Anaerobic phloroglucinol degradation by Clostridium scatologenes

ABSTRACT Polyphenols are abundant in nature, and their anaerobic biodegradation by gut and soil bacteria is a topic of great interest. The O2 requirement of phenol oxidases is thought to explain the microbial inertness of phenolic compounds in anoxic environments, such as peatlands, termed the enzyme latch hypothesis. A caveat of this model is that certain phenols are known to be degraded by strict anaerobic bacteria, although the biochemical basis for this process is incompletely understood. Here, we report the discovery and characterization of a gene cluster in the environmental bacterium Clostridium scatologenes for the degradation phloroglucinol (1,3,5-trihydroxybenzene), a key intermediate in the anaerobic degradation of flavonoids and tannins, which constitute the most abundant polyphenols in nature. The gene cluster encodes the key C-C cleavage enzyme dihydrophloroglucinol cyclohydrolase, as well as (S)-3-hydroxy-5-oxo-hexanoate dehydrogenase and triacetate acetoacetate-lyase, which enable phloroglucinol to be utilized as a carbon and energy source. Bioinformatics studies revealed the presence of this gene cluster in phylogenetically and metabolically diverse gut and environmental bacteria, with potential impacts on human health and carbon preservation in peat soils and other anaerobic environmental niches. IMPORTANCE This study provides novel insights into the microbiota’s anaerobic metabolism of phloroglucinol, a critical intermediate in the degradation of polyphenols in plants. Elucidation of this anaerobic pathway reveals enzymatic mechanisms for the degradation of phloroglucinol into short-chain fatty acids and acetyl-CoA, which are used as a carbon and energy source for bacterium growth. Bioinformatics studies suggested the prevalence of this pathway in phylogenetically and metabolically diverse gut and environmental bacteria, with potential impacts on carbon preservation in peat soils and human gut health.

P olyphenols are structurally diverse metabolites that together constitute one of the most ubiquitous groups of plant natural products (1,2). Polyphenols have histori cally been used in the tanning industry (1,3). Their abundant production by plants, and subsequent degradation by microbes, makes polyphenols an important component of the biological carbon cycle. Certain anoxic biomes, such as peatlands and wetlands, accumulate large quantities of polyphenols and serve as important global carbon sinks (4,5). The enzyme latch hypothesis suggests that the absence of O 2 -dependent phenolic degradation in anaerobic environments leads to the accumulation of polyphe nols and consequent inhibition of microbial growth (4). However, recent studies of anoxic wetland soil microbiota demonstrated their ability to degrade diverse polyphe nols, including condensed tannins, generally thought to be recalcitrant (5). Due to their antioxidant properties, polyphenols have also been used as nutritional supplements (A0A0E3M5J5) and a putative 3-hydroxyacyl-CoA dehydrogenase (A0A0E3GQA2) in close proximity with CsPGR ( Fig. 1D) (17,24). This gene cluster appears to be also conserved in E. ramulus, E. oxidoreducens, and F. plautii  pathway of flavonoids involving formation of phloroglucinol (5,8). (B) The degradation pathway of tannins involving formation of phloroglucinol (13,15).
(C) Proposed microbial metabolic pathway for phloroglucinol (17,18,22). The known enzyme PGR is labeled with a blue ellipse. Each of the unknown enzymes in the pathway is labeled with a question mark in a pink ellipse. (D) The putative gene cluster involved in phloroglucinol degradation in environmental and gut bacteria.
Research Article mBio contains a major family transporter and transcriptional regulator, which are also found in conjunction with the putative PG degradation genes in other bacterial species.

