Simultaneously Quantitative Analysis of Naringin and Its Major Human Gut Microbial Metabolites Naringenin and 3-(4′-Hydroxyphenyl) Propanoic Acid via Stable Isotope Deuterium-Labeling Coupled with RRLC-MS/MS Method

Widespread in citrus fruits, naringin, a natural 2,3-dihydroflavonoid, is of particular interest to scientists and has a broad range of beneficial bioactivities to health. Orally administered naringin remains in the gut tract for a relatively long time because of its low bioavailability. Under the metabolism mediated by human gut microbiota, naringin could be an active precursor for derived metabolites to play important physiological roles. However, naringin and its metabolites are hard to accurately quantify due to severe endogenic interference. In this study, an analytical rapid resolution liquid chromatography tandem mass spectrometry (RRLC-MS/MS) method coupled with stable isotope deuterium-labeling is developed and validated to simultaneously quantify naringin as well as its major human gut microbial metabolites naringenin and 3-(4′-hydroxyphenyl) propanoic acid. By eliminating the matrix interferences, this strategy not only confirms naringenin and 3-(4′-hydroxyphenyl) propanoic acid as the predominant metabolites which contribute to the pharmacological effects of naringin but also provides a suitable choice for other flavonoid pharmacokinetics study.


Introduction
Fruits not only enrich our life with good taste but also contain abundant bioactive constituents to protect human health, since consumption of fruits can help reduce risks of various diseases [1]. Citrus fruits are easily available and popular in daily diets. Among the bioactive contents in citrus fruits, flavonoids are of particular interest to researchers due to their protective effects against inflammation, cardiovascular diseases, and cancers [2,3]. Naringin, a natural 2,3-dihydroflavonoid, presents dominantly in citrus fruits and has been reported to possess a broad range of bioactive effects on health such as protection against oxidation [4,5], hyperlipidemia [6], and neurological disorders [7,8]. In addition, our previous investigations have proven that naringin has extensive beneficial effects against respiratory diseases including inflammation [9,10], coughs [11], phlegm [12], and pulmonary fibrosis [13].
With multiple varieties and potent functions, there are about 10 14 microorganisms, tenfold more than human cells, residing in the gut tract and playing important roles in metabolic, nutritional, physiological, and immunological activities of both health and disease conditions [14]. For most bioactive molecules in food or oral medication, human gut microbiota secret numerous enzymes to get involved in the chemical transformation processes and have certain effects on not only their pharmacokinetic behaviors but also pharmacological characteristics [15]. The results of the metabolism mediated by human intestinal microbes could be positive or negative, which is correlated to bioactivities of the subsequent metabolites [16].
With its low bioavailability, naringin is poorly absorbed into the blood circulation system [17], suggesting that orally administered naringin remains in the gastrointestinal tract for a correspondingly extended time and is inevitably influenced by intestinal microbiota. In our preliminary research, we had systematically profiled the metabolites of naringin mediated by human intestinal microbiota with an ingenious strategy that combined stable isotope-labeling and ultra-fast liquid chromatography-quadruple-time-of-flight tandem mass spectrometry (UFLC-Q-TOF-MS/MS) method [18]. Among the human gut microbial metabolites, naringenin and 3-(4 -hydroxyphenyl) propanoic acid (HPPA) have been considered the major microbial biotransformation products of naringin. Of note is that both naringenin and HPPA have been reported to possess beneficial bioactivities, among which naringenin has been seen to be effective with regards to anticancer [19], anti-inflammation [20], and neuroprotection [21], while HPPA can efficiently suppress influenza infection [22]. Therefore, it seems as though naringin could be an effective precursor in vivo to play bioactive roles via its derived metabolites from the biotransformation mediated by human gut microbiota. Thus, it is of great value to quantitatively investigate the dynamic biotransformation of naringin and its human gut microbial metabolites including naringenin and HPPA. However, it is difficult to achieve an accurate quantification of naringin and its metabolites due to strong endogenic interference.

