siRNA-mediated knockdown of P450 oxidoreductasein rats: a tool to reducemetabolism by CYPs and increase exposure of high clearance compounds

-Supplementary Material-

Rob S. Burke, Inthirai Somasuntharam, Paul Rearden, Duncan Brown, Sujal V. Deshmukh,

Martha A. DiPietro, Jillian DiMuzio, Roy Eisenhandler, Scott E. Fauty, Christopher Gibson,

Marian E. Gindy, Kelly A. Hamilton, Ian Knemeyer, Kenneth A. Koeplinger, Hae Won Kwon,

Traci Q. Lifsted, Karsten Menzel, Mihir Patel, Nicole Pudvah, Deanne Jackson Rudd,

Jessica Seitzer, Walter R. Strapps, Thomayant Prueksaritanont, Charles D. Thompson,

Jerome H. Hochman, Brian A. Carr

PK model development

According to Pang and Rowland, the well-stirred model of hepatic clearance is a good fit in situations where the drug distribution equilibrium is achieved so rapidly that the unbound drug concentrations in the emergent venous blood and liver are equal(27). The well-stirred model is characterized by:

1

(4)

1

where ERH is the hepatic extraction ratio, fb is the fraction of unbound drug in the blood (i.e. the ratio of unbound drug concentration in plasma to whole blood drug concentration), CLint,u is the unbound hepatic intrinsic clearance of drug, and QH is the hepatic blood flow. Instead of a single well-stirred compartment, the parallel tube model assumes the liver is composed of identical cylindrical tubes and is a good fit in circumstances where there is a concentration gradient of drug down the length of the tube (27). In the situation where the unbound plasma concentration of drug (Cu) is much less than the Km of the CYP-mediated hydroxylation reaction that accounts for loss of parent drug, the equation for the parallel tube model reduces to:

1

(5)

1

In both models, the following definitions hold true:

1

(6)

1

(7)

1

where FH is the fraction of drug that passes through the liver without being metabolized and CLb,h is the steady-state hepatic blood clearance of drug. Assuming the unbound concentration of drug in blood cells is equal to the unbound drug concentration in plasma(28), the unbound fraction of drug in blood (fb) can be determined by:

1

(8)

1

whereCb is the whole blood drug concentration, Cp is the plasma drug concentration, fu,p is the fraction of drug unbound in plasma defined as fu,p = Cu / Cp, and (B/P) is the blood-to-plasma ratio defined as (B/P) = Cb / Cp. In a similar fashion, the total blood clearance (CLb) can be determined from in vivoplasma clearance following an intravenous (i.v.) dose:

1

(9)

1

where CLp is the total plasma drug clearance calculated using the dose and the area under the plasma drug concentration-time curve (AUC) and is defined as CLp = dose / AUCp,iv. Additionally, when intrinsic clearance is measured in liver microsomesin vitro, it must be corrected for any effect of binding to microsomal protein to determine the unbound intrinsic clearance as follows:

1

(10)

1

where CLint,app is the apparent intrinsic clearance determined in vitro by eq. 3 and fu,mic is the fraction of drug unbound in the microsomal incubations (29). The above equations focused on clearance are predominantly useful following i.v. administration when the total drug dose can be guaranteed to have access to the systemic blood compartment. However, when drugs are dosed orally (p.o.) it becomes important to understand the bioavailability (F) of the dose (i.e. the fraction of the dose that actually enters systemic blood circulation). Bioavailability can be experimentally determined as the ratio of the dose-normalized oral plasma AUC to the dose-normalized i.v.plasma AUC as follows:

1

(11)

1

When assuming that the liver and the gut are the only sites of drug metabolism, the bioavailability following an oral dose can bedefined as:

1

(12)

1

where FA is the fraction of drug absorbed into the gut following an oral dose and FG is the fraction of the dose that passes through the gut wall without being metabolized (similar to FH defined above). The total area under the blood drug concentration-time curve following a single oral dose (AUCb,oral) can then be calculated from the oral plasma AUC as follows:

1

(13)

1

where AUCp,oral is the experimentally determined oral plasma AUCused in eq. 11. The oral blood AUC is also defined by:

1

(14)

1

Changes to midazolam PK following POR knockdown

Using the parallel tube model to describe i.v. blood clearance, upon substituting eq. 5 into eq. 7 we find:

1

(15)