Validification of CsPGR
The gene CsPGR was amplified by colony PCR and cloned into the pET28a by Gib son assembly. N-terminal His 6 -tagged CsPGR was then overexpressed in Escherichia coli BL21(DE3) and purified to near homogeneity by Ni-NTA affinity chromatography ( Supplementary Fig. 2). Incubation of phloroglucinol and NADPH with CsPGR resulted in time-dependent changes in the absorption spectra, indicating a reaction occurring at the aromatic nucleus ( Fig. 2A and B). Ultraviolet-visible (UV-Vis) difference spectra, obtained by subtracting the spectrum of the starting material from that of the product, exhibited a time-dependent increase in absorbance at 278 nm (Fig. 2B), consistent with the conversion of phloroglucinol into DPG (ε 278 nm = 29,986 M −1 cm −1 ) (23). During the reaction, a concomitant decrease in absorbance at 340 nm, indicating the consumption of NADPH (ε 340 nm = 6,220 M −1 cm −1 ) was also observed (Fig. 2B) Fig. 3).

Identification and characterization of an Mn 2+ -dependent dihydrophloroglu cinol cyclohydrolyase in C. scatologenes
The closest homolog that has been characterized of the "cyclase" enzyme (A0A0E3JZE4, putatively dihydrophloroglucinol cyclohydrolyase, CsDPGC) in the gene cluster is an Mn 2+ -dependent isatin hydrolase (26) from Labrenzia aggregata with only 25% overall sequence identity ( Supplementary Fig. 4A). Structure modeling and comparison showed that the metal binding sites of these two enzymes are highly conserved (Supplementary Fig. 4B and C). However, the essential residues involved in substrate recognition (Ile32 and Leu34) in isatin hydrolase are replaced by Trp20 and Tyr22 in CsDPGC, respectively (Supplementary Fig. 4A and C), suggesting a different substrate. Analogies between the mechanism of the isatin amide C-N hydrolase reaction and that of the proposed DPG retro-Claisen "C-C hydrolase" reaction ( Supplementary Fig. 4D) led us to hypothesize that CsDPGC catalyzes the cleavage of DPG, forming 3-hydroxy-5-oxo-hexanoate (Fig. 3A). The "cyclase" family also includes DpsY, an enzyme proposed to catalyze C-C bond forming polyketide cyclization in the biosynthesis of daunorubicin (27). We purified the recombinant CsDPGC to homogeneity ( Supplementary Fig. 2) and next performed enzyme activity assays with CsDPGC as isolated or in the presence of additional Mn 2+ , or other divalent metals (Fig. 3B). Incubation of CsDPGC with DPG resulted in time-dependent decrease in 278 nm ( Fig. 3C and D). Without additional divalent metal, enzyme activity was barely detectable suggesting that the recombinant protein as purified was mostly in its apo form. Among all tested divalent metals, Mn 2+ was optimal, and the enzyme retained about 25% activity when Co 2+ was used as the cofactor (Fig. 3B). The enzyme activity of DPGC was inhibited by Zn 2+ (Fig. 3B) similar to that of isatin hydrolase (26). The spectrophotometric assays varying substrate concentra tions were carried out to measure enzyme kinetic parameters (k cat = 1.04 ± 0.10 s −1 , K M = 0.19 ± 0.04 mM) (Fig. 3E).

Research Article mBio
CsTfD with CsPGR-CsDPGC reaction mixtures resulted in changes in the absorption spectra ( Fig. 4A and B), UV-Vis difference spectra, obtained by subtracting the spectrum of the starting material (CsPGR-CsDPGC reaction mixtures) from that of the product, exhibited a maximum (λmax) at 281 nm (  6D) much lower than that for NADPH, which could only be estimated to be 1.7 mM ( Supplementary Fig. 6B). The enzyme has a much lower activity for NADPH-dependent reduction of acetoacetate (0.02% relative to TAA) ( Supplementary Fig. 6E), and the K M is 3.69 mM for acetoacetate ( Supplementary Fig. 6F). CsTfD favors (S)-3-hydroxybutyrate Research Article mBio over (R)-3-hydroxybutyrate as its substrate for the reverse oxidation reaction (Fig. 4D), suggesting that the physiological substrate of CsTfD and the product of CsDPGC, 3hydroxy-5-oxohexanoate should be in its (S)configuration. Therefore, CsTfD is an NAD(P) + -dependent (S)-3-hydroxy-5-oxo-hexanoate dehydrogenase.