Method Development and Optimization
For analyzing flavonoid and phenolic compounds from human biological samples, especially in fecal microbiota solution, there is a common challenge that those specimens have shown severe matrix effects from co-eluted peaks [23,24]. Thus, it is urgent to seek for a suitable strategy with  In summary, the above results identified the synthetic products as [2 ,3 ,5 ,6 -D 4 ]naringenin and 3-(4 -hydroxyphenyl)- [2 ,3 ,5 ,6 -D 4 ]propanoic acid, indicating that our synthetic strategy was appropriate with expected future application in other flavonoids and phenolic acids.

Method Development and Optimization
For analyzing flavonoid and phenolic compounds from human biological samples, especially in fecal microbiota solution, there is a common challenge that those specimens have shown severe matrix effects from co-eluted peaks [23,24]. Thus, it is urgent to seek for a suitable strategy with excellent specificity and sensitivity to distinguish target molecules from matrix disturbances. Coupled with LC-MS, there is a stable isotope-labeling technique which has been proven to be a valid alternative for qualitative and quantitative analysis of diverse compounds with its remarkable advantages including safety, sensitivity, and versatility [25][26][27][28][29]. Referring to our previous study [18], we synthesized with the same stable isotope-labeling pattern to replace hydrogen with deuterium in the 2 , 3 , 5 , 6 positions on the B ring of naringenin and HPPA.
As is known to us, multiple reaction monitoring (MRM) detection mode with characteristic precursors and product ions is generally considered as the gold standard for the quantification of molecules from complex biological samples [30]. Combined with the integrated advantages of high resolution, rapid speed, and excellent sensitivity, RRLC coupled with triple quadrupole mass spectrometry (TQ-MS/MS) is a better choice for quantitative analysis of compounds [31]. Hence, we developed an analytical strategy coupled with RRLC and a stable isotope deuterium-labeling method to simultaneously quantify naringin and its major metabolites naringenin and HPPA mediated by human gut microbes.
Sample preparation and RRLC conditions including anaerobic incubation, extraction method, mobile phase system, and so on, were employed according to our previous report with feasible adjustments [18].  Table 1) and MS/MS conditions such as gas temperature, gas flow, nebulizer, and capillary through injecting individual standard solutions of target compounds at a concentration of 10 µg mL −1 . The product ion spectra and MS/MS fragmentation patterns of analytes are shown in Figure 3

Specificity
As seen in Figure 4, typical MRM chromatograms from blank gut microbiota solution spiked with analytes at a lower limit of quantification (LLOQ) concentration and IS show sharp and satisfactory peaks with retention times (RT) at 1.6 min, 2.3 min, 1.7 min, and 2.2 min for D4-NG, D4-NE, D4-HPPA, and IS D5-BA, respectively. At the same time, there are no obvious interferential MRM signals at the RT of all the analytical targets and IS compared with the blank gut microbiota solution. This result indicates that our developed method is specific for analyzing D4-NG, D4-NE, and D4-HPPA simultaneously.

Specificity
As seen in Figure 4, typical MRM chromatograms from blank gut microbiota solution spiked with analytes at a lower limit of quantification (LLOQ) concentration and IS show sharp and satisfactory peaks with retention times (RT) at 1.

Precision and Accuracy
The results of precision and accuracy validation are summarized in Table 2. The relative standard deviations (RSDs) of intra-day were 0.91-3.58%, 1.25-3.49%, and 1.45-9.90% for D 4 -NG, D 4 -NE, and D 4 -HPPA, respectively, while the relative error (RE) of intra-day varied from −9.42 to 8.73 for the three target compounds. For inter-day validation, the RSD were lower than 12% and the RE range was −8.90-9.81% for all the molecules. In summary, the above results suggest suitable precision and accuracy of our method for quantification of D 4 -NG, D 4 -NE, and D 4 -HPPA at the same time.

Extract Recovery
The mean extraction efficiencies of D 4 -NG, D 4 -NE, and D 4 -HPPA at various concentrations were found to be 56.22-58.02%, 53.97-58.50%, and 49.53-70.69%, respectively (see Table 2). As a result, ethyl acetate was considered to be acceptable as a suitable and effective reagent for simultaneously extracting D 4 -NG, D 4 -NE, and D 4 -HPPA in human gut microbiota solutions because of its moderate polarity and solubility.