1

Furthermore, for PBS-treated rats the intrinsic clearance of midazolam is unchanged and equal to CLint,u, while for POR-treated rats there is a 9-fold reduction in the intrinsic clearance of midazolam such that CLint,u,POR = CLint,u / 9. Determining the ratio of blood clearances, we find:

1

(16)

1

Since the limit of the natural exponential function is zero as CLint,u goes to infinity, it is clear from eq. 15 that CLb,h approaches QH at high values of CLint,u for both PBS-treated rats as well as POR-treated rats. From eq. 16, it is clear that the ratio of the blood clearances then approaches 1 at high values of CLint,u. In contrast, using the parallel tube model to describe the bioavailability of a p.o. dose, upon substituting eq. 5 into eq. 6 and then eq. 6 into eq. 12, as well as using FA = FG = 0.95 as described below, we find:

1

(17)

1

Using the same treatment of the 9-fold reduction in midazolam CLint,u,POR described above as well as the mathematical definition that ex / ey = e(x-y), we find:

1

(18)

1

Thus, the ratio of the bioavailability in rats with POR knockdown to the bioavailability in PBS-treated rats increases exponentially with increasing CLint,u.

Table IV. Experimentally measured and calculated parameters for midazolam used as model inputs.

Parameter values are reported as mean ± S.E.M. for those that were experimentally measured. The values with no reported error were either estimated or calculated using the mean value of the experimentally determined parameters. A “-” in the units column denotes a dimensionless parameter.

Table V. Parameters determined for midazolam using two models of liver clearance.

The parameter values were calculated using the mean value of the experimentally determined parameters. A “-” in the units column denotes a dimensionless parameter.

PK model discussion

In order to determine theunbound intrinsic clearance (CLint,u) using eq. 10, the fraction of unbound drug in the microsomal incubations (fu,mic) was determined. For midazolam, it was found that fu,mic = 0.89 ± 0.03 in a microsomal incubation with 0.5 mg/mL of microsomal protein. Similar values can be found for the fu,mic of midazolam in the literature and it is also known that fu,mic increases as the microsomal protein concentration decreases due to the presence of fewer protein binding sites (29, 30). At the microsomal protein concentration of 0.1 mg/mL used in the in vitro incubations in this work the fu,mic should be close to unity, so an estimated value of 0.95 was used to account for the possibility of a small amount of protein binding (Table IV).

Since the intrinsic clearance represents the maximum rate of midazolam metabolism in liver independent of other factors (e.g. hepatic blood flow), any decrease in the intrinsic clearance would lead to lower rates of hepatic metabolism and manifest as a lower hepatic extraction ratio (ERH). The fraction of drug that passes through the liver unmetabolized (FH) is defined by eq. 6 such that any decrease in ERH directly leads to an increase in FH. The total bioavailability (F) defined by eq. 12 is directly proportional to FH, so any increase in FH directly increases F as well. It was assumed here that POR knockdown would have no impact on the amount of midazolam absorbed through the gut following an oral dose and that metabolism of midazolam only occurs in liver and gut tissue. It is clear from the literature that there is significant intestinal metabolism of midazolam following an oral dose in humans (31-33); however, it is known for rats that intestinal metabolism is minimal and that FA and FG for an oral dose of midazolam are close to unity (34-36). In this work, there was no POR knockdown detected in the mucosal layers of the small intestine (Figure 7), which means that the CYP enzymes in the enterocytes should retain their full levels of activity and leads to the assumption that FG is not affected by anti-POR siRNA treatment. Furthermore, when midazolam was incubated with intestinal microsomes, no metabolism was observed even when high microsomal protein concentrations were used. Taken together, the results presented here are consistent with minimal contributions to overall bioavailability from FA and FG such that changes in F will be dominated by variations in FH. Thus, both FA and FG were estimated to be 0.95 to account for nearly complete absorbance and only slight metabolism in the gut (Table IV).

