Identification of phenol- and p-cresol-producing intestinal bacteria by using media supplemented with tyrosine and its metabolites.

To identify intestinal bacteria that produce phenols (phenol and p-cresol), we screened 153 strains within 152 species in 44 genera by culture-based assay using broth media supplemented with 200 µM each of tyrosine and its predicted microbial metabolic intermediates (4-hydroxyphenylpyruvate, DL-4-hydroxyphenyllactate, 3-(p-hydroxyphenyl)propionate, 4-hydroxyphenylacetate and 4-hydroxybenzoate). Phenol-producing activity was found in 36 strains and p-cresol-producing activity in 55 strains. Sixteen strains had both types of activity. Phylogenetic analysis based on the 16S rRNA gene sequences of strains that produced 100 µM or more of phenols revealed that 16 phenol producers belonged to the Coriobacteriaceae, Enterobacteriaceae, Fusobacteriaceae and Clostridium clusters I and XIVa; four p-cresol-producing bacteria belonged to the Coriobacteriaceae and Clostridium clusters XI and XIVa; and one strain producing both belonged to the Coriobacteriaceae. A genomic search for protein homologs of enzymes involved in the metabolism of tyrosine to phenols in 10 phenol producers and four p-cresol producers, the draft genomes of which were available in public databases, predicted that phenol producers harbored tyrosine phenol-lyase or hydroxyarylic acid decarboxylase, or both, and p-cresol producers harbored p-hydroxyphenylacetate decarboxylase or tyrosine lyase, or both. These results provide important information about the bacterial strains that contribute to production of phenols in the intestine.

INTRODUCTION

The more than 100 trillion bacteria in the human intestinal tract form a complicated ecosystem (Bäckhed et al. 2005). These bacteria produce many metabolites that can either harm or benefit host health (Nicholson et al. 2012). Short-chain fatty acids, which are produced mainly through the fermentation of carbohydrates, not only are used as energy sources for the host’s colonocytes but also have anti-inflammatory effects (Verbeke et al. 2015). Polyamines in the intestinal lumen enhance longevity and delay senescence (Kibe et al. 2014). Equol produced by intestinal microbiota reduces the risk of prostate cancer (Sugiyama et al. 2013). In contrast to these beneficial metabolites, intestinal secondary bile acid concentrations are closely related to the incidence of colorectal cancer (Ajouz, Mukherji and Shamseddine 2014), and indole, which is a uremic toxin, promotes the progression of chronic kidney disease (Evenepoel et al. 2009; Ito and Yoshida 2014). Because of the increasing importance of metabolites to host health, many metabolomic analyses have been performed to identify novel factors. For example, it has been found that trimethylamine is a risk factor for cardiovascular disease (Wang et al. 2011). As shown in these studies, we are aware of the role of metabolites in host health, but few studies have attempted to identify the bacteria involved in producing each type of metabolite in the colon. Obtaining information about the bacteria producing these metabolites would provide new clues to our understanding of disease from the perspectives of morbidity risk evaluation and the establishment of prevention methods.

Phenols (phenol and p-cresol) are microbial metabolites produced from tyrosine (Windey, De Preter and Verbeke 2012). Phenol exhibits cytotoxicity and increases paracellular permeability in vitro (Verbeke et al. 2015); it acts as a promoter of skin cancer in an animal model (Boutwell and Bosch 1959). p-cresol exhibits cytotoxicity and genotoxicity and reduces endothelial barrier function in vitro (Andriamihaja et al. 2015; Verbeke et al. 2015). p-cresyl sulfate, a sulfate-conjugate of p-cresol, suppresses Th1-type cellular immune responses in mice (Shiba et al. 2014); an increase in its levels is associated with chronic kidney disease-associated events such as cardiovascular disease (Meyer and Hostetter 2012; Ito and Yoshida 2014). Furthermore, phenol and p-cresol suppress the differentiation of keratinocytes in humans and cause dermal disorders in mice (Iizuka et al. 2009a,b). Although studies focusing on the relationship between phenols and various diseases have been accumulating, to our knowledge there has been no comprehensive study to identify the bacteria contributing to phenol- andp-cresol-production, with the exception of reports focused only on the genus Clostridium or on limited species (Bone, Tamm and Hill 1976; Elsden, Hilton and Waller 1976; Smith and Macfarlane 1996).

Here, we screened bacteria producing phenol or p-cresol, or both, using 153 strains within 152 species in 44 genera—mainly of intestinal bacteria—to determine which strains had the ability to produce phenol or p-cresol or both. Strains that screened positive were analyzed to determine the relationship between the ability to produce phenols and phylogenetic classification. They were then genetically analyzed to predict their metabolic pathways from tyrosine to phenols.

MATERIALS AND METHODS

Chemicals

DL-4-hydroxyphenyllactic acid, 4-hydroxyphenylpyruvic acid, 4-hydroxyphenylacetic acid and 4-hydroxybenzoic acid were obtained from Tokyo Chemical Industry Co., Ltd. (Tokyo, Japan). Tyrosine and 3-(p-hydroxyphenyl)propionate were purchased from Wako Pure Chemical Industries, Ltd. (Osaka, Japan). Substrate solution was prepared by dissolving these compounds together in 18 mM NaOH solution (final 2 mM each) and filtered for sterilization through a 0.20 µm cellulose acetate filter (Toyo Roshi Kaisha, Ltd., Tokyo, Japan).

