Identification and Quantification of Urinary Microbial Phenolic Metabolites by HPLC-ESI-LTQ-Orbitrap-HRMS and Their Relationship with Dietary Polyphenols in Adolescents

This study aimed to develop and validate a liquid chromatography/electrospray ionization-linear ion trap quadrupole-Orbitrap-high-resolution mass spectrometry (HPLC/ESI-LTQ-Orbitrap-HRMS) method to identify and quantify urinary microbial phenolic metabolites (MPM), as well as to explore the relationship between MPM and dietary (poly)phenols in Spanish adolescents. A total of 601 spot urine samples of adolescents aged 12.02 ± 0.41 years were analyzed. The quantitative method was validated for linearity, limit of detection, limit of quantification, recovery, intra- and inter-day accuracy and precision, as well as postpreparative stability according to the criteria established by the Association of Official Agricultural Chemists International. A total of 17 aglycones and 37 phase II MPM were identified and quantified in 601 spot urine samples. Phenolic acids were the most abundant urinary MPM, whereas stilbenes, hydroxytyrosol, and enterodiol were the least abundant. Urinary hydroxycoumarin acids (urolithins) were positively correlated with flavonoid and total (poly)phenol intake. An HPLC-ESI-LTQ-Orbitrap-HRMS method was developed and fully validated to quantify MPM. The new method was performed accurately and is suitable for MPM quantification in large epidemiological studies. Urinary lignans and urolithins are proposed as potential biomarkers of grain and nut intake in an adolescent population.


Introduction
The beneficial health effects of dietary (poly)phenols have been reported in several epidemiological and clinical trials [1][2][3], although their biological activities are not all attributed to their native form. After ingestion, modification by phase I and II metabolic enzymes reduces the concentrations of native (poly)phenols in the systemic circulation [4,5]. More than 80% of dietary (poly)phenols are not absorbed in the small intestine and reach the colon, where they undergo conjugation and are metabolized by gut microbiota through a range of enzymatic reactions (deglycosylation, dehydroxylation, demethylation, deconjugation, epimerization, ring fission, hydrolysis, and chain-shortening) [5][6][7]. The microbial phenolic metabolites (MPM) may be more bioactive than the parental (poly)phenol when they reach the target cells or tissues [8][9][10][11]. Fewer studies have reported MPM in young populations, such as adolescents.
High-resolution mass spectrometry (HRMS) using an Orbitrap mass analyzer is a well-established method for rapid targeted and untargeted identification of (poly)phenols in nutrimetabolomics studies [12]. This equipment provides exact mass information, twostage mass analysis (MS/MS), and multi-stage mass analysis (MS n ), which facilitates the structural elucidation of known and unknown compounds [12][13][14][15]. Therefore, HRMS constitutes a versatile and robust system for quantitative analysis [16][17][18][19]. However, to date, few methods are available to quantify MPM in human biological samples using high performance liquid chromatography (HPLC)/Orbitrap-HRMS.
The aim of this study was to develop and validate a high-performance liquid chromatography/electrospray ionization-linear ion trap quadrupole-Orbitrap-high-resolution mass spectrometry (HPLC/ESI-LTQ-Orbitrap-HRMS) method to identify and quantify urinary MPM in adolescents, and to explore the relationship of MPM with dietary (poly)phenols.

Study Design and Sample Selection
This work was carried out as a cross-sectional analysis within the SI! (Salud Integral) Program for Secondary Schools trial in Spain, a cluster-randomized controlled intervention trial (NCT03504059) aiming to evaluate the impact of a lifestyle educational program on cardiometabolic health in adolescents. A total 1326 participants were recruited in the baseline of the trial. Details of the study design, recruitment procedures, and Commission on Ethics are available elsewhere [20]. Informed consent was obtained for all the parents or caregivers.
For the current study, baseline data of 601 randomly chosen participants (53% girls) with available baseline urine samples were included, equivalent to 45% of the original cohort.

Chemicals and Urine Samples
The provenance of chemicals and standards is listed in the Supplemental data. Urine samples were collected in 2017 and stored at −80 • C until analysis.

