STUDIES IN HUMANS

Phenolic catabolites excreted by fasting subjects with a functioning colon and ileostomists on a low (poly)phenol diet have been investigated. Urine was collected over a 12 h fasting period after adherence to a low (poly)phenol diet for 36 h. UHPLC-HR-MS quantified 77 phenolics. Some were present in the urine of both groups in similar trace amounts and others were excreted in higher amounts by participants with a colon indicating the involvement of the microbiota. Most were present in sub-or low-µmol amounts, but hippuric acid dominated accounting on average for 60% of the total for both volunteer categories indicating significant production from sources other than non-nutrient dietary (poly)phenols. The potential origins of the phenolics associated with the low (poly)phenol diet, include endogenous catecholamines, surplus tyrosine and phenylalanine, and washout of catabolites derived from pre-study intakes of non-nutrient dietary (poly)phenols.


Introduction
Dietary (poly)phenolics, including flavonoids, occur mainly conjugated with sugars such as glucose, and rutinose, while conjugates of cinnamic acid include quinic and tartaric acid derivatives (Crozier et al. 2006). Only very small amounts of a few such conjugates are absorbed unchanged. Deconjugation begins in the upper gastrointestinal tract (GIT) and the released aglycones undergo phase II metabolism in epithelial/hepatic cells appearing in the systemic circulation as the sulphate, glucuronide and methylated derivatives (Williamson et al. 2018). However, the bulk of the native (poly)phenol conjugates pass from the small intestine to the colon (Stalmach et al. 2010a;2010b;Borges et al. 2013) where they are subjected to the action of the resident microbiota. After microbial cleavage of the conjugating moiety, a portion of the released aglycones enters the systemic circulation as phase II metabolites with most being subjected to more extensive catabolism including ring fission. Collectively these transformations yield a substantial range of low molecular weight phenolic catabolites, some of which is absorbed and generally subjected to phase II metabolism, with the remainder being voided in faeces (Jenner et al. 2005;Ottaviani et al. 2016;Clifford et al. 2022).
There is increasing interest in these catabolites because of their potential bioactivity (Verzelloni et al. 2011;Van Rymenant et al. 2017;Williamson et al. 2018;Lonati et al. 2022) and the possibility of them serving as biomarkers of (poly)phenol intake (Clarke et al. 2020;Ottaviani et al. 2020;Xu et al. 2023).
The identification and quantification of phenolics in biofluids and an evaluation of their role in (poly)phenol bioavailability and bioactivity is less straightforward than that of (poly)phenol phase II metabolites of flavonoids. This is because of the number of compounds potentially involved. For instance, Pereira-Caro et al. (2016) identified 65 phenolic derivatives in urine after ingestion of orange juice, while Carregosa et al. (2022) reported that in human intervention studies using diets rich in (poly) phenols 137 low molecular weight phenolics have been detected in plasma. A further complication is that some of the phenolics in plasma and urine do not originate exclusively from non-nutrient dietary (poly)phenols but are endogenous products derived from hepatic catabolism of surplus amino acids such as phenylalanine (1) and tyrosine (2), and catecholamines including dopamine (3). The picture is further complicated as some of the phenolics may also be produced by gut microbiota catabolism of unabsorbed aromatic amino acids. An additional factor is washout from tissues of phenolic catabolites derived from earlier pre-study intakes of non-nutrient dietary (poly)phenols which may also contribute to the urinary phenolic profile.
To obtain further information on the identity, quantity and potential origin of phenolics in urine this paper reports a study in which UHPLC-HR-MS was used to quantify 77 low molecular weight phenolics in urines excreted over a 12 h period by fasting ileostomists and subjects with an intact colon, who for the previous 36 h had followed a low (poly)phenol diet.

