Metabolomic insights into phenolics-rich chestnut shells extract as a nutraceutical ingredient – A comprehensive evaluation of its impacts on oxidative stress biomarkers by an in-vivo study

The present study attempted for the first time to explore the effects of the daily oral intake of a phenolics-rich extract from chestnut shells (CS) on the metabolomic profiling of rat tissues by liquid chromatography coupled to Orbitrap-mass spectrometry (LC-ESI-LTQ-Orbitrap-MS) targeted to polyphenolics and their metabolites and screen potential oxidative stress biomarkers, validating its use as a promising nutraceutical ingredient with outstanding antioxidant properties for the prevention and co-therapy of lifestyle-related diseases triggered by oxidative stress. The results demonstrated new insights into the metabolomic fingerprinting of polyphenols from CS, confirming their absorption and biotransformation by phase I (hydrogenation) and II (glucuronidation, methylation, and sulfation) enzymes. Phenolic acids were the main polyphenolic class, followed by hydrolyzable tannins, flavanols, and lignans. In contrast to the liver, sulfated conjugates were the principal metabolites reaching the kidneys. The multivariate data analysis predicted an exceptional contribution of polyphenols and their microbial and phase II metabolites to the in-vivo antioxidant response of the CS extract in rats, recommending its use as an appealing source of anti-aging molecules for nutraceuticals. This is the first study that explored the relation between metabolomic profiling of rat tissues and in-vivo antioxidant effects after oral treatment with a phenolics-rich CS extract.


Introduction
Inflammation and oxidative stress are potentially major risk factors for the development of chronic diseases, including cancer, neurodegenerative, cardiovascular, and metabolic pathologies (Charles et al., 2021).Oxidative stress results from the imbalance between the overproduction of pro-oxidant reactive species and the antioxidant defense capacity of cells, namely enzymatic (i.e., catalase (CAT), glutathione peroxidase (GSH-Px) and superoxide dismutase (SOD)) and nonenzymatic antioxidants (i.e., reduced glutathione (GSH)), favoring the accumulation of reactive species with deleterious effects in biological tissues (Pinto, Cádiz-Gurrea, Vallverdú-Queralt, et al., 2021).
A high dietary intake of fruits and vegetables enriched in antioxidants (such as polyphenols) has unveiled a close relationship with preventive effects against oxidative stress-mediated diseases (Martins et al., 2016;Pinto, Cádiz-Gurrea, Vallverdú-Queralt, et al., 2021).Polyphenols are plant secondary metabolites with pronounced bioactivity documented in animals and humans (Charles et al., 2021;Martins et al., 2016).The metabolic effects of polyphenols have been extensively evaluated by in-vivo studies through measuring the concentrations of glucose, lipids, and other oxidative stress biomarkers in blood (Martins et al., 2016;Noh et al., 2010Noh et al., , 2011;;Pinto et al., 2023).Even so, these biomarkers may provide inconsistent findings and fail to reflect the full diversity of metabolic effects (Martins et al., 2016).Nevertheless, polyphenols are poorly absorbed, owing to limited bioavailability and extensive metabolism across different organs (A. López-Yerena et al., 2021;Martins et al., 2016).For instance, lower bioactivity has been suggested for polyphenols that undergo phase II metabolism, whereas the biotransformation into aglycones allows better intestinal absorption via gut microbiota (A. López-Yerena et al., 2021;Marhuenda-Muñoz et al., 2019).The interaction polyphenols-gut microbiota may also deliver positive effects against several diseases (Corrêa et al., 2019;Martins et al., 2016).Hence, the hypothesis that polyphenols exhibit a direct scavenging activity on radicals is constrained.Beyond the welldocumented anti-radical properties, numerous mechanisms have evidenced the efficacy of polyphenols in modulating antioxidant and detoxifying enzymes activities, encompassing the first defense system under oxidative stress conditions (Martins et al., 2016;Pinto, Cádiz-Gurrea, Vallverdú-Queralt, et al., 2021).
Metabolomics has arisen as a hot research topic in the pharmaceutical and nutraceutical fields to comprehend the causal role of bioactive molecules in preventing lifestyle-related diseases and attenuating premature aging effects (Charles et al., 2021).Hence, the metabolites may be recognized as potential oxidative stress biomarkers (Charles et al., 2021).Although metabolomic studies in biological tissues are still scarce, this approach comprises a valuable tool for a deep insight into the biotransformation of polyphenols, underlying their in-vivo bioactivity.
The purpose of this study was to explore the effects of the daily oral intake of phenolics-rich CS extract on the metabolomic profiling of rat tissues by liquid chromatography coupled to Orbitrap-mass spectrometry (LC-ESI-LTQ-Orbitrap-MS), valorizing it as a prominent nutraceutical ingredient for the prevention and co-therapy of oxidative stressmediated pathologies.The relationship between phenolic metabolites and potential oxidative stress biomarkers, namely LPO, SOD, CAT, and GSH-Px activities, was ascertained through multivariate data analysis.This is the first study that offers an in-depth assessment of the metabolomic fingerprinting of different tissues from rats orally treated with a nutraceutical extract from CS.

Chestnut shells
Shells were kindly provided by Sortegel (Bragança, Portugal) in October 2018.CS were dehydrated at 40 • C for 24 h, ground to 1 mm particle size (Retsch ZM200 ultra-centrifugal grinder, Germany) and stored at room temperature in the dark.