Sample Preparation and Extraction of (Poly)Phenols
All the spot urine samples were analyzed in a room with filtered light and kept on ice to avoid phenolic oxidation, following the procedure proposed by Martínez-Huelamo et. al., with some modifications [21]. Firstly, 1 mL of urine was acidified with 2 µL of formic acid and centrifuged at 15,000× g at 4 • C for 4 min. After centrifugation, the urine underwent a solid-phase extraction (SPE) and clean-up procedure using Waters Oasis HLB 96-well plates 30 µm (30 mg) (Waters Oasis, Milford, MA, USA). Plates were activated by consecutively adding 1 mL of methanol (MeOH) and 1 mL of 1.5 M formic acid. After loading 1 mL of sample, clean-up was performed with 0.5 mL of 1.5 M formic acid and 0.5% MeOH, and the elution with 1 mL MeOH acidified with 0.1% of formic acid.

Mass Spectrometry Parameters
Accurate mass measurements were performed on an LTQ Orbitrap Velos mass spectrometer (Thermo Scientific, Hemel Hempstead, UK) equipped with an ESI source working in negative mode. Mass spectra were acquired in profile mode with a setting of 30,000 resolution at m/z 400, and the mass range was from m/z 100 to 2000. Operation parameters were as follows: source voltage, 5 kV; sheath gas, 50 units; auxiliary gas, 20 units; sweep gas, 2 units, and capillary temperature, 375 • C.

Validation of the HPLC/ESI-LTQ-Orbitrap-HRMS Method
The method was validated following the criteria of the Association of Official Agricultural Chemists (AOAC) International in terms of linearity, limit of detection (LOD), limit of quantification (LOQ), recovery, intra-and inter-day accuracy and precision, and postpreparative stability [23]. All parameters were examined based on three concentrations (low, medium, and high) of each phenolic compound standard, as shown in Table 1.
LOD and LOQ were estimated for a signal-to-noise (S/N) ratio of 3 and 10, respectively.

Accuracy and Precision
Accuracy was determined by analyzing five replicates of spiked synthetic urine with three known concentrations (Table 1) to evaluate the closeness of agreement between the calculated amount and the nominal amount of analyte. The results were expressed as the percentage of the ratio of the mean concentration observed and the known spiked concentration in the biological matrices. Precision was calculated using relative standard deviation (RSD) between the five spiked urine samples at three different levels on three different days. Intra-and inter-day precision was assessed using five determinations per three concentration levels (Table 1) in a single analytical run or on three different days, respectively.

Recovery and Matrix Effect
Recovery and matrix effects (ME) were evaluated following the procedure described by Matuszewski et al., and Pereira-Caro et al. [19,24], analyzing three synthetic urines spiked at the three standard concentration levels (Table 1). Recoveries were calculated as the ratio between the area responses of standard concentration levels dissolved in preextracted samples and the analyte area responses of post-extracted urine spiked at the same concentrations. The results were expressed as recovery rate.
MEs were determined with the same concentration levels by comparing area responses of the spiked pre-extracted samples with the analyte area responses with neat standards dissolved in the mobile phase. The results were expressed as percentages. ME values above 100% are considered to indicate ion enhancement, and below 100% ion suppression.

Stability
Postpreparative stability and freeze and thaw stability were assessed in this method. Postpreparative stability of the sample extraction process and during the time inside the autosampler at 4 • C were evaluated by injecting the post-extracted synthetic urine spiked with two standard concentrations (low and high) ( Table 1) into the HPLC-ESI-LTQ-Orbitrap-HRMS system at 0 and 24 h. Freeze and thaw stability were assessed by injecting the post-extracted synthetic urine spiked at the same concentration levels into the HPLC-ESI-LTQ-Orbitrap-HRMS system after three freeze (−80 • C) and thaw (room temperature) cycles.

Selectivity
The selectivity of the method was assessed by comparing chromatograms of blank human urine from three individuals and urine spiked with analytes at a known low concentration to discriminate between analytes and other endogenous components in urine.

Quantification of MPM
Calibration curves were constructed with available standards in synthetic urine and subjected to the same procedure as described above. To quantify phase II metabolites (glucuronides and sulfates), calibration curves of the aglycon form were used. Samples with concentrations that exceeded the highest point of the calibration curve were diluted and reinjected into the HPLC-FTMS system. Quantitative data processing was performed using Trace Finder software (LC version 4.1, Thermo Fisher Scientific, San Jose, CA, USA).
MPM concentration was normalized by urinary creatinine concentrations, which were determined using the Jaffé alkaline picrate method adapted to microtiter 96-well plates [26] and expressed as µg MPM/g creatinine.

