(A) Overview of Cynomolgus Monkey Transporter Models

Monkeys are established as a nonclinical species for testing PK, PD, and safety, along with rodents, dogs, and other specialized species (Tang and Prueksaritanont, 2010; Deng et al., 2011; Uno et al., 2011). Monkeys are used because they possess many genetic, physiologic, biochemical, and pharmacological similarities to humans. It is generally presumed that the similarities make the use of monkeys in nonclinical research a considered decision (Table 1). To date, only a few monkey species, such as macaques of the Cercopithecidae family, are available and established as translational models for drug research. These primates share a common ancestor with humans, which is reported to have lived approximately more than 30 million years ago (Perelman et al., 2011). From this family, the species Macaca fascicularis and Macaca mulatta, also known as cynomolgus and rhesus monkeys, respectively, are the most common and best-studied nonhuman primate animal models today (Ebeling et al., 2011). However, usage of rhesus monkeys in drug testing has declined since India banned all rhesus monkey exports to breeding centers across the world. As an alternative to the rhesus, several commercial breeding centers in Indonesia, China, the Philippines, and Mauritius tried to provide a sufficient number of captive-bred cynomolgus monkeys originating from wild-trapped founders. The natural range of cynomolgus monkeys spans the mainland of southern Asia, Indonesia, the Philippines, and more recently Mauritius, where a small number of founder animals was imported on a trading ship during the 15th century (Ferguson et al., 2007; Ebeling et al., 2011). Therefore, the cynomolgus monkey is currently the most widely used nonhuman primate species used for drug PK, PD, and safety testing at pharmaceutical companies.

Over the last decade, the use of cynomolgus monkeys as a surrogate model for transporter-mediated DDI studies has been under intense investigations because it is presumed that the monkey is the most relevant animal species in which the transport of a drug candidate can be studied, with results extrapolatable to humans. Investigators evaluated how cynomolgus monkeys compare with humans in terms of drug transporter composition, amino acid sequence, tissue expression, substrate specificity, inhibitor potency, and substrate–inhibitor interaction. It is generally presumed that in vivo approaches using monkeys reflect complex, integrated, and dynamic processes in humans and can shed mechanistic insight into drug absorption and disposition beyond what are offered using in vitro models. These animal models can bridge the knowledge gap and enable the extrapolation of in vitro data to clinical outcomes (Fig. 1). Furthermore, monkeys have an advantage due to the availability of tissue samples, enabling the measurement of tissue drug concentrations and transporter expression. Bile duct cannulated (BDC) monkeys provide a useful tool to examine biliary excretion. In addition, the recovery of the parent drug in urine can support the determination of renal clearance (CL_r). These methods yield information regarding the excretion processes that transporters are heavily involved in. Determination of systemic, renal, biliary, and tissue clearance can be combined with a transporter modulator to ask a specific transporter-related question. As discussed in the following sections, there are examples of published reports that deployed rifampin and cyclosporin A to inhibit liver organic anion transporting polypeptide (OATP) 1B1 and OATP1B3 in vivo (Shen et al., 2013; Chu et al., 2015; Thakare et al., 2017; Ufuk et al., 2018; Gu et al., 2020; Ogawa et al., 2020a; Eng et al., 2021). Additionally, probenecid, pyrimethamine, elacridar, and curcumin are often co-administered to determine the effects on the inhibition of cynomolgus monkey renal organic anion transporters (OAT) 1 and OAT3 (Tahara et al., 2005; Tahara et al., 2006; Kosa et al., 2018; Shen et al., 2018b), organic cation transporter (OCT) 2, and multidrug and toxin extrusion (MATE) 1 and MATE2-K (Shen et al., 2016b), as well as P-glycoprotein (P-gp) and breast cancer resistance protein (BCRP), respectively (Karibe et al., 2015; Karibe et al., 2018; Kosa et al., 2018). Furthermore, monkeys share great similarities with humans in terms of the nuclear receptors that regulate transporter expression such as the pregnane X receptor (PXR) (Kim et al., 2010). Not surprisingly, rifampin has been used as a PXR agonist in monkey transporter induction research (Niu et al., 2019; Zhang et al., 2020a). Besides in vivo inhibition, the cynomolgus monkey transporter tool kit has expanded to include individual transporter transfected cell lines, recombinant transporter membrane vesicles, primary cells, and sandwich cultures of monkey hepatocytes, as well as in vivo drug probes, endogenous biomarkers, and inhibitors. Such reagents and models can facilitate drug transporter research. There are also examples of publications describing the use of cynomolgus monkey data for rationalizing the scaling factors used in physiologically based pharmacokinetic (PBPK) models (Shida et al., 2015; Morse et al., 2017; Liang and Lai, 2021). Based on such collective data, it can be concluded that considerable advancements have been made characterizing and validating cynomolgus monkey transporter models. Moreover, it is apparent that the cynomolgus monkey is valuable in supporting transporter DDI assessment.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Schematic diagram of translational research using cynomolgus monkey transporter models. Different monkey models are used as nonclinical models of transporter-mediated drug disposition and interaction investigations. PDA, pyridoxic acid.

(B) Cynomolgus Monkey Shares a High Degree of Isoform Composition and Amino Acid Sequence Homology with their Human Orthologs

Compared with mice, rats, or dogs, cynomolgus monkeys have a closer evolutionary relationship to humans. They exhibit greater isoform composition and amino acid sequence homology of transporters than other commonly used nonclinical species. A cornerstone of nonhuman primate research with high impact for drug testing was the publication of the draft whole genome of the cynomolgus monkey, Macaca fascicularis, using the shotgun sequencing approach in 2011 (Ebeling et al., 2011). The main objective of this multicenter investigation was to improve the predictive power of the nonhuman primate models for humans. This investigation mainly focused on identifying 17,387 orthologs of human protein-coding genes in the cynomolgus monkey draft genome. Additionally, this genome facilitates the use of cynomolgus monkeys as a surrogate model for evaluating drug metabolism and conducting transport phenotyping. For example, novel 18 cynomolgus monkey cytochrome P450 (CYP) enzymes were identified in the genome sequence of cynomolgus monkeys, and their amino acid sequences identical to human orthologs were between 94% and 99% (Ebeling et al., 2011).

