schrödinger). The rmsd values of the pyranose ring fittings were 0.237 and 0.097 Å for carminate and orientin, respectively.

Nucleotide Sequence Accession Numbers.

The nucleotide sequence data reported in this paper appear in the DNA Data Bank of Japan/GenBank database under accession no. LC387598 for carA.

Data Availability

Data of the X-ray crystal structure of FAD-dependent C-glycoside oxidase reported in this paper have been deposited in the Protein Data Bank (PDB accession ID 7DVE). All study data are included in the article and/or SI Appendix.