Identification and characterization of a triacetate acetoacetate-lyase in C. scatologenes
Sequence and structure analyses suggested that the as-annotated 3-keto-5-aminohexa noate cleavage enzyme in C. scatologenes (A0A0E3M5J5, denoted here as CsTAL) could have been a Zn 2+ -dependent triacetate acetoacetate-lyase (TAL). Compared with the structure of 3-keto-5-aminohexanoate cleavage enzyme (2Y7F [32]), most of the residues in the substrate binding pocket of CsTAL are highly conserved ( Supplementary Fig. 7). Strikingly, an exception is noted for Glu14 which is required to interact with the 5-amino group of 3-keto-5-aminohexanoate and is replaced by Trp16 in CsTAL (Supplementary Fig. 7A and C). This substitution is consistent with the hypothesis of the reaction substrate being 3,5-dioxohexanoate, forming acetoacetate and acetoacetyl-CoA (Fig. 5A).
To test this hypothesis, CsTAL was purified to near homogeneity by Ni-NTA affinity chromatography ( Supplementary Fig. 2). Incubation of CsTAL with triacetate and acetyl-CoA resulted in the formation of acetoacetate and acetoacetyl-CoA, which can be monitored at 340 nm due to the consumption of NADH in coupling enzyme reactions with D-3-hydroxybutyrate dehydrogenase from Pseudomonas fragi (33) and (S)-3hydroxybutyryl-CoA dehydrogenase from C. acetobutylicum (34), respectively ( Fig. 5A through C). The CsTAL enzyme activity catalyzing the reverse reaction with acetoacetate and acetoacetyl-CoA as substrates was also confirmed in couple with CsTfD ( Fig. 5A and D).

Reconstitution of the phloroglucinol degradation pathway
In vitro reconstitution of the phloroglucinol degradation pathway was carried out with purified recombinant C. scatologenes enzymes (PGR, DPGC, TfD, and TAL) and phloroglu cinol, NADPH, and acetyl-CoA as substrates. The reaction mixture and controls were analyzed by LC-MS (liquid chromatography-mass spectrometry), with separation carried out on a C18 column, and detection by ESI-MS (electrospray ionization mass spectrome try) in the negative ionization mode. Incubation of phloroglucinol (m/z = 125) (Supple mentary Fig. 8A) and NADPH with CsPGR led to a decrease of the phloroglucinol peak, and the appearance of a new peak corresponding to DPG (m/z = 127) (Supplementary Fig. 8B), consistent with the NADPH-dependent reduction of phloroglucinol to DPG. Incubation of phloroglucinol and NADPH with CsPGR and CsDPGC led to a decrease in the DPG peak, and the appearance of a new peak corresponding to (S)-3-hydroxy-5oxohexanoate (m/z = 145) ( Supplementary Fig. 8C), suggesting that CsDPGC catalyzes the ring-opening C-C bond cleavage of DPG. Incubation of phloroglucinol and NADPH with CsPGR, CsDPGC, and CsTfD results in the decrease of the (S)-3-hydroxy-5-oxohexanoate peak, and appearance of a new peak corresponding to triacetate (m/z = 143), consistent with the NADP + -dependent oxidation of (S)-3-hydroxy-5-oxohexanoate to form triacetate by CsTfD ( Supplementary Fig. 8D). Incubation of phloroglucinol, NADPH, and acetyl-CoA with all four enzymes CsPGR, CsDPGC, CsTfD, and CsTAL led to the disappearance of the triacetate peak and appearance of a new peak corresponding to acetoacetate (m/z = 101) ( Supplementary Fig. 8E). Further experiments were conducted to ascertain the identity of the enzymatic reaction products, including dihydrophologlucinol, 3-hydroxy-5-oxo-hexanoate, triacetate, acetoacetyl-CoA, and acetoacetate. Dihydrophloroglucinol, triacetate, and acetoacetyl-CoA were analyzed by high-resolution LC-UV/MS, showing co-elution with commercial standards (Fig. 6A and C through E; Supplementary Fig. 9). 3-Hydroxy-5-oxohexanoate, formed by both dihydrophologlucinol hydrolysis catalyzed by DPGC and triacetate reduction catalyzed by TfD, was also analyzed by high-resolution LC-MS and showed co-elution of the two reaction products ( Fig. 6B; Supplementary Fig. 9).
Acetoacetate, formed from the sequential reactions of DPGC, TfD, and TAL, could not be detected by high-resolution LC-MS, likely due to decarboxylation to form acetone during the workup and ionization processes. However, derivatization with 2,4-dinitrophenylhy drazine (DNPH) followed by high-resolution LC-MS analysis demonstrated the formation of DNPH-acetone, showing co-elution with a commercial standard ( Fig. 6F; Supplemen tary Fig. 9). The LC-MS data are consistent with the proposed phloroglucinol degradation pathway (Fig. 7A).