Matrix Effect
As shown in Table 3, the matrix effects for D 4 -NG, D 4 -NE, and D 4 -HPPA were between 94.73-101.1% and 98.12-107.1% at quality control (QC) low and high concentrations, respectively. In addition, the RSDs of the three targets at the tested concentrations were no more than 3%, indicating that matrix effects were negligible for analysis with the application of stable isotope-labeling.

-NG with Human Gut Microbiota Solution
The established stable isotope deuterium-labeling coupled with RRLC-MS/MS strategy was well applied to quantitative analysis after incubation of naringin with human fecal microbe solutions. As we determined, with the help of deuterium-labeling, there was no baseline interference for analysis [32,33], demonstrating that stable isotope-labeling was effective to eliminate matrix interferences from biological samples and be capable of analyzing flavonoid and phenolic compounds precisely and sensitively.
In order to systematically profile the dynamic biotransformation courses of naringin and its microbial metabolites, we detected their concentrations four sequential times within 24 h. As shown in Figure 5a, D 4 -NG underwent ongoing degradation while D 4 -HPPA continuously increased over time after incubation. Of note is that D 4 -NG was consumed after 4 h, indicating a proposed metabolic pattern of naringin mediated by human gut microbiota ( Figure 6). In general, naringin was first hydrolyzed to its aglycone naringenin under the β-glucosidases secreted by microorganisms [34,35] and then underwent C ring fission to generate 3-(4 -hydroxyphenyl)propanoic acid [36,37]. However, the enzyme profiling which catalyzes C ring fission is still unclear and deserves in-depth research.
in Figure 5a, D4-NG underwent ongoing degradation while D4-HPPA continuously increased over time after incubation. Of note is that D4-NG was consumed after 4 h, indicating a proposed metabolic pattern of naringin mediated by human gut microbiota ( Figure 6). In general, naringin was first hydrolyzed to its aglycone naringenin under the β-glucosidases secreted by microorganisms [34,35] and then underwent C ring fission to generate 3-(4′-hydroxyphenyl)propanoic acid [36,37]. However, the enzyme profiling which catalyzes C ring fission is still unclear and deserves in-depth research.  After 24 h, the total concentration of D4-NG (0.3579 μmol L −1 ), D4-NE (0.4418 μmol L −1 ), and D4-HPPA (0.4605 μmol L −1 ) was 1.2602 μmol L −1 , which was about a 40% ratio compared to the administrated dose of D4-NG. Hence, naringenin and HPPA could be the major metabolites of naringin under the metabolism of human intestinal microbiota, which is in accordance with our best knowledge that gut microorganisms mainly proceed reduction and hydrolysis to produce low molecular weight and polarity compounds [15,38]. Considering the bioactivities of naringenin [19][20][21] and HPPA [22], gut microbiota could in part play an important role in regulating the pharmacological effects of naringin through biotransformation in vivo.
Of note is that the metabolic profile was significantly discrete among various individuals ( Figure  5B). Due to the complexity of intestinal microecology, different compositions of bacteria show diverse physiological activities [39,40]. For decades, although the gut bacterial metabolism of naringin has time after incubation. Of note is that D4-NG was consumed after 4 h, indicating a proposed metabolic pattern of naringin mediated by human gut microbiota ( Figure 6). In general, naringin was first hydrolyzed to its aglycone naringenin under the β-glucosidases secreted by microorganisms [34,35] and then underwent C ring fission to generate 3-(4′-hydroxyphenyl)propanoic acid [36,37]. However, the enzyme profiling which catalyzes C ring fission is still unclear and deserves in-depth research.  After 24 h, the total concentration of D4-NG (0.3579 μmol L −1 ), D4-NE (0.4418 μmol L −1 ), and D4-HPPA (0.4605 μmol L −1 ) was 1.2602 μmol L −1 , which was about a 40% ratio compared to the administrated dose of D4-NG. Hence, naringenin and HPPA could be the major metabolites of naringin under the metabolism of human intestinal microbiota, which is in accordance with our best knowledge that gut microorganisms mainly proceed reduction and hydrolysis to produce low molecular weight and polarity compounds [15,38]. Considering the bioactivities of naringenin [19][20][21] and HPPA [22], gut microbiota could in part play an important role in regulating the pharmacological effects of naringin through biotransformation in vivo.
Of note is that the metabolic profile was significantly discrete among various individuals ( Figure  5B). Due to the complexity of intestinal microecology, different compositions of bacteria show diverse physiological activities [39,40]. For decades, although the gut bacterial metabolism of naringin has , which was about a 40% ratio compared to the administrated dose of D 4 -NG. Hence, naringenin and HPPA could be the major metabolites of naringin under the metabolism of human intestinal microbiota, which is in accordance with our best knowledge that gut microorganisms mainly proceed reduction and hydrolysis to produce low molecular weight and polarity compounds [15,38]. Considering the bioactivities of naringenin [19][20][21] and HPPA [22], gut microbiota could in part play an important role in regulating the pharmacological effects of naringin through biotransformation in vivo.
Of note is that the metabolic profile was significantly discrete among various individuals ( Figure 5B). Due to the complexity of intestinal microecology, different compositions of bacteria show diverse physiological activities [39,40]. For decades, although the gut bacterial metabolism of naringin has been reported [41][42][43][44][45][46], the metabolic association between naringin and microbial groups is yet to be uncovered and is deserving of further research through multiple omics.

