Structure-Activity Relationship and Substrate-Dependent Phenomena in Effects of Ginsenosides on Activities of Drug-Metabolizing P450 Enzymes

Ginseng, a traditional herbal medicine, may interact with several co-administered drugs in clinical settings, and ginsenosides, the major active components of ginseng, may be responsible for these ginseng-drug interactions (GDIs). Results from previous studies on ginsenosides' effects on human drug-metabolizing P450 enzymes are inconsistent and confusing. Herein, we first evaluated the inhibitory effects of fifteen ginsenosides and sapogenins on human CYP1A2, CYP2C9, CYP2C19, CYP2D6 and CYP3A4 enzymes by using commercially available fluorescent probes. The structure-activity relationship of their effects on the P450s was also explored and a pharmacophore model was established for CYP3A4. Moreover, substrate-dependent phenomena were found in ginsenosides' effects on CYP3A4 when another fluorescent probe was used, and were further confirmed in tests with conventional drug probes and human liver microsomes. These substrate-dependent effects of the ginsenosides may provide an explanation for the inconsistent results obtained in previous GDI reports.


Introduction
With the increasing use of alternative medicine and a wide spread of combination therapies for various diseases, there is an increasing interest in determining drug-drug, drug-nutrient, and drug-dietary supplements interactions. For example, Panax ginseng, a traditional herbal medicine used in the Eastern Asia for more than 2000 years [1], is presently being used worldwide as one of the most common complementary alternative medicines. Ginseng, the root of Panax ginseng, has diverse pharmacological activities, including effects on the central nervous system, antineoplastic and immunomodulatory effects [2]. In the clinical settings, however, co-administration of ginseng or its extracts with other therapeutic agents (e.g. warfarin, digoxin and phenelzine) may lead to ginsengdrug interactions (GDIs) [3,4]. Ginsenosides (Fig. 1a), the major active components of ginseng, may account for the GDIs and other adverse effects [5]. It is postulated that most metabolic drugdrug interactions can be attributed to inhibition or induction of drug-metabolizing cytochrome P450 (CYP or P450) enzymes [6]. Therefore, we hypothesized that investigations on the effects of ginsenosides on P450s will help elucidate the mechanism of GDIs.
We also speculated that there are structure-activity relationships (SAR) for the effects of ginsenosides on P450s. Although several ligand-based and structure-based models have been established for substrates and/or inhibitors of P450s [15], the SAR of the effects of the ginsenosides on P450s were rarely addressed. In the present study, we systematically evaluated the effects of 15 ginsenosides and sapogenins (Fig. 1A) on five major human drug-metabolizing P450 enzymes with commercially available VividH substrates to use as probes (Fig. 1B). Subsequently, the SAR for these effects was explored to generate new knowledge about ginsenoside-P450 interactions at the molecular level. Moreover, we also selected different fluorescent and conventional probes to determine whether the effects of different ginsenosides on CYP3A4, a key P450 enzyme with the largest substrate repository, are substrateand analog-dependent.

Results
Ginsenosides and sapogenins affect P450 enzymes in an analog-and enzyme-dependent manner The parameters used to determine the activities of the P450 enzymes are summarized in Table 1. Reported potent inhibitors of respective P450 enzyme were evaluated to validate the efficiency of these fluorescent assays (Table 2). These inhibitors show comparable activities against respective P450s to the documented reports [16][17][18][19]. Additionally, the inhibitory effects of methanol, a vehicle used in the study, were carefully evaluated. Methanol at the final concentration of 1% had no significant inhibitory effect against VividH blue metabolism by CYP1A2 or green metabolism by CYP3A4, and had minimal inhibition against VividH blue metabolism by CYP2C19 (Table 2). However, 1% methanol significantly inhibited VividH CYP2C9 red metabolism (by 42.4%) and blue metabolism by CYP2D6 (27.5%). At the final concentration of methanol of 0.5% or 0.1%, the inhibition against substrate metabolism by methanol was all either negligible or within an acceptable range (Table 2).