Bacterial strains and culture conditions

The 153 bacterial strains and culture conditions used for screening are listed in Table 1. The 153 strains represented 152 species found in the human gut habitat and their phylogenetic relatives; they accounted for about 70% of the common species detected in human feces (Qin et al. 2010). Two types of media (rich medium and poor medium) were used for culture. Rich medium was used for its growth efficiency: modified Gifu anaerobic medium broth (Nissui Pharmaceutical Co., Ltd., Tokyo, Japan) supplemented with 1% glucose; MRS broth (Nissui Pharmaceutical Co., Ltd.); Trypticase soy broth (Becton, Dickinson and Company, Franklin Lakes, New Jersey, USA); or peptone–yeast extract (PY) broth supplemented with 1% glucose was used. The PY broth (1 L) contained 5.0 g peptone, 5.0 g trypticase peptone, 10.0 g yeast extract, 0.5 g L-cysteine HCl • H2O, 4.0 g Na2CO3, 7 mL 0.07% hemin solution, 1.0 mL 0.1% resazurin solution, 0.04 g K2HPO4, 0.04 g KH2PO4, 0.4 g NaHCO3, 0.08 g NaCl, 8 mg CaCl2, 19 mg MgSO4 • 7H2O and 1 mg vitamin K1 (pH 6.9). Basal Medium (Bone, Tamm and Hill 1976), which contains Tripticase Peptone (Becton, Dickinson and Company) instead of casein hydrolysate, was used as poor medium. As glucose supplementation can have critical effects on the production of phenols (Smith and Macfarlane 1996), basal medium that did not contain glucose as a carbon source was selected. The substrate solution described above was added to rich medium or poor medium to prepare test medium (final 200 µM each). Bacterial strains were pre-cultured in 4 mL of rich medium, and aliquots (40 µL) were inoculated into 4 mL of test media and incubated statically at 37°C for 6 days. An anaerobic chamber (N2:CO2:H2 = 88:5:7) was used for culture, except in the case of three strains: Cl. perfringens YIT 6050T and Cl. difficile YIT 10084Twere cultured under O2 free N2 gas, and Staphylococcus epidermidis YIT 6049Twas cultured aerobically.

Table 1.

