Urinary metabolite profiling identifies novel colonic metabolites and conjugates of phenolics in healthy volunteers

2.1 Preparation of fruit puree

The consumed fruit puree consisted of 100 g of each of five fruits: blueberries (Vaccinum spp. variety Georgia Gem), blackberries (Rubus L. subgenus Rubus Watson variety Karaka Black) and raspberries (Rubus idaeus L. variety Himbo Top) were harvested at the Fataca experimental field in Odemira, Portugal; strawberry tree fruits (Arbutus unedo L.) were harvested in the Alentejo region, Portugal and Portuguese crowberries (Corema album L.) were harvested in the Comporta region, Portugal. Fruits were blended together using a domestic food processor for 1 min at room temperature. The puree was prepared on the day of the study and was passed through a sieve to remove seeds before being given to the volunteers. A sample of the puree was also freeze-dried and stored at −80°C until analysis.

2.2 Subjects and study design

This study was conducted according to the guidelines laid down in the Declaration of Helsinki, and all procedures involving human subjects were approved by the ethical committee of the Faculty of Pharmacy, University of Lisbon, Portugal (02/CEECFFUL/2012). The protocol was explained to each volunteer and written informed consent was given before the study. Nine individuals (six females and three males) aged between 23 and 54 years with a body mass index of 23.2 ± 3.5 were recruited. Individuals were all considered healthy by a medical questionnaire and standard blood tests where the levels of glucose, urea, creatinine, total cholesterol, HDL cholesterol, triglycerides, C-reactive protein, and the activity of aspartate aminotransferase, alanine aminotransferase and gamma glutamyl transferase were assessed. Volunteers did not have any history of cardiovascular diseases or any other medical illnesses, were non-smoking, and were not receiving medication or taking vitamins that could interfere with the study.

Volunteers followed a (poly)phenol-free diet for 2 days before the study and throughout the day of the study, and the compliance with this restriction was confirmed through a questionnaire. After an overnight fast, volunteers ingested 500 mL of the above fruit puree with a standard breakfast containing no additional (poly)phenols, consisting of bread, with ham or cheese, yogurt and biscuits.

Urine samples were collected into a container (containing 0.5 g of ascorbic acid) before the ingestion of the puree, and in the periods of 0–2, 2–4, 4–8 and 8–24 h, and the volumes recorded. Sodium azide (0.1% w/v) was added to urine samples which were kept at −20°C until analysis.

2.3 Determination of phenolic compounds in the prepared fruit puree

For determination of (poly)phenols, the dried puree was extracted with ethanol-water (1:1), containing 1 mM butylated hydroxytoluene and 0.1 mM daidzein as internal standard. One part of the extract was used directly for HPLC-DAD analysis of total anthocyanins and caffeoylquinic acids. Another part of the extract was incubated with cellulase (EC. 3.2.1.4, Sigma-Aldrich, St. Louis, MO, USA) and hesperidinase (EC. 3.2.1.40, Sigma-Aldrich) from Aspergillus niger and analysed by HPLC-DAD, as described previously 7.

2.4 Extraction of phenolics from urine

A 2-mL aliquot of urine was spiked with 10 μL of 1 mg/mL rutin as internal standard and was added to 10 mL of ACN. Samples were centrifuged at 3000 × g for 15 min. The supernatant was collected, dried under vacuum and reconstituted in 500 μL of 50% v/v ethanol with 0.1% w/v ascorbic acid and 20 μg/mL of taxifolin as second internal standard for further analysis.

2.5 Prediction of metabolites in human samples

A list of metabolites and conjugates potentially present in human urine after ingestion of the fruits was drawn up based on prediction of metabolic pathways. The structures of 205 compounds were drawn and their exact mass calculated by using MarvinSketch software 5.7.0 from Chemaxon.