Subsequently, we determined the effects of the ginsenosides and sapogenins on five major cDNA-expressed P450 enzymes. The compounds had weak inhibitory effects against CYP1A2, with the exception of Rb 1 , Rg 3 , and Compound K (C-K) which had moderate inhibitory effects with IC 50 values less than 50 mM ( Table 2). Ginsenosides Rg 3 , Rh 2 , and C-K and all sapogenins exhibited moderate inhibition against CYP2C19, while they more potently inhibited CYP2C9 and CYP3A4 (green substrate) Note: a. The percent inhibition of ginsenosides against the respective P450 enzymes is shown when its IC 50 value is greater than the maximum concentration assayed. b. The maximum concentration of ginsenosides evaluated for their effects on CYP2C9 and CYP2D6 were 50 mM due to the marked solvent effect of 1% methanol on these two P450 enzymes (inhibition by 42.4% and 27.5%, respectively). When the final concentration of methanol was decreased to 0.5%, the solvent effects were acceptable for these two enzymes (10.1% and 18.9%, respectively). 1% methanol had no inhibition against CYP1A2 and CYP3A4 and had an acceptable inhibitory effect on CYP2C19 (9.7%). c. Positive control compounds were a-naphthoflavone (for CYP1A2), sulfaphenazole (CYP2C9), miconazole nitrate salt (CYP2C19), quinidine (CYP2D6), and ketoconazole (CYP3A4), respectively. d. A.A. = apparent activation. 100 mM 25-OH-PPD and 25-OH-PPT increased the turnover of VividH CYP3A4 red by more than 100%. doi:10.1371/journal.pone.0002697.t002 (p,0.05). Rg 3 exerted relatively weak inhibitory effects against VividH CYP3A4 green metabolism (Table 2). However, it potently inhibited CYP2D6, for which all of the other ginseng compounds had IC 50 values greater than 50 mM. Ginsenosides Rb 1 , Rb 3 , Rd, Re, Rg 2 and Rg 1 weakly affected all five P450 enzymes, with the exception of Rb 1 and C-K, which moderately inhibited CYP1A2 ( Table 2). All of the sapogenins had weak inhibitory effects on CYP1A2 and CYP2D6 (Table 2).
There is a SAR in the effects of ginsenosides on P450s Considering that the same core structure shared by these ginsenosides, the location, number and type of the glycosyl groups covalently linked to the dammarane scaffold may account for the their varying effects on P450 enzymes ( Fig. 1A and Table 2). For CYP2C9 and CYP2C19, the inhibition patterns of ginsenosides are similar. Ginsenosides that do not have a glycosyl group on R 2 and have two or fewer glycosyl groups on R 1 and R 3 as well as sapogenins generally have moderate to potent effects on CYP2C9 and CYP2C19. As for CYP3A4 (green substrate), the inhibitory activities of the ginsenosides and sapogenins decreased as the number of glycosyl groups increased.

3D-SAR is shown in the effects of ginsenosides on VividH CYP3A4 green metabolism
We applied the HipHop process from Catalyst software, which compares diverse and highly active compounds to derive 3D hypotheses based on common chemical features, without considering biological activities. The final training set consisting of 6 highly active compounds was submitted for pharmacophore building. The best hypothesis (Hypo1), consisting of four features, namely, one hydrogen-bond acceptor, one hydrophobic point, and two hydrogen-bond donors ( Fig. 2A). A representation of the Figure 2. Shape-based pharmacophore model of the inhibition of VividH CYP3A4 green activity by ginsenosides. The model was generated from the 3D structure of PPT and its activities against CYP3A4. PPT fitted into the model was shown in the figure. Blue represents the shape space, black represents carbon atoms, red represents oxygen atoms, and white represents hydrogen atoms. doi:10.1371/journal.pone.0002697.g002 chemical features mapped onto representative studied compound (PPT) is depicted in Fig. 2B. In this model, the hydrogen bond acceptor seems to be mapped to the oxygen atom substituted at atom C20, and the hydrophobic group fit well with the aliphatic chain attached as side-chain to this moiety. The analysis shows that hydroxyl substitutions at atom C3 and C12 are crucial. Hydrogen-bond donors at these atomic positions will increase the bioactivity. From Fig. 2B, it is evident that the hydroxyl group substituted at atom C3 and C12 serves to be hydrogen-bond donors (HD1 and HD2). Thus replacement with other substituent group at oxygen atoms of C3-OH or C12-OH will result in decreased potency.