Bacterial strains used in this study, and culture conditions

No.	Species	Registration No.	Medium for culture
1	Acidaminococcus fermentans	YIT 6071T = ATCC 25085T	modified GAM + 1% Glucose broth
2	Acinetobacter baumannii	YIT 12295T = JCM 6841T	Trypticase Soy broth
3	Akkermansia muciniphila	YIT 11774T = ATCC BAA-835T	modified GAM + 1% Glucose broth
4	Anaerococcus hydrogenalis	YIT 12837T = JCM 7635T	modified GAM + 1% Glucose broth
5	Anaerococcus vaginalis	YIT 11698T = DSM 7457T	modified GAM + 1% Glucose broth
6	Anaerostipes caccae	YIT 10168T = DSM 14662T	modified GAM + 1% Glucose broth
7	Anaerostipes hadrus	YIT 10092T = DSM 3319T	modified GAM + 1% Glucose broth
8	Bacteroides caccae	YIT 10226T = JCM 9498T	modified GAM + 1% Glucose broth
9	Bacteroides dorei	YIT 12192	modified GAM + 1% Glucose broth
10	Bacteroides eggerthii	YIT 10227T = DSM 20697T	modified GAM + 1% Glucose broth
11	Bacteroides fragilis	YIT 6158T = ATCC 25285T	modified GAM + 1% Glucose broth
12	Bacteroides ovatus	YIT 6161T = ATCC 8483T	modified GAM + 1% Glucose broth
13	Bacteroides plebeius	YIT 12661	modified GAM + 1% Glucose broth
14	Bacteroides stercoris	ATCC 43183T	modified GAM + 1% Glucose broth
15	Bacteroides thethaiotaomicron	YIT 6163T = JCM 5827T	modified GAM + 1% Glucose broth
16	Bacteroides uniformis	YIT 6164T = JCM 5828T	modified GAM + 1% Glucose broth
17	Bacteroides vulgatus	YIT 6159T = ATCC 8482T	modified GAM + 1% Glucose broth
18	Bifidobacterium adolescentis	YIT 4011T = ATCC 15703T	modified PYG broth
19	Bifidobacterium animalis subsp. lactis	YIT 4121T = DSM 10140T	modified PYG broth
20	Bifidobacterium angulatum	YIT 4012T = ATCC 27535T	modified PYG broth
21	Bifidobacterium bifidum	YIT 4039T = DSM 20456T	modified PYG broth
22	Bifidobacterium breve	YIT 4014T = ATCC 15700T	modified PYG broth
23	Bifidobacterium catenulatum	YIT 4016T = ATCC 27539T	modified PYG broth
24	Bifidobacterium longumsubsp. infantis	YIT 4018T = ATCC 15697T	modified PYG broth
25	Bifidobacterium longum subsp. longum	YIT 4021T = ATCC 15707T	modified PYG broth
26	Bifidobacterium pseudocatenulatum	YIT 4072T = JCM 1200T	modified PYG broth
27	Blautia coccoides	YIT 6035T = JCM 1395T	modified GAM + 1% Glucose broth
28	Blautia hansenii	YIT 12129T = DSM 20583T	modified GAM + 1% Glucose broth
29	Blautia hydrogenotrophica	YIT 10080T = DSM 10507T	modified GAM + 1% Glucose broth
30	Blautia producta	YIT 6141T = JCM 1471T	modified GAM + 1% Glucose broth
31	Blautia schinkii	YIT 6177T = DSM 10518T	modified GAM + 1% Glucose broth
32	Butyrivibrio crossotus	YIT 10152T = DSM 2876T	modified GAM + 1% Glucose broth
33	Citrobacter freundii	YIT 6045T = JCM 1657T	Trypticase Soy broth
34	Citrobacter koseri	YIT 10117T = JCM 1658T	Trypticase Soy broth
35	Clostridium aminophilum	YIT 6167T = DSM 10710T	modified GAM + 1% Glucose broth
36	Clostridium aminovalericum	YIT 10174T = JCM 11016T	modified GAM + 1% Glucose broth
37	Clostridium asparagiforme	YIT 12840T = DSM 15981T	modified GAM + 1% Glucose broth
38	Clostridium bifermentans	YIT 6053T = JCM 1386T	modified GAM + 1% Glucose broth
39	Clostridium butyricum	YIT 10073T = JCM 1391T	modified GAM + 1% Glucose broth
40	Clostridium celerecrescens	YIT 6168T = DSM 5628T	modified GAM + 1% Glucose broth
41	Clostridium clostridioforme	YIT 6051T = JCM 1291T	modified GAM + 1% Glucose broth
42	Clostridium cochlearium	YIT 12837T = JCM 1396T	modified GAM + 1% Glucose broth
43	Clostridium cocleatum	YIT 6036T = JCM 1397T	modified GAM + 1% Glucose broth
44	Clostridium difficile	YIT 10084T = JCM 1296T	modified GAM + 1% Glucose broth
45	Clostridium ghonii	YIT 11479T = JCM 1400T	modified GAM + 1% Glucose broth
46	Clostridium glycolicum	YIT 6058T = JCM 1401T	modified GAM + 1% Glucose broth
47	Clostridium hathewayi	YIT 12259T = DSM 13479T	modified PYG broth
48	Clostridium hylemonae	YIT 12258T = DSM 15053T	modified PYG broth
49	Clostridium indolis	YIT 10077T = JCM 1380T	modified GAM +1% Glucose broth
50	Clostridium innocuum	YIT 10151T = DSM 1286T	modified GAM + 1% Glucose broth
51	Clostridium leptum	YIT 6169T = DSM 753T	modified GAM + 1% Glucose broth
52	Clostridium limosum	YIT 6061T = JCM 1427T	modified GAM + 1% Glucose broth
53	Clostridium malenominatum	YIT 12839T = JCM 1405T	modified GAM + 1% Glucose broth
54	Clostridium nexile	YIT 6170T = ATCC 27757T	modified GAM + 1% Glucose broth
55	Clostridium orbiscindens	YIT 10060T = DSM 6740T	modified GAM + 1% Glucose broth
56	Clostridium oroticum	YIT 6037T = JCM 1429T	modified GAM + 1% Glucose broth
57	Clostridium paraputrificum	YIT 10074T = JCM 1293T	modified GAM + 1% Glucose broth
58	Clostridium perfringens	YIT 6050T = JCM 1290T	modified GAM + 1% Glucose broth
59	Clostridium ramosum	YIT 10062T = JCM 1298T	modified GAM + 1% Glucose broth
60	Clostridium saccharolyticum	YIT 12747T = DSM 2544T	modified GAM + 1% Glucose broth
61	Clostridium scindens	YIT 6171T = JCM 6567T	modified GAM + 1% Glucose broth
62	Clostridium sordellii	YIT 6065T = JCM 3814T	modified GAM + 1% Glucose broth
63	Clostridium sphenoides	YIT 6059T = JCM 1415T	modified GAM + 1% Glucose broth
64	Clostridium spiroforme	YIT 10342T = JCM 1432T	modified GAM + 1% Glucose broth
65	Clostridium sporogenes	YIT 6060T = JCM 1416T	modified GAM + 1% Glucose broth
66	Clostridium symbiosum	YIT 11480T = JCM 1297T	modified GAM + 1% Glucose broth
67	Clostridium tetanomorphum	YIT 12841T = DSM 4474T	modified GAM + 1% Glucose broth
68	Clostridium xylanovorans	YIT 12130T = DSM 12503T	modified PYG broth
69	Collinsella aerofaciens	YIT 10235T = DSM 3979T	modified GAM + 1% Glucose broth
70	Coprococcus eutactus	YIT 10160T = ATCC 27759T	modified GAM + 1% Glucose broth
71	Cronobacter sakazakii	YIT 10246T = JCM 1233T	Trypticase Soy broth
72	Dorea formicigenerans	YIT 10093T = DSM 3992T	modified GAM + 1% Glucose broth
