This is an open access article published under an ACS AuthorChoice License, which permits copying and redistribution of the article or any adaptations for non-commercial purposes.

Article

Cite This: J. Agric. Food Chem. 2019, 67, 11099−11107 pubs.acs.org/JAFC

Identification of Novel Urolithin Metabolites in Human Feces and Urine after the Intake of a Pomegranate Extract Rocío García-Villalba, María V. Selma, Juan C. Espín, and Francisco A. Tomas-Barberá ń* Laboratory of Food & Health, Research Group on Quality, Safety, and Bioactivity of Plant Foods, Department of Food Science and Technology, CEBAS-CSIC, P.O. Box 164, Campus de Espinardo, Murcia 30100, Spain

*S Supporting Information

ABSTRACT: Urolithins are bioactive gut microbiota metabolites of . Here, we have identified four unknown urolithins in human feces after the intake of a pomegranate extract. The new metabolites occurred only in 19% of the subjects. 4,8,9,10-Tetrahydroxy urolithin, (urolithin M6R), was unambiguously identified by 1H NMR, UV, and HRMS. Three metabolites were tentatively identified by the UV, HRMS, and chromatographic behavior, as 4,8,10-trihydroxy (urolithin M7R), 4,8,9-trihydroxy (urolithin CR), and 4,8-dihydroxy (urolithin AR) urolithins. Phase II conjugates of the novel urolithins were detected in urine and confirmed their absorption, circulation, and urinary excretion. The production of the new urolithins was not specific of any of the known urolithin metabotypes A and B. The new metabolites needed a bacterial 3-dehydroxylase activity for their production, and this is a novel feature as all the previously known urolithins maintained the hydroxyl at 3 position. The ability of production of these “R” urolithins can be considered an additional metabolic feature for volunteer stratification. KEYWORDS: metabolism, gut microbiota, urolithins, interindividual variability, metabotypes

■ INTRODUCTION samples from some volunteers showed that several unknown Urolithins (hydroxylated dibenzo[b,d]-pyran-6-one-deriva- metabolites with the mass of tetrahydroxy-, trihydroxy-, and dihydroxy-urolithins were present together with the main tives) are produced in vivo by the gut microbiota of humans metabolite, , although in smaller amounts. This and different animals after the intake of and 1−4 prompted us to characterize these new urolithin metabolites, ellagic acid. The main final metabolites found in plasma, although this was a challenging task because of the small tissues, and excreted in urine and feces, include urolithin A, amount present and their presence in a very complex matrix as , and isourolithin A, and their corresponding phase- it is feces. II conjugates. Other intermediate metabolites have been fi identi ed and they follow the sequential loss of hydroxyls MATERIALS AND METHODS from the pentahydroxy urolithin (urolithin M5) which is the ■ fi Chemicals. The internal standard 6,7-dihydroxycoumarin was Downloaded via CSIC on October 10, 2019 at 10:26:33 (UTC). rst urolithin produced by opening one of the two lactone 5 from Sigma-Aldrich (St. Louis, MO, USA). Urolithin standards were rings of ellagic acid and its subsequent decarboxylation. These 16 final urolithin metabolites are much better absorbed than the obtained as previously described. Purity was higher than 95% for all 6 tested compounds. Organic solvents such as methanol and acetonitrile ellagitannins and ellagic acid found in food products, reaching were from Baker (Deventer, The Netherlands). Formic acid and HCl fi 2,3 signi cant concentrations in blood and urine. were obtained from Panreac (Barcelona, Spain). The Milli-Q system Urolithin metabolites also show different biological activities (Millipore Corp., Bedford, MA) ultrapure water was used throughout See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles. and have been suggested to be responsible for the health the study. All chemicals and reagents were of analytical grade. effects observed after the intake of ellagitannin-rich foods such Human Study Design. This trial (“The POMEcardio study”) was as pomegranates, walnuts, and berries.4,7,8 This has fueled the conformed to ethical guidelines outlined in the Declaration of studies to understand the production of urolithins in the gut Helsinki and its amendments. The protocol included in the − and the mechanisms behind their biological effects.8 11 BACCHUS Project (FP7-KBBE-2012) was approved by the Spanish National Research Council’s Bioethics Committee (Spain) and Urolithins have also been suggested as biomarkers for the 4,12 registered at clinicaltrials.gov (NCT02061098). Protocol details intake of ellagic acid-containing foods. As the production of were reported elsewhere.15 Briefly, 49 overweight-obese healthy urolithins in our body depends on the occurrence in the gut of individuals (32 men and 17 women) aged between 40 and 65 years specific bacteria,10 and because of the known large variability participated in a double-blind, crossover, dose−response, randomized, of gut microbiota composition between individuals, it is and placebo-controlled trial.15 In the first phase, they consumed daily expected that not all subjects produce the same urolithins and one capsule of PE (160 mg phenolics) for 3 weeks, and after a in the same quantity.13 Thus, three different consistent washout period, they consumed daily four capsules of PE (640 mg urolithin metabotypes, A, B, and 0 have been described after the analysis of these metabolites in human urine.14 Received: July 15, 2019 Under a previous human intervention study with a Revised: September 6, 2019 pomegranate extract (PE) rich in ellagitannins and ellagic Accepted: September 8, 2019 acid,15 the HPLC−DAD−MS analysis of urolithins in fecal Published: September 8, 2019