Dietary (Poly)Phenols
Dietary intake was estimated using a semiquantitative food frequency questionnaire [27]. Dietary (poly)phenol intake was assessed by matching data from the Phenol-Explorer database v.3.6. [28]. Flavonoids, phenolic acids, stilbenes, lignans, phenolic acids, tyrosols, and other minor (poly)phenols, such as alkylphenols and alkylmethoxyphenols, were included in this analysis. Total (poly)phenol intake was estimated as the sum of individual (poly)phenol intakes and categorized into tertiles. Energy-adjusted (poly)phenol intake was calculated by the residual method established by Willet et al. [29].

Data Analysis
General characteristics of the studied population are presented as means (standard deviation (SD)) or median (interquartile range (IQR)) for quantitative variables and percentages (number) for categorical variables. MPM concentrations are presented as the mean, standard error of the mean (SEM). Student's t-test was used to compare mean values of general characteristics between girls and boys, but also to compare MPM and postpreparative stability.
For the statistical analysis, MPM levels below the LOQ were set to values corresponding to half the LOQ. Pearson correlation was used to assess the relationship between urinary MPM and dietary (poly)phenols, as well as polyphenol-rich food sources. The false discovery rate (FDR) method was applied to adjust p-values for multiple correlations [30]. Data were normalized with the inverse normal distribution before the analysis.
The overall urinary MPM pattern and tertiles of total phenolic intake were assessed using principal component analysis (PCA) and presented as biplots in which eigenvectors were plotted as lines and the scores of individual samples as points. Beforehand, MPM data were standardized to unit variance.
Statistical analyses were conducted using the Stata statistical software package version 16.0 (StataCorp, College Station, TX, USA) and R v.4.1.1 (https://www.r-project.org, accessed on 1 April 2022). Statistical tests were two-sided, and p-values below 0.05 were considered significant.

Optimization of the HPLC/ESI-LTQ-Orbitrap-HRMS Method
Several SPEs solutions, as well as two SPE cartridges (Table S1 and Figure S1), were tested in order to obtain optimum recoveries. Two reverse-phase chromatographic columns were tested: Kinetex F5 (50 × 4.6 mm i.d., 2.6 µm) (Phenomenex, Torrance, CA, USA) and Atlantis T3 C18 (100 × 2.1 mm i.d., 3 µm) (Waters, Milford, MA, USA), obtaining better recoveries with Kinetex F5 and SPE 2 procedure ( Figure S1). Different percentages of formic acid (from 0.05% to 0.1%) in mobile phases were tested to achieve desirable peak shapes and compound separation. The best results were obtained with 0.05% formic acid (data not shown). Two injection volumes (5 and 10 µL) were also tested to ensure optimum separation and detection of the analytes, and 5 µL sample injection gave the best results (data not shown). Details of the analytical conditions tested are available in the Supplemental Data.

Linearity, LOD, and LOQ
The HPLC/ESI-LTQ-Orbitrap-HRMS method provided quadratic responses with coefficients of determination (R 2 ) above 0.995 for all standards (Table S2). Weighted factors (1/x statistical weight) were used to obtain the most reliable calibration curves.
The sensitivity of the method was evaluated by determining the LOD and LOQ of a synthetic urine sample spiked with standards. The LOD ranged from 0.02 to 3.29 µg/L, and LOQ from 0.06 to 10.96 µg/L.

Precision and Accuracy
Intra-and inter-day precision varied in the ranges of 0−15% and 1−16%, respectively, in accordance with the values proposed by the AOAC (RDS < 15%) [23]. However, interday precision values for the lowest concentration of gallic acid, 3-hydroxytyrosol, and 3,4-dihydroxyphenylpropionic acid were 58, 31, and 26%, respectively (Table S3), possibly due to early elution, which leads to a lower resolution peak when the concentration is low. Pereira et al. reported an intra-day precision of less than 15% (0% to 10%) for flavan-3-ols and their metabolites in a study using ultra high-performance liquid chromatography (UHPLC)-HRMS [19].
The accuracy was within the accepted limits of the AOAC guidelines [23] at all tested concentration levels for 89% of the metabolites analyzed, ranging from 80 to 120%. However, at the lowest concentrations the inter-day accuracy of gallic acid, 3,4-dihydroxyphenylpropionic acid, m-coumaric acid, and urolithin-A fell outside this range (Table S3).

