*2.2. E*ff*ect of Polymorphisms on PAC Metabolism*

Next, we sought to examine the effect that these polymorphisms have on PAC turnover rates. Previous studies reported a poor solubility of PAC that precluded *Vmax* determination for in vitro CYP2C8 [33], and so we analyzed our data based on the time linearity. Linearity was established for the hydroxylation of 70 μM PAC to 6α-hydroxypaclitaxel (PAC-OH) over a 20 min period. We can estimate the catalytic efficiency of the PAC metabolism among the variants by fitting the data to Equation (1)

$$[S] = [S\_0] \varepsilon^{-kt} \tag{1}$$

where [*S*0] is the initial concentration of the substrate and *k* = *kcat Km* [*E*] (Figure S1) [34]. The catalytic efficiencies of these variants range from 0.207 to 1.02 min−<sup>1</sup> nM−<sup>1</sup> (Figure S1, Table 1), which are lower than previously reported rates for CYP2C8-mediated in vitro PAC metabolism using a lipidreconstituted system [33]. CYP2C8\*2 showed a marked decrease in turnover rate (47.2% WT), consistent with earlier reports of its inefficient PAC metabolism [13,14]. Compared to WT, CYP2C8\*3 had slightly, albeit not significantly, lower PAC turnover rates. These results for CYP2C8\*3 agree with the findings

of Yu et al. [14] and Soyama et al. [18]. However, they contradict the findings of Dai et al. [13], who were unable to measure PAC metabolism, and contradict the findings of Kaspera et al., who reported higher PAC metabolism [19]. The CYP2C8-K399R variant metabolized PAC with rates similar to WT, but interestingly the CYP2C8-R139K variant showed a remarkable increase in PAC turnover (265% compared to WT). Therefore, the effects of the CYP2C8\*3 single mutations do not additively contribute to the CYP2C8\*3 activity. Overall, we observed that the CYP2C8\*3 polymorphism does not significantly affect PAC turnover, but CYP2C8\*2 is half as efficient as WT. Since these polymorphisms do not occur in the active site of CYP2C8, it is unlikely that they directly affect PAC binding. Therefore, we further probed other steps of the CYP catalytic cycle to determine the mechanism through which these polymorphisms affect CYP2C8 activity.

**Table 1.** Paclitaxel (PAC) metabolism by CYP2C8 variants. Linear rates of 70 μM PAC metabolism by each CYP2C8 variant to PAC–OH was determined over a 20 min period. Estimates of the catalytic efficiencies were determined using Equation (1) as stated in the text. Error represents the SEM of three experiments.


#### *2.3. Polymorphisms in CYP2C8 Lead to Greater HOOH Uncoupling*

To assess the uncoupling efficiency of PAC metabolism, we next measured the rate of HOOH production over time by each variant. None of the variants showed a significant difference in the HOOH production rates in the presence of 70 μM PAC compared to without. This is likely due to the high amounts of HOOH produced compared to PAC turnover (Figure S2). The overall amount of HOOH linearly decreased for all CYP2C8 variants over time, indicating a burst of activity at the initiation of the reaction followed by decomposition of HOOH (Figure 3). For all time points, CYP2C8\*3 displayed nearly 200% higher HOOH production compared to WT. Therefore, the CYP2C8\*3 polymorphism leads to a greater ROS production and catalytic uncoupling. PAC and ROS have both been implicated in the pathogenesis of neuropathic pain [35,36], and therefore it would be interesting to see if the increase in HOOH production contributes to the neuropathy observed with CYP2C8\*3 [15,16]. There was an increase in HOOH production with the CYP2C8-K399R variant at the start of the reaction and a minor decrease in HOOH production over time with the CYP2C8-R139K variant. Therefore, likewise to the PAC metabolism experiments, the individual mutations of the CYP2C8\*3 do not additively contribute to the CYP2C8\*3 phenotype. CYP2C8\*2 did not have a significant difference in the HOOH production compared to WT, but this polymorphism also displayed half the PAC turnover as WT. By normalizing the amount of HOOH produced to the activity of the enzyme by looking at the HOOH produced per PAC turnover, we see that both CYP2C8\*2 and CYP2C8\*3 produce almost 200% more HOOH per PAC turnover than WT (Table 2). The CYP2C8\*2 and CYP2C8\*3 polymorphisms, therefore, are about twofold more uncoupled (i.e., produce twofold more ROS), with CYP2C8\*3 producing significantly more HOOH than WT. ROS uncoupling can be caused by altered redox kinetics during the CYP2C8 catalytic cycle or electron transfer from CPR to CYP2C8 (Figure 1). Therefore, we next proceeded to determine if these polymorphisms alter the intrinsic redox potential of the CY2C8 heme.


**Table 2.** Hydrogen peroxide (HOOH) production per PAC turnover. The amount of HOOH formed at 20 min was divided by the amount of PAC–OH produced at 20 min. These values are compared to WT.

**Figure 3.** HOOH production rates. The rate of HOOH production by each CYP2C8 variant was measured using an Amplex Red peroxidase kit at 10, 15, and 20 min reaction times, in the presence of 70 μM PAC. Error represents the SEM of 3–4 experiments. Statistical significance was determined by comparing experiments to their respective WT controls. \*\* *p* < 0.01; \*\*\* *p* = 0.0001; \*\*\*\* *p* < 0.0001.