Study design
The study was conducted with the prior approval of the University of Ulster Ethical Committee (REC 19/0097) and with the informed consent of participants and in accordance with the Declaration of Helsinki and the Human Tissue Act. Eight healthy subjects (4 males and 4 females, mean age 35.1 ± 11 yr) and 10 ileostomists (7 males and 3 females mean age 45.8 ± 11 years) participated in the study. All were non-smokers and the ileostomists had undergone terminal ileostomies at least 1.5 years post-operative prior to the commencement of the study. Both groups followed a low (poly)phenol diet for 36 h avoiding tea and coffee, both regular and decaffeinated, alcohol, fruit juices, fruits and vegetables, chocolate and cocoa-based products, wholemeal bread/grains, spices such as curry, herbs and olive oil. Foods that could be consumed were white bread, butter, vegetable oil, excluding olive oil, pasta, white rice, meat, eggs, fish, potatoes, mushrooms, milk, plain yogurts and cheese. In the final 36-48 h period all participants fasted and urine excreted over the 12 h period was collected the volume and pH were measured and aliquots stored at −80 °C prior to analysis.

Analysis by UHPLC-HR-MS
Aliquots of frozen urine were defrosted, vortexed and centrifuged at 16000 g for 10 min. Then, 5 µL volumes were analysed by UHPLC-HR-MS using an Ultimate 3000 RS UHPLC system (Dionex, San José, CA, USA) described previously by Ordoñez et al. (2020). Briefly, the UHPLC separation was achieved using a Zorbax SB-C18 RRHD column (100 × 2.1 mm i.d., 1.8 µm (Agilent, Santa Clara, CA, USA) preceded by a guard precolumn of the same stationary phase and maintained at 40 °C. The flow rate was set to 0.2 mL/min with a 26 min gradient of phase A: deionised water with 0.1% formic acid and B: acetonitrile with 0.1% formic acid. The gradient started at 3% B, was held for 2 min, before rising to 65% B in 18 min, before increasing to 80% B in 1 min and being maintained for 6 min. The column was then equilibrated with 3% B for 10 min prior to the analysis of a further sample. The Exactive Orbitrap mass spectrometer was fitted with a heated electrospray ionisation probe (ThermoFisher Scientific, San José, CA, USA) and operated in negative ionisation mode (scanning from 100 to 1000 m/z). The capillary temperature and the heater temperature were set to 300 °C and 150 °C, respectively. The sheath gas and the auxiliary gas flow rate was 20 units, the sweep gas was 3 units, and the spray voltage was 4.00 kV. Xcalibur (3.0 software) was used for data acquisition and data processing.
Phenolic compounds were identified by comparing the exact mass and the retention time with available reference compounds. In the absence of standards, phenolic compounds were putatively identified by comparing the theoretical exact mass of the molecular ion with its measured accurate mass and referred to databases or libraries containing HRMS spectral information such as Phenol Explorer (http://phenol-explorer.eu/), Phytohub (http:// phytohub.eu/) and Metlin (https://metlin.scripps.edu/ landing_page.php?pgcontent=mainPage) databases. Identifications were categorised according to the annotation described by Sumner et al. (2007) using the MSI level (see Supplementary Table S1).
Compounds were quantified by selecting the theoretical exact mass of the molecular ion by reference to 0.1 to 100 ng µL −1 standard curves. The analytes showed good linear relation with a coefficient of determination >0.9824 in the regression analysis of all analytes in a urine matrix. The intra-day precision, expressed as the relative standard deviation, was <2.4%. Inter-day precision was <5.6%. The recoveries for urine presented an average of 129%, while the matrix effect for urine was 64%. In absence of reference compounds, phenolics were quantified using the calibration curve of a closely related parent compound.

Statistical analysis
The results are expressed as mean values ± standard error. Two-way ANOVA and Tukeýs honestly significant difference (HSD) post-hoc test was applied to identify the differences among samples using Statistix software (v. 3.6.3, R Core Team, Vienna, Austria).

Discussion
There are six potential sources for the phenolics in the urine of ileostomists and subjects with a colon on a protein-replete, low (poly)phenol diet who are not receiving medication such as aspirin. These are summarised in Table 3 as categories 1-6.