73	Edwardsiella tarda	YIT 10118T = JCM 1656T	Trypticase Soy broth
74	Eggerthella lenta	YIT 6077T = ATCC 25559T	modified GAM + 1% Glucose broth
75	Enterobacter aerogenes	YIT 6042T = JCM 1235T	Trypticase Soy broth
76	Enterobacter cloacae	YIT 6041T = JCM 1232T	Trypticase Soy broth
77	Enterococcus avium	YIT 10255T = JCM 8722T	MRS broth
78	Enterococcus durans	YIT 2036T = GIFU 9960T	MRS broth
79	Enterococcus faecalis	YIT 2031T = ATCC 19433T	MRS broth
80	Enterococcus faecium	YIT 2032T = ATCC 19434T	MRS broth
81	Enterococcus gilvus	YIT 11114T = DSM 15689T	MRS broth
82	Enterococcus hirae	YIT 2004T = ATCC 8043T	MRS broth
83	Enterococcus malodoratus	YIT 11175T = JCM 8730T	MRS broth
84	Enterococcus mundtii	YIT 11176T = JCM 8731T	MRS broth
85	Enterococcus pseudoavium	YIT 11177T = JCM 8732T	MRS broth
86	Enterococcus raffinosus	YIT 11178T = JCM 8733T	MRS broth
87	Escherichia coli	YIT 6044T = JCM 1649T	Trypticase Soy broth
88	Eubacterium biforme	YIT 6076T = ATCC 27806T	modified GAM + 1% Glucose broth
89	Eubacterium cellulosolvens	YIT 12261T = ATCC 43171T	modified GAM + 1% Glucose broth
90	Eubacterium cylindroides	YIT 10236T = DSM 3983T	modified GAM + 1% Glucose broth
91	Eubacterium dolichum	YIT 10081T = DSM 3991T	modified GAM + 1% Glucose broth
92	Eubacterium eligens	YIT 10078T = DSM 3376T	modified GAM + 1% Glucose broth
93	Eubacterium hallii	YIT 10064T = DSM 3353T	modified GAM + 1% Glucose broth
94	Eubacterium rectale	YIT 6082T = ATCC 33656T	modified GAM + 1% Glucose broth
95	Eubacterium siraeum	YIT 10049T = DSM 3996T	modified GAM + 1% Glucose broth
96	Eubacterium uniforme	YIT 12318T = ATCC 35992T	modified GAM + 1% Glucose broth
97	Eubacterium ventriosum	YIT 10066T = ATCC 27560T	modified GAM + 1% Glucose broth
98	Faecalibacterium prausnitzii	YIT 10067T = ATCC 27768T	modified PYG broth
99	Fusobacterium necrogenes	YIT 10362T = ATCC 25556T	modified GAM + 1% Glucose broth
100	Fusobacterium necrophorum subsp. necrophorum	YIT 10343T = JCM 3718T	modified GAM + 1% Glucose broth
101	Fusobacterium nucleatum subsp. nucleatum	YIT 6069T = JCM 8532T	modified GAM + 1% Glucose broth
102	Fusobacterium russii	YIT 10363T = ATCC 25533T	modified GAM + 1% Glucose broth
103	Fusobacterium varium	YIT 11855 = JCM 3722	modified GAM + 1% Glucose broth
104	Hafnia alvei	YIT 10121T = JCM 1666T	Trypticase Soy broth
105	Holdemania filiformis	YIT 12717	modified GAM + 1% Glucose broth
106	Klebsiella oxytoca	YIT 10122T = JCM 1665T	Trypticase Soy broth
107	Klebsiella pneumoniae	YIT 6046T = JCM 1662T	Trypticase Soy broth
108	Lactobacillus acidophilus	YIT 0070T = ATCC 4356T	MRS broth
109	Lactobacillus brevis	YIT 0076T = ATCC 14869T	MRS broth
110	Lactobacillus casei	YIT 0180T = ATCC 334T	MRS broth
111	Lactobacillus fermentum	YIT 0081T = ATCC 14931T	MRS broth
112	Lactobacillus fructivorans	YIT 0235T = JCM 1117T	MRS broth
113	Lactobacillus gasseri	YIT 0192T = DSM 20243T	MRS broth
114	Lactobacillus plantarum	YIT 0102T = ATCC 14917T	MRS broth
115	Lactobacillus reuteri	YIT 0197T = JCM 1112T	MRS broth
116	Lactobacillus ruminis	YIT 0221T = JCM 1152T	MRS broth
117	Lactobacillus sakei subsp. sakei	YIT 0247T = JCM 1157T	MRS broth
118	Lactococcus garvieae	YIT 2071T = NCFB 2155T	MRS broth
119	Lactococcus lactis subsp. lactis	YIT 2008T = ATCC 19435T	MRS broth
120	Lactococcus plantarum	YIT 2061T = ATCC 43199T	MRS broth
121	Lactococcus raffinolactis	YIT 2062T = ATCC 43920T	MRS broth
122	Megasphaera elsdenii	YIT 6063T = JCM 1772T	modified GAM + 1%Glucose broth
123	Morganella morganii	YIT 10124T = JCM 1672T	Trypticase Soy broth
124	Olsenella uli	YIT 12014T = JCM 12494T	modified GAM + 1%Glucose broth
125	Parabacteroides distasonis	YIT 6162T = JCM 5825T	modified GAM + 1%Glucose broth
126	Parabacteroides johnsonii	YIT 12680	modified GAM + 1%Glucose broth
127	Parabacteroides merdae	ATCC 43184T	modified GAM + 1%Glucose broth
128	Peptoniphilus asaccharolyticus	YIT 10026T = GIFU 7656T	modified GAM + 1%Glucose broth
129	Porphyromonas gingivalis	YIT 12766T = JCM 12257T	modified GAM + 1%Glucose broth
130	Prevotella denticola	YIT 6131 = JCM 8528	modified GAM + 1%Glucose broth
131	Prevotella intermedia	YIT 12886T = JCM 11150T	modified GAM + 1%Glucose broth
132	Prevotella melaninogenica	YIT 6039T = ATCC 25845T	modified GAM + 1%Glucose broth
133	Prevotella oris	YIT 6134T = JCM 8540T	modified GAM + 1%Glucose broth
134	Proteus mirabilis	YIT 6047T = JCM 1669T	Trypticase Soy broth
135	Proteus penneri	YIT 10252T = JCM 3948T	Trypticase Soy broth
136	Proteus vulgaris	YIT 10335T = DSM 13387T	Trypticase Soy broth
137	Providencia alcalifaciens	YIT 10128T = JCM 1673T	Trypticase Soy broth
138	Providencia rettgerii	YIT 10108T = DSM 4542T	Trypticase Soy broth
139	Pseudomonas aeruginosa	YIT 6108T = IFO 12689T	Trypticase Soy broth
140	Romboutsia lituseburensis	YIT 10059T = JCM 1404T	modified GAM + 1% Glucose broth
141	Roseburia faecis	YIT 11921T = DSM 16840T	modified GAM + 1% Glucose broth
142	Roseburia hominis	YIT 11920T = DSM 16839T	modified GAM + 1% Glucose broth
143	Roseburia intestinalis	YIT 10172T = DSM 14610T	modified GAM + 1% Glucose broth
144	Ruminococcus bromii	YIT 6078T = ATCC 27255T	modified GAM + 1% Glucose broth
145	Ruminococcus gnavus	YIT 6176T = ATCC 29149T	modified GAM + 1% Glucose broth
146	Ruminococcus lactaris	YIT 10225T = ATCC 29176T	modified GAM + 1% Glucose broth
147	Ruminococcus obeum	YIT 6085T = ATCC 29174T	modified GAM + 1% Glucose broth
148	Ruminococcus torques	YIT 10159T = ATCC 27756T	modified GAM + 1% Glucose broth
149	Staphylococcus epidermidis	YIT 6049T = ATCC 14990T	Trypticase Soy broth
150	Streptococcus mitis	YIT 2069T = GIFU 12458T	MRS broth
151	Streptococcus salivarius	YIT 10260T = JCM 5707T	MRS broth
152	Streptococcus thermophilus	YIT 2037T = ATCC 19258T	MRS broth
153	Veillonella parvula	YIT 6072T = GIFU 7884T	modified GAM + 1% Glucose broth