To understand how the reduced midazolam intrinsic clearance measured in vitro following POR knockdown would impact PK parametersin vivo, midazolam was dosed i.v. and p.o. two weeks after administration of LNPs or PBS. Various PK parameters were measured and other PK parameters were calculated using pharmacokinetic modeling (Tables IV and V). Two models (the well-stirred model and the parallel tube model) were used to describe the function of the liver with respect to drug metabolism and clearance (27). In order to use eq. 5 to describe the parallel tube model, the conditions that Cu « Km and that fb remains constant must be met (27). Midazolam is metabolized by CYP enzymes (primarily of the CYP3A family) into both 1-hydroxymidazolam and 4-hydroxymidazolam, with the 1-hydroxymidazolam formation reaction having a Km ~ 4 μM and the 4-hydroxymidazolam formation reaction having a Km approximately 10-fold higher in several species including rats (37-40). The fraction unbound in plasma (fu,p) was measured to be 0.04 for midazolam (Table IV) and is consistent with other reported values (36, 41). Since Cu = Cp·fu,p and the maximum Cp (from Figure 6) is approximately 1 μM, the maximum value of Cu is ~0.04 μM. Thus, even using the lowest Km of 4 μM the maximum value of Cu is 100-fold lower than the Km, which satisfies the Cu « Km constraint. This constraint also provides that fb can be assumed to remain constant over the length of the tube by preventing a high enough drug concentration to saturate protein binding sites, which means both conditions are satisfied and eq. 5 is a reasonable representation of the parallel tube model.

As for liver blood flow rate (QH), a value of 70 mL/min/kg was used in the models (Table IV). This value of QH has been used in other reports (41), but it is slightly lower than the 84 mL/min/kg value that is typically used for QH in rats. During analysis of the models, it was determined that the lower value of QH seemed to fit the data more accurately. It is interesting to note that Uhing et al. have shown that anesthesia with ketamine and xylazine can slow down liver blood flow for several days post-surgery, which can have profound effects on the hepatic clearance of drugs (35). For the animals that were pretreated with LNPs or PBS prior to dosing of midazolam (Figure 6 and Tables IIIIV), jugular vein cannula surgery was performed in-house using ketamine and xylazine less than 24 h before dosing midazolam. In contrast, the animals that were dosed with midazolam in the dose-response experiment (Figure 5 and Table II) were pre-cannulated by the vendor approximately one week before dosing with midazolam. The plasma clearance of midazolam in the pre-cannulated animals averaged ~60 mL/min/kg (Table II) while the plasma clearance of midazolam in the in-house cannulated animals (pretreated with PBS or scramble LNP) averaged ~50 mL/min/kg (Table III), so it could be possible that the anesthesia slowed down the liver blood flow and reduced hepatic clearance of midazolam. Since data from the in-house cannulated animals pretreated with LNP or PBS were also used in the PK modeling (Table IVV), the use of 70 mL/min/kg for QH was deemed appropriate. Even if a higher value of QH is used in the analysis, the absolute values of calculated parameters may change, but the trends and overall conclusions remain the same.

The key findings from the in vivo studies were that POR knockdown leads to a 25% decrease in plasma clearance that drives a less than 2-fold increase in the plasma AUC following i.v. administration of midazolam, while POR knockdown leads to a 14-fold increase in bioavailability that drives a 20-fold increase in plasma AUC following p.o. administration of midazolam. The large difference in response to POR knockdown depending on the route of administration for midazolam was explored using the two PK models described above. In order to calculate the total blood clearance (CLb) from the total plasma clearance (CLp) using eq. 9, the blood-to-plasma ratio (B/P) must be determined. The (B/P) for midazolam was measured to be 0.81, which is consistent with other reported values (42). The CLb values (Table IV) were then compared to the values for hepatic blood clearance (CLb,h) calculated using the two PK models (Table V). The parallel tube model produced CLb,h values that more closely matched the CLb values than those derived from the well-stirred model, which is not surprising because midazolam is a high extraction ratio drug (ERH > 0.7) and the parallel tube model generally predicts behavior more accurately for high ERH drugs. In this work, it was assumed that the vast majority of the clearance of midazolam is caused by hepatic metabolism, so CLb,h should be nearly equal toCLb. The fact that both models generated values of CLb,h that were significantly lower than CLb for the animals with POR knockdown highlights a deficiency in the models. The bioavailability for midazolam in untreated and PBS-treated animals was determined to be F = 0.02, which is consistent with the literature (41). POR knockdown led to a significant increase in midazolam bioavailability, and both models predicted that F would increase upon POR knockdown. However, the absolute values of the bioavailability and the increase in bioavailability following POR knockdown compared to PBS treatment were not accurately predicted by the models. Similarly, the models predicted increased AUCb,oral following POR knockdown, but again the absolute values and the magnitude of the predicted changes were slightly different from experimental measurement.