Phloroglucinol was used as a carbon and energy source for C. scatologenes growth
Next, we decided to investigate the physiological role of the gene cluster in utilizing phloroglucinol as a carbon and energy source. C. scatologenes was anaerobically grown in a defined medium with glucose or phloroglucinol as a carbon source. The C. scatolo genes cells grew robustly when glucose was provided, and the culture turned turbid with the optical density at 600 nm reaching 1.4 within 2-3 d. By contrast, cells grew much slower when phloroglucinol was provided as a carbon source, the OD 600 reached 0.32 after prolonged incubation at 37°C for 7-8 d (Supplementary Fig. 10A). The OD 600 of the  Fig. 11). The transcript levels of PGR, DPGC, TfD, and TAL genes in C. scatologenes were investigated using reverse transcription quantitative polymerase chain reaction (qPCR ; Table S2; Supple mentary Fig. 12). Cells grown on phloroglucinol as the sole carbon source were insuffi cient for RNA isolation; therefore, cells were grown on glucose and phloroglucinol instead. Compared with growth on glucose alone, growth on glucose and phloroglucinol led to a 4.2-to 5.7-fold induction of the four genes.

Presence of the phloroglucinol degradation pathway in diverse bacteria
To examine the prevalence of this pathway, we employed cblaster (version 1.

Research Article mBio
Supplementary data 3). The genome neighborhoods of the DPGC homologs were examined using the Enzyme Function Initiative Genome Neighborhood Tool (37), revealing the presence of gene clusters encoding this phloroglucinol degradation pathway in phylogenetically and metabolically diverse bacteria from different environ ments ( Fig. 8; Supplementary data 4). The gene cluster was present in several bacteria that are reported to degrade phloroglucinol, including strict anaerobic Gram-positive butyrate-producing fermenting bacteria from the environment and digestive system of humans and herbivorous animals (E. ramulus, E. oxidoreducens, F. plautii, and C. scatologenes reported here), and the facultative anaerobic phototrophic Betaproteobacterium Rubrivivax gelatinosus isolated from food wastewater (38). In addition, the gene cluster was also present in organisms Research Article mBio that have not previously been reported to degrade phloroglucinol, including marine alpha-and gamma-proteobacteria, anaerobic Gram-negative and Gram-positive sulfatereducing bacteria, and sulfate-reducing archaea belonging to the genus Archaeoglobus, which may play roles in polyphenol mineralization in marine, coastal, and estuarine sediments. Presence of the gene cluster in Mesorhizobium loti and Bradyrhizobium sp., which were previously reported to degrade flavonoids to produce phloroglucinol, suggests that certain strains of these soil bacteria and plant symbionts or pathogens may be able to further degrade phloroglucinol. Presence of the gene cluster in various Mycobacterium species, which are inhabitants of soil and water and are also human and animal pathogens, suggests an ability to degrade polyphenols. Notably, the gene cluster in Mycobacterium species also contains phloretin hydrolase, which was previously reported in Mycobacterium abscessus (39). Collectively, these observations suggest a central role for this phloroglucinol degradation pathway in mobilization of phenolic carbons by diverse bacteria in different environments.