© 2019 American Chemical Society 11099 DOI: 10.1021/acs.jafc.9b04435 J. Agric. Food Chem. 2019, 67, 11099−11107 Journal of Agricultural and Food Chemistry Article phenolics).15 Feces and urine samples were collected after each phase (NMR) analysis. Finally, the whole 500 μL was transferred to a 5 and were frozen at −80 °C until further analysis. In a previous study, mm standard NMR tube. urolithins were identified in feces samples, and volunteers were 1H NMR Analysis. For the identification of the new urolithin stratified as metabotype A (UM-A) (production of urolithin A metabolites, a Bruker AVANCE III HD spectrometer equipped with derivatives), metabotype B (UM-B) (production of isourolithin A an Ascend TM magnet, 11.7 T (1H operating frequency 500 MHz) and/or urolithin B in addition to urolithin A), and metabotype 0 and a cryoprobe (Cryoplatform Prodigy BBO) for increased (UM-0) (urolithins were not detected).15 sensitivity was used. One of the new metabolites was isolated in Sample Clean-up for LC Analyses. Feces samples (1 g) were enough amount to allow the 1H NMR analysis. The isolated new 1 defrosted and homogenized with 10 mL of MeOH/DMSO/H2O tetrahydroxy-urolithin was analyzed by H NMR (500 MHz in δ (40:40:20) with 0.1% HCl using an Ultra-Turrax for 1 min at 24 000 DMSO-d6) with chemical shift values ( ) in ppm: 6.78 (d; 1H, J = 8.2 rpm. The mixture was centrifuged at 5000×g for 10 min at room Hz, H-3); 6.96 (t, 1H, JH3−H2 = 8.20 Hz; JH1−H2 = 8.53 Hz, H-2); 7.03 temperature, and the supernatant was filtered through a 0.22 μm (s, 1H, H-7); 8.33 (d, 1H, JH2−H1 = 8.53 Hz, H-1) (Supporting PVDF filter before analysis. Information Figure S1). HPLC−DAD−ESI MS and UPLC−QTOF−MS Analyses. Ur- olithins were analyzed in feces as explained elsewhere.16 The analyses ■ RESULTS AND DISCUSSION were performed using an high-performance liquid chromatography Urolithin Analysis in Fecal Samples after the Intake (HPLC) system (1200 Series, Agilent Technologies) equipped with a of a PE. photodiode-array detector and a single quadrupole mass spectrometer The urolithin metabolites present in fecal samples (n = 47) after the intake of an ellagitannin-rich PE were analyzed detector in series (6120 Quadrupole, Agilent Technologies). − − The chromatographic separation was carried out on a Poroshell by HPLC DAD MS. As previously reported, the urolithin 120 EC-C18 column (3 × 100 mm, 2.7 μm) (Agilent Technologies) base level in biological samples prior to pomegranate intake using water with 0.5% formic acid (A) and acetonitrile (B) as the was negligible for most volunteers because of the dietary mobile phases with a flow rate of 0.5 mL/min. The gradient profile restriction of ellagitannin-containing foods.15 The metabolites was: 0−7 min, 5−18% B, 7−17 min, 18−28% B, 17−22 min, 28−50% identified in previous studies16,17 were easily detected with the B, 22−27 min, and 50−90% B, and this percentage was maintained specificdifferences for the characteristic metabotypes.14 In for 1 min and then came back to the initial conditions. A volume of 5 μ some volunteers, however, new metabolites with UV spectra L of sample was injected onto the column operating at room similar to those of urolithins and consistent masses were also temperature. Ultraviolet (UV) chromatograms were recorded at 360 and 305 nm. detected (Figure 1). In this case, one unknown metabolite with MS in selective ion-monitoring mode and negative polarity was used to confirm the identification. Optimal electrospray ionization (ESI) parameters using nitrogen as nebulizer gas were: capillary voltage 3500 V; drying gas flow 10 L/min; nebulizer pressure 45 psi, and drying T 300 °C. Metabolites were identified using their UV spectral properties and molecular mass, and whenever possible by comparison with authentic standards. To help in the identification of the new urolithins detected, some fecal samples were analyzed in an Agilent 1290 Infinity UPLC system coupled to a 6550 Accurate-Mass Quadrupole time-of-flight (UPLC- QTOF-MS) using the methodology previously reported.16 QTOF- MS provides the possible molecular formulae for the compounds based on the accurate mass and isotopic pattern. Besides, MS/MS analyses were also performed using the m/z range of 50−800, collision energy of 35 V and acquisition rate of 4 spectra/s, providing fragmentation information very useful to improve the confidence in the identification process. Isolation of the New Urolithin. One of the chromatographic fi peaks, identi ed as an unknown urolithin, was isolated by analytical Figure 1. HPLC−UV chromatogram (305 nm) of an extract from a HPLC with a semipreparative purpose. For this, the sample was fecal sample from volunteer 38 after the intake of a pomegranate previously subjected to a clean-up process and concentration. Two extract. (1) Ellagic acid, (2) urolithin M6 (3,8,9,10-tetrahydroxy grams of feces were extracted with 20 mL of MeOH/H2O (80:20) × urolithin), (3) urolithin C (3,8,9-trihydroxy urolithin), (4) unknown with 0.1% HCl. The mixture was centrifuged at 5000 g for 10 min at tetrahydroxy urolithin (urolithin M6R) (4,8,9,10-tetrahydroxy room temperature, and the methanol of the supernatant was urolithin), (5) unknown trihydroxy urolithin (urolithin CR) (4,8,9- evaporated in a speed vacuum concentrator. The aqueous remnant trihydroxy-urolithin), (6) unknown trihydroxy urolithin (urolithin was cleaned up with 3 mL of hexane that was discarded after M7R) (4,8,10-trihydroxy urolithin), (7) urolithin A (3,8-dihydroxy centrifugation to remove the nonpolar fraction, and the residue was × urolithin), and (8) unknown dihydroxy urolithin (urolithin AR) (4,8- extracted with 4 mL of ethyl acetate. After centrifugation at 10 000 g dihydroxy urolithin). for 10 min, the supernatant was evaporated in a speed vacuum concentrator, redissolved in 500 μL of methanol, and filtered through a 0.22 μm filter before the analysis and isolation by using an analytical the mass of a tetrahydroxy-urolithin (4) ([M − H]− m/z 259), HPLC system. The same HPLC-DAD-ESI-MS/MS instrument, with two unknown trihydroxy-urolithins (5 and 6) ([M − H]− m/z the same column and chromatographic conditions described in the 243), and one unknown dihydroxy-urolithin (8) ([M − H]− previous section, was used for the isolation. The injection volume was μ fi m/z 227) were detected. increased to 10 L. The new peaks identi ed at 305 nm were Identification of the New Tetrahydroxy Urolithin. manually collected. The isolation was done trying to obtain the The compound as pure as possible, avoiding other interferences. After five unknown tetrahydroxy urolithin (4 in Figure 1) showed a Rt at injections, the samples collected were taken to dryness in a speed 12.64 min that did not coincide, under the same assay vacuum concentrator and the residue was reconstituted in 500 μLof conditions, with the already known tertrahydroxy urolithins, deuterated DMSO (DMSO-d6) for nuclear magnetic resonance urolithin D (Rt 10.02 min), urolithin E (Rt 10.48 min), and