The effects of the ginsenosides on CYP3A4 were substrate-dependent Considering potential influences of substrate selection on the evaluation of the effects of the ginsenosides on the catalytic activity of CYP3A4, another fluorescent probe, VividH CYP3A4 red (Fig. 1B), was used. The assay conditions are also summarized in Table 1. The solvent effect of methanol (1%) on CYP3A4 red metabolism was negligible. The IC 50 value of ketoconazole on the CYP3A4 red assay was 0.97 mM ( Table 2). As for the ginsenosides and sapogenins, the inhibition profiles using the red substrate and green substrate were not consistent. For the same saponin, the inhibition potency determined using the red substrate was greater than using the green substrate (Table 2) (p,0.05). However, the pattern was the opposite for sapogenins: the inhibition potency was overestimated using the green substrate compared to the red substrate (Table 2). Strikingly, 25-OH-PPD and 25-OH-PPT activated the catalytic activity of CYP3A4 when using red substrate. The representative influence profiles of PPD, 25-OH-PPD, 25-OCH 3 -PPD and Rh 2 on CYP3A4 activities using either red or green substrate are shown in Fig. 3. In the VividH red assay (Fig. 3A), 25-OH-PPD potently activated CYP3A4 (about twofold) at a low concentration (10 mM). It was notable that at higher concentrations 25-OH-PPD activated CYP3A4 red metabolism in a saturated profile. At 10 mM, 25-OCH 3 -PPD weakly activated CYP3A4 (about 20%), but weakly inhibited the enzyme at high concentrations. PPD and Rh 2 inhibited CYP3A4 in a dosedependent manner. However, the effect of Rh 2 (IC 50 = 9.8 mM) was more potent than PPD (IC 50 = 43.1 mM) (p,0.01). In the VividH green assay (Fig. 3B), however, all four tested compounds inhibited CYP3A4 in a dose-dependent manner, with the potency decreasing in the order: PPD.25-OCH 3 -PPD.25-OH-PPD.Rh 2 .
The effects of the ginsenosides and sapogenins on CYP3A4 were also substrate-dependent when using human liver microsomes and different conventional probes. CYP3A4-catalyzed reactions (carbamazepine 10,11-epoxidation and nifedipine oxidation) were used to evaluate the potential ginsenoside-drug interactions. We employed high performance liquid chromatography/tandem mass spectrometry (HPLC-MS-MS) analysis to detect the reaction products carbamazepine 10,11-epoxide (CBZ-E) and oxidized nifedipine (ONF) of respective drug substrates carbamazepine (CBZ) and nifedipine (NF).
As for NF in vitro metabolism, Rh 2 and PPD exerted a weak activation effect at low concentrations (1,10 mM). However, the high concentration (50 mM) of Rh 2 weakly inhibited CYP3A4catalyzed NF oxidation. The effects of 25-OCH 3 -PPD and 25-OH-PPD on NF metabolism were negligible (Fig. 4B). a-Naphthoflavone (10 mM) did not significantly alter in vitro NF metabolism, whereas ketoconazole (10 mM) inhibited CYP3A4catalyzed NF metabolism almost completely (decreased by 94.9%), compared to the vehicle control.

Discussion
Potential GDIs in the clinical settings make it necessary to investigate the influences of major ginseng components on drugmetabolizing enzymes. To that end, we studied the effects of 15 ginsenosides and sapogenins on P450-mediated drug metabolism. Compared to microsomal assays with conventional probe drugs using HPLC analysis, fluorescent methods for determining P450 activity are more high-throughput and reproducible [20][21][22][23][24] and have been extensively employed in screening for inhibitors and inducers of P450s [24][25][26][27]. Using the fluorescent probes, we found that ginsenoside Rb 1 , Re and Rg 1 had no significant effects on major P450s, with the exception that Rb 1 moderately inhibited CYP1A2 in our assay. A previous report [28] indicated that ginsenoside Rd inhibits CYP3A4-mediated testosterone 6bhydroxylation with an IC 50 value of 62 mM, however, we did not observe any significant influence of Rd on any of the P450s tested. More recently, Liu et al. found that ginsenoside metabolites (C-K, PPD and PPT) are more potent P450 inhibitors than natural ginsenosides [10]. In our present study, we also found that sapogenins are more potent than saponins in their effects on CYP3A4, despite the fact that the IC 50 values we obtained for prosapogenin C-K and Rh 2 were lower. It is noteworthy, however, that several saponins in our assay were found to have activities comparable to (Rg 3 , Rh 2 and C-K against CYP2C9 and CYP2C19) or even greater than (Rb 1 , Rg 3 and C-K against CYP1A2, and Rg 3 against CYP2D6) sapogenins (p,0.05). These findings indicate that natural ginsenosides still have the potential to inhibit certain P450 enzymes and possibly lead to GDIs in vivo. Information about the circulating levels of these ginsenosides and sapogenins in humans would be helpful to determine whether these interactions occur in vivo. However, except for only a few ginsenosides (Rb 1 , Rg 1 [29] and Rg 3 [30]), such information is not available in the literature due to limited pharmacokinetic studies of ginsenosides in human.