DISCUSSION
Through a combination of bioinformatics, biochemical, and biophysical studies, we have identified and characterized the enzymes involved in the anaerobic phloroglucinol degradation pathway in C. scatologenes. In this pathway, the key retro-Claisen C-C cleavage of DPG to (S)-3-hydroxy-5-oxo-hexanoate is catalyzed by DPGC, a member of the "cyclase" family. NAD(P)-dependent oxidation of (S)-3-hydroxy-5-oxo-hexanoate is  Research Article mBio catalyzed by a stereospecific NAD(P) + -dependent dehydrogenase to form triacetate, followed by the second C-C cleavage catalyzed by TAL, a member of the "beta-keto acid cleavage enzyme" family (40). The gene cluster also contains a homolog of thiolase, which catalyzes the conversion of acetoacetyl-CoA into acetyl-CoA, and acetyl/butyryl-CoA:acetoacetyl-CoA transferase, which converts acetoacetate to acetoacetyl-CoA. Acetoacetyl-CoA and acetyl-CoA are intermediates in the butyrate fermentation pathway, allowing for the production of ATP via phosphate acetyl/butyryl-transferase and acetate/butyrate kinase and further energy conservation via crotonyl-CoA reduction (Fig. 7A). Previously, the anaerobic phloroglucinol degradation has been studied in the Gram-positive bacterium E. oxidoreducens (17) and the Gram-negative bacterium P. acidigallici (whose genome is currently not available) (18). The degradation process is believed to proceed through similar pathways in both bacterial strains. However, while P. acidigallici degrades phloroglucinol to produce a stoichiometric amount of acetate, E. oxidoreducens requires an external electron donor, such as H 2 or formate, to grow on phloroglucinol and generates both acetate and butyrate. The differences in the need for an external electron donor are attributed to the divergent mechanisms employed by the two bacterial strains to regenerate NADPH, a requisite for phloroglu cinol reduction. Specifically, while the oxidation of 3-hydroxy-5-oxo-hexanoate in P. acidigallici produces NADPH, the oxidation of the 3-hydroxyacid or 3-hydroxyacyl-CoA intermediate in E. oxidoreducens is thought to generate NADH, necessitating NADPH generation via hydrogenase or formate dehydrogenase. Our experiments with recombi nant C. scatologenes TfD demonstrate its preference for NADH over NADPH and that growth on phloroglucinol generates both acetate and butyrate, similar to E. oxidoredu cens. In our experiments, growth of C. scatologenes required the inclusion of amino acids in its growth medium, which could function as the external electron donor instead of H 2 or formate.