As expected, cynomolgus monkey exhibits excellent agreement with the human in transporter protein sequence homology (Table 2). Major cynomolgus monkey drug transporter genes of the SLCO1B, ABCB, ABCC, ABCG, SLC22A, and SLC47A subfamilies have been cloned by different groups recently. They include OATP1B1, OATP1B3, OATP2B1, P-gp, multidrug resistance-associated protein 2 (MRP2), BCRP, OCT1, OCT2, OAT1, OAT2, OAT3, MATE1, and MATE2-K (Tahara et al., 2005; Yasunaga et al., 2008; Shen et al., 2015; Shen et al., 2016b; Kosa et al., 2018). These drug transporters in monkeys display a high degree of similarity (92–97%) in amino acid sequences with their human counterparts (Table 2). For example, we cloned the cDNAs of cynomolgus monkey OATP1B1, OATP1B3, and OATP2B1 using the reverse transcription-polymerase chain reaction (RT-PCR) approach with primers derived from the human genes (Shen et al., 2013). The OATP1B1, OATP1B3 and OATP2B1 transporter amino acid sequences between cynomolgus monkeys and humans were 91.9%, 93.5%, and 96.9% identical, respectively (Shen et al., 2013). In contrast, the mouse, rat and dog transporters showed poorer homology compared with humans, ranging from 60% to 70% (Table 2). More importantly, the OATP1B subfamily consists of two highly conserved proteins in both humans and cynomolgus monkeys but not in other species. OATP1B1 and OATP1B3 recognize acidic drugs such as statins as substrates. In contrast, Oatp1b2 and Oatp1b4 are the only ortholog in rodents and dogs, respectively (Hagenbuch and Gui, 2008, Gui and Hagenbuch, 2010). Murine Oatp1b2 and canine Oatp1b4 show 64% and 69% identity of amino acid sequence compared with human OATP1B1, respectively (Table 2). Moreover, although the ortholog to human MATE2-K transcript (NM_001099646) was identified in cynomolgus monkeys (Shen et al., 2016b), no human MATE2-K counterpart has been identified in rodents (Table 2). Identification of the cynomolgus monkey organic cation transporter sequences demonstrate that these mRNAs show very high sequence identities between cynomolgus monkeys and humans, with 99.5%, 99.8%, and 99.6% for OCT2, MATE1, and MATE2-K, respectively (Table 2) (Shen et al., 2016b). Additionally, a high degree of similarity in amino acid sequences was observed in a cynomolgus monkey OAT1 and OAT3 model, i.e., 97% and 96% identical to human OAT1 and OAT3, respectively (Tahara et al., 2005). Similar profiles [92% to 97% sequence homology (Table 2)] were documented in a cynomolgus monkey efflux transporter model for P-gp (Kosa et al., 2018), BCRP (Kosa et al., 2018), MRP2 (Yasunaga et al., 2008), OCT1 (in-house data), and OAT2 (in-house data). In conclusion, the orthologous transporter proteins are extremely similar, with the amino acid sequence of the major drug transporter proteins commonly differing by less than 10% between humans and cynomolgus monkeys.

(C) Cynomolgus Monkey Transporter Expression Compared with Humans

Transporter activity for a drug is determined by drug affinity and transporter abundance (or expression). Differences in the transporter activity between species can be due to differences in drug affinity and/or transporter expression. Therefore, understanding transporter expression in cynomolgus monkeys and the difference versus humans is a prerequisite to extrapolate monkey data to humans.

Over the years, knowledge on cross-species analysis of transport protein expression has increased. Major studies of interspecies similarities and differences in the transporter expression have been compared by quantifying the expression of transporter mRNA and protein using RT-PCR and immunoblotting, respectively. However, the mRNA expression of transporters is not always correlated with protein abundance of transporters, in particular OATP1B1 and OATP1B3 proteins (Ohtsuki et al., 2012; Prasad et al., 2013; Couto et al., 2020). Furthermore, immunoblotting, an antibody-based quantification assay, has the disadvantage that the resultant band intensities depend on the antibody binding affinity. As a result, it does not allow the comparison of the transporter expression across species due to different antibody binding affinities and lack of cross-species standards (Li et al., 2009b; Hammer et al., 2021). Recent protein quantification by liquid chromatography–tandem mass spectrometry has revealed interspecies differences in the absolute abundance of transporters of the liver, kidney, and other tissues in humans, cynomolgus monkeys, beagle dogs, and Sprague-Dawley rats (Fig. 2) (Li et al., 2009a; Li et al., 2009b; Qiu et al., 2013; Wang et al., 2015; Nakamura et al., 2016; Hammer et al., 2021; Morse et al., 2021). In addition, the expression of various hepatobiliary transporters in cryopreserved hepatocytes is similar to those of the liver tissue. Cross-species comparison by proteomics of the liver indicates that the absolute protein abundance of the transporters of preclinical species is generally greater than that of humans (Fig. 2A). Although species difference may exist, the monkey is most similar to human in terms of transporter abundance. Takahashi et al. determined the cross-species expression of efflux transporters in the liver and intestine by real-time RT-PCR using glyceraldehyde 3-phosphate dehydrogenase as an endogenous calibration gene (Takahashi et al., 2008). The expression levels of mRNA of P-gp, BCRP, and MRP2 were significantly higher in cynomolgus monkeys than those in humans. Among hepatobiliary transporters, OATPs are the most abundant transporter across species. The absolute protein abundance of OATPs in the liver tissue of cynomolgus monkeys and dogs has been determined to be 5.7- and 3.3-fold higher than that of humans (27.4 and 15.7 vs. 4.8 fmol/µg membrane protein) (Wang et al., 2015). The next most abundant hepatobiliary transporter in humans and cynomolgus monkeys was OCT1, accounting for 32% and 27% of the total abundance of transporters expressed in cynomolgus monkey and human liver, respectively, whereas Mrp2 and Na⁺-taurocholate co-transporting polypeptide (Ntcp) in dogs and rats, respectively, and Oct1 accounted for only 2% of total hepatobiliary transporters in the liver of dogs (Fig. 2A) (Wang et al., 2015; Morse et al., 2021). Li et al. reported that the amount of Mrp2 protein in the monkey liver was comparable to that in humans but was approximately 10-fold lower than that in rats (Li et al., 2009b). The abundance of BCRP protein in liver tissues among species was ranked in an order of dog > rat > monkey ≈ human (Li et al., 2009a), but BCRP has been detected in the liver tissue from humans and dogs but not from monkeys and rats in a different study (Wang et al., 2015). Similarly, the hepatobiliary OATP2B1 protein level of cynomolgus monkeys is less than that of humans (0.30 ± 0.10 vs. 1.7 ± 0.6 fmol/μg membrane protein) (Wang et al., 2015).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Summary of available published cross-species uptake and efflux transporter expression data in the liver (A and B) and kidneys (C and D). Mean expression data of transporters in the tissues of humans (blue box), cynomolgus monkeys (red box), Sprague-Dawley rats (green box), beagle dogs (brown box), and mice (yellow box) were obtained from Wang et al., 2015, and Basit et al., 2019. Boxes represent the ranges of transporter protein expressions.