Extraction and preparation of phenols from culture

Phenols were extracted by using a previously reported method, with partial modification (Niwa 1993). The bacterial culture was centrifuged at 20,400g for 5 min at 4°C, and the supernatant was filtered through a 0.20 µm cellulose acetate filter. Filtrates were diluted if necessary, and 225 µL of filtrate was mixed with 0.3 g sodium chloride, 180 µL of 1 N hydrochloride, 45 µL of 200 µM 4-isopropylphenol as an internal control and 450 µL of ethyl acetate, then vigorously vortexed for 30 s. The mixture was centrifuged at 2,350g for 5 min at room temperature. The ethyl acetate layer was filtered by using 0.45 µm PTFE filter vials (Thomson Instrument Company, Oceanside, California, USA), and the filtrate was subjected to HPLC analysis.

HPLC conditions

HPLC analysis was performed under the following conditions: pump: PU-2080 Plus (JASCO Corporation, Tokyo, Japan); column: L-column (Chemicals Evaluation and Research Institute, Tokyo, Japan); detector: FP-2025 Plus (excitation wavelength 260 nm and emission wavelength 305 nm); column temperature: 40°C; mobile phase: 0.1% phosphoric acid: acetonitrile (75:25) mixture; flow rate: 1 mL/min; sample injection volume: 6 µL.

Statistical analysis

Bacterial culture was performed three times independently. Bacterial strains were judged positive on screening if the concentrations of phenols in their cultures were significantly higher than those in uninoculated controls as background levels. Results were analyzed by using Student’s t-test, and strains were considered positive if the P-value was less than 0.05.

Phylogenetic analysis

Sequences of the 16S rRNA genes of bacterial strains identical to, or the same species as, the strains used in this study were collected from the Ribosomal Database Project (http://rdp.cme.msu.edu/index.jsp) or GenBank (http://www.ncbi.nlm.nih.gov/genbank/). Sequences were aligned by using Clustal X 2.1 (Larkin et al. 2007) and analyzed by using the Neighbor Joining method (Saitou and Nei 1987). The phylogenetic tree was visualized by using the TreeView 32 program (ver.1.6.6) (Page 1996). The 16S rRNA sequence of Desulfovibrio desulfuricans ATCC 29577T was used as an outgroup.

Search for homologous protein

Files on the proteins that phenol- or p-cresol-producing bacteria were expected to have were obtained from the National Center for Biotechnology Information (NCBI) database (http://www.ncbi.nlm.nih.gov/); the accession numbers of the derived genomes are listed in Tables 2 and 3. The amino acid sequences of tyrosine phenol-lyase (TPL) from Citrobacter freundii MT-10419 (Iwamori et al. 1991), TyrB (tyrosine aminotransferase) from Escherichia coli K-12 substr. MG1655 (Accession No. NP_418478), and ThiH (tyrosine lyase) from E. coli K-12 (Accession No. NP_418417) were used as queries. Homology searches between queries and obtained protein lists were performed by using GENETYX ver.11. Searches for proteins homologous to KpdB, KpdC and KpdD (Klebsiella pneumoniae decarboxylase) from K. pneumoniae NCTC 418 (Accession Nos. AAY57854, AAY57855 and AAY57856, respectively); HpdA, HpdB and HpdC (p-hydroxyphenylacetate decarboxylase) from Cl. difficile DSM 1296T (Accession Nos. AJ543427, AJ543425 and AJ543426, respectively); FldH (phenyllactate dehydrogenase); FldBC (phenyllactate dehydratase); AcdA (acyl-CoA dehydrogenase) and PorA (pyruvate:ferredoxin oxidoreductase A) were performed by using MultiGeneBlast (Medema, Takano and Breitling 2013) with the default parameters. Amino acid sequences encoded by gene clusters consisting of fldL, fldA, fldI, fldB, fldC, acdA, etfB, etfA, permease and fldH from Cl. sporogenes ATCC 15579T (Accession Nos. EDU39251 to 39261) were used as queries to identify homologs of FldH, FldBC and AcdA. Similarly, amino acid sequences encoded by porA from Cl. sporogenes ATCC 15579T (Accession Nos. EDU39094 to 39096) were used to search for homologous proteins.

Table 2.

Predicted proteins homologous to enzymes involved in metabolic pathways from tyrosine to phenol

			% of identity / E-value
Species	Strains	Genome (Accession No.)	TPLa)	TyrBb)	HadBc)	HadCd)	HadDe)
Citrobacter freundii	YIT 6045T	NZ_JMTA00000000	99/0.0	90/0.0	83/6e−98	97/0.0	87/3e−37
Clostridium saccharolyticum	YIT 12747T	NC_014376	70/0.0	–	–	–	–
Cronobacter sakazakii	YIT 10246T	NZ_CP011047	–	85/0.0	82/3e−91	93/0.0	88/7e−38
Enterobacter aerogenes	YIT 6042T	NC_015663	–	88/0.0	92/2e−106	98/0.0	92/7e−40
Enterobacter cloacae	YIT 6041T	NC_014121	–	87/0.0	90/4e−103	96/0.0	94/6e−38
Fusobacterium necrophorum	YIT 10343T	NZ_FMXX00000000	76/0.0	–	–	–	–
Fusobacterium russii	YIT 10363T	NZ_ARMK00000000	82/0.0	–	–	–	–
Klebsiella pneumoniae	YIT 6046T	NZ_AJJI00000000	–	84/0.0	100/2e−114	100/0.0	100/5e−42
Morganella morganii	YIT 10124T	NZ_BCZU00000000	90/0.0	66/0.0	–	–	–
Olsenella uli	YIT 12014T	NC_014363	–	–	48/6e−127	40/9e−12	48/1e−48

Iwamori et al. 1991

Accession No. NP_418478

Accession No. AAY57854

Accession No. AAY57855

Accession No. AAY57856

Table 3.

Predicted proteins homologous to enzymes involved in metabolic pathways from tyrosine to p-cresol

			% of identity / E-value
Species	Strains	Genome (Accession No.)	ThiHa)	TyrBb)	HpdAc)	HpdBd)	HpdCe)
Blautia hydrogenotrophica	YIT 10080T	NZ_ACBZ00000000	–	–	57/2e−108	55/0.0	42/5e−17
Clostridium difficile	YIT 10084T	NZ_AUOX00000000	36/8e−85	–	99/0.0	100/0.0	100/9e−47
Olsenella uli	YIT 12014T	NC_014363	–	–	56/6e−109	55/0.0	34/4e−8
Romboutsia lituseburensis	YIT 10059T	NZ_FNGW00000000	35/2e−82	–	68/9e−133	76/0.0	59/2e−28

Accession No. NP 418417

Accession No. NP_418478

Accession No. AJ543427

Accession No. AJ543425

Accession No. AJ543426

RESULTS

Evaluation of phenol-producing ability

We determined the phenol concentrations in cultures of the 153 strains. The cultures of 36 strains had higher phenol concentrations than the background level (Fig. 1A). Of these 36 strains, 16 (Cl. malenominatum YIT 12839T, Cl. tetanomorphum YIT 12841T, Fusobacterium varium YIT 11855, Morganella morganii YIT 10124T, Cl. cochlearium YIT 12837T, Cl. saccharolyticum YIT 12747T, Citrobacter koseri YIT 10117T, K. pneumoniae YIT 6046T, Olsenella uli YIT 12014T, Enterobacter aerogenes YIT 6042T, Citrobacter freundii YIT 6045T, Cronobacter sakazakii YIT 10246T, K. oxytoca YIT 10122T, En. cloacae YIT 6041T, F. necrophorum subsp. necrophorum YIT 10343T and F. russii YIT 10363T) exhibited phenol production at 100 µM or more in their cultures (Fig. 1C). They were calculated to convert at least half of the supplemented substrates, even if only one of the substrates were metabolized. The remaining 20 strains produced less than 100 µM of phenol in their cultures (Fig. 2A, blue).