Subject to the well-documented person-to-person variation it would be expected that both groups of volunteers would excrete similar quantities of category 1 and 2 catabolites unless some also fall in category 3, for example 3-(4′-hydroxyphenyl)propanoic acid (13) produced by the gut microbiota from unabsorbed tyrosine and its phase-II conjugates, as shown by the results presented in Table 1.
Catabolites in categories 4-6 have received little critical attention, being viewed merely as something to be avoided in volunteer studies designed to quantify phenolics associated with a test meal or beverage. Compounds in these categories are derived from (poly)phenol substrates consumed prior to the commencement of the low (poly)phenol diet stage of studies. Category 4 metabolites are derived from substrates which have bound strongly to the gut mucosa, such as ellagitannins, and which remain available for gut microbiota catabolism leading to excretion of urolithins for at least 80 h post-consumption (Truchado et al. 2012). Similar behaviour can be reasonably anticipated for the proanthocyanidins and black tea thearubigins and even some simple flavonoids. For example, it is clear from a bioavailability study using 207 µmol of 2-14 C-labelled (-)-epicatechin that volunteers with an intact colon may still be voiding labelled substrate and/or catabolites in faeces until at least 6 days post-consumption (Ottaviani et al. 2016). Categories 5 and 6 catabolites have entered the plasma comparatively rapidly post-consumption followed by sequestration in the tissues and/or binding to human serum albumin thus delaying their excretion. These tissue-associated phenolics are potentially responsible for any biological effect associated with consumption of the foods/beverages containing the (poly)phenol substrates. As such, they merit more extensive investigation. Prominent among them are the hydroxybenzenes, which are known from studies with isotopically-labelled substrates to be minor metabolites of aromatic amino acids and catecholamines (Curtius 1973, Curtius et al. 1975, 1976Alonso et al. 1982). Because they are also found in the urine of ileostomists (Table 1) the gut microbiota are not essential, suggesting that the duodenal fragmentation of flavonols, such as quercetin, and anthocyanins, plus hydrolysis of various gallate esters and gallic acid decarboxylation may be important sources (Schantz et al. 2010;Kahle et al. 2011;Pimpão et al. 2014;Barnes et al. 2016;Feliciano et al. 2016). Likewise, cinnamic acids, 3-(phenyl)propanoic acids and 3-hydroxy-3-(phenyl)propanoic acids with substituents at both 3′ and 4′ of the phenyl ring. These are characteristic metabolites of caffeoylquinic and feruloylquinic acids (coffee, artichoke, many fruit) and some flavonoids (many fruits and vegetables). The excretion of such metabolites after volunteers have followed a low (poly)phenol diet has been discussed previously . Some fraction of the C 6 -C 2 metabolites potentially derived from gut microbiota catabolism of flavonols would fall in category 6, whereas category 5 would include C 6 -C 1 metabolites derived from fragmentation of anthocyanins. C 6 -C 1 metabolites could be produced from endogenous β-oxidation of C 6 -C 3 precursors, for example 4-hydroxybenzoic acid (25) derived from 3-(4′-hydroxyphenyl)propanoic acid (13) could be categories 5 and 6 depending on whether it originated in hepatic catabolism of surplus tyrosine or gut microbial metabolism of unabsorbed tyrosine, the latter explaining the greater excretion by volunteers with an intact colon (Table 1) (Lang et al. 2013;Pereira-Caro et al. 2017;Clifford et al. 2022).