Basit et al. (2019) analyzed cross-species kidney cortical transporter protein abundances in mice, rats, dogs, monkeys, and humans using a targeted proteomics approach (Basit et al., 2019). Generally, higher protein expression of all 19 transporters was found in nonclinical species compared with humans except for P-gp, OCT3, and organic cation/carnitine transporter 1 (OCTN1) (Fig. 2B). However, in cynomolgus monkeys, the total abundance of 12 transporters for which proteomics data were available was approximately threefold compared with that in humans (Basit et al., 2019). For example, the abundance of human P-gp was similar to that of monkeys, dogs, and rats, but it was 2.5-fold higher than mouse P-gp. Some of these transporters, such as OAT1, OAT3, OCT2, MATE1, and P-gp, are clinically relevant. These results suggest that cynomolgus monkeys may be a suitable model to support renal transporter research and serve as an in vivo bridge to humans.

Despite an increase in literature data, important knowledge gaps are identified while compiling these proteomics data. One of the reasons is the scarcity of tissue, especially human tissues. Transporter protein expression data in the gastrointestinal tract and brain are limited (Drozdzik et al., 2014; Harwood et al., 2015; Akazawa et al., 2018; Billington et al., 2019; Harwood et al., 2019; Storelli et al., 2021). The absolute abundance of transporter and UDP-glucuronosyltransferase proteins in the small intestine of cynomolgus monkeys were determined by using by liquid chromatography–mass spectrometry and were then compared with the corresponding levels in humans (Akazawa et al., 2018). The expression levels of P-gp and BCRP in the jejunum and ileum in monkeys has been determined to be comparable to that of humans. Of note, BCRP displays the most abundant expression among apical drug efflux transporters in monkeys and humans (Akazawa et al., 2018). Ito et al. measured the absolute abundance of P-gp and other drug transporter proteins at the BBB of cynomolgus monkeys using targeted proteomics (Ito et al., 2011). They discovered that the expression of the various transporters examined in cynomolgus monkeys was similar to those reported for humans in an earlier study (Uchida et al., 2011). For example, a 1.3-fold difference in the P-gp protein expression at the BBB between monkeys and humans was observed. In contrast, the expression level of P-gp in humans was 2.3-fold smaller than that in mice (Uchida et al., 2011). Pronounced species differences were observed in the brain concentrations and brain-to-plasma concentration ratios of [¹¹C]verapamul, [¹⁸F]altanserin, and [¹¹C]GR205171, known P-gp substrates, with higher brain distribution in humans and monkeys than in rats and guinea pigs (Syvanen et al., 2009). It is arguable whether a rat model is a suitable nonclinical model for the prediction of drug brain distribution in humans. Additionally, proteomics-based extrapolation from the in vitro transporter data to the in vivo brain distribution was established for P-gp substrates in cynomolgus monkeys (Uchida et al., 2014).

(D) Cynomolgus Monkey Transporters Display Similar Transport Kinetics and Inhibition Potency Compared with the Corresponding Human Forms

To allow for a meaningful translation of results from cynomolgus monkey studies to clinical outcomes, it is, therefore, of utmost importance to know about the specific transport properties of cynomolgus monkeys compared with humans in terms of transport kinetics and inhibition potency. Recently, the kinetics of cynomolgus monkey transporters have been increasingly evaluated using transfected cell lines and hepatocytes. For example, Shen et al. evaluated substrate specificity between cynomolgus monkey and human hepatic OATP transporters using stably transfected human embryonic kidney (HEK) 293 cells that individually overexpress cynomolgus monkey and human OATP1B1, OATP1B3, and OATP2B1 (Shen et al., 2013). It was revealed that cynomolgus monkey and human hepatic OATPs were qualitatively similar in terms of transport rates of the three chosen substrates (estradiol-17β-glucuronide, cholecystokinin-8, and estrone-3-sulfate). Subsequently, the uptake of three human OATP1B1 and OATP1B3 substrates (pitavastatin, pravastatin, and rosuvastatin) and one selective OATP1B3 substrate (telmisartan) was investigated using transfected cell lines (Takahashi et al., 2019). Pitavastatin, pravastatin, and rosuvastatin were demonstrated to be substrates of monkey OATP1B1 and OATP1B3, and telmisartan was suggested to be a substrate of monkey OATP1B3, in a manner similar to human OATP1B. Good agreement between the transport of prototypical substrates by cynomolgus monkey and human OATP1B transporters was observed by other laboratories (Chu et al., 2015; Takahashi et al., 2016). Two studies in both cynomolgus monkey and human hepatocytes investigated the uptake of 9 and 16 OATP1B substrates, respectively (De Bruyn et al., 2018, Eng et al., 2021). Total uptake clearance and passive diffusion were determined in vitro from initial uptake rates in the absence and presence of the OATP inhibitor rifamycin SV. The study suggested that total uptake clearance values in hepatocytes ranged over three orders of magnitude in both species, with a similar rank order and good agreement in the relative contribution of active transport to the total uptake between cynomolgus monkeys and humans. Similar findings were reported by several groups (Shen et al., 2013, Ufuk et al., 2018), who observed comparable Michaelis-Menten constant (K_m) values of OATP1B substrates in hepatocytes between cynomolgus monkeys and humans (Table 3).