2.6 Urine analysis using LC-Orbitrap MS

A small proportion of each aliquot of urine from each time point from all volunteers were pooled together just for the LC-Orbitrap MS analysis. Phenolics were extracted as described above. Samples were separated on an HPLC Accela 600 HPLC system (Thermo Scientific, Bremen, Germany) using a C18 Synergi Hydro RP18 column (Phenomenex, Macclesfield, UK) 4 μm particle size and dimensions 2 mm ID × 150 mm. The column was fitted with a Security GuardTM guard system containing an Aqua 10 μm C18 Guard Cartridge (2 mm ID × 4 mm; Phenomenex) and eluted over a gradient of 100% solvent A (95% H₂O, 5% ACN with 0.1% v/v formic acid) to reach 15% B (95% ACN, 5% H₂O with 0.1% v/v formic acid) at 10 min, 25% B at 30 min, 60% B at 60 min, 100% B at 63 min, 100% B at 68 min at a flow rate of 0.26 mL/min. Analysis was done on an LTQ Orbitrap™ XL hybrid mass spectrometer (Thermo Scientific).

MS analysis was performed using data-dependent N^th order double play analysis comprising full scan mass range 80–2000 amu, 30 000 resolution, data-type centroid and data-dependent MS/MS (60 s of exclusion duration) on the top three most intense ions detected above threshold automatically in the independent scan event. ESI settings were as follows: source voltage, 3.4 kV; the capillary temperature was 300°C with a sheath gas at 40 psi and auxiliary gas at 5 psi.

MS data handling software (Xcalibur QualBrowser software, Thermo Electron Corp.) was used to search for predicted metabolites by their appropriate m/z value. All peaks were checked for m/z value and fragmentation products.

2.7 Preparation of liver cytosolic fractions

For preparation of liver cytosolic fractions, ox and pig liver were rinsed with 250 mM sucrose solution and a sample of 13 g was weighed. Samples were minced and homogenised with 50 mL Tris-HCl buffer 50 mM pH 7.5 using a polytron for 2 min on ice. Samples were centrifuged for 10 min at 17 400 × g at 6°C. The supernatant was recovered and centrifuged for 90 min at 35 000 × g at 6°C. The supernatant was recovered and stored at −80°C until use.

2.8 Preparation of liver microsomes

For preparation of microsomes, pig liver was rinsed with 250 mM sucrose solution and a sample of 13 g was weighed. Samples were minced and homogenised with 50 mL of the sucrose solution using a polytron for 2 min on ice. Samples were centrifuged for 10 min at 17 400 × g at 6°C, supernatant was recovered and 0.2 mL of 88 mM CaCl₂ was added to 20 mL of supernatant. The solute was left to stand on ice for 5 min with occasional swirling. The mixture was centrifuged at 27 000 × g for 15 min at 6°C, and after discarding the supernatant, the pellet was resuspended in 2.5 mL of 50 mM Tris-HCl buffer, pH 7.5, containing 10 mM EDTA and 20% v/v glycerol. Microsomal fractions were stored at −80°C until use.

2.9 Sulphation of phenolic substrates

Individual phenolic compounds used as substrates were incubated at a concentration of 300 μM with 35 μM adenosine 3′-phosphate 5′-phosphosulfate and 5.1 mM p-nitrophenyl sulphate as co-factors, 1 mM DTT, 10 mM sodium sulphite and 100 μM ascorbic acid in 100 mM potassium phosphate buffer, pH 7.4. To this solution, 25 μL of liver cytosolic fraction was added and the reaction was incubated at 37°C for 4 h. The reaction was stopped with addition of 200 μL of ice cold, ACN containing 500 mM HCl. Samples were centrifuged twice for 15 min at 17 000 × g and the supernatant was stored at −20°C until analysis.

2.10 Glucuronidation of phenolic substrates

The protocol is based on the modification of the method followed by Dueñas et al. 8. Individual compounds used as substrates were incubated at a concentration of 800 μM with 25 mM HEPES buffer, pH 7.2, containing 1 mg/mL alamethicin, 2 mM uridine 5′-diphosphoglucuronic acid trisodium salt and 1 mM saccharolactone in a final volume of 110 μL. The reaction was started with the addition of 5 μL of pig liver microsomes and incubation was performed at 37°C for 4 h. Reaction was stopped by the addition of 200 μL of methanol containing 1.6 mM ascorbic acid. Samples were centrifuged twice for 15 min at 17 000 × g, and the supernatant was stored at −20°C until analysis.

2.11 Chemical synthesis of hydroxycinnamate metabolites

The hydroxycinnamate and dihydroxycinnamate-sulphates and glucuronides (glcrnds) used as standards were chemically synthesised and characterised as described previously 9 and were kindly provided by Prof. Denis Barron, NIHS, Lausanne, Switzerland.