11100 DOI: 10.1021/acs.jafc.9b04435 J. Agric. Food Chem. 2019, 67, 11099−11107 Journal of Agricultural and Food Chemistry Article

Figure 2. Unknown and known urolithin metabolites. (Blue) previously known urolithins; (Red) new urolithins discovered in the present study; (Black) urolithins not reported yet. (URO) Urolithin. urolithin M6 (Rt 11.56 min) (Figure 2), suggesting a new discovered so far that maintained the hydroxyl at 3 position, as metabolite. Its UV spectrum, with a very predominant BII at metabolites produced by the activity of an urolithin 3- 258 nm, suggested that this urolithin metabolite should have a dehydroxylase had not been described in previous studies. hydroxyl at position 9.17 The UPLC-Q-TOF-MS analysis Urolithin M6R was thus potentially more unusual as it confirmed that the molecular formula of this metabolite was involved the dehydroxylation at position 3. Therefore, as no − fi C13H8O6 with a score of 97.66 and an error of 0.22. Its UV identi cation was feasible with the available information (UV, 1 spectrum was closer to that of urolithin M6 rather than to Rt, and high-resolution mass spectrometry), an H NMR study those of urolithin D or urolithin E (Table 1), suggesting a close was attempted. The main challenge for this analysis was the structure to that of urolithin M6. Its high-resolution MS−MS small amount of sample available, and the nature of the sample fragmentation analysis (Supporting Information Table 1) also itself as it was a very complex matrix with many metabolites in showed that the fragments of the new metabolite were similar relatively small amounts (Figure 1). The whole fecal sample to those observed for urolithin M6. Only two possible was extracted, and the extract was then used to isolate the undiscovered tetrahydroxy-urolithin structures were foreseen metabolite by analytical reversed phase HPLC with a after the catabolism of ellagic acid and urolithin M5 (Figure 2) semipreparative approach. The purified compound 4 was and were named urolithin F and urolithin M6R. Urolithin F then submitted to 1H NMR, and the spectrum clearly showed would follow the usual trend for all the urolithin metabolites that the unknown urolithin had one singlet (integrating 1H)

11101 DOI: 10.1021/acs.jafc.9b04435 J. Agric. Food Chem. 2019, 67, 11099−11107 Journal of Agricultural and Food Chemistry Article

Table 1. R , UV Spectra and Structures of the Different Urolithins with Comments for the Tentative Identification and the t a Rules to be Followed

aMetabolites colors: (Red) new metabolites; (blue) already known metabolites; (black) synthetic metabolite. for an isolated proton consistent with a H at position 7 and grouping as it would happen in urolithin M6R (Table 2, three vicinal protons consistent with the H1, H2, and H3 Supporting Information Figure S1), and this grouping had not