Chemicals and Reagents
Potassium carbonate, sodium sulfate, sodium hydroxide, methanol, and sulfuric acid were purchased from the Guangzhou Chemical Reagent Factory (Guangzhou, China). Acrylic acid, bis(triphenylphosphine)palladium(II) chloride, palladium on carbon (30 wt%), and LC-MS grade formic acid were obtained from Sigma-Aldrich Co.

NMR, Infra-Red Spectra and Melting Point Analysis
1 H (500 MHz) and 13 C (125 MHz) NMR spectra were acquired on a Bruker AV-500 spectrometer (Bruker, Billerica, MA, USA). Chemical shifts were reported in parts per million (ppm) with values relatively compared with the internal methanol-D (3.31 and 49.00 ppm for 1 H and 13 C, respectively). Abbreviations of signal coupling are s, d, t, q, and m for singlet, doublet, triplet, quartet, and multiplet, respectively. IR spectra were recorded using an FT-IR spectrometer (Thermo Fisher Scientific, Waltham, MA, USA) with an infra-red microscope (Nicolet 6700, Thermo Fisher Scientific, Waltham, MA, USA) while m.p. was measured using a Mettler Toledo Melting Point System (MP 50) (Mettler-Toledo AG, Schwerzenbach, Swizerland).

Preparation of Stock Solutions, Calibration Standards, and Quality Control Samples
All the analytical stock solutions of D 4 -NG, D 4 -NE, D 4 -HPPA, and the internal standard D 5 -BA were prepared in dissolution with methanol/water (60%:40%, v/v) at concentrations of 1 mg mL −1 and stored at 4 • C before use. All stock solutions were diluted to 10 µg mL −1 to investigate the most optimal MS/MS parameters including MRM transition, dwell time, fragmentor voltage, and collision energy.

Recruitment of Human Participants
Under authorization and supervision by the Ethics Committee of the School of Life Sciences, Sun Yat-sen University (Guangzhou, China), a total of 30 healthy adult participants were recruited according to our previous study [18]. The investigation enrolled 16 female and 14 male subjects, with ages ranging from 20 to 40 and body mass indexes (BMIs) between 19 to 25 kg m −2 . None of the participants used any alcohol, medication, or antibiotic therapies, nor did they have any accurate or chronic gastrointestinal diseases during the last three months before recruitment. Experimental design and criteria were introduced to the recruited participants in detail and signatures of informed consent were obtained.

Feces Sample Collection and Gut Microbiota Solution Preparation
Feces samples were freshly collected and processed for no longer than 2 h. For gut microbiota solution preparation, 1.0 g fecal samples were thoroughly homogenized with 4 mL sterilized saline solution and the suspensions were centrifuged at 100× g at 4 • C for 10 min to filter out food residue. One milliliter supernatant was mixed with 9 mL sterile GAM broth with a gentle vortex, followed by anaerobic (2%, 20%, and 78% for H 2 , CO 2 , and N 2 , respectively) co-incubation at 37 • C for 24 h.