2.12 LC-MS/MS analysis

Individual urine samples were analysed by LC-MS/MS. The HPLC system comprised an Agilent 1200 series micro degasser, SL binary pump, SL autosampler with chilled sample compartment (8°C), column oven (30°C) and diode array detector (Agilent Technologies, Cheadle, UK). The system was controlled and data processed by Agilent MassHunter software (version B.01.03). An Atlantis T3 Column, 100 Å, 3 μm, 2.1 mm ID × 100 mm HPLC column (Waters, Hertfordshire, UK) was used for chromatographic separations at a flow rate of 0.26 mL/min over a gradient of 100% solvent A (95% H₂O, 5% ACN with 0.5% v/v formic acid) for 10 min, reaching 15% B (95% ACN, 5% H₂O with 0.5% v/v formic acid) from 10 to 20 min. Solvent B increased to 25% at 40 min and to 100% B at 43 min where it was maintained for 5 min returning to 0% in 2 min. MS analysis was performed with an Agilent 6410a triple quadrupole LC-MS-MS with electrospray source at 350°C, a source voltage of 4 kV and N₂ drying gas flow rate of 11 L/min at a pressure of 30 psi (Peak Scientific, NM30LA, Inchinnan, UK). The analysis was performed in negative multiple reaction monitoring (MRM) mode and optimisation for ion transmission was achieved by repeated injections of individual standards.

Relative amounts of compounds were estimated after normalizing to internal standard (rutin) and corrected by urinary volume.

2.13 Statistical analysis

Pharmacokinetic excretion profile of urine metabolites was constructed using GraphPad Prism 5. This package was also used for statistical analysis. Box-and-Whiskers plots for minimum and maximum values were produced. Comparisons in relation to the baseline were performed with two-tailed Wilcoxon matched pairs test with a confidence level of 95%.

3.1 Identification and quantification of phenolic compounds in fruit puree

The puree containing blueberries, raspberries, blackberries, Portuguese crowberry and strawberry tree fruits was characterised by HPLC-DAD for the major compounds (caffeoylquinic acids and anthocyanins, Supporting Information Table 1) and the aglycones were also quantified after multi-enzyme hydrolysis of the glycosides (Table 1). The most abundant compounds were anthocyanins, chlorogenic acids and gallic acid. Caffeic acid was abundant after hydrolysis and possibly resulted from esterase activity on caffeoylquinic acids plus glycosidase activity towards glycosides of caffeic acid.

3.2 Initial determination of metabolites in urine using exact mass data

Compounds were identified in pooled urine samples for each time point by comparing their exact mass and fragmentation patterns (when available) obtained using an LC- Orbitrap MS with predicted metabolites. By comparing chromatograms of urine samples before ingestion and after ingestion of the fruit puree, it was observed that several metabolites appeared after ingestion. Only metabolites which increased in abundance after ingestion of the fruit puree were selected for confirmation to exclude compounds that could result from endogenous metabolism (Supporting Information Tables 2–6). Anthocyanins, which were very abundant in the fruit puree, were not evident by LC-Orbitrap MS analysis. After this initial analysis, the most abundant compounds were selected for confirmation by comparison with enzymically or chemically synthesised standards in individual samples using the LC-triple quadrupole MS.

3.3 Enzymatic synthesis of phenolic compounds

For production of conjugated compounds, (poly)phenol aglycones were incubated in vitro with pig liver microsomes and uridine 5′-diphosphoglucuronic acid trisodium salt to generate glcrnd conjugates, and with an ox liver cytosolic fraction in the presence of the co-factors adenosine 3′-phosphate 5′-phosphosulfate and p-nitrophenyl sulphate to generate the formation of mono-sulphated compounds. Ox liver cytosolic fraction was used instead of pig liver as it was observed to be more efficient in generating sulphated conjugates. The resulting conjugates were identified by their mass and fragmentation pattern (MRM transitions), and are shown in Table 2. When multiple conjugates resulted from the reaction owing to the presence of multiple available hydroxyl groups, it was not possible to distinguish the exact position of the conjugation, with exception for the compounds synthesised chemically.