11102 DOI: 10.1021/acs.jafc.9b04435 J. Agric. Food Chem. 2019, 67, 11099−11107 Journal of Agricultural and Food Chemistry Article

Table 2. 1H NMR Analysis of Urolithins

urolithin metabolite H1 H2 H3 H4 H7 H9 H10 references 1H NMR urolithin M6R 8.33 6.96 6.78 7.03 (s) 500 MHz, (4,8,9,10-OH) (d, J = 8.5) (dd, J = 8.5; 8.2) (d, J = 8.2) DMSO-d6 urolithin M6 8.80 6.75 6.67 7.28 (s) Rupiani et al., 400 MHz, 20 (3,8,9,10-OH) (d, J = 9.0) (dd, J = 9.0, 1.8) (d, J = 1.8) 2016 DMSO-d6 urolithin M6 8.91 6.76 6.70 7.36 (s) Hong et al., 700 MHz 18 (3,8,9,10-OH) (d, J = 9.1) (dd, J = 8.4, 2.1) (d, J = 2.1) 2017 CD3OD urolithin M7 8.76 6.81 6.76 7.80 8.57 Pottie et al., 500 MHz 21 (3,8,10-OH) (d, J = 9.3) (dd, J = 9.3, 2.6) (d, J = 2.6) (d, J = 1.1) (d, J = 1.1) 2011 CDCl3 urolithin C 7.84 6.81 (dd, 8.4, 2.1) 6.71 7.58 (s) 7.43 Hong et al., 700 MHz 18 (3,8,9-OH) (d, J = 8.4) (d, J = 2.1) (s) 2017 CD3OD urolithin M5 8.47 6.79 (d, J = 8.9) 7.41 (s) Cozza et al., 300 MHz 19 (3,4,8,9,10-OH) (d, J = 8.9) 2011 CD3OD

Figure 3. Proposed catabolic pathways for the conversion of ellagic acid into the new urolithins. In red, the new catabolic branch. been previously reported in any other urolithin metabolite. and therefore this was supposed to be a very unusual ellagic The observed NMR spectrum for compound 4 was consistent acid catabolic activity for the gut microbes. with the spectra already reported for trihydroxy- tetrathydroxy- Tentative Identification of the Remaining Unknown − and pentahydroxy urolithins (Table 2).18 21 The other Urolithin Metabolites. The two unknown trihydroxy − possible tetrahydroxy urolithin metabolite (urolithin F; urolithins ([M − H] at m/z 243) differed from urolithin C − 3,4,9,10-tetrahydroxy urolithin, Figure 2) would had two H (Rt 12.44 min) and urolithin M7 (Rt 13.59 min) (5 and 6 in H groupings (H7−H8 and H1−H2) that would have an NMR Figures 1 and 2). The first unknown trihydroxy urolithin eluted spectrum with two doublets, which were not observed in the with Rt 13.17 min and showed a UV spectrum, suggesting that spectrum of the isolated metabolite, and therefore the it had hydroxylation at position 9 (low absorbance and short unknown tetrahydroxyurolithin (4) was unambiguously wavelength maximum for BI while a high absorbance for the identified as 4,8,9,10-tetrahydroxy urolithin (urolithin M6R), BII).17 In the present study, we describe that the UV a new metabolite. Besides, another criterion that excluded the absorbance BI/BII ratio can clearly discriminate between urolithin F structure was that the new tetrahydroxy urolithin urolithin metabolites with a hydroxyl at position 9 (BI/BII metabolite was identified in the two metabotypes, UM-A and ratio between 0.12 and 0.17) from those that lack a hydroxyl at UM-B, whereas urolithin F could only have been detected in position 9 (BI/BII ratio between 0.27 and 0.46) (Figure 4, UM-B because UM-A lacks 8-hydroxylase activity (Figure 3). Table 1). Thus, the first eluting unknown trihydroxy urolithin The identification of this new metabolite (M6R) means more metabolite 5 (Figure 1) clearly shows a hydroxyl at 9 position than just the discovery of a new urolithin as it opens the and a UV spectrum similar to that of urolithin C (Table 1). On possibility of discovering a new family of urolithins produced the other hand, the other unknown trihydroxy metabolite 6 after the removal of the hydroxyl at 3 position (3- (Figure 1) eluted at Rt 14.19 min, and showed a UV spectrum dehydroxylase activity), that so far had not been documented, similar to that of urolithin M7 and its BI/BII ratio (0.33), was

