SEN (98%) was purchased from Phytolab (Phytolab GmbH and Co. KG, Germany) and was prepared in DMSO purchased from Acros Organics (Geel, Belgium). Acetonitrile (ACN, UPLC/MS grade) and methanol were purchased from Biosolve (Valkenswaard, the Netherlands). Di-potassium hydrogen phosphate trihydrate (K₂HPO₄.3H₂O) and potassium dihydrogen phosphate (KH₂PO₄) were purchased from Merck (Darmstadt, Germany). The reduced form of nicotinamide adenine dinucleotide phosphate (NADPH) was purchased from Carbosynth (Carbosynth, United Kingdom). L-glutathione-reduced (GSH, purity ≥98%) was purchased from Sigma-Aldrich (St. Louis, MO, United States). Pooled rat liver S9 from male Sprague-Dawley (SD) rats was purchased from Corning (Amsterdam, the Netherlands).

The rat liver S9 incubation conditions for 7-GS-DHP formation were similar to the conditions previously used to measure SEN depletion (Ning et al., 2019a; Ning et al., 2019b; Widjaja et al., 2022a) but with the addition of GSH. Briefly, the incubation was performed in a total volume of 100 µl containing (final concentrations) 0.1 M potassium phosphate (pH 7.4), 2 mM NADPH, 4 mM GSH, 1 mg/ml rat liver S9, and 0.5–50 µM SEN (added from 50 times concentrated stock solutions in DMSO). Controls were performed without the addition of NADPH. Upon 5 min preincubation with NADPH in a shaking water bath at 37°C, the reaction was started by the addition of SEN. After 60 min incubation, the reaction was terminated by adding 25 µl (20% v/v) ice-cold ACN followed by centrifugation at 16,000 g for 5 min at 4°C and supernatants were immediately analyzed by LC-MS/MS.

7-GS-DHP was quantified by LC-MS/MS using a Shimadzu Nexera XR LC-20AD XR UHPLC system coupled with a Shimadzu LCMS-8040 MS (Kyoto, Japan). A 1 µl aliquot was loaded onto a reverse phase C18 column (Phenomenex 1.7 µm 2.1 × 50 mm). The flow rate was 0.3 ml/min and the mobile phase was made with ultrapure water with 0.1% (v/v) formic acid and ACN containing 0.1% (v/v) formic acid. A linear gradient was applied from 0% to 5% ACN in 8 min and was further increased to 100% ACN in 6 min. This percentage was kept for 0.5 min and was then reduced to the starting conditions in 0.1 min. The column was equilibrated for another 4 min at the starting conditions before the next injection. Under these conditions, 7-GS-DHP eluted at 8.7 min.

For detection, a Shimadzu LCMS-8040 triple quadrupole with an ESI interface was used. The instrument was operated in a positive ionization mode in the multiple reaction monitoring (MRM) mode with a spray voltage of 4.5 kV. The 7-GS-DHP was monitored at the [M + H]⁺ of precursor to products of 443.2 → 425.15 (CE = −7 eV), 443.2 → 118.1 (CE = −24 eV) and 443.2 → 247.2 (CE = −15 eV) m/z. The peak area of the total ion chromatogram (TIC) was used for quantification (Ning et al., 2019a; Ning et al., 2019b). Quantification was done by comparing the peak area of the 7-GS-DHP formed in the incubation samples to the calibration curve of 7-GS-DHP ranging from 0.078 to 10 µM (r ² = 0.991) previously reported by Ning et al. using the same conditions and LC MS-MS instrument (Ning et al., 2019a; Ning et al., 2019b).