In-vivo studies
Male Wistar rats (200 ± 50 g, 5-6 weeks old) supplied by Jackson Laboratory (Bar Harbor, ME, USA) were acclimatized for 1 week, housed in polypropylene cages under standard conditions (temperature: 21 ± 2 • C; relative humidity: 45-55%; light/dark cycle: 12 h/12 h) and fed ad libitum.Animal procedures were performed as described by Pinto et al. (Pinto et al., 2023).The animals were randomly separated into three groups (n = 6 per group): a control group orally administered with water (4 mL/kg body weight (b.w.)) and two treatment groups orally treated with two doses of CS extract (50 and 100 mg/kg b.w.) dispersed in water.The solutions were administered per os by gastric gavage once daily, for 7 days, after a fasting period of 4 h.Acute toxicity test was D. Pinto et al. conducted to ascertain the toxic effects of the two doses of extract by examining the health status, body weight, and behavior.A minimum number of animals was used following a commitment to the 3R's policy (Replacement, Reduction & Refinement).All procedures were approved by the Local Ethical Committee from the Animal Welfare Body at i3S -Institute for Research & Innovation in Health (protocol number BSm_2017_10) and accomplished following FELASA, ARRIVE guidelines and European Directive 2010/63/EU for animal experiments.The general health status was monitored, and humane endpoints were defined in the case of any toxic effect and impairment of animal welfare.At the end of the experiments, the animals were euthanized by pentobarbital overdose (50 mg/kg b.w.).
2.4.1.2.Tissues pre-treatment.Liver and kidneys of rats were dissected and handled on ice, in a room with light-filter.The tissue homogenates were prepared with 50 mM potassium phosphate (pH 7.0) at 1:5 (w/v) ratio, using a T10 basic Ultra-Turrax® (IKA laboratory technology, Staufen, Germany).After centrifuging at 20,000 g for 10 min (4 • C), the supernatants were kept at − 80 • C. The protein precipitation was performed as described in previous studies (López-Yerena, Vallverdú-Queralt, et al., 2021;Pinto et al., 2023).Briefly, tissue homogenates were thawed and centrifuged at 11,000 g for 10 min (4 • C).The upper layer (100 µL) was added to ice-cold ACN with 2% formic acid in 1:4 (v/ v) ratio.Preliminary tests were done to ascertain the most adequate ratio.Samples were homogenized for 1 min, stored at − 20 • C for 20 min, and centrifuged at 11,000 g for 10 min (4 • C).Finally, 100 µL of the organic phase was transferred to vials for analysis.
The instrumental conditions were set as described by Pinto et al. (Pinto et al., 2023).Samples were examined in full scan mode at a resolving power of 30,000 at m/z 600 and data-dependent MS/MS events were acquired at a resolving power of 15,000.Most intense ions were detected in FTMS mode triggered data-dependent scanning.Ions not sufficiently intense for a data-dependent scan were explored in MSn mode.Precursors were fragmented by collision-induced dissociation using a C-trap with normalized collision energy (35 V) and activation time of 10 ms.
The polyphenols were identified using commercial standards, whereas the identification of their metabolites was based on the elution time, chemical formula, and MS/MS fragmentation and compared to similar compounds.For the identification of phenolic metabolites in rat tissues, literature data and human metabolome database (https://hmdb.ca) were consulted.In the extract, the remaining compounds (of which the standards were not available) were identified by comparison with previous data from our research group (Escobar-Avello et al., 2021;Sasot et al., 2017), considering chemical formula, mass spectrometry fragmentation, and retention times, as well as confirmed through databases (https://foodb.caand https://phenol-explorer.eu).Previous studies followed identical procedures for the identification of polyphenols in foods, plant extracts and biological tissues (Escobar-Avello et al., 2021;Sasot et al., 2017).MS n patterns were acquired to determine the fragment ions produced in the linear ion trap.The elemental composition of the phenolic metabolites was assessed by accurate masses and isotopic patterns.In FTMS mode, the mass range was set from m/z 100 to 800.
The quantification of polyphenols and their metabolites was achieved by plotting the calibration curves in the respective tissues (concentration range = 0.1-3 µg/mL, R 2 > 0.992) for the polyphenols previously identified as major compounds in CS extract, prepared using the same extraction technique and conditions, i.e., GA, ellagic acid, methyl gallate, protocatechuic acid, and pyrogallol (Pinto, Vieira, et al., 2021).Semi-quantification was attained for the remaining polyphenols by comparison with the respective standards.The semi-quantification of phenolic metabolites was also performed by comparing with the standards of parent compounds through calibration curves plotted, owing to the lack of standards for phenolic metabolites available on the market.The use of the respective tissues (i.e., rat liver and kidney) to plot the calibration curves is of uttermost importance to eliminate interferences from matrix effects, providing the accurate identification and quantification of polyphenolics and their metabolites (López-Yerena, Domínguez-López, et al., 2021;López-Yerena, Vallverdú-Queralt, et al., 2021); Pinto et al., 2023).Besides identification, recent studies have successfully employed LC-ESI-LTQ-Orbitrap-MS for quantitative analysis targeted to polyphenols and their metabolites, using appropriate methods, databases, and standards for all compounds to be quantified, and applied to a high variety of complex samples, such as human and rat blood serum, human urine, rat organs (e.g., liver, kidney, intestine, lung, and heart), and rat tissues (e.g., muscle and skin), confirming its reproducibility and feasibility (Laveriano-Santos, Marhuenda-Muñoz, et al., 2022;Laveriano-Santos, Quifer-Rada, et al., 2022;(López-Yerena, Domínguez-López, et al., 2021;López-Yerena, Vallverdú-Queralt, et al., 2021); Pinto et al., 2023;Sasot et al., 2017).System control and data treatment were performed using XCalibur v3.0 software (ThermoFisher Scientific, Hemel Hempstead, UK).The results were expressed in nmol of each polyphenol equivalents per mg of tissue (nmol/g tissue).

Data statistics
The results were presented as mean ± SD of three independent experiments.IBM SPSS Statistics v24.0 software (Chicago, IL, USA) was used for one-way ANOVA and Tukey's HSD tests.A denoting significance was set for p < 0.05.Principal component analysis (PCA) and Pearson correlations between in and vivo antioxidant activity (data published in our previous paper (Pinto et al., 2023)) and metabolic profile of rat tissues were accomplished using GraphPad Prism v9 software (La Jolla, CA, USA).PCA was performed to allow the identification of variables that most significantly affect the samples clustering.Two principal components (PC) were used to establish the model after confirming the normal distribution of the variables.Prior to the PCA, an ANOVA analysis and a Bartlett's test for sphericity were carried out to ascertain all the assumptions needed for a suitable PCA application (Bailey, 2012;Spiegelberg & Rusz, 2017).A scores plot was designed to identify the trends among treatment groups, while a biplot diagram was used to disclose the contribution of different variables.