To explore the substrate specificity between monkey and human renal organic Cation transporters, cynomolgus monkey OCT2, MATE1 and MATE2-K were independently expressed in HEK-293 cells (Shen et al., 2016b). Based on the results obtained with the six chosen substrates and nonsubstrates (metformin, 1-methyl-4-phenylpyridinium, tetraethylammonium, cimetidine, estrone-3-sulfate, and methotrexate), it is concluded that cynomolgus monkey and human hepatic organic cation transporters are qualitatively similar. In addition, the pH-dependence experiments were performed with the buffer at any pH between 5 and 11. The transport of metformin mediated by the monkey and human MATE1 was comparable at various extracellular pH values. Both cynomolgus monkey and human MATE1-mediated uptake increased with increasing extracellular pH, reaching a maximum of pH 9, and decreased when pH was raised from 9 to 11 (Shen et al., 2016b). MATE2-K-mediated metformin uptake also demonstrated comparable pH dependence between monkeys and humans. In contrast, both monkey OCT2- and human OCT2-mediated uptake appeared considerably less sensitive to the increase of pH. Furthermore, available published cross-species K_m values for 20 transporter substrates, such as rosuvastatin (OATP1B1, OATP1B3, OATP2B1, and NTCP) (Shen et al., 2013; Shen et al., 2015; Ufuk et al., 2018; Takahashi et al., 2019), p-aminohippurate (OAT1), benzylpenicillin (OAT3) (Tahara et al., 2005; Tahara et al., 2006), and metformin (OCT2, MATE1, and MATE2-K) (Shen et al., 2016b), showed good agreement within a threefold difference between cynomolgus monkey and human transporters (Table 3 and Fig. 3A).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Comparison of available published K_m (A and B) and IC₅₀ values (C and D) of reference substrates and inhibitors between cynomolgus monkey and human transporters. Summary of the data presented in Table 3 (K_m values of 20 substrates) and Table 4 (Inhibition constant_i and IC₅₀ values of 72 inhibitors). The solid and dotted lines represent the line of unity and the 33–300% range of the observed value, respectively.

Inhibitor potency between monkey and human drug transporters has also been compared using single transfected SLC in HEK-293 cells or using membrane vesicles expressing ABC transporters. For example, more than 20 known transporter inhibitors were chosen, which rendered a wide range of IC₅₀ values (Table 4) (Shen et al., 2013; Takahashi et al., 2013; Chu et al., 2015; Shen et al., 2015; Shen et al., 2016b; Takahashi et al., 2016; Kosa et al., 2018; Zhang et al., 2019). All except for a few compounds inhibited both cynomolgus monkey and human transporters almost equally, and their IC₅₀s were in good agreement for each transporter (less than threefold difference) (Fig. 3B). For example, imipramine was a less potent inhibitor of monkey OCT2 than human OCT2 (IC₅₀ of 12.3 and 2.1 µM, respectively) (Shen et al., 2016b). Rifampin was a less potent inhibitor of monkey BCRP compared with human BCRP (IC₅₀ of 79 and 14 µM, respectively) (Kosa et al., 2018), but KO143 is a more potent monkey BCRP inhibitor than human BCRP (IC₅₀ of 0.19 and 1 µM, respectively). Taken together, these results suggest that cynomolgus monkey transporters often present transport characteristics similar to those of human transporters.

(E) Effects of In Vivo Inhibitors on Transporter Drug Probe PK in Cynomolgus Monkeys

To extend our findings from in vitro to in vivo, many transporter substrate-inhibitor pairs were chosen to be investigated as transporter-mediated DDIs in cynomolgus monkeys. The in vitro transport parameters (K_m and IC₅₀) generated with cynomolgus monkey transporters and primary hepatocytes were comparable to those derived from human transporters and primary hepatocytes (Tables 3 and 4). Consistently, the mean area under plasma concentration–time curve (AUC) values of drug and endogenous probes in cynomolgus monkeys generally increased to a similar extent following coadministration of a reference inhibitor compared with those in humans despite differences in the dose and route of administration (Table 5 and Fig. 4). Collectively, these results demonstrate the human relevance of studying the transporter-mediated DDIs in monkeys both in vitro and in vivo (Tahara et al., 2006; Shen et al., 2013; Takahashi et al., 2013; Chu et al., 2015; Shen et al., 2015; Shen et al., 2016a; Shen et al., 2016b; Thakare et al., 2017; Karibe et al., 2018; Kosa et al., 2018; Shen et al., 2018b; Ufuk et al., 2018; Zhang et al., 2019; Gu et al., 2020; Wiebe et al., 2020; Cheng et al., 2021; Eng et al., 2021).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Comparison of available published pharmacokinetic DDI mediated by OATP1B (A), BCRP (B), OAT1/3 (C), and OCT2-MATE inhibition (D), as well as renal clearance between cynomolgus monkey and human transporters. Summary of the area under the plasma concentration–time curve ratios (AUCR) and CL_r) ratios of victim drug substrates presented in Table 5. Open blue circle and red square data points represent the AUC ratio and CL_r ratio of individual cynomolgus monkey and human study, respectively. Box and error bar depict mean and S.D., respectively.

It is important to note that the PK profiles of hepatic OATP1B substrates, including rosuvastatin (Shen et al., 2013; Shen et al., 2015), pitavastatin (Takahashi et al., 2013; Kosa et al., 2018; Ufuk et al., 2018; Eng et al., 2021), atorvastatin (Chu et al., 2015), coproporphyins (CPs) (Shen et al., 2016a; Takehara et al., 2019; Gu et al., 2020), and bile acids as well as their conjugates (Chu et al., 2015; Thakare et al., 2017), have been extensively characterized after coadministration of a single dose of rifampin and cyclosporin A, known OATP inhibitors, in cynomolgus monkeys. In this instance, the AUCs of OATP1B substrates were significantly altered because of the OATP1B transporter inhibition. In general, these changes were in agreement with reported clinical DDIs (Fig. 4). However, Chu et al. demonstrated that the atorvastatin AUC change following rifampin administration in monkeys is 1.5-fold (Chu et al., 2015). Such an AUC ratio is lower than that reported in the clinical studies (five- to sevenfold) described by Lau et al. (Lau et al., 2007) and He et al. (He et al., 2009). On the other hand, a single dose of 20 to 30 mg/kg rifampicin or 75 mg/kg cyclosporin A increased the AUCs of pitavastatin and rosuvastatin by approximately 10- to 39-fold in discrete and cassette DDI studies (Takahashi et al., 2013; Kosa et al., 2018), demonstrating stronger DDI compared with clinical studies (Lai et al., 2016). These results suggest that there is a variability in transporter-mediated statin DDIs in cynomolgus monkeys. Additionally, it might be more difficult to interpret the data when probe drugs are evaluated in cassette PK study.