Kinetic constants were obtained from the SEN concentration dependent rate for 7-GS-DHP formation in incubations with rat liver S9 with SEN concentrations varying from 0.5 to 50 μM, performed as previously reported for lasiocarpine and riddelliine (Ning et al., 2019a; Ning et al., 2019b). The concentration of 7-GS-DHP formed was determined as the concentration detected in full incubations minus the concentration detected in the blanks (without NADPH). Since 7-GS-DHP formation from SEN follows first-order reaction kinetics, the slope of formation rate versus substrate concentration directly represents k_cat, which equals V_max and K_m values as such, for which can be derived from the Michaelis-Menten equation for substrate concentrations at [S] << K _m , thus 1 + K _m /[S] equals K _m /[S]. Consequently, the Michaelis-Menten curve in the range where [S] << K _m becomes: v = V max 1 + K m S = V max K m S = V max K m   S = k c a t   S

In this equation v is the rate of reaction, V _max the apparent maximum rate of reaction, K _m the apparent Michaelis-Menten constant, and [S] the substrate (SEN) concentration. The slope k_cat was determined by fitting the data to the linear regression model using GraphPad (GraphPad Prism software version 5.04, San Diego California United States). The in vitro k_cat expressed in ml min^-1 mg^-1 S9 or the slope value was scaled to an in vivo value expressed in L h^-1 by using an S9 protein yield of 143 mg g^-1 rat liver, 34 g kg^-1 bw rat liver and 0.25 kg bw (Punt et al., 2008).

The PBK model for SENO with a submodel for SEN in rats previously developed and evaluated (Widjaja et al., 2022a) was built in Berkeley Madonna version 9.1.18 and was run with the Rosenbrock (stiff) method as ordinary differential equations solver. The model was extended to include GSH-scavenging of the reactive pyrrole intermediate through 7-GS-DHP formation from SEN (Ning et al., 2019a; Ning et al., 2019b), with 7-GS-DHP being the major metabolite formed in this reaction (Tamta et al., 2012; Ning et al., 2019a; Ning et al., 2019b) (Figure 3). All other physiological, physicochemical, and kinetic parameters remained unchanged from the previous code. The updated model code can be found in the Supplementary Material. To include 7-GS-DHP formation the equation for the change in the amount of SEN in the liver was extended to read as follows (in bold the parts that were added): A L S E N ′ = Q L * C B S E N − C V L S E N + k b 1 * A S I S E N + k b 2 * A L I S E N + A L M 1 ′ –   A L M 2 ′ –   A L M 4 ′ + A L M 4 ′ where AL_SEN’ is the change in the total amount of SEN in liver tissue, QL*(CB_SEN - CVL_SEN) is the net amount of SEN going into the liver from SEN that enters from the arterial blood and leaves to the systemic blood circulation, k_b1*ASI_SEN is the amount of SEN entering the liver via the portal vein from the small intestine while k_b2*ALI_SEN reflects the uptake of SEN from the large intestine where it is formed from SENO via reduction by gut microbiota, ALM1′ is the amount of SEN formed by SENO reduction in the liver, and ALM2′ is the amount of SEN metabolized or cleared in the liver. This part of the equation is similar to what was previously included in the model code.

SEN clearance proceeds by several different pathways including N-oxidation, N-glucuronidation, hydrolysis and oxidation (He et al., 2021). In rat liver S9 incubations containing the cofactor NADPH particularly, N-oxidation, hydrolysis and oxidation are included. Oxidation by cytochromes P450 in these rat liver S9 incubations results in formation of the reactive intermediate DHP and subsequent 7-GS-DHP formation and is thus included in this overall clearance reaction (described by the term -ALM2′). However, to enable quantification of 7-GS-DHP formation the model should also describe this 7-GSH-DPH formation in the liver separately. Therefore, the differential equation that describes the change in the amount of SEN over time should include a term that describes the conversion of SEN into 7-GSH-DPH (the term ALM4′). Given that this conversion described by the term ALM4′, is also included in the overall clearance of SEN (described by the term -ALM2′), one also has to add a term + ALM4′ to the equation to avoid that this part of the clearance is subtracted twice when modeling the change in the amount of SEN in the liver. Thus, subtracting and adding the reaction that forms 7-GS-DHP from the entire SEN clearance (-ALM4′and +ALM4′) allows quantification of 7-GS-DHP formation without disturbing the mass balance of the model. This has previously been published by Ning et al. (2019) (Ning et al., 2019a; Ning et al., 2019b) and the same concept is now applied in the present study.