Metabolomic profiling of chestnut shells extract
Plant-based foods and their by-products have been indicated as vital allies in the promotion of a healthy lifestyle by preventing chronic diseases, protecting against premature aging, and providing many other health benefits (Rudrapal et al., 2022).A comprehensive assessment of the biological effects of plant-based foods depends not only on their nutritional and phytochemical composition, but also on an in-depth understanding of their metabolism and in-vivo bioactivity (Martins et al., 2016;Rudrapal et al., 2022).Most of the pro-healthy properties of plant-based foods, such as CS, are attributed to their richness in polyphenols (Lameirão et al., 2020;Pinto & Cádiz-Gurrea et al., 2020;Pinto, Silva, et al., 2021;Pinto, Vieira, et al., 2021).
Recently, our research group performed a preliminary study on the phytochemical composition of the CS extract prepared using the same extraction technology and conditions by LC/ESI-MS and reported its richness in ellagic acid, GA, protocatechuic acid, methyl gallate, and pyrogallol (Pinto, Vieira, et al., 2021).Nonetheless, several compounds remained unknown.This study evaluated the full metabolomic profiling of CS extract targeted on polyphenols, especially considering the ones already identified in the previous study.Notwithstanding, a comprehensive examination regarding the unknown compounds was also performed before the metabolomic approach in liver and kidney.Table 1 presents the polyphenols and metabolites identified in CS extract and rat tissues.Fig. 1 depicts the polyphenols quantified in CS extract.
A total of 37 compounds were identified in the extract, with a total content of 107.80 µmol/g dw.The different polyphenolic classes are present in the following order: phenolic acids (89.31%) > tannins (2.38%) > flavonoids (0.88%).Among phenolic acids, hydroxybenzoic acids (HBAs) represent 57.74% of the total content, while 30.04% correspond to hydroxycinnamic acids (HCAs).A small amount of hydroxyphenylpropanoic acids (HPPAs) (1.53%) was also detected.Protocatechuic acid (48.47 µmol/g dw) was the major compound identified, followed by CinnamAc (22.38 µmol/g dw), 4-HBA (11.32 µmol/g dw), and pyrogallol (10.09 µmol/g dw).Some of these compounds may result from the thermal degradation of more complex  phenolic acids and flavonoids during the extraction at high temperatures (220 • C/30 min) (Pinto, Vieira, et al., 2021).For instance, pyrogallol may be generated via decarboxylation of GA, while protocatechuic acid is a catechin metabolite formed by its thermal decomposition (Bento-Silva et al., 2020).Ellagic acid was the only hydrolysable tannin identified, probably derived from ellagitannins degradation (Tomás- Barberán et al., 2014).Only two flavonoids were detected, (epi)catechin and (epi)gallocatechin, whose presence may be explained by the thermal decomposition of (epi)catechin-O-gallate and (epi)gallocatechin-Ogallate through the loss of GA.Secoisolariciresinol was the only lignan n.i., non-identified.n.q., non-quantified.Results are expressed as mean ± standard deviation, n = 6 in each group.Different letters (a and b) in the same line within the same rat tissue indicate significant differences between groups (p < 0.05).Different superscript numbers (1 and 2) in the same line within the same treatment group indicate significant differences between rat tissues (p < 0.05).
The use of polyphenols as key ingredients for adjuvant therapy of numerous lifestyle-related chronic diseases has been targeted on latest studies (Martins et al., 2016).Recent advances suggested that phenolic acids, including HCAs and HBAs, may serve as valuable molecules for the treatment of oxidative stress-related cardiovascular diseases, such as atherosclerosis, coronary heart disease, dyslipidemia, hypertension, stroke, and cancer, due to their strong antioxidant potential, protective effects against oxidative damage and signaling pathways modulation (Rudrapal et al., 2022).Beyond the antioxidant and antiradical properties, HCAs are natural molecules used in the management of lipid metabolism, obesity, and inflammatory metabolic diseases, i.e., CA, FA, p-CoumAc, ChlorogenAc and CinnamAc, via i) inhibiting the macrophage infiltration and nuclear factor κB (NF-κB) activation in adipose tissues; ii) preventing the expression of proinflammatory adipokines tumor necrosis factor-α (TNF-α), monocyte chemoattractant protein-1 (MCP-1), and plasminogen activator inhibitor type-1 (PAI-1); and iii) promoting the secretion of adiponectin capable of regulating the glucose levels, lipid metabolism, and insulin sensitivity owing to its anti-fibrotic, anti-inflammatory, and antioxidant effects (Alam et al., 2016).Therefore, polyphenols are potential candidates for the development of novel nutraceuticals with potential use in chronic pathologies.