In vivo studies documented that probenecid treatment resulted in decreased CL_r and increased AUC of furosemide (OAT1 and OAT3 substrate) and famotidine (OAT3 substrate) in cynomolgus monkeys, as compared with control animals (2.0- to 4.1-fold) (Tahara et al., 2006; Thakare et al., 2017; Shen et al., 2018b). These findings could be explained by the renal transporter OAT1 and/or OAT3 inhibition by probenecid. In contrast, the systemic exposure and renal clearance CL_r of cimetidine, a substrate of renal organic cation transporters, were not affected by probenecid in monkeys (Tahara et al., 2006). These results are in line with the clinical studies (Gisclon et al., 1989; Inotsume et al., 1990; Shen et al., 2019a). Although monkey OAT1 and OAT3 shared greatly similar transport characteristics with human orthologs and exhibited suitability in assessing OAT1/3-mediated DDIs in humans, the clinical DDI between famotidine and probenecid could not be reproduced in rats as the Oat3 transport activity and Oct1 expression in the rodent kidney are different to those of humans (Tahara et al., 2006).

There is also emerging evidence that cynomolgus monkeys are considered suitable models for predicting human DDIs involving renal organic cation transporters including OCT2, MATE1, and MATE2-K (Shen et al., 2016b). Consistent with the in vitro inhibition of OCT2-, MATE1-, and MATE2-K-mediated transport of metformin by pyrimethamine, intravenous pretreatment of monkeys with pyrimethamine, a known inhibitor of renal organic cation transporters, decreased the CL_r and increased the AUC of metformin by approximately 50% and 123%, respectively (Shen et al., 2016b), which is in agreement with the clinical DDI data (Weidekamm et al., 1982).

The impact of P-gp and BCRP on drug disposition has been well characterized due to the availability of P-gp and BCRP knockout mice and rats. However, loss of P-gp or Bcrp mRNA and protein expression in the transporter-deficient animals may lead to compensatory changes in the expression of other genes involved in drug metabolism and disposition. Additionally, the expression and transporter activity of P-g and Bcrp in rodents may be different from those of humans. Therefore, the potential for misestimating the influence of P-gp and Bcrp knockout mice and rats cannot be ruled out. For example, characterization of P-gp (Mdr1a) and Bcrp knockout rats, using paclitaxel and sulfasalazine as a corresponding probe of P-gp and BCRP, showed that the AUCs of paclitaxel and sulfasalazine were increased by 8.5- and 23.3-fold in comparison with wild-type animals, respectively (Zamek-Gliszczynski et al., 2012). These changes are greater than the magnitudes of DDIs observed in the clinic. Some have resorted to using specific P-gp and BCRP drug substrates and inhibitors in cynomolgus monkey transporter models. Toward this end, investigators have employed P-gp and BCRP inhibitors such as elacridar and curcumin. Although elacridar is a dual inhibitor of monkey P-gp and BCRP, curcumin is thought to be a more selective BCRP inhibitor owing to its low affinity for CYP enzymes and other transporters such as P-gp and MRP2 (Karibe et al., 2018; Kosa et al., 2018). Karibe et al. evaluated if Bcrp knockout mice and BCRP inhibitor-treated cynomolgus monkeys can be used as an animal model to assess the clinical BCRP-mediated DDI (Karibe et al., 2015). The deletion of the Bcrp gene resulted in marked effects on sulfasalazine with a 150-fold increase of AUC following oral administration of sulfasalazine in Bcrp knockout mice compared with wild-type controls. In contrast, curcumin, a potent BCRP inhibitor, increased the AUC of sulfasalazine 2.0- and 3.2-fold at a micro-dose and the therapeutic dose in humans (Kusuhara et al., 2012). The bioavailability changes in mice, which corrected the effect of systemic clearance by Bcrp knockout, correlated better with the AUC changes in humans (i.e., 8.4-fold vs. 2.2- to 3.2-fold). Curcumin increased the AUC of sulfasalazine in cynomolgus monkeys by 6.1-fold (Karibe et al., 2015). Similar increases of sulfasalazine AUC were observed in cynomolgus monkeys when administrated with BCRP inhibitors (i.e., 2.9- to 3.2-fold) (Karibe et al., 2018; Kosa et al., 2018). Overall, these results support the utility of cynomolgus monkeys as a surrogate model that could enable DDI risk assessment before advancing a new molecular entity (NME) to the development stage, as well as providing mechanistic insights regarding transporter-mediated DDI.

(A) Assessment of Transporter-Mediated DDI Risk for NMEs

Because cynomolgus monkeys exhibited comparable transport characteristics and the known human transporter DDIs are recapitulated in the cynomolgus monkey as described above, cynomolgus monkeys serve as an appropriate model for predicting human DDIs involving transporters. Several studies have documented that the cynomolgus monkey has been used for preclinical assessment of transporter DDIs with NMEs as test perpetrators. Ogawa et al. recently investigated a transporter-mediated DDI using an in vitro cell system and in vivo cynomolgus monkey model (Ogawa et al., 2020a, Ogawa et al., 2020b). Oral admiration of either rifampicin or cyclosporine A significantly increased the AUC of intravenously administered pemafibrate, a drug recently approved by Japanese Pharmaceuticals and Medical Devices Agency, by 7.4- and 4.9-fold, respectively, in cynomolgus monkeys. In agreement with this, clinical studies with pemafibrate showed that rifampicin (600 mg) and cyclosporine (600 mg) increased the AUC of pemafibrate by 11- and 14-fold, respectively (Pemafibrate; Japanese prescribing information 2018; http://www.pmda.go.jp/). These results suggested that pemafibrate was actively taken up by hepatocytes via OATP1B and cleared from plasma similarly in cynomolgus monkeys and humans.