There are a few explanations for the discrepancies between the models and the observations, including the assumptions regarding QH mentioned above as well as the possibility that the CLint,u measured in vitro underestimates the actual CLint,u of the enzymes in vivo (the models assume that these values are equivalent). For both models, the CLb,h asymptotically approaches liver blood flow with increasing CLint,u, which means that large changes in CLint,u only lead to small changes in CLb for i.v. doses of high clearance compounds (parallel tube model shown in Figure 8). Thus, if the CLint,u for the hepatic CYP enzymes in vivo were actually higher than those measured in vitro, the same fold reduction in CLint,uwould result in a smaller impact on CLb. Since the CLint,u plays a dominant role in determining FH, any underestimation of CLint,u would also lead to incorrect predictions of F and AUCb,oral. Despite the deficiencies of the models, the overall trends from the data are predicted by the parallel tube model: POR knockdown reduces CYP-mediated metabolism and leads to a reduction in ERH, which leads to an increase in FH. The increase in FH drives only small reductions in blood clearance following an i.v. dose of midazolam, but generates large increases in oral bioavailability and blood AUC following a p.o. dose of midazolam.

REFERENCES

27.K.S. Pang and M. Rowland. Hepatic clearance of drugs. I. Theoretical considerations of a "well-stirred" model and a "parallel tube" model. Influence of hepatic blood flow, plasma and blood cell binding, and the hepatocellular enzymatic activity on hepatic drug clearance. Journal of pharmacokinetics and biopharmaceutics. 5:625-653 (1977).

28.J. Yang, M. Jamei, K.R. Yeo, A. Rostami-Hodjegan, and G.T. Tucker. Misuse of the well-stirred model of hepatic drug clearance. Drug metabolism and disposition: the biological fate of chemicals. 35:501-502 (2007).

29.C. Giuliano, M. Jairaj, C.M. Zafiu, and R. Laufer. Direct determination of unbound intrinsic drug clearance in the microsomal stability assay. Drug metabolism and disposition: the biological fate of chemicals. 33:1319-1324 (2005).

30.R.S. Obach. Prediction of human clearance of twenty-nine drugs from hepatic microsomal intrinsic clearance data: An examination of in vitro half-life approach and nonspecific binding to microsomes. Drug Metabolism and Disposition. 27:1350-1359 (1999).

31.K.E. Thummel, D. O'Shea, M.F. Paine, D.D. Shen, K.L. Kunze, J.D. Perkins, and G.R. Wilkinson. Oral first-pass elimination of midazolam involves both gastrointestinal and hepatic CYP3A-mediated metabolism. Clinical pharmacology and therapeutics. 59:491-502 (1996).

32.M.F. Paine, D.D. Shen, K.L. Kunze, J.D. Perkins, C.L. Marsh, J.P. McVicar, D.M. Barr, B.S. Gillies, and K.E. Thummel. First-pass metabolism of midazolam by the human intestine. Clinical pharmacology and therapeutics. 60:14-24 (1996).

33.M.F. Paine, M. Khalighi, J.M. Fisher, D.D. Shen, K.L. Kunze, C.L. Marsh, J.D. Perkins, and K.E. Thummel. Characterization of interintestinal and intraintestinal variations in human CYP3A-dependent metabolism. The Journal of pharmacology and experimental therapeutics. 283:1552-1562 (1997).

34.F. Higashikawa, T. Murakami, T. Kaneda, A. Kato, and M. Takano. Dose-dependent intestinal and hepatic first-pass metabolism of midazolam, a cytochrome P450 3A substrate with differently modulated enzyme activity in rats. The Journal of pharmacy and pharmacology. 51:67-72 (1999).

35.M.R. Uhing, D.W. Beno, V.A. Jiyamapa-Serna, Y. Chen, R.E. Galinsky, S.D. Hall, and R.E. Kimura. The effect of anesthesia and surgery on CYP3A activity in rats. Drug metabolism and disposition: the biological fate of chemicals. 32:1325-1330 (2004).

36.T. Kotegawa, B.E. Laurijssens, L.L. Von Moltke, M.M. Cotreau, M.D. Perloff, K. Venkatakrishnan, J.S. Warrington, B.W. Granda, J.S. Harmatz, and D.J. Greenblatt. In vitro, pharmacokinetic, and pharmacodynamic interactions of ketoconazole and midazolam in the rat. The Journal of pharmacology and experimental therapeutics. 302:1228-1237 (2002).

37.F. Higashikawa, T. Murakami, T. Kaneda, and M. Takano. In-vivo and in-vitro metabolic clearance of midazolam, a cytochrome P450 3A substrate, by the liver under normal and increased enzyme activity in rats. The Journal of pharmacy and pharmacology. 51:405-410 (1999).