Because it is not feasible to determine the precise origin of all catabolites recorded in Table 1, it is not possible to define the amount derived from dietary (poly) phenols (categories 4-6) excreted 36-48 h after commencement of the low (poly)phenol diet, but summation of those which have not been observed as metabolites of catecholamines, phenylalanine and tyrosine indicates that it was at least 110 μmol and 60 μmol, respectively, for volunteers with an intact colon and Table 2. urinary excretion of hippuric acids by individual participants with (n = 8) and without a colon (n = 10), who had been on a low (poly)phenol diet for 0-36 h, after which they fasted for 12 h. urine collected in the 36-48 h period of the study was analysed by uHPlc-Hr-MS. ileostomists. Such washout is exponential and will be significantly less in the next 24 h, so using a two-day washout rather than a three-day or longer washout might not greatly inflate the total amount of metabolites excreted following consumption of any test meal/beverage. However, it might distort the profile if for example, urolithins derived exclusively from ellagitannins, or 3-hydroxy-3-(3′-hydroxy-4′-methoxyphenyl)propanoic acid (31) a characteristic catabolite of both hesperetin and caffeoylquinic acid intakes, appeared in urine associated with a test meal/beverage which did not contain these substrates. The washout period on a low (poly) phenol diet must be set with care, but clearly should be longer rather than shorter if such distortions are to be avoided. The amounts of categories 4-6 excreted in the 0-36-hour period were not recorded in the current study but would have been appreciably larger because of the exponential character of the washout. More information on the nature of this washout is necessary because of their potential bioactivity, and their potential to provide insights as to the nature of the recent diet. Hippuric acid (11) was the dominant metabolite excreted 36-48 h after commencement of the low (poly)phenol diet accounting for 60% of the total metabolites for both volunteer groups (Table 1). Clinical studies using labelled test substances have recorded hippuric acid yields in 48-hour urines of 12.8% from 207 µmol of 14 C-labelled (-)-epicatechin (Ottaviani et al. 2016) and 1.2% from 1,114 µmol of 13 C 5 -cyanidin-3-O-glucoside (de Ferrars et al. 2014).
Clinical studies with labelled phenylalanine and tyrosine reporting excretion of the associated hippuric acids are scarce. After intravenous injection of two phenylketonuric volunteers with 14 C-phenylalanine, Grümer (1961) reported a mean 4.7% recovery of labelled hippuric acid in urine collected for a 6 h period after dosing. This can be taken as a guide to the hepatic fate of surplus endogenous phenylalanine which is independent of the presence or absence of a colon. Jones et al. (1978) orally dosed 11 healthy males with 14 C-phenylalanine (606 μmol/kg) and recorded 2167 ± 724 μmol of hippuric acid (11) in 24-hour urine. Volunteer body weight was recorded as 76 ± 10 kg giving a mean recovery of hippuric acid of 4.7%, albeit with considerable person-to-person variation. Curtius et al. (1973) orally fed a subject 2 H-labelled tyrosine (2) (820 μmol/kg) and monitored the recovery in urine over two days, recording 4.7 mmol of hippuric acid (11) on day 1 and 10.9 mmol on day 2, making a 48 h total of 15.6 mmol. The increased excretion of hippuric acid on day 2 suggests a relatively slow voiding of the unabsorbed tyrosine. Assuming a 65 kg body weight, the overall recovery of labelled-hippuric acid from 2 H-tyrosine was ca. 29%. Production of hippuric acid was obliterated by oral dosing the volunteer with neomycin (200 mg/kg) for three days prior to the 2 H-tyrosine load. This clearly indicates that the gut microbiota are responsible for at least the initial step in the tyrosine catabolism and, by inference, ileostomists would produce little if any hippuric acid from dietary tyrosine. Munro (1976) estimated that ingestion of mixed dietary protein by North American subjects contained ca. 4.8% phenylalanine (1) and ca. 2.6% tyrosine (2), and that for a typical daily intake of 100 g protein, some 70 g are lost in urine and 10 g in faeces. This suggests a surplus of 3.8 g (23 mmol) of phenylalanine and 2.1 g (11.6 mmol) of tyrosine which after deamination may lead into C 6 -C 3 , C 6 -C 1 catabolites and associated hippuric acids. Assuming respective conversions of 4.7% and 29% of surplus phenylalanine and tyrosine to hippuric acid, based on the data of Jones et al. (1978) and Curtius et al. (1973), such surpluses could deliver ca. 4.5 mmol of hippuric acid for volunteers with an intact colon but only ca. 0.9 mmol for ileostomists because in the absence of gut microbiota tyrosine does not yield benzoic