The second example involves a drug candidate in discovery and early development. During the routine DDI risk assessment screening of transporter DDIs risk in drug discovery, it was determined that a Bristol Myers Squibb (BMS) compound was a potent inhibitor of OATP1B1 and OATP1B3 with the IC₅₀ of 0.24 and 0.65 µM, respectively. An IVIVE analysis using the mechanistic static model indicated a ratio of the victim AUC in the presence and absence of perpetrators of 1.4, which suggests a potential to impede the uptake of OATP1B substrates into hepatocytes from the portal circulation. Because the OATP1B DDI prediction with a static model is conservative and the binding of the BMS compound to serum proteins is high (99.8% and 99.6% in human and monkey, respectively), a pharmacokinetic DDI study was performed to determine if there is any DDI in monkeys. Oral DDI studies in cynomolgus monkeys were performed at two perpetrator doses to evaluate the potential for interactions between the known OATP1B substrate rosuvastatin and the compound. The two perpetrator doses for the BMS compound were necessary due to a difference in the elimination pathways of the compound between monkeys and humans as well as the difference in the hepatic blood flow between the two species. When the oral dose of the compound for the monkey DDI was selected based on achieving similar systemic concentrations in monkeys as in humans, a marked DDI was observed (11-fold increase in the AUC of rosuvastatin). However, when the dose for the DDI was chosen based on achieving similar portal vein concentrations in monkeys as in humans, only a moderate interaction was seen (2.2-fold increase in the AUC of rosuvastatin). The above-described monkey analysis subsequently triggered in vivo clinical DDI evaluations. The clinical DDI study using rosuvastatin as an OATP1B probe showed that the extent of the plasma AUC increases in the human DDI study was comparable to that in the monkey DDI study in which the dose was selected based on achieving similar portal vein concentrations in monkeys as in humans at a therapeutically relevant dose. These investigations suggest that the cynomolgus monkey has utility in predicting OATP-mediated DDIs in the clinic, with the need of careful selection on perpetrator doses.

The third example using the cynomolgus monkey for the DDI evaluation was on OAT1/3 inhibition. With the goal of predicting a human DDI, Ball et al. studied the extent of interaction between S44121 as victim and probenecid as an inhibitor of OAT1 and OAT3 because the monkey has previously been reported to be a good preclinical model for renal organic anion transporter inhibition (Ball et al., 2017). The S44121 AUC ratios in monkeys were between 3.0 and 3.5, which was slightly higher than the clinical observations (2.2). Additionally, an AUC ratio of 1.6 was predicted for S44121 in the presence of probenecid using a PBPK model, which was slightly smaller than the results obtained from the clinical DDI study. However, it is worth noting that the elimination pathway of S44121 in cynomolgus monkeys is different than in humans (Ball et al., 2017), which may explain the overprediction in monkeys.

(B) Prediction of Clearance Involving Transporters

Cynomolgus monkey has been evaluated as a preclinical model to predict OATP1B-mediated clearance by assessing IVIVE and applying cross-species extrapolation because of the excellent agreement with humans in OATP1B protein sequence homology and transport activity (Kimoto et al., 2017; De Bruyn et al., 2018; Matsunaga et al., 2019; Liang et al., 2020; Eng et al., 2021). Such studies would increase confidence in the subsequent extrapolation of human in vitro transporter-mediated uptake to clinical hepatic clearance. De Bruyn et al. recently demonstrated the utility of this preclinical model to delineate drug hepatic uptake and predict human in vivo intrinsic hepatic clearance (De Bruyn et al., 2018). The uptake of 9 OATP1B drug substrates: bosentan, cerivastatin, fexofenadine, repaglinide, rosuvastatin, pitavastatin, pravastatin, telmisartan, and valsartan in cynomolgus monkey and human hepatocytes were evaluated. The in vitro uptake parameters including total and active uptake clearance, passive diffusion, the contribution of active transport to total uptake in hepatocytes, and in vitro intracellular binding across a wide range of values, displayed a similar rank order and agreement between cynomolgus monkeys and humans. The in vivo hepatic clearance values of 9 OATP1B drug substrates determined by intravenous pharmacokinetics studies in cynomolgus monkeys exhibited comparable results to human parameters. In addition, the in vivo monkey and human hepatic clearance values were well predicted by corresponding hepatocyte uptake data with mean 2.7- and 3.8-fold bias, respectively (De Bruyn et al., 2018). Additionally, the application of cross-species empirical scaling factors either as mean or individual drug values improved the prediction of human hepatic clearance from human hepatocyte data. Subsequently, studies were performed with the same drugs in beagle dogs and Sprague-Dawley rats by the same group (Matsunaga et al., 2019). Large interspecies differences in hepatocyte-to-medium concentration ratio for total and unbound drug values were seen with the general rank order of cynomolgus monkey ≈ human < dog < rat for the nine drugs (Matsunaga et al., 2019). Furthermore, the use of monkey empirical scaling factors for human hepatic clearance prediction resulted in a prediction bias of threefold, with about 70% of drugs predicted within twofold of the in vivo hepatic clearance values. In contrast, both the prediction accuracy and percentage of drugs within the same error threshold were lower using dog- and rat-specific scalars (Matsunaga et al., 2019). Comparable in vitro uptake clearance and fraction transported by OATP1B between monkey and human hepatocytes were also observed for 16 large lipophilic acid compounds with molecular weight approximately 400 to 730 Da, logP 3.5 to 8, and acid pK_a < 6 (Eng et al., 2021). Furthermore, Kimoto et al. evaluated the dose of 6 OATP1B drug substrates excreted in the bile of nonclinical species including cynomolgus monkeys, dogs, and rats (Kimoto et al., 2017). They found that the biliary excretion clearance values determined with BDC monkeys were comparable to those in humans. As a result, the minimal species differences observed for the OATP1B-mediated transport between cynomolgus monkey and human hepatocyte uptake and excretion, both in vitro and in vivo, support the use of cynomolgus monkey as a nonclinical model to predict drug hepatic clearances.