New enzymes characterized in this study
Besides the anaerobic phloroglucinol degradation pathway described above, there exist several other known mechanisms for affecting anaerobic phenolic ring cleavage (41). These pathways typically involve either reductive or oxidative dearomatization, followed by ring cleavage via a retro-Claisen reaction. Degradation of phenol, catechol, and hydroquinone involves the formation of a benzoyl-CoA intermediate (41), which undergoes reductive dearomatization catalyzed by benzoyl-CoA reductase, followed by hydrolytic ring cleavage. In some bacteria, resorcinol degradation takes place via reductive dearomatization catalyzed by resorcinol reductase (42), followed by hydrolytic ring cleavage. Alternatively, in other bacteria, resorcinol degradation proceeds through hydroxylation to form hydroxyhydroquinone (43), followed by oxidative dearomatization, and then hydrolytic cleavage. The degradation of pyrogallol involves its conversion into phloroglucinol by the molybdoenzyme pyrogallol-phloroglucinol transhydroxylase, followed by degradation through the anaerobic phloroglucinol degradation pathway. Phloroglucinol and resorcinol have hydroxyl groups that are located in the meta position with respect to each other. As a result, their keto tautomers (1,3,5-trioxocyclohexane and 1,3-dioxocyclohexene) are more stable, which could explain why they can be reduced relatively easily by the NADPH-dependent PGR and the ferredoxin-dependent resorcinol reductase, respectively (44,45). By contrast, the reduction of benzoyl-CoA is believed to follow a mechanism similar to Birch reduction, requiring a much stronger reductant. In class I benzoyl-CoA reductase, this reduction is driven by stoichiometric ATP hydrolysis, while in class II benzoyl-CoA reductase, it is driven by electron bifurcation (45).
Our bioinformatics analysis revealed the presence of gene clusters encoding the phloroglucinol degradation pathway in metabolically diverse bacteria, including strict anaerobic Gram-positive fermenting bacteria, strict anaerobic sulfate-reducing bacteria and archaea, and facultative anaerobic Gram-negative bacteria. These organisms inhabit diverse environmental niches ( Fig. 8; Supplementary data 4), and some are plant or animal commensals or pathogens, suggesting that microbial phloroglucinol degradation ability is widespread. Identification of these gene clusters will facilitate further investiga tion of the fate of phenolic carbon in different ecologically important biomes.
The findings described here are also of great relevance to polyphenol biochemistry in the human intestinal microbiome. Studies of gut microbial polyphenol metabolism have focused on the production of bioactive polyphenol derivatives, such as the conversion of isoflavones to equol (46) and ellagitannins to urolithin (47). These involve reduction, dehydroxylation, and decarboxylation reactions, but not phenolic ring opening. By contrast, the phloroglucinol degradation pathway allows for the mobilization of phenolic carbons for fermentative energy metabolism, which is particularly important for the microbiota inhabiting the nutrient-scarce distal gut. Moreover, elucidation of this pathway reveals enzymatic mechanisms for the conversion of polyphenols into shortchain fatty acids (acetate and butyrate), important mediators of gut health (48,49), providing another route through which microbial polyphenol metabolism can influence human health.

Materials and general methods
Lysogeny broth (LB) was purchased from Sangon Biotech (Shanghai, China). Ultrapure deionized water from Millipore Direct-Q was used in this work. HIS*BIND RESIN (69670-10 mL) was purchased from EMD Millipore Corp. (USA). All protein purification chromato graphic experiments were performed on gravity columns. Phloroglucinol dihydrate was purchased from Aladdin (Shanghai, China). DPG was synthesized by chemical reduction of phloroglucinol with NaHB 4 according to the method of Patel et al. (23). The purity and identity of DPG were confirmed by UV spectroscopic and LC-MS analyses. Methyl 3,5-dioxohexanoate was purchased from Bidepharm (Shanghai, China), and triacetate was obtained by the ester hydrolysis of methyl 3,5-dioxohexanoate with 5N LiOH in 75% methanol system for 5 h. Other chemicals unless otherwise specified, including acetyl-CoA (A281-10MG) and acetoacetyl-CoA (A1625-5MG) and NAD(P)(H), were purchased from Sigma-Aldrich. UV-Vis spectroscopic measurements were monitored using a Biotek synergy 2 reader for 96-well plates and using a NANODROP ONE (Thermo Fisher Scientific, MA, USA ) otherwise. Kinetic parameters for the enzyme assays were extracted using GraphPad Prism 5.0.