Identification and quantification of polyphenols and metabolites in the liver
Metabolomics combines high-throughput analytical methodologies and statistics to characterize the metabolic profiling of bioactive compounds and its effects on bioactivity, providing new insights into mechanisms of action and proposing novel biomarkers for the proven effects (López-Yerena, Domínguez-López, et al., 2021); Rudrapal et al., 2022).Several studies applied metabolomic approaches in blood or urine (López-Yerena, Domínguez-López, et al., 2021); Rudrapal et al., 2022;Sasot et al., 2017), however, just few studies have explored the metabolic profile in biological tissues (Anallely López-Yerena et al., 2021).In this regard, there is only one study evaluating the metabolomic profiling of blood serum from rats orally treated with CS extract (Pinto et al., 2023).Noteworthy, this is the first study that provides a comprehensive assessment of the metabolomic profile of tissues from rats orally treated with CS extract and correlate it with previously published outcomes of the in-vivo protective effects against oxidative stress.
All metabolites from the parent compounds identified in CS extract were searched in rat tissues.A total of 50 compounds were identified in the liver (Table 1).Beyond phenolic acids that represented 65% of the compounds detected, one lignan and three metabolites, two flavonoid metabolites, and one hydrolysable tannin, along with nine metabolites, were also present.The phenolic acids identified are generally weak basic molecules with low molecular weight (<250 g/mol) and mild lipophilicity (log P ~ 0.9), facilitating their absorption through passive transport (López-Yerena, Domínguez-López, et al., 2021); Pinto et al., 2023).The detection of unmetabolized compounds in liver and kidney tissues confirmed their absorption by passive diffusion or through carriers located in the intestine, while phase II metabolites were secreted owing to their high polarity and molecular weight (Marhuenda-Muñoz et al., 2019;Pinto et al., 2023).Almost 81% of the metabolites derived from phase II reactions, whereas 11% represent microbiota metabolites and the remaining 8% resulted from other phase I reactions.Furthermore, twelve compounds identified are parent compounds, while the remaining represent their metabolites.Considering HBAs, eight compounds were detected in liver.Although GA was not identified, two metabolites from phase I (hydrogenation) and II reactions (methylation and sulfation) were found.Likewise, Bhat et al. (Bhat et al., 2020) also reported dihydro-GA in rice bran.Unmetabolized SyrAc and two phase II metabolites from methylation and sulfation, as well as the sulfated conjugate of HBA, were detected.A fraction of protocatechuic acid reached the liver unchanged, along with its methylated and sulfated conjugates.HippurAc was also detected in liver.
Regarding HCAs, twelve compounds were identified.Small fractions of unmetabolized CA and ρ-CoumAc reached the liver, while FA was not detected.The phase I metabolites of CA and FA obtained by hydrogenation, DHCA and DHFA, were also identified, along with glucuronidated conjugates.Unmetabolized SinapAc and ChlorogenAc and two phase II metabolites of ρ-CoumAc were detected.A major fraction of unmetabolized CinnamAc accumulated in the liver, along with its glucuronidated conjugate.
Unmetabolized HPPA also reached the liver, along with sulfated and glucuronidated-sulfated conjugates, while HPAA-O-glucuronide was the only HPAA metabolite identified.
A small fraction of unmetabolized ellagic acid also reached the liver, along with its dimethylated conjugate and urolithins which are gut microbiota metabolites of ellagic acid (Tomás- Barberán et al., 2014).UroC and UroD were identified accompanied by methylated-sulfated and dimethylated-sulfated conjugates of UroC.Sulfated and methylated metabolites of UroA and UroB were also found.
Unlike the CS extract, epi(gallo)catechins and derivatives were not present in the liver, probably due to their high excretion and poor absorption rates (López-Yerena, Domínguez-López, et al., 2021).Only twophase II metabolites of catechin were detected in liver.
Catechol was also found in liver, along with its methylated, sulfated and glucuronidated conjugates.A trace fraction of unmetabolized pyrogallol reached the liver, together with its glucuronidated conjugate.
Nonetheless, it was only possible to quantify some phenolic metabolites owing to their low concentrations, which hinders their quantification.Table 2 summarizes the polyphenols and metabolites quantified in rat tissues.
HCAs and HPPAs are the main classes of polyphenols and their circulating metabolites in the liver of CS extract-treated rats.CinnamAc was the major polyphenol identified (471.09 and 508.45 nmol/g tissue for 50 and 100 mg/kg b.w., respectively), followed by HPPA (114.53 and 94.82 nmol/g tissue).Furthermore, HPPA-O-sulfate was the main metabolite (242.83 and 226.79 nmol/g tissue), followed by HPPA-Oglucuronide-sulfate (19.18 and 19.59 nmol/g tissue) and CinnamAc-Oglucuronide (8.76 and 7.45 nmol/g tissue).Among HBAs, only two GA and one SyrAc metabolites were quantified.Even though HippurAc was the only metabolite quantified in control (29.96 nmol/g tissue), only trace levels (p < 0.05) were determined compared to treatment groups (77.88 and 68.94 nmol/g tissue, respectively, for 50 and 100 mg/kg b. w.).HippurAc is an endogenous metabolite found after the consumption of whole grains, cereals, and vegetable oils (Luzardo-Ocampo et al., 2017).This may explain its presence in the control group since rats fed a standard pellet diet composed by corn starch and soybean oil that may D. Pinto et al. be rich in polyphenols, mainly phenolic acids.Thus, HippurAc may also result, in a lower extent, from the metabolization of these phenolic acids into benzoic acid by phase I enzymes, with further conjugation with glycine (endogenously produced from amino acids delivered by diet) (Luzardo-Ocampo et al., 2017).