(C) Improvement of Transporter PBPK Modeling

To devise an effective mechanistic IVIVE strategy, the incorporation of transporter-mediated uptake and efflux into a PBPK model is necessary. Consequently, transporter parameters including expression levels and transport kinetics are determined for each process and used in conjunction with system and physiology parameters to predict drug behavior in vivo. When integrating drug-related parameters into a model, the key transport kinetics, including active kinetics, are often described by the K_m and the maximal flux rate. Alternatively, the uptake clearance, assuming substrate concentrations well below the K_m, can be used. Although significant progress has been made, the application of PBPK for transporter-mediated DDI in drug development is relatively uncommon, as the predictive performance of PBPK models for transporter-mediated DDI is not well established compared with CYP-mediated DDI (Taskar et al., 2020). The major challenge of transporter PBPK is that it requires a detailed understanding of the complex role of drug transporters in the organs (e.g., liver, kidney, gut, and brain), transporter–enzyme interplay, and the physiologic parameters that define the membrane barriers. For example, a range of physiologic factors essential to the gut function can markedly impact drug absorption. These include gastric emptying, intestinal transit time, gastrointestinal pH, fluid dynamics, and intestinal segmental blood flow. In addition, empirical scaling factors are specifically required for in vitro active uptake parameters determined in primary hepatocytes to compare with the in vivo plasma clearance in PBPK modeling efforts when evaluating the prediction of hepatic clearance governed by OATP1B- and OCT1-mediated uptake in nonclinical species and humans (Jones et al., 2012). The ability to include liver drug concentrations for PBPK modeling and estimate in vitro to in vivo prediction of human liver drug accumulation is restricted, as human liver partitioning cannot be directly measured. However, a validated IVIVE of drug liver partitioning in cynomolgus monkeys, through the simultaneous liver and plasma sampling and in vitro monkey hepatocyte uptake, is thought to be a suitable approach for the assessment of liver accumulation (Morse et al., 2015, Morse et al., 2017). For example, Morse et al. assessed the in vitro to in vivo liver partitioning for OATP1B (rosuvastatin and bosentan) and OCT1 substrates (metformin) utilizing an ultrasound-guided liver sampling technique in cynomolgus monkeys (Morse et al., 2017). They also incorporated the liver along with plasma data for determination of in vivo transport parameters with PBPK modeling. In turn, the scaling factors obtained from the IVIVE validated from the prediction in cynomolgus monkeys can further be applied for the prediction of human hepatic clearance and assessment of human liver accumulation using in vitro human hepatocyte data (Morse et al., 2017). Such assessments of in vitro to in vivo liver partitioning and incorporation of liver and plasma concentrations for the determination of in vivo transport parameters with PBPK modeling have been expanded to the other OATP1B drug substrates such as bromfenac, carotegrast, and pemafibrate (Ogawa et al., 2019, Cheng et al., 2021). As a result, it is envisioned that the predictability of a PBPK model and the scaling factor for the IVIVE need to be validated with measured monkey liver concentration and/or pharmacokinetic DDI data. A human PBPK model incorporating the in vitro human transporter data and scaling factors obtained from monkey PBPK modeling validation would be valuable in drug discovery and early development (Liang and Lai, 2021).

(D) Identification, Characterization, and Validation of Transporter Endogenous Biomarkers

The last decade has witnessed important efforts toward the identification, characterization, and validation of endogenous transporter biomarkers, with numerous examples of researchers employing the cynomolgus monkey as a model species. Although conventional approaches to assess clinical DDI risk using a probe drug that is concomitantly administered with an investigational drug has limitations, such as the safety concerns of specific drug probes, uncertain therapeutically efficacious dose in the early development phase, and often conservative prediction by using US Food and Drug Administration and EMA mechanistic static model method, the use of circulating and urinary endogenous biomarkers that are selective for individual or combinations of hepatic and renal transporters may solve the dilemma. Using endogenous biomarkers could circumvent the need to dose substrate drugs and thus reduce pill burden in clinical trials. The biomarker data generated in phase I studies together with model-based predictions would complement regulatory decision tree-based approaches by minimizing false-positive predictions. Consequently, the investigation of endogenous clinical biomarkers of transporters has increased exponentially in the last few years (Chu et al., 2018; Rodrigues et al., 2018; Mochizuki et al., 2021). In this regard, cynomolgus monkey transporter models have played a pivotal role in the identification and validation of novel transporter endogenous biomarkers.

The suitability of CPI and CPIII as endogenous biomarkers of OATP1B was initially evaluated in nonclinical studies (Shen et al., 2016a). CPI and CPIII were shown to be substrates of cynomolgus monkey and human OATP1B1 and OATP1B3 using HEK-293 cells overexpressing isoform transporter of monkey or human OATP1B subfamily, as well as monkey and human hepatocytes. Oral administration of 100 mg/kg cyclosporin A increased the AUC of CPI and CPIII by 2.6- and 5.2-fold, respectively, and 15 mg/kg rifampicin increased the AUCs by 2.7- and 3.6-fold, respectively (Shen et al., 2016a). In agreement, the AUC of rosuvastatin, a probe drug, in cynomolgus monkeys increased 6.3- and 2.9-fold by the cyclosporin A and rifampin treatments, respectively, in the same animals (Shen et al., 2013; Shen et al., 2015). Furthermore, in South Asian Indian, black, white, and Hispanic subjects, administration with 600 mg rifampin orally increased the AUC of CPI and CPIII by 2.4- to 3.7-fold, respectively, indicating similar CP changes between cynomolgus monkeys and humans (Fig. 4 and Table 5) (Lai et al., 2016; Shen et al., 2018a). The absorption and disposition of CPI in cynomolgus monkeys following oral and intravenous administration of CPI-d8 were further examined (Gu et al., 2020). CPI-d8 showed low absolute oral bioavailability (i.e., 3.2%), implying limited intestinal absorption and enterohepatic circulation of CPI as it is metabolically stable (Gu et al., 2020). Consistently, a BDC monkey study indicated that CPI and CPIII did not undergo significant enterohepatic circulation as the plasma CP concentrations were not altered in BDC monkeys compared with normal animals (Takehara et al., 2019). Additionally, by leveraging the cynomolgus monkey data, Thakare et al. were able to identify five bile acid 3-O-sulfate conjugates as novel endogenous biomarker candidates for OATP1B (Thakare et al., 2017). These biomarker candidates were then validated in clinical studies (Takehara et al., 2018, Mori et al., 2020). Furthermore, Chu et al. also evaluated cynomolgus monkeys as a translatable model to assess OATP1B-mediated DDI and identified total and unconjugated bilirubin as well as several bile acids as potential endogenous probes of OATP1B inhibition (Chu et al., 2015). Cynomolgus monkeys orally received different doses of cyclosporin A (4, 20, and 100 mg/kg) and rifampin (1, 3, 10, and 30 mg/kg) which yielded dose-dependent increases of the AUC of the biomarker candidates such as CPs and glycochenodeoxycholate-3-sulfate, demonstrating the in vivo sensitivity of the endogenous compounds as probes of OATP1B (Thakare et al., 2017; Gu et al., 2020). These nonclinical studies greatly help identify and evaluate endogenous biomarkers of OATP1B inhibition.