Figure 1.

Evaluation of phenol and p-cresol production ability in 153 screened strains

One-hundred fifty three strains were cultured in rich or poor medium for 6 days. The counts of (A) phenol-positive strains and (B)p-cresol-positive strains are shown as histograms. The concentrations of (C) phenol and (D)p-cresol produced in culture by high producers are shown. White bars indicate results using rich-medium; gray bars indicate those using basal medium. Error bars indicate standard deviations. Asterisks represent P < 0.05 as analyzed by Student’s t-test (increased compared with uncultured control medium).

Figure 2.

Phylogenetic analysis of phenol or p-cresol producing bacteria

DNA sequences of 16S rRNA from 153 strains were subjected to phylogenetic analysis using Clustal X 2.1 and phylogenetic trees were constructed. (A) Phenol- or (B)p-cresol-producing strains are colored red (strains that produced at least 100 µM product) or blue (strains that produced less than 100 µM product). Strains in black font are phenol non-producers. Cluster no. represents the Clostridium 16S rRNA phylogenic cluster number (Collins et al.1994). Accession numbers used for analysis are displayed according to the name of each species, respectively.

We then determined the p-cresol concentrations in the cultures of the 153 strains. The p-cresol concentrations in the cultures of 55 strains were higher than the background level (Fig. 1B). Blautia hydrogenotrophica YIT 10080T, Cl. difficile YIT 10084T, O. uli YIT 12014T and Romboutsia lituseburensis YIT 10059T produced at least 100 µM of p-cresol (Fig. 1D). These four strains had markedly higher p-cresol production than the other 51, which produced less than 10 µM (Fig. 2B, blue).

Fourteen strains produced both phenol and p-cresol (Anaerostipes hadrus YIT 10092T, Bacteroides caccae YIT 10226T, B. ovatus YIT 6161T, B. vulgatus YIT 6159T, Cl. celerecrescens YIT 6168T, Cl. clostridioforme YIT 6051T, Cl. cochlearium YIT 12837T, Cl. indolis YIT 10077T, Cl. innocuum YIT 10151T, Cl. saccharolyticum YIT 12747T, Cl. sphenoides YIT 6059T, F. varium YIT 11855, O. uli YIT 12014T and Veillonella parvula YIT 6072T). Of these strains, only O. uli YIT 12014T produced both products at more than 100 µM; the others produced phenol or p-cresol, or both, at less than 10 µM.

Phylogenetic analysis of phenol-producing strains

All strains used in the screening were phylogenetically analyzed on the basis of the DNA sequences of the 16S rRNA gene. Phylogenetic tree analysis indicated that the phenol-producing strains were widely distributed in the Enterobacteriaceae, Coriobacteriaceae, Bacteroidaceae, Prevotellaceae, Porphyromonadaceae, Fusobacteriaceae, Enterococcaceae and Lactobacillaceae, as well as Clostridium clusters XVIII, XVI, IX, I and XIVa (Fig. 2A). The 16 strains that produced high levels of phenol (100 µM or more) belonged to specific families, namely the Coriobacteriaceae, Enterobacteriaceae and Fusobacteriaceae, along with Clostridium clusters I and XIVa. p-cresol-producing strains were dispersed across the Bifidobacteriaceae, Coriobacteriaceae, Bacteroidaceae, Fusobacteriaceae and Lactobacillaceae, along with Clostridium clusters XVI, IX, IV, I, XI, XIII and XIVa (Fig. 2B). Among them, four high p-cresol producers (100 µM or more) belonged to the specific family Coriobacteriaceae, or to Clostridium clusters XI and XIVa. The 14 strains that produced both phenol and p-cresol fell into the Fusobacteriaceae, Coriobacteriaceae or Bacteroidaceae, or Clostridium clusters XVI, IX, I and XIVa (Fig. 2). O. uli YIT 12014T, which had strong ability to produce phenol and p-cresol, belonged to the Coriobacteriaceae.