Our data showed that a SAR exists in the effects of ginsenosides on almost all of the P450 enzymes examined, especially on CYP2C9, CYP2C19 and CYP3A4 (green substrate) ( Table 2). Apparently, the similarities and differences among the activities of tested ginsenosides against each of the P450s are consistent with the fact that these compounds are analogs and differ in only four substituent groups. Furthermore, a pharmacophore (Fig. 2) was established based on the data set from the CYP3A4 green assay to explore more information about effects of ginsenosides and sapogenins on this key enzyme, which participates in the metabolism of more than half of pharmaceuticals. The model shows that hydroxyl substitutions at atom C3 and C12 play a key role in the inhibitory potency of ginsenosides and sapogenins against VividH CYP3A4 green metabolism. Presence of hydrogenbond donors at these atomic positions will increase the bioactivities of ginsenosides and sapogenins on CYP3A4. Therefore, substitution with glycosyl group(s) at oxygen atoms of C3-OH will result in decreased potency of the compounds.
The inconsistencies existing in the previous reports (e.g. effects of Rd and Rb 2 on CYP2C19 [7,8]) and conflicts between our data and previous reports (e.g. effects of saponins on several P450s [10]) on the inhibition of P450s by ginsenosides can most likely be attributed to the different substrates and assay systems used to determine the P450 activities. It is recommended that multiple probes should be used when P450-mediated drug interactions are evaluated [24]. This is especially the case for CYP3A4, because its drug interaction patterns were substrate-dependent [31]. We therefore evaluated the effects of the ginsenosides and sapogenins on the activity of human CYP3A4 by using another fluorescent probe and two conventional drug substrates. Interestingly, the influence patterns of these compounds against CYP3A4 using the red substrate were distinct from those observed using the green substrate. In subsequent microsomal studies with CBZ and NF as substrates, these substrate-dependent and analog-dependent phenomena were further validated. CBZ in vitro metabolism was potently activated by PPD, 25-OCH 3 -PPD and 25-OH-PPD, even at low concentrations (5,10 mM). CBZ is an anticonvulsant agent used in the treatment of epilepsy, acute mania and bipolar disorder [32]. Due to co-administration of ginseng products among some epilepsy patients [33], ginseng or ginseng-based products have the potential to promote the metabolism of CBZ, and thus decrease the bioavailability of this drug. However, whether PPD, 25-OCH 3 -PPD, or 25-OH-PPD can activate CBZ in vivo metabolism needs further investigation.
The substrate-dependent phenomena and atypical kinetics of P450-mediated drug-drug interactions, especially for CYP3A4, were documented in several previous in vitro studies [31,34,35]. There have also been several reports concerning the activation of P450s. Tea and tea polyphenols were found to heteroactivate CYP1A1 with a fluorescent probe [36]. Endogenous steroids were reported to activate CYP3A4-mediated CBZ 10,11-epoxidation [37,38]. As for ginsenosides, Rc was documented to activate CYP2C9 activity, while Rf activated CYP3A4 in fluorescent assays [7]. These substrate-dependent activation events may be due to the large active site of P450s, especially CYP3A4, as evidenced by its crystal structure [39]. CYP3A4 can bind multiple ligands simultaneously, a property that may contribute to its complex cooperative effects [40][41][42][43]. Thus, different inhibition profiles of the test compounds against CYP3A4 result when probes with different structures are used. As to the mechanism of heteroactivation of CBZ metabolism by endogenous steroids, CBZ may be more easily catalyzed due to direct interaction with steroids in the active site, which may change the activity of CYP3A4 [44]. Interestingly, ginsenoside sapogenins that have stimulatory effects on CBZ metabolism (PPD, 25-OCH 3 -PPD and 25-OH-PPD) have structures with similarity to steroids, and may activate CYP3A4 activity by interacting with CBZ in the active site. However, saponin ginsenosides (such as Rh 2 ), which are of relatively large size, may hinder the co-binding events and consequently lead to no or inhibitory effects on CYP3A4. Nevertheless, the molecular and structural basis of these substrateand analog-dependent effects is still not clear and needs further investigation.