Identification and quantification of polyphenols and metabolites in the kidney
A total of 50 compounds were identified in kidney, with phenolic acids and their metabolites representing the major polyphenolic class (Table 1).Five lignans and six ellagic acid metabolites were also detected.Almost 83% of the metabolites are phase II metabolites, while 10% correspond to microbiota metabolites and the remaining 7% to other phase I metabolites.Only nine compounds identified are parent compounds, while the remaining correspond to their metabolites.Concerning HBAs, ten compounds were identified.As proven for the liver, GA was also not noticed in the kidney.Only GA methylated and sulfated conjugates were found.Dihydro-GA was not detected in kidney, oppositely to liver.Unmetabolized 4-HBA, protocatechuic acid, and SyrAc reached the kidney, corroborating that a considerable fraction of these phenolic acids from the CS extract remained unchanged, being capable of exerting their pro-healthy effects in-vivo without structural modifications on their molecules.Methylated and sulfated conjugates of HBA, protocatechuic acid, and SyrAc also accumulated in the kidney.Similar to liver, HippurAc was also detected in the kidney.
Among HCAs, fourteen compounds were identified, with 71% representing CA and FA metabolites, including phase I (DHCA and DHFA) and phase II conjugates (sulfated, methylated and, in a lower extent, glucuronidated).Unmetabolized CA and FA were not detected, in contrast to unmetabolized ρ-CoumAc and SinapAc that accumulated in the kidney along with their sulfated conjugates.Unlike liver, Cinna-mAc and ChlorogenAc were not detected in kidney, having probably been metabolized into simpler phenolic acids or formed conjugates.
As stated for liver, unmetabolized HPPA and its sulfated and glucuronidated-sulfated conjugates also accumulated in kidney.HPAA-O-sulfate was also found, accompanied by a higher fraction of its unmetabolized form.Unlike liver, no catechin metabolites were identified in kidney.
In contrast to liver, ellagic acid did not reach kidney.However, UroA, UroC, UroD, and methyl-UroB, and sulfated conjugates of UroA and UroB accumulated in kidney.
EntL and its phase II conjugates were the main lignan metabolites identified, together with EntD-O-sulfate.
Unmetabolized catechol accumulated in kidney, along with its methylated, sulfated and glucuronidated conjugates.Pyrogallol was not detected in its unmetabolized form, in contrast to liver.Nevertheless, dihydropyrogallol (not present in liver) was noticed in kidney, suggesting that a fraction of pyrogallol was metabolized by phase I enzymes.Pyrogallol conjugates arising from methylation and sulfation were also detected.
As shown in Table 2, HPPAs was the most representative polyphenolic class in kidney, representing 75 and 78% of total content, respectively, in 50 and 100 mg/kg b.w.groups, followed by HBAs (13 and 16%) and HPAAs (3.5 and 4%).HPPA was the main polyphenol quantified (739.48 and 870.47 nmol/g tissue, respectively, for 50 and 100 mg/kg b.w.groups), followed by HippurAc (205.18 and 171.54 nmol/g tissue).Regarding metabolites, HPPA-O-sulfate (1319.31 and 1572.23 nmol/g tissue) and HBA-O-sulfate (211.53 and 228.41 nmol/g tissue) were the main phase II conjugates, while dihydropyrogallol was the principal phase I metabolite (62.48 and 57.61 nmol/g tissue).Among HBAs, only sulfated and methylated conjugates from GA and SyrAc were quantified, with concentrations varying between 0.32 and 6.59 nmol/g tissue.Similar contents (p > 0.05) of GA metabolites, 4-HBA, HBA-O-sulfate, and HippurAc were determined in both groups.As ascertained in liver, HippurAc was the only metabolite in the control group (59.77nmol/g tissue), although only trace levels (p < 0.05) were quantified when compared to CS extract groups (205.18 and 171.54 nmol/g tissue for 50 and 100 mg/kg b.w.groups).Ten HCAs metabolites were quantified in kidney, with concentrations increasing in the following order: DHCA in a total of 80 nmol/g tissue.High concentrations of unmetabolized HPPA and HPAA reached kidney, along with their sulfated metabolites, representing 2250 and 110 nmol/ g tissue, respectively.Urolithins arouse as ellagic acid metabolites, with concentrations increasing as follows: methyl-UroB < UroD < UroA-Osulfate < UroC, in a total amount of 4.73 and 3.81 nmol/g tissue for 50 and 100 mg/kg b.w.CS extract groups, respectively.Sulfated conjugates of EntL were the only lignan metabolites quantified (≈0.90 nmol/g tissue), while catechol and its metabolites corresponded to 14.65 and 13.03 nmol/g tissue, respectively, for 50 and 100 mg/kg b.w.groups.Catechol-O-sulfate was the main metabolite, while a substantial fraction (9-15%) of unmetabolized catechol also reached kidney.In summary, sulfated conjugates correspond to 60% of the total metabolites content.Higher concentrations of HPPAs (2442.70 nmol/g tissue), HCAs (80.51 nmol/g tissue), and lignans metabolites (0.94 nmol/g tissue) were achieved for 100 mg/kg b.w.CS extract-treated rats.Considering the 50 mg/kg b.w.CS extract group, higher levels of HBAs (426.80 nmol/g tissue), HPAAs (113.60 nmol/g tissue), pyrogallol (62.48 nmol/g tissue), catechol (14.65 nmol/g tissue) and ellagic acid metabolites (4.73 nmol/g tissue) were determined.Overall, higher concentrations of polyphenols and their metabolites were accumulated in the kidney of   100 mg/kg b.w.group, evidencing an interesting metabolomic profiling that endorses the strong in-vivo antioxidant effects proved in our previous study.
There was no direct proportionality between the concentration of phenolic metabolites and the increase of CS extract dose from 50 to 100 mg/kg b.w.The discrepancies on the results between the two extract doses tested may be explained by: i) the saturation of carriers with other molecules, hindering their transport and absorption through intestinal barrier; and ii) the poor reproducibility in the animal research related to high heterogenization of responses and biological variance (López-Yerena, Domínguez-López, et al., 2021;Voelkl et al., 2020).

Hydroxycinnamic acids
Three possible pathways were proposed for HCAs in this study: i) a minor fraction was absorbed intact in the upper gastrointestinal tract; ii) a small ratio was directly metabolized into glucuronidated, methylated, and sulfated conjugates by phase II enzymes, i.e., uridine diphosphate (UDP)-glucuronosyltransferases, catechol-O-methyltransferases and sulfotransferases in liver, kidney and intestine; and iii) a major portion reached the colon, where were metabolized by microbial esterases and reductases or biotransformed into smaller phenolic acids before absorption and conjugation into phase II metabolites (Piazzon et al., 2012;Sova & Saso, 2020).Fig. 2 presents a metabolic scheme of the CS extract.Fig. 3 depicts the metabolic pathways of the polyphenols identified in rat tissues.
Beyond a small fraction that may be absorbed in the upper gastrointestinal tract, caffeoylquinic, feruloylquinic, and coumaroylquinic acids (identified in CS extract) are hydrolyzed in colon via microbial esterases, releasing CA, FA and CoumAc, respectively, which are further conjugated with methyl, sulfate, and glucuronic acid in intestine or liver (Sova & Saso, 2020).This explains the absence of quinic acid derivatives in rat tissues.Additionally, the quinic acid originates benzoic acid (dehydroxylation) that conjugates with acyl glycine producing Hip-purAc, which was detected in both tissues (Sova & Saso, 2020).
Likewise, phenolic glycosides detected in CS extract, i.e., CA-Oglucoside, FA-O-glucoside, CoumAc-O-glucoside and SinapAc-O-glucoside, are partially hydrolyzed in the stomach by glycosidases (regardless the low pH), biotransformed into aglycones in the small intestine (by lactase-phlorizin hydrolase or cytosolic β-glucosidase), converted into phase II metabolites that are transported into liver (where undergo further metabolization) and, finally, enter the bloodstream or return to the digestive tract by enterohepatic circulation (López-Yerena, Domínguez-López, et al., 2021;Sova & Saso, 2020).These mechanisms may also explain the presence of sulfated, methylated, and glucuronidated conjugates of CA, FA, CoumAc, CinnamAc and SinapAc in rat tissues, along with traceable concentrations of their free forms.
Another fraction of HCAs may reach the colon and be metabolized into reduced forms, DHCA and DHFA, by the intestinal microbiota (Alam et al., 2016;Sova & Saso, 2020).These colonic metabolites may be further absorbed through colonic epithelium and then conjugated by phase II enzymes in intestine, liver, and kidney (Sova & Saso, 2020).This metabolic pathway corroborates the high concentrations of DHCA and DHFA and their metabolites detected in rat tissues, along with p-CoumAc, CinnamAc and their conjugates.Otherwise, colonic metabolites may undergo enzymatic bioconversion into smaller phenolic acids (Alam et al., 2016;Sova & Saso, 2020).For instance, the dehydroxylation of DHCA generates HPPA, while the loss of carbons originates 3,4-dihydroxyphenylacetic acid.
In summary, HPPAs, HPAAs, and HBAs are among the colonic metabolites of HCAs, followed by their reduced forms (DHCA and DHFA), along with phase II conjugates, which explain their high concentrations in rat tissues, underlying the importance of the gut microbiota in HCAs metabolism.3-HPPA was indicated as a major metabolite of caffeoylquinic acids in human fecal samples, while DHCA-O-sulfate is a potential biomarker of coffee intake, representing 24% of the metabolites excreted (Stalmach et al., 2010).Sulfated conjugates were the major phase II metabolites, followed by methylated and, to a lower extent, glucuronidated conjugates, which agrees with this study findings.These different pathways may extend the permanence of HCAs in the human body contributing to a prolonged in-vivo bioactivity, owing to enterohepatic recirculation or colonic absorption.Concerning the bioactivity, previous studies revealed effective scavenging potential against nitric oxide and DPPH radicals for HCAs metabolites (Alam et al., 2016;Piazzon et al., 2012).