Matrix Effect and Recovery
The average ME was 83%, with ranges from 53% to 126%, except those of urolithins-A and -B, which were below 35%. Minor ion suppression was also reported by Ordoñez et al. and Pereira-Caro et al. [18,19]. Ion enhancement was observed for 4-hydroxybenzoic acid and 3-hydroxyphenyacetic acid (Table S3).
The average recovery of the three concentration levels was 89%, ranging between 70% and 99%. The lowest recovery was for gallic acid and 3-hydroxytyrosol, which was 70% at the lowest concentration (Table S3). Similarly, Ordoñez et al. reported a mean recovery of 73% of urinary (poly)phenols extracted by an HLB cartridge and using an HPLC-HRMS method, obtaining a good recovery rate of 79% to 104% for free phenolic and glucuronide derivatives [18]. Better recoveries were reported by Pereira-Caro et al., with values ranging from 95% to 102% for 34 flavan-3-ol and its metabolites in rat urine samples analyzed by UHPLC-HRMS [19].

Stability
The postpreparative stability assay showed no significant variation of analyte concentration in the urine matrix 24 h post-extraction at both low and high concentrations, except for 3-hydroxyphenylacetic acid, which was the analyte with the highest reduction (13%) ( Figure S2). The freeze and thaw stability assay showed a signal decline of 14% for most analytes after the third freeze-thaw cycle. Likewise, Martínez-Huélamo et al., described a 12.9% reduction in signal for 3-hydroxyphenylacetic acid [21].

Selectivity
Selectivity was confirmed by the absence of endogenous peaks in chromatograms at the same retention time as the analytes in three human urine samples. The method was, therefore, found to be selective for analytes at low concentrations and was able to discriminate between analytes and other components in urine.

General Characteristics of the Study Population
Out of the 601 randomized participants selected in this cross-sectional analysis, 546 had available information of food intake. The general characteristics of participants are presented in Table 2. The average age and body mass index (BMI) were 12.0 (0.4) years and 20.9 (4.2) kg/m 2 , respectively. The mean energy-adjusted (poly)phenol intake was 683.5 (335.3) mg/day. No differences were observed between boys and girls in terms of BMI and total (poly)phenol intake ( Figure S3). Higher mean intakes of energy (p-value = 0.002), carbohydrates (p-value = 0.001), total fat (p-value = 0.010), and proteins (p-value < 0.001) were observed in boys ( Figure S3).