11103 DOI: 10.1021/acs.jafc.9b04435 J. Agric. Food Chem. 2019, 67, 11099−11107 Journal of Agricultural and Food Chemistry Article

behavior (ionization and hydrophilic character) and the UV 22 spectra. Thus, urolithin B eluted with shorter Rt than the isomeric 8-hydroxy urolithin, and they also showed differences in the UV spectra, particularly in the BI, which only had a maximum in urolithin B (332 nm) while showed a maximum and a shoulder in 8-hydroxy urolithin (350sh, 338 nm) (Table 1). Therefore, it seems that metabolites without a hydroxyl at 3 position (i.e., 8-hydroxy urolithin) have a longer Rt that is consistent with a lower hydrophilic character because of a poorer ionization related to the less acidity of the hydroxyl at 8 position compared with that of the corresponding 3-hydroxy isomer. These characteristics seem to be appropriate for distinguishing those metabolites without 3-hydroxylation as their UV spectra and chromatographic mobility characteristics are also consistent with those of the new urolithin M6R that was unambiguously identified by 1H NMR, compared with those of urolithin M6 and other tetrahydroxy urolithin isomers Figure 4. UV spectra of urolithins. Calculation of the BI/BII ratio as (Table 1). diagnostic for 9-hydroxylation. The two new trihydroxy urolithins detected (5 and 6 in Figure 1), differed from the two known trihydroxylated derivatives, urolithin C (Rt 12.44 min) and urolithin M7 (Rt consistent with the lack of hydroxylation at position 9 (Figure 13.59). The first unidentified trihydroxy urolithin (5) eluted 4). The UPLC-Q-TOF-MS analyses showed that both with a Rt of 13.17 min and had a UV spectrum with a BI at 332 metabolites had a molecular formula of C13H8O5 with a nm and a very high absorbance BII at 250 nm similar to that of score of 98.57 and an error of −1.47. Their MS−MS fragments urolithin C. Its BI/BII ratio (0.15) indicated that the new were similar to those of urolithin C for the trihydroxy urolithin metabolite had a hydroxyl at position 9 (Figure 4), and the 5 and close to urolithin M7 and urolithin C for the trihydroxy shape of the UV spectrum (particularly that of the BI) and the urolithin 6 (Supporting Information Table S1). For these higher Rt compared with that of urolithin C suggested that this unknown metabolites, eight different possible structures were metabolite did not have a hydroxyl at position 3. Therefore, initially considered (Figure 2), although the assignment of the from the potential unknown trihydroxy urolithins (Figure 2), most feasible one was not possible with the available data. the data suggest that the most feasible one was 4,8,9-trihydroxy Unfortunately, their purification for NMR analysis was not urolithin and it was named urolithin CR, a new urolithin possible because of the small amount of sample, and to the fact metabolite. that these metabolites were only minor ones in the extract as The second unknown trihydroxy urolithin had a relatively revealed in the HPLC chromatogram (Figure 1). longer Rt (14.19 min), suggesting the lack of hydroxyl at When the potential dihydroxy urolithins were analyzed, one position 3, a UV spectrum with a BI/BII absorbance ratio unknown metabolite was detected. Urolithin A (Rt 16.01 min) (0.33) consistent with the lack of a hydroxyl at position 9, and fi and isourolithin A (Rt 15.95 min) were identi ed, and the new the UV BI with a maximum and a shoulder (360sh, 350) metabolite showed a higher Rt at 17.22 min (8 in Figure 1). Its supported the lack of hydroxyl at position 3. The longer UPLC-Q-TOF-MS analysis showed a molecular formula of wavelength for the maxima of BI also supported hydroxylation C13H8O4 consistent with a dihydroxy urolithin and a score of at position 8. Therefore, this new metabolite was tentatively 97.7 and an error of −1.40. Its MS/MS fragmentation was identified as 4,8,10-trihydroxy urolithin, and the name of similar to that of urolithin A, suggesting a similar structure urolithin M7R was given (an isomer of urolithin M7, without (Supporting Information Table S1). For this new metabolite, hydroxyl group at position 3). eight structures were taken into account (Figure 2), although Regarding the new dihydroxy urolithin detected (Rt 17.22 the assignation of the most feasible one was not possible with min) (8 in Figure 1), this was compared with the available the data available. As it happened with the trihydroxy dihydroxy urolithin standards urolithin A (Rt 16.01 min) and derivatives, its purification and concentration for NMR analysis isourolithin A (Rt 15.95 min), and eight potential unknown was not possible (Figure 1). metabolites were considered (Figure 2). Its long Rt suggested Because of the limited amount available of the metabolites 5, that this could also be a metabolite without hydroxyl at 6, and 8 in the fecal extracts, an indirect approach for the position 3. The lack of 3 hydroxylation was also confirmed by tentative identification of these urolithins was attempted. the two responses for BI (350sh, 344 nm), the lack of hydroxyl Therefore, differences in their UV spectra, their HPLC at position 9 by the BI/BII ratio (0.31), and the presence of chromatographic behavior (Rt), and their MS fragmentation hydroxyl at position 8 was supported by the high BI maximum patterns were explored (Table 1 and Supporting Information wavelength (350 nm), closer to that of urolithin A rather than Table S1). that of isourolithin A. Therefore, this metabolite was The comparison of the naturally occurring 3-hydroxy tentatively identified as 4,8-dihydroxy urolithin, a new urolithin (urolithin B) and the synthetic 8-hydroxy isomer metabolite that we gave the name of urolithin AR to indicate was a key couple of metabolites to explore the effect of the that this was an isomer of urolithin A without a hydroxyl at presence or absence of the 3-hydroxylation in the urolithin position 3 that was located at position 4. molecule (Table 1). The discriminant features can be based on Detection of the New Metabolites in Different differences in the acidity of the hydroxyl at 3 position and that Volunteers. The new metabolites were only found in 9 out of the hydroxyl at 8 position that affect the chromatographic of the 47 fecal samples analyzed (19%) (Table 3), confirming