Flavanols
Besides acting as radical scavengers, catechins modulate essential signaling pathways, influencing inflammatory, oxidative or cell proliferation processes, and revert metabolic changes induced by high-fat diets (Márquez Campos et al., 2019;Shang et al., 2017).Nonetheless, it is important to comprehend the metabolic pathways of catechins to ascertain their bioactivity (Fig. 3B).Catechin gallates generally maintain stable during gastric passage and are biotransformed by gut microbiota, prior to their detoxification in the liver (Márquez Campos et al., 2019).At intestine, (epi)gallocatechin-O-gallate and (epi)catechin-O-gallate (present in CS extract) are hydrolyzed by microbial esterases, releasing GA, (epi)gallocatechin and (epi)catechin.Free epi (gallo)catechins are metabolized in the colon through C-ring opening, producing diphenylpropan-2-ol intermediates, further converted into PVL and PVA via A-ring fission.Smaller phenolic acids, including HPPAs and HBAs, are formed by the gut microbiota through successive loss of carbons via β-oxidation (Márquez Campos et al., 2019;Shang et al., 2017).These phenolic acids are easily absorbed and may exert even higher physiological effects than the parent flavonoids (Shang et al., 2017).PVL, PVA and phenolic acids may be bioconverted into phase II metabolites, which are excreted in urine (Márquez Campos et al., 2019;Shang et al., 2017).These metabolic pathways may explain the presence of methylated and sulfated GA, free HPPA and its glucuronidated and sulfated metabolites in liver and kidney.Free catechol and pyrogallol may result from decarboxylation and dehydroxylation of GA, respectively (Shang et al., 2017).HPPA may be metabolized into benzoic acid (dehydroxylation and demethylation) and further conjugated with acyl glycine, originating HippurAc (Shang et al., 2017).This may be also the metabolic source of 4-HBA and its sulfated conjugate.
A small ratio of (epi)catechin was absorbed and directly metabolized by phase II enzymes, explaining the presence of (epi)catechin-O-glucuronide and methyl-(epi)catechin in liver (Márquez Campos et al., 2019;Shang et al., 2017).PVL and PVA were not identified in rat tissues, having probably been converted into phenolic acids and their metabolites that were detected in liver and kidney.Beyond the health benefits of HBAs, HippurAc provides antibacterial, anticancer, antifungal, and antiviral properties (Shang et al., 2017).

Hydroxybenzoic acids
HBAs are major metabolites of more complex polyphenols, including HCAs and flavanols (Bento-Silva et al., 2020).Beyond its abundance as free compound, GA may be released from gallotannins and further absorbed, or undergo microbial metabolism via methylation, dehydroxylation and decarboxylation, producing methyl-GA, DHBA and pyrogallol, respectively (Fig. 3C) (Bento-Silva et al., 2020).This metabolic pathway corroborates the presence of GA, methyl gallate and pyrogallol in CS extract, while only two phase I and three phase II metabolites were detected in liver and kidney.

Hydrolysable tannins
Besides its release by hydrolysis of ellagitannins (previously D. Pinto et al. identified in CS), ellagic acid may be also present as free compound which could be absorbed at the upper digestive tract and biotransformed by phase II enzymes, explaining the presence of ellagic acid and its dimethylated conjugate in liver (Tomás-Barberán et al., 2014).Otherwise, ellagic acid is converted into urolithins by intestinal microbiota (Tomás-Barberán et al., 2014).UroD is produced through hydrolysis of one lactone and reduction of carboxylic acid into a semi-hydroquinone, followed by dehydroxylation and decarboxylation.Subsequent dehydroxylation produces UroC, UroA and UroB (Fig. 3D) (Tomás-Barberán et al., 2014).Urolithins may be further conjugated with methyl, sulfate, and glucuronic acid, explaining their presence in liver and kidney (Tomás-Barberán et al., 2014).Beyond the pro-healthy properties of ellagic acid, urolithins have attested anti-inflammatory, anticancer, immunomodulatory, neuroprotective, cardioprotective, osteoprotective, and microbiota modulatory effects in animals and humans, highlighting its potential use as effective molecules in the prevention and co-treatment of certain pathological disorders (Laveriano-Santos, Marhuenda-Muñoz, et al., 2022;Laveriano-Santos, Quifer-Rada, et al., 2022).

Lignans
Secoisolariciresinol was the only lignan identified in CS extract.Even though trace levels were detected in liver, the highest fraction was metabolized into enterolignans via intestinal microbiota (Fig. 3E).
Secoisolariciresinol was metabolized into EntD via O-demethylation and dehydroxylation and, then, into EntL by dehydrogenation which were further conjugated with sulfate and glucuronic acid (Li et al., 2022).This metabolic pathway justifies the presence of EntD, EntL and sulfated and glucuronidated metabolites in liver and kidney (Li et al., 2022).An invitro study using intestinal cells detected EntL-O-sulfate, EntL-O-glucuronide and EntD-O-glucuronide after exposure to lignans (Jansen et al., 2005).EntD and EntL are indicated as the main metabolites responsible for the pro-healthy effects ascribed to lignans, including antioxidant properties and modulation of hormone metabolism, protecting against certain cancers and hair loss (Laveriano-Santos, Marhuenda-Muñoz, et al., 2022;Li et al., 2022).