Sample Preparation
Ten microliters of internal standard solution (10 µg mL −1 ) were mixed with 100 µL co-incubated solution and then the mixture was extracted by 1 mL ethyl acetate with vortexing for 3 min and centrifugation at 13,000× g for 15 min at 4 • C. After the liquid-liquid extraction, 900 µL supernatant was evaporated under a gentle nitrogen stream at 37 • C. The residue was re-dissolved by 100 µL 60% methanol (v/v) followed by vortexing for 3 min and centrifugation 15,000× g for 10 min at 4 • C. Five microliter supernatant was injected for RRLC-MS/MS analysis.

RRLC-MS/MS Conditions
A connected system of 1200 RRLC-6410 triple quadrupole (QQQ) mass spectrometers (Agilent Technology, Santa Clara, CA, USA) with an ESI was employed for RRLC-MS/MS analysis. A Poroshell 120 EC-C18 column (Agilent Technology, 3.0 × 50 mm, 2.7 µm) was applied to the chromatographic separation. The mobile phase system consisted of deionized water (A) and methanol (B). Both of the solvents A and B were with 0.1% formic acid (v/v). The separation was conducted with a linear gradient elution profile using 0-3 min: 60-100% B with a 7 min post run for system equilibration. The flow rate was kept at 0.3 mL min −1 and the column temperature was maintained at 40 • C.
For higher sensitivity, optimized MS/MS parameters including gas temperature, gas flow, nebulizer, and capillary were set at 350 • C, 10 L min 1 , 25 psi, and 4000 V, respectively. MRM detection mode was applied to quantify the target compounds and the optimal parameters including MRM transition, dwell time, fragmentor voltage, and collision energy, which are shown in Table 1.

Method Validation
This quantitative analysis method was fully validated according to the guidance of the Chinese Pharmacopoeia Commission for Bioanalytical Method Validation criteria on specificity, linearity, precision and accuracy, extract recovery, and matrix effects, which included various aspects to evaluate a new method. In addition, we applied the Guide to the Expression of Uncertainty in Measurement (GUM), International Organisation for Standardisation and the method reported by Konieczka et al. [47] to further possess detailed calculations of the different uncertainties. The results will be reported in due course.
For specificity validation, chromatograms of blank gut microbiota solutions from six individual sources were compared with samples spiked with analytes an LLOQ concentration and IS standard. The responses of any interferences should be below the 20% responses of the target analytes and the 5% response of IS.
Precision and accuracy were evaluated using both intra-day and inter-day analysis with four calibrated standards including LLOQ and QC samples in six duplicates with concentrations at 10, 20, 200, and 1500 ng mL −1 for D 4 -NG; 5, 10, 100, and 750 ng mL −1 for D 4 -NE; and 2.5, 7.5, 50, and 375 ng mL −1 for D 4 -HPPA. The precision was expressed as RSDs (%) and the accuracy was determined using RE values (%) from measured concentrations to nominal concentrations. The accepted criteria of precision RSD was less than 15% for QC samples and 20% for LLOQ samples while the qualified requirement of accuracy was RE within ±15% and ±20% for QC and LLOQ samples, respectively.
Extract recovery of analytical compounds was validated by comparing the responses of LLOQ and QC samples prepared in accordance with analytical procedures and the blank gut microbiota solution extracted samples post-spiked with working standard solutions with corresponding concentrations in six duplicates.
Matrix effects were determined by comparison of the matrix factors normalized by IS from six individual gut microbiota solution extracted samples post-spiked with low and high concentrations to corresponding neat working standard solutions in three duplicates.
Stability validation was carried out through measuring low and high QC samples in three duplicates under conditions including freeze and thaw (three cycles from −80 • C to 25 • C), short-term storage (24 h at 25 • C), and long-term storage (3 months at −80 • C).

Conclusions
In summary, this research has quantitatively analyzed naringin and its major metabolites mediated by human gut microbiota with efficiency and precision by utilizing our newly developed and validated strategy, which combines stable isotope deuterium-labeling and RRLC-MS/MS method together. As expected, our strategy not only eliminated strong matrix interferences but also showed its potential application in the research of flavonoid and phenolic acid molecules. In addition, this work has revealed that bioactive naringenin and HPPA are major metabolites of naringin under the atmosphere of human gut microorganisms, indicating that microbiota play a crucial role in regulating the pharmacological effects of naringin. In addition, our results will be of value to complement and guide the study of mass balance and pharmacokinetics in clinical research.