Prediction of metabolic pathways in phenol-producing strains

Three enzymes are involved in the initial or final steps of metabolic pathways from tyrosine to phenol: TPL, which metabolizes tyrosine to phenol in one step; TyrB, which metabolizes tyrosine to 4-hydroxyphenylpyruvate; and Had (hydroxyarylic acid decarboxylase), which metabolizes 4-hydroxybenzoate to phenol (Fig. 3A and B). Their activities were examined by using TPL from C. freundii MT-10419 (Iwamori et al. 1991), TyrB from E. coli strain K-12 (Kuramitsu et al. 1985), and Had from K. pneumoniae NCTC 418 (Lupa 2005), respectively. We then analyzed 10 strains with high phenol-producing ability, namely C. freundii YIT6045T, Cl. saccharolyticum YIT 12747T, C. sakazakii YIT 10246T, En. aerogenes YIT 6042T, En. cloacae YIT 6041T, F. necrophorum subsp. necrophorum YIT 10343T, F. russii YIT 10363T, K. pneumoniae YIT 6046T, M. morganii YIT 10124T and O. uli YIT 12014T, the draft genomes of which had already been sequenced, to determine whether homologous proteins of TPL, TyrB or Had were encoded in their genomes. A search for homologs of TPL derived from C. freundii MT-10419 revealed that homologs were encoded in the genomes of C. freundii YIT 6045T (99% identity of amino acid sequences), Cl. saccharolyticum YIT 12747T(70%), F. necrophorumsubsp.necrophorum YIT 10343T(76%), F. russii YIT 10363T(82%) and M. morganii YIT 10124T (90%) (Table 2). Similarly, we found that homologs of TyrB from E. coli strain K-12 were encoded in the genomes of C. freundii YIT 6045T (90% identity of amino acid sequences), C. sakazakii YIT 10246T (85%), En. aerogenes YIT 6042T(88%), En. cloacae YIT 6041T (87%), K. pneumoniae YIT 6046T (84%) and M. morganii YIT 10124T (66%) (Table 2). Had activity depended on three clusters encoded in the hadBCD operon and a cell lysate of E. coli transformed with kpdBCD; the hadBCD operon derived from K. pneumoniae NCTC 418 can metabolize 4-hydroxybenzoate to phenol (Lupa 2005). Thus, homologs of KpdBCD were found to be encoded in the genome of C. freundii YIT 6045T, C. sakazakii YIT 10246T, En. aerogenes YIT 6042T, En. cloacae YIT 6041T and K. pneumoniae YIT 6046T with more than 80% identity of amino acid sequences; in the case of O. uli YIT 12014T there was 40% to 48% identity (Table 2). The three homologs were encoded on these genomes in the order of HadB, HadC and HadD, except in the case of O. uli YIT 12014T, the three homologs of which were encoded on the genome in the order of hadC, hadD and hadB; the ORF encoding cation transporter was inserted between hadD and hadB (Fig. S1A, Supporting Information). FldBC homologs and AcdA homologs were not detected in the genomes of these six hadBCD-operon-positive strains (data not shown).

Figure 3.

Metabolic pathways from tyrosine to phenol and p-cresol

Metabolic pathways from tyrosine to phenol (A, B) and p-cresol (C, D) are shown as indicated by previous reports (Enei et al.1973; Gelfand and Steinberg 1977; Kriek et al.2007; Windey, De Preter and Verbeke 2012; Dodd et al.2017). Known enzymes—tyrosine phenol-lyase (TPL), tyrosine aminotransferase B (TyrB), phenyllactate dehydrogenase (FldH), phenyllactate dehydratase (FldBC), acyl-CoA dehydrogenase (AcdA), hydroxyarylic acid decarboxylase (Had), tyrosine lyase (ThiH), pyruvate:ferredoxin oxidoreductase A (PorA) and hydroxyphenylacetate decarboxylase (Hpd)—are shown near the arrows for each step. Steps with unidentified enzymes are indicated by dotted lines. Compounds used in this study are marked with asterisks.

Prediction of metabolic pathways in p-cresol-producing bacteria

TyrB and Hpd, which metabolize 4-hydroxyphenylacetate to p-cresol, and ThiH, which metabolizes tyrosine to p-cresol in one step, are metabolic enzymes that act in metabolic pathways from tyrosine to p-cresol (Fig. 3C and D). We therefore examined whether TyrB, Hpd or ThiH homologous proteins were found in all four strains (B. hydrogenotrophica YIT 10080T, Cl. difficile YIT 10084T, O. uli YIT 12014T and R. lituseburensis YIT 10059T) with high p-cresol-producing ability. We used information already reported on their draft genome sequences. No proteins with more than 30% amino acid sequence identity to TyrB of E. coli strain K-12 were found. In Cl. difficile DSM 1296T, three enzymes—HpdA, an activating enzyme; HpdC, a large subunit; and HpdC, a small subunit—are responsible for Hpd activity and are encoded in the hpdBCA operon (Andrei et al. 2004). Homologs of HpdBCA were identified in all four strains, with more than 30% identity of amino acid sequences (Table 3). In all four strains, the three homologs were encoded in a line in the order of hpdB, hpdC and hpdA (Fig. S1B, Supporting Information). ThiH from E. coli strain K-12 metabolizes tyrosine to dehydroglycine as the first step of the thiamine synthesis pathway, and p-cresol is formed as a by-product of this step (Kriek et al. 2007). We then found ThiH homologs encoded by the genome of Cl. difficile YIT 10084T (36% amino acid sequence identity) and R. lituseburensis YIT 10059T (35%) (Table 3). Analysis of homologs of other enzymes involved revealed that all four strains harbored FldH or PorA or both (data not shown). FldBC homologs were identified in B. hydrogenotrophica YIT 10080T and O. uli YIT 12014T. No AcdA homologs were identified in any strain (data not shown).

DISCUSSION

Screening conditions

To identify phenol- and p-cresol-producing bacteria, we used two major strategies. First, we supplemented the culture media with metabolic intermediates. Some of the supplemented intermediates—for example, 4-hydroxyphenyllactate and 4-hydroxyphenylacetate—are formed by intestinal bacteria in vitro (Smith and Macfarlane 1996; Beloborodova et al. 2012), suggesting that phenol- and p-cresol-producing bacteria further metabolize these intermediates in the intestinal environment. Cl. difficile YIT 10084T and O. uli YIT 12014T, which lacked a gene encoding TyrB in their genomes, might produce phenols from 4-hydroxyphenylpyruvate, 4-hydroxyphenyllactate, 3-(p-hydroxyphenyl)propionate, 4-hydroxybenzoate or 4-hydroxyphenylacetate as initial substrates in the intestine (Fig. 3B and D). Considering the complicated nature of the intestinal ecosystem, adding predicted metabolic intermediates to the culture media for screening was an effective strategy.

Second, we considered that other factors in the media might have affected phenol-production ability. Enei et al. (1973) reported that the presence of glucose in culture media suppressed TPL production in Erwinia herbicola ATCC 21434. Indeed, some TPL-positive strains, as represented by Cl. saccharolyticum YIT 12747T and F. russii YIT 10363T, produced much more phenol in glucose-limited media (poor media) than glucose-supplemented media (rich media) (Fig. 1C). On the other hand, glucose-limited media might be disadvantageous to growth. For example, O. uli YIT 12014T produced less p-cresol in poor medium than in rich medium (Fig. 1D), possibility because of this growth limitation. Thus it is a reasonable strategy to use both rich and poor media supplemented with tyrosine metabolites.