In the present work, two methods were used to derive the REP value of SENO relative to SEN as presented in Figure 1 and the equations below: M e t h o d   1   R E P = A U C S E N   d o s e   S E N O A U C S E N   d o s e   S E N M e t h o d   2   R E P = A m o u n t p y r r o l e − p r o t e i n   a d d u c t s   d o s e   S E N O A m o u n t p y r r o l e − p r o t e i n   a d d u c t s   d o s e   S E N

The in vivo values for the AUC for SEN and the AUC for the pyrrole-protein adducts required to calculate these two REP values were extracted from Yang et al. (Yang et al., 2017a) (Table 1), while the current PBK modeling predicts the AUC for SEN and the amount for the pyrrole-protein adducts to calculate the REP values. Since the PBK model does not include clearance of the pyrrole-protein adducts, Method 2 uses the total amount instead of the AUC of pyrrole-protein adducts formed. Given that the REP value is calculated as a ratio, the outcome will not be affected regardless of the chosen parameter (as AUC or as amount). Eventually, amount can be converted into AUC and the conversion factor would cancel out when calculating the ratio.

Considering that only a fraction of SEN will be bioactivated to the reactive pyrrole intermediate, and assuming that the bioactivated fraction not binding to DNA is converted to either reactive 7-GS-DHP or pyrrole-protein adducts, the following equations hold: A m o u n t 7 − G S − D H P   P A − N − O x i d e + A m o u n t p y r r o l e − p r o t e i n   a d d u c t s   P A − N − O x i d e = F ∗   f ∗   R E P M e t h o d   1 *   I n i t i a l   d o s e P A − N − O x i d e A m o u n t 7 − G S − D H P   P A + A m o u n t p y r r o l e − p r o t e i n   a d d u c t s   P A = F ∗ f ∗   I n i t i a l   d o s e P A where a m o u n t 7 − G S − D H P is the amount of 7-GS-DHP at 24 h (µmol), a m o u n t p y r r o l e − p r o t e i n   a d d u c t is the amount of pyrrole-protein adducts at 24 h (µmol), initial dose is the administrated oral dose level (µmol), F is the oral bioavailability amounting to 8.20% (as reported in literature for SEN (Wang et al., 2011) or 100% (for comparison), f is the fraction bioactivated to reactive pyrrole intermediates not binding to DNA, REP _{Method 1} is the REP value of SENO relative to SEN at corresponding dose level based on the ratio of AUC_SEN from 0–24 h, PA-N-oxide signifies oral dosing of SENO, and PA signifies oral dosing of SEN. These equations were used to calculate the amount of pyrrole-protein adducts, which were needed to calculate the REP value by Method 2. Upon SENO dosage, SEN will be formed as a result of SENO reduction. Subsequently, formed SEN is bioactivated to its active pyrrole metabolites that give rise to 7-GS-DHP and pyrrole-protein adducts (Figure 1). Similarly, upon SEN dosing, SEN also results in reactive pyrrole metabolites that give rise to 7-GS-DHP and pyrrole-protein adducts. In this approach, pyrrole-DNA adduct formation is considered not to influence the balance between formation of 7-GS-DHP and pyrrole-protein adducts, and to fall outside the bioactivation that is assumed to lead to 7-GS-DHP and pyrrole protein adducts.