Our observations of substrate-dependent effects of ginsenosides and sapogenins on CYP3A4 suggest that GDIs may also be substrate-dependent. In contrast with reports mentioned in the 'Introduction' section, several in vivo studies suggested that ginseng or ginseng extracts have little effect on P450 activities. Coadministration of ginseng did not alter the pharmacokinetics and pharmacodynamics of warfarin in human subjects [45]. Gurley et al found that supplementation of Panax ginseng did not have any significant effects on CYP1A2, 2D6, 2E1 or CYP3A4 activities in vivo when their activity was examined by use of probe-drug cocktails [46]. Our studies may provide an explanation for the inconsistencies in clinical GDI reports. Inhibition or activation, and the extent of these effects of ginseng or ginseng products on P450s, are relative and based on the substrate(s) selected in the assay. Translation from data obtained with only one 'probe' to other drugs metabolized by the same P450 enzyme will probably lead to ex parte conclusion on GDI, and may lead researchers to overlook potential interactions of ginseng with drugs other than the probe. Therefore, multiple probe drugs are needed for careful in vitro and in vivo evaluation of P450-mediated GDIs, and drugdrug interactions to a larger extent, especially in case of CYP3A4.
In summary, we evaluated the in vitro effects and the SAR of fifteen ginsenosides and sapogenins on five major human drugmetabolizing P450 enzymes. In addition, we also found that substrate-dependent phenomena exist in the effects of the ginsenosides and sapogenins when employing different fluorescent and conventional probes. To the best of our knowledge, this study is the first report on the SAR and substrate-dependence of the effects of ginsenosides on P450s. The information derived will enhance our understanding of GDIs, and provide a possible explanation for the inconsistent results obtained in previous reports.
In brief, the assays were carried out in CostarH black-wall 96well plates with ultra thin clear bottoms (Corning Inc, Corning, NY, USA) in a kinetic assay mode. Stock solutions (10 mM) of ginsenosides in methanol were prepared and diluted to various concentrations (2.56 100, 50, 10, 5, 1 mM). For each well, 40 mL of solution of test compounds or vehicle was incubated with 50 mL pre-mixture (mixture of BACULOSOMESH reagent, regeneration system and reaction buffer) at 37uC for 20 min. The reaction was initiated by adding 10 mL of a mixture of substrate and NADP + per well with a respective concentration of VividH substrate ( Table 1). The plate was read immediately for fluorescence changes every 0.5 min at 37uC for 30 min using a FlexStation II384 fluorometric plate reader (Molecular Devices, Sunnyvale, CA, USA) with respective excitation and emission wavelengths for each P450 enzyme ( Table 1). The final methanol volume in the reaction was less than or equal to 1%. Inhibitory potencies of test compounds (IC 50 values) were compared by using Student' t-test. Difference was considered to be significant when the two-tailed pvalue was less than 0.05.

Pharmacophore generation
The compounds were built using Catalyst (Accelrys corporation, San Diego, CA, USA) 2D-3D sketcher, and a family of representative conformations was generated for each ginsenoside or sapogenin using the best conformational analysis method. A maximum number of 250 conformations of each compound were selected using ''best conformer generation'' option with a constraint of 20 kcal/mol energy thresholds above the global energy minimum to ensure maximum coverage of the conformational space. Based on the conformations for each compound, Catalyst 4.10 software package was employed to construct possible pharmacophore models. When generating a hypothesis, catalyst attempts to minimize a cost function consisting of two terms. Analysis of functional groups on each compound in the training set revealed that three chemical features, hydrogen-bond acceptor (HA), hydrogen-bond donor (HD), and hydrophobic group (HY), could effectively map all of the critical chemical features. Hence, the three features were selected to form the essential information in this hypothesis generation process. The best predictive hypothesis (Hypo1), produced by HipHop process encoded in Catalyst 4.10, has four features: one hydrogen-bond acceptor, one hydrophobic point, and two hydrogen-bond donor, which was characterized by the highest cost difference, the lowest rms divergence, and the best correlation coefficient. Remarkably, the highest active compound (PPT) can be nicely mapped onto the Hypo1 model by the best fit values.

