Role of SLC transporters in toxicity induced by anticancer drugs

Abstract

Introduction.

Membrane transporters are integral components to the maintenance of cellular integrity of all tissue and cell types. While transporters play an established role in the systemic pharmacokinetics of therapeutic drugs, tissue specific expression of uptake transporters can serve as an initiating mechanism that governs the accumulation and impact of cytotoxic drugs.

Areas covered.

This review provides an overview of organic cation transporters as determinants to chemotherapy-induced toxicities. We also provide insights into the recently updated FDA guidelines for in vitro drug interaction studies, with a particular focus on the class of tyrosine kinase inhibitors as perpetrators of transporter-mediated drug interactions.

Expert opinion.

Studies performed over the last few decades have highlighted the important role of basolateral uptake and apical efflux transporters in the pathophysiology of drug-induced organ damage. Increased understanding of the mechanisms that govern the accumulation of cytotoxic drugs has provided insights into the development of novel strategies to prevent debilitating toxicities. Furthermore, we argue that current regulatory guidelines provide inadequate recommendations for in vitro studies to identify substrates or inhibitors of drug transporters. Therefore, the translational and predictive power of FDA-approved drugs as modulators of transport function remains ambiguous and warrants further revision of the current guidelines.

1. Introduction

It has become evident that the expression and localization of drug transporters contribute to a dynamic interplay between the maintenance of cellular integrity via the removal of toxic metabolites or entry of essential nutrients [1]. Moreover, members of the ATP-binding cassette (ABC) and solute carrier (SLC) superfamilies are recognized as important determinants governing the therapeutic response of many drugs [1–3]. Disruption of these sensitive transport systems through changes in gene expression, genetic variations or drug-drug interaction, has the potential to contribute to chemoresistance, endogenous toxin-mediated diseases, diminished therapeutic efficacy, and increased sensitivity to drug-induced organ damage [4–7]. The following review focuses on specific examples within the SLC subfamily of organic cation transporters (OCTs) and multidrug and toxin extrusion proteins (MATEs) as critical determinants of the dose-limiting toxicity profiles associated with chemotherapeutic agents.

2. Organic cation-type transporters

ABC and SLC family members comprise of over 400 genes encoded in the human genome [7,8]. While the function of these gene families is to regulate the entry of essential nutrients and the removal of endogenous toxins, the SLC family also contributes to the cellular and systemic distribution profile of many therapeutic drugs [3,9,10]. Specifically, OCT1 [SLC22A1], OCT2 [SLC22A2], OCT3 [SLC22A3], OCTN1 [SLC22A4], and OCTN2 [SLC22A5] are the major organic cation-type transporters encoded by the SLC22A subfamily that facilitate drug accumulation. Due to their expression on the basolateral membranes of enterocytes, hepatocytes, proximal tubules and other endothelial peripheral tissues, OCTs play a critical role in the vectorial transport of organic cations [3,11–13]. Independent of Na⁺ and H⁺, OCTs translocate substrates across a biological membrane driven by membrane potential and the electrochemical gradient of the protonated molecule [3]. Additionally, the subfamily of MATEs, including MATE1 [SLC47A1] and MATE2-K [SLC47A2], mediate the efflux of diverse substrates consisting primarily of organic cations, but also certain anions and zwitterions, from the liver and kidney [14,15]. MATEs have emerged as clinically relevant efflux transporters that work in concert with OCTs to mediate apical efflux into the bile and urine. Unlike OCTs, the bi-directionality of MATE transport is dependent on the proton concentration gradient, and driven by exchange of protons in the bile or urine [16]. It is important to note that other members within this family, including OCTL2 [SLC22A14], OCT6 [SLC22A16], OCTN3 [SLC22A21], and UST6 [SLC22A25] [17,18] as well as members of the ABC family such as MDR1 [ABCB1], MRP1 [ABCC1], MRP2 [ABCC2], MRP4 [ABCC4], and BCRP [ABCG2] [19,20], are also capable of governing intracellular levels of organic cations. Since 40% of FDA-approved therapeutic drugs exist as organic cations under physiological conditions [2], OCTs and MATEs are recommended for evaluation during drug development by the Food and Drug Administration (FDA) [21].