Influence of metabolism in the bioavailability and bioactivity of polyphenols
The relationship between polyphenols and gut microbiota is bidirectional, with microbiota metabolizing polyphenols and, in turn, polyphenols modifying the microbiome by favoring the growth of beneficial bacteria and inhibiting pathogenic species (Corrêa et al., 2019).The microbiota-modulating effect of polyphenols may prevent or accelerate the recovery of metabolic diseases by strengthening the immune system (Bento-Silva et al., 2020;Corrêa et al., 2019).This is particularly important for HCAs, re-establishing microbiota, reducing pro-inflammatory cytokines secretion and pro-oxidant species, and relieving oxidative stress (Bento-Silva et al., 2020).Additionally, HPAA and HPPA were correlated to Bifidobacterium growth with beneficial gut effects (Bento-Silva et al., 2020).Noteworthy, it has been suggested that circulating phenolic metabolites may undergo enzymatic deconjugation at the site of action, despite this mechanism remains still unexplained (Bento-Silva et al., 2020).
Latest advances have attested in-vivo bioactivity of plasma and tissues from animals and humans after intake of polyphenols-rich foods, suggesting that phenolic metabolites still retain strong antioxidant properties (Martins et al., 2016;Piazzon et al., 2012).Glucuronidation and sulfation provide more hydrophilic compounds that may affect the bioavailability and site of action (López-Yerena, Domínguez-López, et al., 2021).Piazzon et al. (2012) described identical antioxidant activity for glucuronidated and sulfated metabolites from FA and CA compared to parent compounds.Conjugation reactions may also enhance other biological activities, namely hypolipidemic properties for catechin-methylated conjugates, and antimicrobial properties for glucuronidated and methylated conjugates from 4-HBA, p-CoumAc and CinnamAc.Nevertheless, there are few studies about the bioactivity of phenolic metabolites owing to the lack of accurate identification and available of commercial standards (Piazzon et al., 2012).
The production of phenolic acids by colonic metabolism contribute to the bioavailability of polyphenols in a greater extent than phase I and II metabolites.Microbial metabolites represent 45% and 70% of the total content in liver and kidney, respectively, of which 80% correspond to colonic metabolites conjugated with methyl, sulfate, and glucuronic acid.

Screening of potential oxidative stress biomarkers
PCA has three key assumptions: i) sphericity or existence of the identity matrix; ii) sampling adequacy or an appropriate number of observations relative to the number of variables under analysis; and iii) positive determinant of the correlation or variance-covariance matrices (Bailey, 2012;Spiegelberg & Rusz, 2017).For the first assumption, a Bartlett's test for sphericity was performed to ascertain the veracity of the null hypothesis before proceeding with PCA.The null hypothesis is that the variables are not correlated to each other; if it is true, the PCA is not appropriate since it relies on the construction of a linear combination of the variables (Spiegelberg & Rusz, 2017).The results of Bartlett's test revealed p < 0.05, corroborating that at least two of the variables are correlated to each other and, thus, rejecting the null hypothesis (Supplementary Fig. S3).Regarding the second assumption, the sampling adequacy was higher than 0.9 for both liver and kidney tissues which is much higher than the minimum acceptable for PCA application (>0.5) (Spiegelberg & Rusz, 2017).Considering the third assumption, the strong correlations between most variables (demonstrated by the heatmap correlation diagram) underline the rejection of null hypothesis owing to the lack of identity matrix and presence of collinearity among variables creating a correlation matrix (Spiegelberg & Rusz, 2017).Based on the results, the three assumptions were ensured, concluding that it is reasonable to apply a dimension-reduction method to these data capable of reducing data dimensionality and noise.As a valuable statistical tool, PCA (Fig. 4) allows the screening of the phenolic metabolites that contribute to the in-vivo antioxidant responses, comprehending the differences among groups regarding metabolomic fingerprinting (evaluated in this study), antioxidant enzymes' activities (SOD, CAT and GSH-Px) and LPO prevention (data previously published (Pinto et al., 2023)).

Liver
The scores plot (Fig. 4A) reveals an evident separation of the treatment groups with 46.91% of explained variance, emphasizing different in-vivo antioxidant responses of liver from 50 and 100 mg/kg b. w.CS extract groups.The PC1 explains 31.96% of the results with an eigenvalue of 10.55 (Supplementary Fig. S1).The biplot (Fig. 4B) indicates a strong role of the metabolomic profiling in the in-vivo antioxidant properties of liver

Kidney
The scores plot (Fig. 4C) evidences a clear separation between the two treatment groups.A cumulative variance of 60.24% is pointed out with PC1 explaining 37.04% variance, underlining markedly different in-vivo antioxidant responses for 50 and 100 mg/kg b.w.CS extract groups.High eigenvalues were determined for PC1 and PC2 (12.22 and 7.66, respectively), corroborating the explained variance results.The biplot (Fig. 4D) suggests a strong effect of the metabolomic profile on the in-vivo antioxidant activity of kidney.The in-vivo antioxidant response of kidney from 50 mg/kg b.w. group (marked green) is closely correlated to CAT activity, LPO prevention and contents of UroC, UroD, methyl-UroB, HPPA, HPPA-O-sulfate, DHCA-O-sulfate, EntL-O-disulfate, methyl-DHFA-O-sulfate, HPAA-O-sulfate, methyl-SyrAc, 4-HBA, CA-O-sulfate, FA-O-sulfate, dimethyl-GA-O-sulfate, p-CoumAc, CoumAc-O-sulfate, DHCA, DHFA, DHFA-O-sulfate, and EntL-O-sulfate.Otherwise, the invivo antioxidant efficacy of 100 mg/kg b.w. group was confirmed by GSH-Px and SOD activities and contents of HPAA and dimethyl-SyrAc-Osulfate.Overall, LPO prevention and CAT activity are the major variables contributing to the in-vivo bioactivity of kidneys from 50 mg/kg b. w. group possibly associated to phenolic acids and metabolites (mainly HBAs, HCAs, HPPAs and HPAAs), along with microbial metabolites from lignans and hydrolysable tannins.The in-vivo bioactivity of kidneys from 100 mg/kg b.w. group is related to HPAAs and one SyrAc metabolite.
In general, the PCA model pointed out a higher heterogeneity among kidney results when compared to liver.Recently, Pinto et al. (2023) proved marked differences in metabolic profiles and in-vivo antioxidant effects of blood serum from rats treated with the same two doses of the CS extract (50 and 100 mg/kg b.w.) analyzed in this study.Phenolic acids, lignans and flavanols, along with their metabolites, were proposed as the major compounds endorsed in the in-vivo antioxidant efficacy.Hence, phenolic metabolites and antioxidant enzymes activities may be potential oxidative stress biomarkers whose concentration and activity may vary in response to an increased redox stress.This study proposes that the bioactive molecules from CS extract and their metabolites exert protective effects against oxidative damages, particularly in liver and kidney, induced by pro-oxidant reactive species into biomolecules (i.e., DNA, lipids, and proteins).