11104 DOI: 10.1021/acs.jafc.9b04435 J. Agric. Food Chem. 2019, 67, 11099−11107 Journal of Agricultural and Food Chemistry Article

Table 3. Presence of the New Urolithins in the Different unknown metabolites were not associated exclusively to Volunteers Stratified by Metabotypes14,16 volunteers either from UM-A or UM-B, indicating that the production of these new metabolites was somehow a Uro M6R Uro CR Uro M7R Uro AR volunteer metabotype (4) (5) (6) (8) transversal catabolic activity additional to those of UM-A fi 2B and UM-B and not associated with the speci c discriminant 3A bacteria of these metabotypes. The discovery of urolithin metabolites without 3-hydrox- 5A fi 6B ylation is a relevant nding as it opens the possibility of detecting and quantifying members of this new family of 7A fl 8 A XXXXurolithin metabolites in biological uids. The production of ff 9A these new metabolites follows a di erent and unique branch in 10 A the catabolic pathway of ellagitannins and ellagic acid (Figure 11 A 3) and can be considered a discriminant characteristic of some − fi 12 A individuals R+ or R . These metabolites can also have speci c 13 B biological properties complementary to those of the known 14 A urolithins already studied. It is particularly relevant that the 15 A different members of this type of “R” metabolites are all 16 A observed together in specific volunteers (Table 3), showing a 17 A particular gut microbiota metabolic activity and therefore a 18 A specific gut bacterial composition. No correlation of the 19 B occurrence of the new metabolites and the body mass index 20 A was detected (unpublished results). 21 A Identification of the New Metabolites in Human 22 A X X Urine. The determination of the occurrence of the new 23 B metabolites in human urine is of interest in order to determine 24 B the potential biological relevance of these new metabolites. 26 A The urine samples of these volunteers producing the new 27 A metabolites in feces were analyzed under the same chromato- 28 B X X X graphic conditions. Most of the volunteers showed the 29 A metabolites at very low concentrations in urine. As could be 30 A expected, the volunteers that had higher production of these 31AXXXXmetabolites in feces also showed the highest concentration of 32BXXXXthe metabolites in urine. Therefore, the HPLC chromatogram 33 A of urine from volunteer 38 (Figure 5) showed phase II 34 A metabolites (glucuronides and sulfates) of the new dihydroxy- 35 0 and trihydroxy urolithins. The chromatogram was dominated 36 B by the urolithin A metabolites including the glucuronide- 37 A X X X sulfate (1′) with [M − H]− at m/z 483, the glucuronide (3′)at 38AXXXXm/z 403, the sulfate (8′)atm/z 307, and the free urolithin A 39 B X (9′)atm/z 227. The chromatogram also showed small 40 B X amounts of two trihydroxy urolithin glucuronide isomers (2′ 41 A X X X and 5′) m/z 419 and the corresponding sulfate (4′) m/z 323. 42 A 43 A Their UV spectra clearly showed that they were conjugates of urolithin M7R as their UV spectra were very similar to that of 44 A fi 45 0 urolithin M7R (Table 1, Figure 5) slightly modi ed with shifts 46 B in the maxima wavelengths to shorter values after the conjugation when compared with the spectrum of the free 47AXXXX 16 48 B urolithin M7. In addition, a dihydroxy urolithin glucuronide ′ 49 A (7 ) m/z 403 with UV spectrum consistent with that of 50 B urolithin AR was also detected (Figure 5). A peak with a UV spectrum similar to that or urolithin M6R was also observed that they were rather unusual metabolites. Nevertheless, it (6′), although the MS analysis did not show a clear spectrum. should be taken into account that the population sample was This study confirms the potential of the analysis of the UV rather small, and all of them were over 40 years and spectra of urolithins and their chromatographic behavior for overweight-obese. Therefore, the relative abundance of these the characterization of these metabolites as was previously 16,17 metabolites should also be explored in other demographic shown. Further studies should be carried out to confirm groups (normoweight and younger individuals). Besides, all the structures of the metabolites tentatively identified by NMR the four new metabolites were present in five out of the nine methods and the occurrence of these metabolites and their samples that produced some of the new metabolites, derived Phase II conjugates in human plasma and to identify supporting a link between the gut bacterial composition and the specific gut bacterial strains responsible for their the biochemical activities (3-dehydroxylase activity) respon- production. Further studies on these new urolithin metabolites sible for the production of these “R” metabolites. The are guaranteed.