A sensitivity analysis was performed to assess which parameters of the PBK model have the largest impact on the predicted a m o u n t p y r r o l e − p r o t e i n   a d d u c t s , which is the parameter for calculating the Method 2 REP value. A sensitivity analysis for the AUC of SEN, which was the parameter used for calculating the Method 1 REP value, was previously presented (Widjaja et al., 2022a). This previous sensitivity analysis only did not include the k_cat for 7-GS-DHP formation. Using the extended model of the present study, the normalized sensitivity coefficients (SC) for this k_cat appeared to be 0. Normalized SCs were calculated using the equation below: S C = C ′ − C /   C P ′ − P /   P where C is the initial value of the model output, C′ is the modified value of the model output with a 5% increase of an input parameter, P is the initial parameter value, and P′ is the parameter value with an increase of 5%. Only one parameter was changed each time, while the other parameters were kept at their initial values. A large SC value indicates that the respective parameter has a large impact on the predicted a m o u n t p y r r o l e − p r o t e i n   a d d u c t s . An equimolar dose of 55 μmol kg^-1 bw (either 19.33 mg kg^-1 bw SENO or 18.45 mg kg^-1 bw SEN) reflecting the dose levels used in the animal study of Yang et al. (Yang et al., 2017a) was used to perform the sensitivity analysis in the rat PBK model assuming a bioavailability of 8.20%.

The rate of formation of 7-GS-DHP in rat liver S9 incubations with increasing concentrations of SEN is shown in Figure 4. The results obtained reveal first-order kinetics with an in vitro k_cat of 0.0023 mL min^-1 mg^-1 S9. When scaled to an in vivo value, the k_cat amounted to 0.1677 L h^-1. This in vitro k_cat value was integrated and converted to an in vivo k_cat in the PBK model to describe the in vivo SEN concentration dependent formation of 7-GS-DHP from either an oral dose of SENO or SEN. Since the initial input parameter was an in vitro k_cat value, subsequent results will also be presented based on the in vitro k_cat value which can be converted to the corresponding in vivo k_cat value using the scaling factor of 72.91 L min h^-1 mg S9 mL^-1 that was also implemented in the PBK model. Given that 7-GS-DHP formation is linear (i.e., not saturated) over the SEN concentration range tested up to 50 μM, it is concluded that limitations in 7-GS-DHP formation are unlikely to result from saturation of the kinetics of this conjugation. Limitation of the GSH pyrrole-scavenging capacity may thus rather be attributed to internal GSH depletion as reported to occur in liver cells upon PA or PA-N-oxide exposure (Griffin and Segall, 1987; Neuman et al., 2007; Chen et al., 2009). Therefore, to mimic this GSH depletion and its consequences for 7-GS-DPH adduct formation, the in vitro k_cat in the SEN submodel was decreased in simulations at high dose levels.

Figure 5 shows the effect of a reduction in the k_cat for 7-GS-DHP adduct formation, when dosing the PA, on the calculated REP values at an equimolar dose of 55 μmol kg^-1 bw of SENO and SEN as used by Yang et al. (Yang et al., 2017a) in their in vivo rat study. The REP values were calculated for two bioavailability values, namely, 8.20% bioavailability (Figure 5B) which is in line with literature reported data for the bioavailability of SEN (Wang et al., 2011) and also 100% bioavailability (Figure 5A), the later for comparison. Additional assumptions for the calculations included: a fraction of the dose bioactivated of 0.20, an in vitro k_cat for 7-GS-DHP formation upon an oral SENO dose of 0.0023 ml min^-1 mg^-1 S9, and a value for the in vitro k_cat for 7-GS-DHP formation upon an oral SEN dose varying from 0.0023 ml min^-1 mg^-1 S9 (no GSH depletion and k_cat reduction) down to 0.00105 ml min^-1 mg^-1 S9 (46% assumed residual k_cat due to GSH depletion). At a fraction bioactivated of 0.20 and a k_cat for 7-GS-DHP formation of 0.0023 ml min^-1 mg^-1 S9, the ratio between 7-GS-DHP and pyrrole-protein adducts formation is 1:1. This implies that it is assumed that the reactive pyrrole intermediate reacts equally well with GSH and protein-SH groups (Mattocks and Jukes, 1992a; Mattocks and Jukes, 1992b). The results obtained reveal that under these conditions, the REP value calculated by Method 2 decreases upon increasing the level of GSH depletion with a concomitant reduction in k_cat. In addition, with increasing depletion of GSH and thus further reduction of the k_cat value for 7-GS-DHP adduct formation upon dosing SEN, the discrepancy between the REP values from Method 1 and 2 becomes more pronounced.