Screening of correlations between metabolic profiling and in-vivo antioxidant response
Several studies have ascribed the in-vitro antioxidant activity of plant derivatives to their phenolic composition (Lameirão et al., 2020;Pinto & Cádiz-Gurrea et al., 2020;Pinto, Silva, et al., 2021;Pinto, Vieira, et al., 2021;Rodrigues et al., 2015).However, there is still a lack of studies regarding this correlation on animal experiments.This study provides for the first time a comprehensive assessment of the correlations between in and vivo antioxidant activity (evaluated in our previous paper (Pinto et al., 2023)) and metabolomic fingerprinting of tissues from rats orally treated with CS extract through correlation heatmaps (Fig. 5).
Considering the in-vivo antioxidant response of liver, a weak positive correlation (r 2 = 0.447) was highlighted between SOD and GSH-Px activities.A negative correlation was disclosed between SOD and CAT activities (r 2 = − 0.609).Likewise, GSH-Px activity was negatively correlated to LPO (r 2 = − 0.812), denoting that a rise on GSH-Px activity is closely related to a decrease on LPO rate.The SOD and GSH-Px activities of liver were mainly ascribed to the free polyphenols and microbial metabolites identified (r 2 = 0.402 and r 2 = 0.847, respectively, for SOD and GSH-Px), especially to EntD, UroD and CA (r 2 > 0.467).The protective effects against LPO were also attributed to the free polyphenols and microbial metabolites detected in liver owing to a remarkable inverse relation (r 2 = − 0.971), particularly to EntD, dimethyl-UroB, UroD and CA (r 2 > − 0.510).Additionally, CAT activity was not correlated to the metabolomic profile.
These outcomes emphasize a good correlation level within some variables, outlining an exceptional contribution of the polyphenols and their microbial and phase II metabolites to the in-vivo antioxidant responses of liver and kidney through upmodulating antioxidant enzymes' activities and downregulating LPO, which corroborates the PCA results.
Altogether, these correlations seem to be efficient indicators in predicting that phenolic acids and microbial metabolites from lignans and ellagic acid are the major compounds contributing to the in-vivo antioxidant response of rats orally treated with CS extract, motivating its use as a nutraceutical ingredient useful in the prevention and co-treatment of lifestyle-related pathologies (i.e., Alzheimer's and Parkinson's diseases, cancer, diabetes, neurological, cardiovascular, and metabolic pathologies) associated to oxidative damages.

Conclusion
The current study attempted to pursue the validation of a nutraceutical extract from CS as an appealing source of antioxidants embracing in-vivo pro-healthy effects.A metabolomic approach provides novel insights into the bioavailability of polyphenols from CS and the identification of circulating metabolites.Following an in-depth assessment of in-vitro and in-vivo biological effects accomplished in recent studies, the metabolomic profile of different tissues from rats orally treated with phenolics-rich CS extract corroborated the promising outcomes regarding its in-vivo bioactivity.Unmetabolized polyphenols, along with their phase I and II metabolites, were noticed in both tissues with higher accumulation in kidney.Phenolic acids were the major polyphenolic class in rat tissues, followed by hydrolysable tannins, flavanols and lignans.Sulfated conjugates were the main metabolites reaching kidney, while liver contained identical concentrations of glucuronidated, methylated, and sulfated metabolites.The multivariate data analysis predicted an outstanding contribution of polyphenols and their microbial and phase II metabolites to the in-vivo antioxidant efficacy of CS extract in rats, proposing its use as a prominent source of antiaging molecules for nutraceuticals with health benefits in the prevention and co-therapy of lifestyle-related diseases triggered by oxidative stress.Further studies will be focused on designing a nutraceutical product incorporating CS extract and evaluate its efficacy and safety.

D
.Pinto et al.

Fig. 2 .
Fig. 2. Schematic representation of the metabolism of chestnut shells extract orally administered to rats.

Fig. 5 .
Fig. 5. Correlation heatmap diagram of in vivo antioxidant activity and metabolomic profile of (A) liver and (B) kidney from rats treated with chestnut shells extract targeted on phenolic compounds.

Table 1
Identification of phenolic compounds and their metabolites detected in liver and kidney tissues from rats orally treated with CS extract explored by LC-ESI-LTQ-Orbitrap-MS.

Table 1
(continued ) +, compound identified in samples; − , compound not identified in samples.D.Pinto et al.

Table 2
Quantification of phenolic compounds and their metabolites in liver and kidney tissues from rats orally treated with water (control group), 50 mg/kg b.w. and 100 mg/ kg b.w. of CS extract explored by LC-ESI-LTQ-Orbitrap-MS.
. A positive correlation is denoted for variables closest and farthest from the diagram origin, while variables situated oppositely are negatively correlated.The plots suggest that the in-vivo CinnamAc, methyl-UroC-O-sulfate, and HPAA-Oglucuronide are strongly correlated with its in-vivo antioxidant response.Overall, the LPO prevention is the main variable contributing to the invivo antioxidant activity of liver from 50 mg/kg b.w.group, which is related to phenolic acids and metabolites (mainly HBAs, HCAs and HPPAs), two lignan metabolites and one hydrolysable tannin metabolite.The in-vivo antioxidant capacity of liver from 100 mg/kg b.w. group is mainly attributed to GSH-Px activity, corroborated by the microbial metabolites from lignans and hydrolysable tannins, and phenolic acids metabolites from HBAs, HCAs, HPPAs and HPAAs.Additionally, CAT and SOD activities had a small influence on the in-vivo bioactivity of liver.