11105 DOI: 10.1021/acs.jafc.9b04435 J. Agric. Food Chem. 2019, 67, 11099−11107 Journal of Agricultural and Food Chemistry Article

Figure 5. HPLC−UV chromatogram (305 nm) of urine samples from volunteer 38 after the intake of a pomegranate extract (A) and UV spectra and [M − H]− m/z of the main urolithins detected (B). (1′) urolithin A (3,8-dihydroxy urolithin) sulfoglucuronide; (2′) urolithin M7R glucuronide (4,8,10-trihydroxy urolithin glucuronide); (3′) urolithin A glucuronide (3,8-dihydroxy urolithin glucuronide); (4′) urolithin M7R sulfate (4,8,10-trihydroxy urolithin sulfate); (5′) urolithin M7R glucuronide (4,8,10-trihydroxy urolithin glucuronide); (6′) urolithin M6R (4,8,9,10-tetrahydroxy urolithin) conjugate? (7′) urolithin AR-glucuronide (4,8-dihydroxy urolithin glucuronide); (8′) urolithin A sulfate (3,8- dihydroxy urolithin sulfate); (9′) urolithin A (3,8-dihydroxy urolithin). ■ ASSOCIATED CONTENT ■ AUTHOR INFORMATION *S Supporting Information Corresponding Author *E-mail: [email protected]. Phone: +34-968396200. Fax: The Supporting Information is available free of charge on the +34-968396213. ACS Publications website at DOI: 10.1021/acs.jafc.9b04435. ORCID 1H NMR spectrum of urolithin M6R (4,8,9,10- Juan C. Espín: 0000-0002-1068-8692 tetrahydroxy urolithin) (*) Contaminants. UPLC- Francisco A. Tomas-Barberá n:́ 0000-0002-0790-1739 QTOF-MS/MS fragments of the main urolithins at Funding collision energy 30 V, in bold, the most intense This work has been supported by the Projects BACCHUS fragments, and in red, the fragments with the highest (FP7-KBBE-2012-6-single stage, European Commission Grant intensity. (PDF) Agreement 312090), 19900/GERM/15 (FundacionSé neca,́

11106 DOI: 10.1021/acs.jafc.9b04435 J. Agric. Food Chem. 2019, 67, 11099−11107 Journal of Agricultural and Food Chemistry Article