and hippuric acid (see Supplementary Information). Table 3. Six potential sources for the phenolics in the urine of subjects with an intact colon and ileostomists on a protein-replete, low (poly)phenol diet. Source Individuals with intact colon Ileostomists 1. Hepatic metabolism of surplus tyrosine and phenylalanine obtained from the protein-replete basal diet Yes Yes 2. endogenous metabolism of catecholamines as part of the normal physiological processes Yes Yes 3. Gut microbiota catabolites of unabsorbed tyrosine and phenylalanine obtained from the protein-replete basal diet Yes no 4. Washout of gut microbiota metabolites derived only from (poly)phenol substrates consumed prior to commencement of the low (poly)phenol diet but which have bound to the gut mucosa and are still to be voided in faeces Yes no 5. Washout of metabolites derived only from (poly)phenol substrates absorbed in the stomach and/or small intestine prior to commencement of the study and sequestered in the tissues or bound to human serum albumin and other proteins Yes Yes 6. Washout of gut microbiota metabolites derived only from (poly)phenol substrates absorbed in the colon prior to commencement of the study and sequestered in the tissues or bound to human serum albumin and other proteins

Yes no
The dominance of hippuric acid in urine after two (38%, 84%) or three days (86%) on a low (poly)phenol diet has previously been observed during washout, but its origin and significance was not discussed (Rios et al. 2003;Kahle et al. 2011;Castello et al. 2018).
Knowledge of the phenolic catabolites of dietary (poly)phenols is of importance because of their potential to impact human health, but it is essential to discriminate between the yield from metabolism of catecholamines, phenylalanine and tyrosine, and that from dietary (poly)phenols in the test meal or beverage. This is particularly important because low (poly) phenol diets will typically be protein-rich and because there is a possibility that (poly)phenols in the test meal/beverage could reduce protein digestibility, as well established for domestic animals, (Jansman et al. 1995;Yu et al. 1995), and potentially also modulate catecholamine catabolism.
Feeding studies with isotopically-labelled substrates provide the necessary discrimination but raise ethical considerations and have significant limitations of substrate availability, cost and the availability of appropriate analytical protocols and alternative approaches are required (Di Pede et al. 2023). One such is the incubation of (poly)phenol substrates and analysis of phenolic catabolites produced by specific microbiota and/or faecal slurries (Pereira-Caro et al. 2015 but it is unclear to what extent this represents the human situation in vivo as it does not address mammalian metabolism post-absorption by steps including β-oxidation, phase-II conjugation, and glycination (Williamson and Clifford 2010).
An alternative is a refinement of the approach examined in the current study-a feeding study involving participants with a full GIT and ileostomists, who pursue a low (poly)phenol diet for at least 48 h prior to and at least 24 h after test substance, meal or beverage intake with collection of urine every 12 h. Analysis of this urine by UPLC-HRMS provides information on the excretion of relevant metabolites during washout, such as those in Table 1.
After transformation to a μmol/h basis the data obtained in the post-feeding period are corrected by subtraction of the data obtained for the final 12-hour period of washout, thus minimising the risk of over-estimating phenolics produced from the test sample. Ideally, the phenylalanine and tyrosine content of the low (poly)phenol diet should be determined, phenolic metabolites in urine should be quantified using authentic standards, and analytical data should be archived for later re-interrogation if required (Clifford and Kuhnert 2022). Characterisation of each volunteer's gut microbiota would be a useful refinement.
Comparison of the data for volunteers with an intact colon and ileostomists allows discrimination of catabolites produced by the gut microbiota. Summation of the total phenolic washout levels provides an insight into tissue accumulation and scrutiny of the washout profile can be used as a guide to a volunteer's recent diet.

Conclusions
Metabolites of non-nutrient dietary (poly)phenols accumulating in tissues and their washout on low (poly)phenol diets deserve thorough investigation. In assessing the yield of catabolites from dietary (poly) phenols it is recommended that low molecular weight phenolics characteristic of catecholamines, phenylalanine and tyrosine are placed in a separate category because of uncertainty about their origin, and that hippuric acid is excluded from the total.