Identification and Quantification of Urinary MPM
Identification of MPM according to classes of (poly)phenols (lignans, hydroxybenzoic acids, hydroxycinnamic acids, hydroxyphenylacetic acids, hydroxyphenylpropanoic acids, stilbenes, hydroxycoumarins, and tyrosols) are presented in Table S4. A total of 54 MPM were identified in urine. Enterolactone and urolithin diglucuronides were determined only in one sample.
Concentrations of MPM are summarized in Table 3. Excretion of urinary MPM varied highly between participants, and the majority of MPM were detected in the form of glucuronides and sulfates. Consistent with our results, Ordónez et al. reported that an HPLC-HRMS method was suitable for the analysis of phase II metabolites [18]. The most abundant MPM in the urine of all participants were phenolic acids, namely 3hydroxyphenylacetic acid, hydroxyphenylacetic sulfate and glucuronide, protocatechuic acid sulfate-I, 3,4-dihydroxyphenylpropionic acid sulfate, hydroxybenzoic acid sulfate, and vanillic acid sulfate. These results are in agreement with Zamora-Ros et al., who detected phenolic acids as the most abundant urinary MPM in adult participants in the European Prospective Investigation into Cancer and Nutrition (EPIC) study [31]. Similarly, Hurtado-Barroso et al. found phenylacetic acids to be among the most abundant urinary MPM in young adults [32].  Urinary concentrations of stilbenes (dihydroresveratrol), tyrosols (3-hydroxytyrosol), and lignans (enterodiol) were low, with mean values below 10 µg/g of creatinine. Those reported by Zamora-Ros et al. in adults from the EPIC study were also low, being less than 5 µg/24 h [31]. These levels could be explained by a low dietary intake of stilbenes, tyrosols, and lignans, as reported in the food frequency questionnaire.
Interindividual variations in MPM could be explained by the gut microbiota profile, which is affected by age, gender, hormonal status, dietary habits, and other lifestyle variables [33]. In this study, the gut microbiota profile was not analyzed and thus the influence of the microbial family on MPM production was not determined.
Differences in urinary MPM between boys and girls are shown in Figure 1. Boys had higher values of 3,4-dihydroxyphenylpropionic, dihydroxyphenylpropionic sulfate, gallic acid, gallic acid sulfate, p-coumaric acid, vanillic acid glucuronide and sulfate, hydroxybenzoic acid glucuronide-I and sulfate-I, protocatechuic acid sulfate, 3-hydroxyphenylacetic acid and hydroxyphenylacetic acid glucuronide and sulfate than girls. Our findings are in line with those of Zamora et al., who observed that the median urinary concentrations of tyrosol, vanillic acid, and 4-hydroxyphenylacetic acid were at least 1.4-fold higher in men than women [31]. Similarly, Mumford et al., found higher values of enterodiol and enterolactone in females than males [34]. As sex hormones may be responsible for these differences [8,35], a limitation of the current study is that the follicular phase of the menstrual cycle was not considered during the urine collection to minimize bias related to the hormonal status of the participants.

Conclusions
In conclusion, an HPLC-ESI-LTQ-Orbitrap-HRMS method was developed and fully validated to quantify urinary MPM in terms of linearity, sensitivity, recovery, accuracy, and precision. To our knowledge, this is the first time that several MPM have been identified and quantified in urine samples of an adolescent population using an HPLC-ESI-LTQ-Orbitrap-HRMS method on a large scale. Variations in MPM were observed between participants, which were associated with variability in dietary (poly)phenol intake and sex. Finally, some MPM were found to be potential dietary biomarkers of specific food groups, namely lignans for whole grains and urolithins for nuts. Further investigations are needed to explore the relationship between MPM and dietary sources of (poly)phenols.

Supplementary Materials:
The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/antiox11061167/s1. Standards and chemicals. Analytical condition testing before validation HPLC/ESI-LTQ-Orbitrap-HRMS method. Figure S1: Recovery obtained according to different solid phase extraction and reverse-phase chromatographic columns (Kinetex F5 (50 × 4.6 mm i.d., 2.6 µm) and Atlantis T3 C18 (100 × 2.1 mm i.d., 3 µm). Figure S2: Postpreparative stability. Mean concentrations (µg/L) of phenolic compounds recovered at the start (t = 0) and at 24 h with two standard concentrations prepared in synthetic urine. Figure S3: General characteristics of participants according to gender. Figure S4: Principal component (PC) biplot of subclass of microbial phenolic metabolites (MPM) according to tertiles of total polyphenol intake (n = 546). Figure S5: Heatmap of the Pearson correlation between subclass of microbial phenolic metabolites and polyphenol-rich food intake in adolescents. Table S1: Recovery obtained in Oasis HLB and PRiMe HLB. Table S2: Validation data: Linearity ranges, coefficient of determination, and low limits of detection and quantification of microbial phenolic metabolite. Table S3: Intra-and interday precision and accuracy, matrix effect and recovery results for three concentration levels (high, medium, and low); RSD (%) was calculated for the recovery values for three replicates. Table S4: Identification of microbial phenolic metabolites in urine by HPLC/ESI-LTQ-Orbitrap-HRMS. Table S5: Pearson correlation coefficients between microbial phenolic metabolites and dietary polyphenols in adolescents. References [37,38] are cited in the Supplementary Materials. ico, Instituto Cervantes, Alburquerque, USA; Instituto Cervantes, Milan, Italy, Instituto Cervantes, Tokyo, Japan, Lilly Laboratories, Spain, and Wine and Culinary International Forum, Spain, and non-financial support to organize a National Congress on Nutrition, as well as feeding trials with product from Grand Fountain and Uriach Laboratories, Spain. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results. The other authors declare no conflict of interest.