Spain), and AGL2015-64124 and AGL2015-73107-EXP microflora from ellagic acid and related compounds. J. Agric. Food (MINECO, Spain). Chem. 2005, 53, 5571−5576. ́ ́ ́ Notes (14) Tomas-Barberan, F. A.; García-Villalba, R.; Gonzalez-Sarrías, fi A.; Selma, M. V.; Espín, J. C. Ellagic acid metabolism by human gut The authors declare no competing nancial interest. microbiota: Consistent observation of three urolithin phenotypes in intervention trials, independent of food source, age, and health status. ■ ACKNOWLEDGMENTS J. Agric. Food Chem. 2014, 62, 6535−6538. (15) Gonzalez-Sarrías,́ A.; García-Villalba, R.; Romo-Vaquero, M.; The authors are grateful to Dr. J.E. Yuste for running the NMR ̈ analysis. Alasalvar, C.; Orem, A.; Zafrilla, P.; Tomas-Barberá n,́ F. A.; Selma, M. V.; Espín, J. C. Clustering according to urolithin metabotype explains ■ REFERENCES the interindividual variability in the improvement of cardiovascular ́ ́ ́ ́ risk biomarkers in overweight-obese individuals consuming pome- (1) Cerda, B.; Llorach, R.; Ceron, J. J.; Espín, J. C.; Tomas-Barberan, granate: A randomized clinical trial. Mol. Nutr. Food Res. 2017, 61, F. A. Evaluation of the bioavailability and metabolism in the rat of 1600830. , and antioxidant polyphenol of pomegranate juice. Eur. J. (16) García-Villalba, R.; Espín, J. C.; Tomas-Barberá n,́ F. A. Nutr. 2003, 42,18−28. ́ ́ ́ Chromatographic and spectroscopic characterization of urolithin for (2) Cerda, B.; Espín, J. C.; Parra, S.; Martínez, P.; Tomas-Barberan, their determination in biological samples after the intake of foods F. A. The potent in vitro antioxidant ellagitannins from pomegranate containing ellagitannins and ellagic acid. J. Chromatogr. A 2016, 1428, juice are metabolized into bioavailable but poor antioxidant hydroxy- 162−175. 6H-dibenzopyran-6-one derivatives by the colonic microflora in ́ ́ − (17) Gonzalez-Barrio, R.; Truchado, P.; Ito, H.; Espín, J. C.; Tomas- healthy humans. Eur. J. Nutr. 2004, 43, 205 220. Barberan,́ F. A. UV and MS identification of urolithins and nasutins, (3) Seeram, N. P.; Henning, S. M.; Zhang, Y.; Suchard, M.; Li, Z.; the bioavailable metabolites of ellagitannins and ellagic acid in Heber, D. Pomegranate juice ellagitannin metabolites are present in different mammals. J. Agric. Food Chem. 2011, 59, 1152−1162. human plasma and some persists in urine up to 24 hours. J. Nutr. (18) Hong, S. S.; Suh, H.-J.; Oh, J. S. Phenolic chemical constituents 2006, 136, 2481−2485. ́ ́ ́ of the stem bark of Robinia pseudoacacia. Chem. Nat. Comp. 2017, 53, (4) Cerda, B.; Tomas-Barberan, F. A.; Espín, J. C. Metabolism of 359−361. antioxidant and chemopreventive ellagitannins from strawberries, (19) Cozza, G.; Gianoncelli, A.; Bonvini, P.; Zorzi, E.; Pasquale, R.; raspberries, walnuts and oak-aged wine in humans: Identification of Rosolen, A.; Pinna, L. A.; Meggio, F.; Zagotto, G.; Moro, S. Urolithin biomarkers and individual variability. J. Agric. Food Chem. 2005, 53, − as a converging scaffold linking ellagic acid and coumarin analogues: 227 235. Design of potent protein kinase CK2 inhibitors. ChemMedChem 2011, (5) Espin, J. C.; Larrosa, M.; García-Conesa, M. T.; Tomas-́ − ́ 6, 2273 2286. Barberan, F. A. Biological significance of urolithins, the gut microbial (20) Rupiani, S.; Guidotti, L.; Manerba, M.; Di Lanni, L.; Giacomini, ellagic acid-derived metabolites: The evidence so far. J. Evidence-Based E.; Falchi, F.; Di Stephano, G.; Roberti, M.; Racanatini, M. Synthesis Complementary Altern. Med. 2013, 2013,1−15. ́ of natural urolithin M6, a galloflavon mimetic, as a potential inhibitor (6) Gonzalez-Sarrías,́ A.; García-Villalba, R.; Nuń̃ez-Sanchez,́ M. A.; − ́ ́ ́ of lactate dehydrogenase. A. Org. Biomolec. Chem. 2016, 14, 10981 Tome-Carneiro, J.; Zafrilla, P.; Mulero, J.; Tomas-Barberan, F. A.; 10987. Espín, J. C. Identifying the limits for ellagic acid bioavailability: A (21) Pottie, I. R.; Nandalaru, P. R.; Bowell, G. J. An Inverse Crossover pharmacokinetic study in healthy volunteers after Electron-Demand Diels-Alder-Based Total Synthesis of Urolithin M- consumption of pomegranate extracts. J. Funct. Foods 2015, 19, 7. Synlett 2011, 2245−2247. 225−235. ́ ́ (22) Lemanska, K.; Szymusiak, H.; Tyrakowska, B.; Zielinski, R.; (7) Larrosa, M.; García-Conesa, M. T.; Espín, J. C.; Tomas- Soffers, A. E.; Rietjens, I. M. The influence of pH on antioxidant Barberan,́ F. A. Ellagitannins, ellagic acid and vascular health. Mol. − properties and the mechanisms of antioxidant action of hydroxy- Asp. Med. 2010, 31, 513 539. flavones. Free Radical Biol. Med. 2001, 31, 869−881. (8) Tomas-Barberá n,́ F. A.; Gonzalez-Sarrías,́ A.; García-Villalba, R.; Nuń̃ez-Sanchez,́ M. A.; Selma, M. V.; García-Conesa, M. T.; Espín, J. C. Urolithins, the rescue of “old” metabolites to understand a “new” concept; metabotypes as a nexus between phenolic metabolism, microbiota dysbiosis and host health status. Mol. Nutr. Food Res. 2017, 61, 1500901. (9) Ryu, D.; Mouchiroud, L.; Andreux, P. A.; Katsyuba, E.; Moullan, N.; Nicolet-dit-Felix,́ A. A.; Williams, E. G.; Jha, P.; Lo Sasso, G.; Huzard, D.; Aebischer, P.; Sandi, C.; Rinsch, C.; Auwerx, J. Urolithin A induces mitophagy and prolongs lifespan in C. elegans and increases muscle function in rodents. Nat. Med. 2016, 22, 879−888. (10) Selma, M. V.; Beltran,́ D.; García-Villalba, R.; Espín, J. C.; Tomas-Barberá n,́ F. A. Description of urolithin production capacity from ellagic acid of two human intestinal Gordonibacter species. Food Funct. 2014, 5, 1779−1784. (11) Xu, J.; Yuan, C.; Wang, G.; Luo, J.; Ma, H.; Xu, L.; Mu, Y.; Li, Y.; Seeram, N. P.; Huang, X.; Li, L. Urolithin attenuate LPS-induced neuroinflammation in BV2Microglia via MAPK, Akt, and NFkB signaling pathways. J. Agric. Food Chem. 2018, 66, 571−580. (12) Tulipani, S.; Urpi-Sarda, M.; García-Villalba, R.; Rabassa, M.; Lopez-Uriarte,́ P.; Bullo,́ M.; Jauregui,́ O.; Tomas-Barberá n,́ F.; Salas- Salvado,́ J.; Espín, J. C.; Andres-Lacueva,́ C. Urolithins are the main urinary microbial-derived phenolic metabolites discriminating a moderate consumption of nuts in free-living subjects with diagnosed metabolic syndrome. J. Agric. Food Chem. 2012, 60, 8930−8940. (13) Cerda,́ B.; Periago, P.; Espín, J. C.; Tomas-Barberá n,́ F. A. Identification of urolithin A as a metabolite produced by human colon

11107 DOI: 10.1021/acs.jafc.9b04435 J. Agric. Food Chem. 2019, 67, 11099−11107