Gallic acid, although theoretically having only two different hydroxyl groups for conjugation, yielded three glucuronidated metabolites (Supporting Information Fig. 1A). Also, vanillic acid, possessing a single hydroxyl group available for conjugation, originated one sulphate as expected, but apparently two glcrnds (Supporting Information Fig. 1B). The one sulphated metabolite of 4-O-methylgallic acid was produced as expected; however, protocatechuic acid yielded only one peak (Supporting Information Fig. 1C), even though two hydroxyl groups are available for conjugation. These events suggest a positional specificity of sulfotransferases for these compounds if only one metabolite is being produced, or alternatively overlapping of two metabolites in one peak.

(−)-Epicatechin yielded three sulphates, although five hydroxyl groups are theoretically available for sulphation or glucuronidation. However, glucuronidation of (−)-epicatechin yielded five peaks (Supporting Information Fig. 1D) but with large differences in efficiency of conjugation for the different positions.

Conjugates of catechol, 4-methylcatechol, pyrogallol and its O-methylated forms, 1-O-methylpyrogallol and 2-O-methylpyrogallol, are summarised in Table 2 and occurred as expected. The only exception was 4-methylcatechol which, although having two available hydroxyl groups, yielded only one glcrnd and one sulphate.

Therefore, enzymic synthesis of sulphated compounds is not strictly regio-selective for most compounds, but reveals a preference for certain positions, as revealed by differences in peak areas for synthesised isomers of the same parent compound.

3.4 Confirmation and kinetics of metabolites in human urine

Individual urine samples from each volunteer and each time point were analysed using the triple quadrupole LC-MS/MS in MRM mode, using conditions optimised for the standard compounds, to confirm if compounds initially identified by exact mass scanning matched the synthesised standards. The presence of aglycones and conjugates in urine is indicated in Table 2. The aglycones protocatechuic and dihydrocaffeic (DHC) acids, and gallic acid-O-glcrnd, were not detected by the LC-Orbitrap MS due to low levels of the compounds combined with the lower sensitivity of this method, but were, however, detected with LC-triple quadrupole MS. On the other hand, some aglycones (such as catechol, 4-methylcatechol, ferulic and isoferulic acids) were tentatively identified using the Orbitrap-MS but were not found by the MRM method. This might result from technical limitations such as misleading identification of the peak due to other compounds with the same exact mass, or due to non-intentional in-source fragmentation of the conjugated compounds resulting in the detection of a ‘ghost’ aglycone.

Therefore, the presence of sulphated and glucuronidated conjugates was confirmed in individual urine samples (Table 2). The levels of compounds which increased in urine after consumption of the fruit puree are shown in Figs. 1-7.

Gallic acid, found in the free form, appeared in urine at early time points, reaching a maximum between 0 and 4 h post ingestion (Fig. 1A). Phase II metabolites of gallic acid (glcrnd, methyl and methyl-sulphate) reached a maximum excretion in urine between 2 and 4 h, slightly later than gallic acid (Table 2; Fig. 1 B–D). O-Glucuronidated metabolites of gallic acid were not observed with the initial screening using the LC-Orbitrap MS/MS, but were found using the triple quadrupole LC-MS. One O-sulphated conjugate of protocatechuic acid was also found in urine samples (Table 2) and matched the single conjugate synthesised in vitro. Protocatechuic acid-O-sulphate and phase II metabolites of vanillic acid had an intermediate time of excretion, peaking between 4 and 8 h (Fig. 2). The excretion profiles of phase II metabolites of hydroxycinnamic acids (Table 2) are shown in Figs. 3 and 4.

Several (−)-epicatechin conjugates were enzymically produced in vitro, but only two O-sulphated and two O-glucuronidated metabolites were found in urine (Table 2), thus suggesting a more selective mechanism of conjugation in vivo by phase II enzymes. The time course in urine of (−)-epicatechin conjugates is consistent with a rapid excretion (Fig. 5).

Both O-glcrnds and O-sulphates of catechol and 4-methylcatechol were detected and had very similar profiles of appearance in urine (Fig. 6), starting from low levels at initial time points, and increasing especially after 4 h. The same excretion profile was observed for phase II metabolites of pyrogallol (Fig. 7). Only one sulphated metabolite of 1-O-methylpyrogallol was identified, although two metabolites were obtained in vitro.