Expression and purification of CsPGR, CsDPGC, CsTfD, and CsTAL
E. coli BL21 (DE3) cells were transformed with the pET-28a (+) plasmids encoding CsPGR, CsDPGC, CsTfD, and CsTAL genes and plated on LB agar supplemented with 50 µg/mL kanamycin. Transformants were grown in the LB medium (200-300 mL) at 37°C in a shaking incubator at 200 rpm. When OD 600 reached 0.6-0.8, the temperature was decreased to 18°C, and 0.3 mM isopropyl β-D-1-thiogalactopyranoside was added to induce the production of the protein of interest. After 16-20 h, cells were harvested by centrifugation (6,000 × g for 10 min at 4°C). Cells (CsPGR, CsTfD, and CsTAL) were resuspended in 20 mL of lysis buffer [50 mM Tris-HCl, pH 8.0, 1 mM phenylmethanesul fonyl fluoride (PMSF), 0.2 mg/mL lysozyme, 0.03% Triton X-100, and 0.02 mg/mL DNase I]. The cell suspension was frozen in a −80°C freezer and then thawed and incubated in a 25°C water bath for 30 min to allow for cell lysis. The cells containing CsDPGC were resuspended in 20 mL of lysis buffer (20 mM Tris-HCl, pH 7.5, 200 mM KCl, 0.03% Triton X-100, and 1 mM PMSF) and were ruptured by probe sonication. About 2 mL of 11% streptomycin sulfate (dissolved in water) was added to the cell lysate followed by gentle mixing and centrifugation (20,

UV-Vis spectrophotometric assays of CsPGR
A 40-µL reaction mixture, containing 50 mM Tris-HCl, pH 7.5, 2 mM phloroglucinol, 2 mM NAD(P)H, and 1 µM of CsPGR was incubated for 0-6 min at room temperature (RT). The absorbance from 190 to 850 nm was monitored. To measure the Michaelis-Menten kinetic constants, the absorbance at 340 nm of a 200-µL reaction mixture, containing 50 mM Tris-HCl, pH 7.5, 0.05 µM of CsPGR, 0-3.0 mM of phloroglucinol, and 0-0.8 mM NADPH in a 96-well plate, was monitored at 15 s intervals for 2 min at RT. ΔA 340 nm and the extinction coefficient of NADPH (6,220 M −1 cm −1 ) were used to calculate the reaction rates.  [23]) were used to calculate the reaction rates.

UV-Vis spectroscopic assays of CsTfD
In a typical end point assay, a 40-µL reaction mixture, containing 50 mM Tris-HCl, pH 7.5, 1 mM triacetate, 1 mM NADPH, and 0.1 µM of CsTfD, was incubated for 10 min at RT, followed by UV-Vis absorbance scan from 190 to 850 nm. A 10-fold less CsTfD (10 nM) was used when NADPH was replaced by NADH as the reductant. To measure the Michaelis-Menten kinetic constants, the concentration of CsTfD was varied from 5 to 300 nM, the concentration of the substrate was varied in the range of 0-1.0 mM for NADPH or 0-0.75 mM for NADH, in the presence of a saturating concentration of 2.0 mM triacetate. In another set of experiments, the concentration of NADPH was fixed at 1.0 mM, while the triacetate concentration was varied from 0 to 5 mM, or acetoacetate concentration was varied from 0 to 50 mM. The reaction system was monitored at 5-60 s intervals for 1-10 min at RT. ΔA 340 nm and the extinction coefficient of NAD(P)H (6,220 M −1 cm −1 ) were used to calculate the reaction rates.