Cynomolgus monkeys have also been used to investigate endogenous biomarkers for renal transporters (Shen et al., 2018b). Specifically, cynomolgus monkeys were pretreated with an intravenous administration of 40 mg/kg probenecid that is an inhibitor of OAT1 and OAT3. The pretreatment increased the AUC of furosemide, a known OAT1 and OAT3 substrate, by 4.1-fold in monkeys. These increases are similar to the changes reported in clinical studies (4.1-fold vs. 3.1- to 3.7-fold) (Smith et al., 1980; Shen et al., 2019a). Additionally, an untargeted metabolomics analysis was applied to plasma samples to screen endogenous compounds that were associated with OAT1 and OAT3 inhibition in monkeys. Plasma concentrations of pyridoxic acid and homovanillic acid concentrations were significantly increased at 1 or 3 hours after probenecid treatments compared with the vehicle-treated animals. Targeted liquid chromatography–tandem mass spectrometry analysis further confirmed that the pyridoxic acid and homovanillic acid AUCs were increased by approximately two- to threefold by probenecid pretreatments. A clinical study examined and validated pyridoxic acid as a plasma-based endogenous biomarker of OAT1 and OAT3 in humans (Shen et al., 2019a). Administration of probenecid markedly increased the AUC of pyridoxic acid by 3.1- to 3.2-fold compared with the prestudy and furosemide groups. Transporter phenotyping experiments using singly stable transporter transfected cell lines indicated that pyridoxic acid and homovanillic acid were substrates for human OAT1, OAT3, OAT2 (homovanillic acid), and OAT4 (pyridoxic acid), but not OCT2, MATE1, MATE2K, OATP1B1, OATP1B3, and NTCP (Shen et al., 2018b). Collectively, these results suggest that cynomolgus monkeys are useful in identifying and examining endogenous biomarker candidates for a better understanding of their translational value.

(E) Elucidation of the Mechanism of Complex PK and Toxicity Involving Transporter

Mechanistic understanding of complex drug disposition, DDIs, and toxicity in the clinic, with potential involvement of numerous elimination pathways and diverse enzyme–transporter interplay has been challenging. Using cynomolgus monkey models, Zhang et al. examined in vivo effects of rifampicin, a potent PXR agonist, on the activity and expression of hepatic OATP1B (Zhang et al., 2020a). The study reported that plasma levels of CPI and CPIII, OATP1B endogenous biomarkers, were not changed in monkeys after multiple rifampicin treatments compared with nontreatment conditions and that the expressions of OATP1B1 and OATP1B3 genes were not induced in the liver (0.85- to 1.3-fold). In contrast, the plasma concentrations of 4β-hydroxycholesterol, a CYP3A endogenous biomarker, were increased by 3.9-fold, and the hepatic and intestinal CYP3A8 mRNA levels were markedly increased by multiple treatments of rifampicin in the same animals (3.7- to 5.0-fold). These results obtained from the cynomolgus monkey model validate that OATP1B1 and OATP1B3 are not induced by rifampicin in cynomolgus monkeys. These findings provide a better understanding of the clinical results of OATP1B drug substrates such as statins with PXR inducers (Zhang et al., 2020a). This is important, as the effect of commonly used inducers on the expression and functional activity of OATP1B is poorly understood and the reports that describe the OATP1B induction as decreased systemic exposure of probe substrates and increased mRNA and protein levels are limited and conflicting. Moreover, there are two major clinical challenges for OATP1B induction investigations. First, the selectivity of OATP1B drug probes is not well established as most statins are substrates of induced CYP enzymes (e.g., CYP3A4) and efflux transporters (e.g., MRP2). Second, it is not practical to include a tissue biopsy procedure in a clinical study protocol, although the determination of transporter expression in the gut and liver tissues following treatment with PXR inducers has been considered the most direct way to assess transporter and drug-metabolizing enzyme induction. Zhang et al. evaluated the effect of multiple administration of rifampicin on gene expression of OATP1B1, OATP1B3, and the major drug transporters in cynomolgus monkey liver, small intestine, and kidneys (Zhang et al., 2020a). OATP1B1 and OATP1B3 mRNA levels in the liver were not altered by the rifampin treatment. These results are in line with an in vitro hepatocyte induction study that showed a minimal change of OATP1B mRNA in cynomolgus monkey and human hepatocytes treated with rifampin (Niu et al., 2019).

Cynomolgus monkeys and primary cultures of monkey renal proximal tubule epithelial cells have been used to evaluate the role of renal transporters in the cytotoxicity of three Pfizer exploratory drug candidates with varying degrees of nephrotoxicity in cynomolgus monkeys (Cai et al., 2009). Compound PF-1 was nephrotoxic in cynomolgus monkeys. Therefore, the nephrotoxicity of PF-1 was qualitatively compared with two other leads, PF-2 and PF-3. The severity of kidney toxicity was ranked as PF-2 > PF-1 > PF-3 in monkeys, with PF-2 inducing mortality at all doses and PF-3 eliciting only mild nephrotoxicity in cynomolgus monkeys. The cytotoxicity was then characterized using lactate dehydrogenase release assay in primary cultures of monkey and human renal proximal tubule epithelial cells for comparisons. The rank of cytotoxicity in monkey kidney proximal tubule cells was PF-2 > PF-3 > PF-1, whereas that in human kidney proximal tubule cells is PF-2 > PF-1 > PF-3 (Cai et al., 2009). Transporter phenotyping studies using transporter-transfected cell lines indicated that these organic anions were substrates for human OAT1 (PF-1 and PF-3) and OAT3 (PF-1, PF-2, and PF-3). Consequently, pretreatment of cynomolgus monkeys with probenecid increased the AUC of PF-3 by approximately twofold compared with untreated monkeys. As a result, OAT1 and/or OAT3 may be responsible for the uptake of these drugs in kidney renal proximal tubule epithelial cells, which directly or indirectly lead to nephrotoxicity in vivo (Cai et al., 2009).