Identification of strains producing phenols

This study newly found 29 strains with phenol-producing potential and 51 with p-cresol-producing potential. Of the 36 phenol-positive strains, three—Cl. malenominatum YIT 12839T, Cl. tetanomorphum YIT 12841T and Cl. cochlearium YIT 12837T—have already been reported to produce phenol (Elsden, Hilton and Waller 1976). Moreover, K. pneumoniae YIT 6046T, En. cloacae YIT 6041T and M. morganii YIT 10124T are known as phenol-producing bacteria at the species level (Patel and Grant 1969; Valkova et al. 2001; Matsui et al. 2006; Iizuka et al. 2009b). The phenol-producing ability of C. freundii YIT 6045T had not been reported but had been surmised, because the phenol-forming activity of the purified TPL gene product from C. freundii species has been well characterized (Chandel and Azmi 2013). To our knowledge, the remaining 29 strains were identified here for first time as phenol producing. Among the 55 p-cresol-producing strains identified in this study, B. longumsubsp.infantis YIT 4018T, Cl. difficile YIT 10084T, Cl. paraputrificum YIT 10074T and F. necrogenes YIT 10362T have already been examined for their ability to produce p-cresol (Bone, Tamm and Hill 1976; Elsden, Hilton and Waller 1976; Smith and Macfarlane 1996). Here, we identified, for the first time, the remaining 51 strains as p-cresol-producing bacteria.

An abundance of strong producers of phenols in the intestine could affect the host’s health. The 16 phenol producers with high activity belonged to the Fusobacteriaceae, Enterobacteriaceae or Coriobacteriaceae, or to Clostridium clusters I and XIVa, and the four p-cresol producers with high activity belonged to the Coriobacteriaceae or to Clostridium cluster XI or XIVa. Kaur, Das and Mande (2017) have reported a relationship between the abundance of specific bacterial groups or specific putrefaction pathways in the intestine and the host’s stage of colorectal cancer. The information from our study could be a new clue to understanding diseases associated with phenols (Boutwell and Bosch 1959; Iizuka et al. 2009a; b, Windey, De Preter and Verbeke 2012; Ito and Yoshida 2014; Shiba et al. 2014; Andriamihaja et al. 2015; Verbeke et al. 2015). For this purpose, we need to examine whether fecal concentrations of phenols are related to the intestinal counts of phenol- and p-cresol-producing clusters. Furthermore, clinical studies are needed to investigate whether the occurrence of diseases associated with phenols is relevant to the abundance of intestinal producers of phenols.

Metabolic pathways from tyrosine to phenols

The metabolic pathways by which bacteria produce phenols are linked to the possession of pathway-related metabolic enzymes. In the genomes of 10 of the strong phenol producers analyzed here (Table 2; genome information for the remaining six was not available in the public database), homologs of TPL or Had were encoded, suggesting that each strain used pathways relevant to the enzymes they possessed (Fig. 3A and B). Cl. saccharolyticum YIT 12747T, F. necrophorum subsp. necrophorum YIT 10343T, F. russii YIT 10363T, and M. morganii YIT 10124T used TPL-dependent pathways; C. sakazakii YIT 10246T, En. aerogenes YIT 6042T, En. cloacae YIT 6041T, K. pneumoniae YIT 6046T and O. uli YIT 12014T used Had-dependent pathways, and C. freundii YIT 6045T used both TPL- and Had-dependent pathways. None of the Had-positive strains harbored FldBC homologs, indicating that these strains could use 3-(p-hydroxyphenyl)propionate or 4-hydroxybenzoate as initial metabolic substrates. More detailed analysis is needed to clarify the enzymes involved in the unknown parts of the Had-dependent pathways (Fig. 3B).

All four strong p-cresol-producing bacteria are predicted to harbor homologs of ThiH or Hpd that are involved in the final steps of p-cresol production. (Fig. 3C and D). This result suggests that ThiH or Hpd, or both, are key enzymes in producing p-cresol in these strains. We can predict from the genomic analysis that B. hydrogenotrophica YIT 10080T and O. uli YIT 12014T could utilize Hpd-dependent pathways, whereas Cl. difficile YIT 10084T and R. lituseburensis YIT 10059T could use both Hpd- and ThiH-dependent pathways. The lack of TyrB homologs and the presence of Hpd homologs in the four abovementioned strains suggest that these strains utilize tyrosine metabolites such as 4-hydroxyphenylpyruvate, 4-hydroxyphenyllactate, 3-(p-hydroxyphenyl)propionate or 4-hydroxyphenylacetate as initial substrates (Fig. 3D). This information could be a clue to identifying the metabolic scheme of p-cresol formation.

Revealing overall metabolic pathways is important for understanding intestinal microbial ecology. Draft genome sequencing of six strains not analyzed in this study (Cl. malenominatum YIT 12839T, Cl. tetanomorphum YIT 12841T, F. varium YIT 11855, Cl. cochlearium YIT 12837T, C. koseri YIT 10117T and K. oxytoca YIT 10122T) is needed. We also need to identify the currently unknown enzymes involved in the metabolism of phenols.

Limitations of this study

This screening took into account the intestinal environment, but there were three major limitations. First, the number of strains examined was limited from the perspective of the diversity of intestinal bacteria. Second, because the ability to produce phenols was evaluated in only one representative strain of each species, we did not consider variations in the ability to produce phenols among strains within a species. Third, the results of this in vitro screening might not always reflect the ability to produce phenols in the intestinal environment. Despite these limitations, this study was meaningful in that we were able to relate producers of phenols to clusters by phylogenetic analysis. This should give new insights into production of phenols in the intestine from the perspective of molecular genetics.

CONCLUSION AND FUTURE PERSPECTIVES

We identified 36 phenol-producing bacteria and 55 p-cresol-producing bacteria. Strong phenol producers belonged to the Coriobacteriaceae, Enterobacteriaceae, Fusobacteriaceae and Clostridium clusters I and XIVa, and strong p-cresol producers belonged to the Coriobacteriaceae and Clostridium clusters XI and XIVa. Such information on phenol- and p-cresol-producing bacteria should help identify the relationships between microbiota and host disease, as well as the underlying mechanisms.

Click here for additional data file.