Growth of C. scatologenes with phloroglucinol
C. scatologenes (DSM 757, ATCC 25775) was purchased from DSMZ (German Collection of Microorganisms and Cell Cultures). The rich medium was prepared by dissolving 10 g beef extract, 15 g casitone, 2.5 mL 1 N NaOH, 0.5 g yeast extract, 0.5 g K 2 HPO 4 , 50 µL 0.1% wt/vol Na-resazurin solution, 0.4 g D-glucose, 0.1 g cellobiose, 0.1 g maltose, 0.1 g starch (soluble), and 0.5 g/L L-cysteine hydrochloride in 100 mL distilled water and adjusting the pH to 7.0. To prepare the defined medium (56) O, 20 µg biotin, 20 µg folic acid, 100 µg pyridoxine-HCl, 50 µg thiamine-HCl, 50 µg riboflavinHCl, 50 µg nicotinic acid, 50 µg pantothenic acid, 10 µg cyanocobalamin, 50 µg p-amino benzoic acid, and 50 µg lipoic acid were dissolved in 1 L distilled water. And add the following L-amino acids to a final concentration of 1 mM each: glycine, valine, leucine, isoleucine, methionine, histidine, arginine, phenylalanine, tyrosine, and tryptophan. After all supplements were added, 5% wt/vol cysteine HCI (pH 6.0) was added until reduction was effected (as judged by the discoloration of the resazurin). Cells were inoculated into the rich growth medium and cultivated anaerobically at 37°C for 2 d, spun down at 6,000 × g and resuspended in the defined medium. About 100 µL portions of the starter culture were transferred into three anaerobic bottles each containing 5.0 mL defined medium with none, 25 mM glucose, or 25 mM phloroglucinol, respectively. After 3-8 d of incubation at 37°C, the cultures with added glucose or phloroglucinol became turbid, indicating bacterial growth. The OD 600 of the negative control was less than 0.08.

LC-MS detection of products of phloroglucinol fermentation
The formation of acetate and butyrate during phloroglucinol fermentation was detected by LC-MS analysis using an Agilent 6420 Triple Quadrupole instrument. The 10 µL samples were chromatographed on an Agilent ZORBAX SB-C18 column (4.6 × 250 mm 2 , product number 880975-902) with a flow rate of 0.5 mL/min, eluting with 0.1% formic acid in water (solvent A) and 0.1% formic acid in acetonitrile (solvent B). A gradient of 0-4 min 0%−60% B, 4-8 min 60%-100% B, and 8-12 min 100% B was used. The mass spectrometry detection was performed under positive electrospray ionization mode [ESI (+)], and the analytes were monitored in multiple-reaction monitoring mode (57) .

Protein identification by SDS-PAGE and mass spectrometry
Cells were harvested by centrifugation, lysed by boiling in Laemmli loading buffer, and analyzed on a 12.5% SDS-PAGE gel. Prominent protein bands induced by growth on phloroglucinol were manually excised. After in-gel digestion and extraction, the peptide mixtures were analyzed twice on an Orbitrap Fusion with EASY-nLC 1200 (Thermo Fisher Scientific). Resulting tandem mass spectra were searched against an appropriate protein database (retrieved from IMG) using Mascot (Matrix Science) and Proteom Discoverer V1.3 (Thermo Fisher Scientific) with "Trypsin" enzyme cleavage, static cysteine alkylation by chloroacetamide, and variable methionine oxidation. The peptide mass fingerprint search against the C. scatologenes (DSM 757) protein database GCF_000968375.1 by Proteome Discoverer (version 1.3) algorithm was used to identify proteins based on probability-based Mowse scoring algorithm with a confidence level of 95%.

Real-time fluorescence quantitative PCR analyses
RNA was extracted from cultures that utilized either glucose or phloroglucinol + glucose as carbon sources, using RNAprep pure Cell/Bacteria Kit (TIANGEN). To perform reverse transcription, 600 ng RNA was added to a 20-µL reaction mixture containing 1× Hiscript III qRT SuperMix (Vazyme), according to the manufacturer's recommended protocol. The reverse transcriptase was then inactivated at 85°C for 5 s, and the resulting cDNA was stored at −70°C. For a typical real-time fluorescence qPCR reaction, a 20-µL reaction mixture was prepared, containing 2 µL of 5× diluted cDNA, 0.2 µM genespecific forward, and reverse primers and 1× AceQ Universal SYBR qPCR Master Mix (Vazyme). qPCRs were performed on an Applied Biosystems 7500. The primers were designed using Primer-BLAST (Table S2). All reagents and consumables for the experiments were RNase-free.