Microsomal studies
The effects of the ginsenosides and sapogenins on the in vitro metabolism of CBZ were evaluated using HLM. Triplicate microsomes (50 mL, 20 mg/ml) were treated sequentially with 20 mL 1 mM CBZ 10% DMSO (v/v) aqueous solution, 20 mL aqueous solutions of test ginsenosides containing 25% methanol (v/v) at different concentrations (1, 5, 10, and 50 mM) or anaphthoflavone (10 mM) or ketoconazole (10 mM), and 850 mL 0.1 M Na 3 PO 4 solution, and pre-incubated at 37uC for 5 min. Vehicle control samples were treated with methanol at a final concentration of 0.5% instead of test compounds. The reaction was initiated by adding 60 mL pre-warmed NADPH-regenerating system (50 mL BD solution A and 10 mL BD solution B). After incubation at 37uC for 30 min, the reaction system was quenched by adding 0.4 ml ice-cold acetonitrile, which was mixed well and centrifuged at 4uC (10,000 g, 10 min). The supernatant (0.4 ml) was collected, mixed with 0.8 ml acetonitrile and then centrifuged at 4uC (10,000 g, 10 min). The supernatant (30 mL) was diluted with 970 mL ddH 2 O and 50 mL of the mixture was injected onto the high performance liquid chromatography/tandem mass spectrometry (HPLC-MS-MS) system for analysis. The calibration samples were prepared with standard solutions of CBZ-E (0.00457, 0.0137, 0.0412, 0.123, 0.370, 1.11, and 3.33 mM) and processed identically to the samples before analysis, except that solution A was added after microsomal proteins were precipitated by acetonitrile. Microsomal studies using NF as the substrate were conducted as described above for CBZ with several modifications. The NF concentration was 1 mM. The supernatant (0.4 ml) of the reaction system after quenching and centrifugation was also combined with 0.8 ml acetonitrile and centrifuged, and 10 mL of the resultant supernatant was directly injected onto the HPLC-MS-MS system for analysis.

LC-MS-MS analysis
The HPLC was performed using an Agilent 1200 system (Palo Alto, CA, USA) equipped with a binary pump and an autosampler. For CBZ-E analysis, an Agilent Zorbax SB-C18 column (4.66150 mm, 5 mm) was used, and the mobile phase consisted of 50% eluent A (aqueous solution containing 0.5% acetic acid) and 50% eluent B (acetonitrile) at the flow rate of 0.4 ml/min. The elution lasted for 8 min. For ONF analysis, an Agilent Zorbax EclipseH XDB-C8 column (4.66150 mm, 5 mm) was used; the mobile phase consisted of eluent A (10 mM ammonium acetate) and eluent B (acetonitrile) at the flow rate of 0.6 ml/min. The gradient elution program was as follows: started with 70% B for 1.5 min; changed to 95% B within 1 min; held at 95% B for 5.5 min; returned rapidly to 70% B; and held at 70% B for 2 min.
The MS analysis was performed on a 4000Q Trap TM triple quadrupole mass spectrometer (Applied Biosystems/MDS Sciex Instruments, Concord, Ontario, Canada) equipped with a Turbo V ion source (TurboIonSpray TM probe was used), a Dell TM computer, and Analyst TM software. Electrospray ionization in the positive mode was employed to analyze CBZ-E and ONF, respectively. The compound parameters for CBZ-E and ONF were: declustering potential, 65 and 95 units, and collision energy, 39 and 40 units, respectively. The ion source parameters were as follows: curtain gas, 15 units; collision gas, medium; ion spray voltage, 4000 and 5000 V, respectively; temperature, 650 and 350uC, respectively; ion source gas 1, 60.0 and 65.0 units, respectively; ion source gas 2, 45.0 units. The fragmentation transitions for multiple reaction monitoring (MRM) were m/z 253.1R180.2 for CBZ-E and m/z 345.1R284.3 for ONF. The representative calibration curve for CBZ-E is: Y = 147X+358 (r = 0.9993) with 1/X 2 weighting, and the representative calibration curve for ONF is: Y = 3570X22410 (r = 0.9976) with 1/X weighting, where Y represents the peak area of the analyte, and X represents the analyte concentration.