All predictions in Figures 5A,B were compared to the actual REP values derived from the in vivo data (Table 1). Figure 5B reveals a similar trend when assuming 8.20% bioavailability and a fraction bioactivated of 0.20. At 8.20% bioavailability, the REP values obtained by Method 1 and 2 are close to the in vivo values of 0.88 and 0.61 at 63% residual k_cat when dosing SEN, hence a k_cat that amounted to 0.00145 mL min^-1 mg^-1 S9 (Figure 5B).

Figure 6 further shows the PBK-simulated REP values when modifying the calculations with the fraction bioactivated from 0.125 to 0.400 at a dose level of 55 μmol kg^-1 bw. In this calculation, the amount of 7-GS-DHP was assumed to remain constant at the value obtained with a fraction bioactivated of 0.200 and with a k_cat of 0.0023 and 0.00145 ml min^-1 mg^-1 S9 for 7-GS-DHP formation when dosing respectively SENO and SEN. As the fraction bioactivated increases, more pyrrole-protein adduct formation occurs when the amount of 7-GS-DHP remains constant. At a low value of fraction bioactivated, the constant level of 7-GS-DHP formation will be higher than the pyrrole-protein adduct formation. In contrast, as the fraction bioactivated increases, more pyrrole-protein adducts are being formed from both oral SENO and SEN dosage. As a result, the REP value increases when the fraction bioactivated increases. Comparison of all the outcomes in Figure 6 reveals that, in agreement with the chosen parameters for Figure 5B, a fraction bioactivated of 0.20 results in REP values that match best with the reported in vivo data.

Next, the consequences for the REP values of modifying the administered dose were studied, in order to obtain insight in REP values not only at the relatively high dose level used in the animal experiment (55 μmol kg^-1 bw) but also at more realistic low human dietary exposure levels. These simulations were performed selecting the conditions where the REP values at 55 μmol kg^-1 bw matched the in vivo data best, including 8.20% bioavailability, 0.20 fraction bioactivated, and an in vitro k_cat for an oral SENO and SEN dose of 0.0023 ml min^-1 mg^-1 S9 and 0.00145 ml min^-1 mg^-1 S9, respectively.

To mimic the gradually lower extent of GSH depletion expected upon lowering the dose of SEN, the k_cat was linearly increased with reducing dose from 63% assumed residual k_cat at 55 μmol kg^-1 bw to 100% of the original value at 0.1 μmol kg^-1 bw. In reverse, the k_cat was linearly decreased at dose levels above 55 μmol kg^-1 bw until being 0.01% of the original value at a dose level of 200 μmol kg^-1 bw. Figure 7 shows the dose-dependent change in the REP values calculated by Method 1 (based on AUC_PA) and Method 2 (based on amount_{pyrrole-protein adducts}). Both Method 1 and 2 show PBK modelling predicted dose-dependent changes in the REP values. At low dose levels, when GSH depletion does no longer play a role, regardless of the chosen method or endpoint, the REP values predicted based on the two endpoints are similar. This is to be expected given that at low dose GSH levels and thus also the k_cat for GSH conjugation of the pyrrole adducts is unaffected.

To evaluate which parameters are influential to the outcome, normalized SCs were calculated for both the SENO and SEN model at high dose (55 μmol kg^-1 bw) in rat resulting in the data shown in Figure 8. Similar results were obtained for both models, where the parameters of influence on the model predictions for the a m o u n t p y r r o l e − p r o t e i n   a d d u c t are those involved in the SEN clearance (VmaxLM2c and KmLM2) and 7-GS-DHP formation (Lslope2c). In the SENO model, parameters other than those related to these two are also shown, albeit with significantly lower influence. It is worth noting that the parameters that are negatively related to 7-GS-DHP are positively related to pyrrole-protein adducts and vice versa.