2.1. Organic cation transporter 1 (OCT1).

OCT1 is the most abundant cationic uptake transporter expressed in the liver in humans [13]. The localization of OCT1 is predominantly expressed on the sinusoidal membrane (blood facing) of hepatocytes, and it serves as the rate-determining step in the translocation of endogenous [22,23] and xenobiotic cationic compounds from the systemic circulation into the liver [24], and thus contributes as an initial step in hepatic metabolism and excretion. In addition to OCT1, efflux transporters of ABC superfamily of transporters such as MDR1 [19], and MATE1 [16] work in a concerted manner to assist in the biliary excretion and detoxification of endogenous cationic toxins. OCT1 has been implicated in drug-drug interactions, and contributes to pharmacokinetic and pharmacodynamic variability of substrates in humans [25,26]. OCT1 is also expressed on the bronchial epithelial cells [27], endothelial cells of the heart and brain, and enterocytes [3,28]. In contrast to the dominant OCT1 expression in human livers, rodents also express Oct1 on enterocytes [29] and epithelial cells found in the S1 and S2 segments of proximal tubular cells. In the rodent kidney, Oct1 is redundant with Oct2, and these proteins jointly fulfill a function that is equivalent to that of OCT2 in humans. Targeted disruption of Oct1 in rodents reduced liver accumulation of prototypical transport substrates such as tetraethylammonium (TEA) and the neurotoxin 1-methyl-4-phenylpyridinium (MPP) [30,31].

2.2. Organic cation transporter 2 (OCT2)

Unlike OCT1, OCT2 presents a more restricted expression pattern. OCT2 is highest on the basolateral membrane (blood facing) along S2 and S3 segments of proximal tubule cells [12,30,31]. OCT2 plays a key role in facilitating the translocation of cationic compounds from systemic circulation into renal proximal tubule cells and working in concert with efflux transporters (e.g., primarily MATE1 and MATE2-K) to mediate excretion of substrates into the urine. Although OCTs share many overlapping substrates, OCT2 also efficiently transports monoamine neurotransmitters, as evidenced by moderate expression across different types of neuronal cells [3,28]. Due to its high expression in proximal tubule cells, OCT2 has been implicated in mediating drug-drug interactions and it therefore recommended for evaluation during development by the FDA [21].

2.3. Organic cation transporter 3 (OCT3)

In contrast to the wealth of knowledge accumulated in recent years on OCT1 and OCT2, relatively little is known about the third member of the SLC22A family, OCT3. OCT3 exhibits a promiscuous expression pattern in most major organs and peripheral tissues, with highest expression in the heart, placenta and skeletal muscles [12,32,33]. Although OCT3 is generally not considered a significant contributor to drug-drug interactions, recent studies have highlighted OCT3 as an important transporter in the heart and cardiovascular disease [34], and carriers of polymorphic variants in OCT3 (rs2048327, rs1810126, rs1810126 and rs3088442) are at reduced risk of coronary artery disease development [35–37]. While OCT3 is present in most tissues, the genetic deletion of Oct3 in mice does not affect viability and fertility, and does not lead to any overt and/or abnormal pathology [38]. Oct3 deficiency in mice is associated with diminished accumulation of prototypical substrates such as metformin and MPP in homogenized hearts [34,38].

2.4. Organic cation/carnitine transporter 1 and 2 (OCTN1 and OCTN2)

In comparison with OCT1–3, OCTN1 and OCTN2 can transport substrates either dependently or independently of Na⁺. Driven by a proton gradient, OCTNs can recognize zwitterions as well as organic cations and even certain anions, and share a high degree of membrane topology [39,40]. OCTN1 is the only known transporter of ergothioneine, whereas OCTN2 is believed to be an essential carnitine transporter. Both OCTN1 and OCTN2 are widely expressed in human tissues. Localization of OCTNs in proximal tubular cells is restricted to the apical membrane (urine side) [3,12], where it may function in the reabsorption of essential cations from the urine. OCTNs are also expressed on the apical membrane of enterocytes, and OCTN2 is expressed in skeletal muscles, heart, lung and eye, where it may be involved in regulating L-carnitine uptake and tissue homeostasis [41–43]. Genetic deletion of Octn1 provides viable offspring [44], whereas Octn2 deletion displays embryonic lethality without L-carnitine supplementation [45]. The clinical relevance of OCTN1- and OCTN2-mediated drug-drug interactions remains largely unstudied but is likely limited due to the narrow substrate specificity of these transporters.

2.5. Multidrug and extrusion protein 1 and 2-K (MATE1 and MATE2-K)

MATE1 and MATE2-K are poly-specific efflux transporters that translocate organic cations, and certain anions, in a Na⁺ independent and pH-dependent manner [3,10]. MATE1 is apically expressed on the cannicular and brush-border membrane of hepatocytes as well as on proximal tubular cells, and mediates the efflux of organic cations into the bile and urine, respectively [46]. MATE2-K, a splice variant, is restricted to kidneys in humans and the redundant expression of this additional antiporter on the luminal membrane of proximal tubules in humans suggests that MATEs evolutionary adapted to mediate the clearance of endogenous toxins or xenobiotics [16]. The tissue distribution of Mate1 in rodents is generally consistent with that observed for MATE1 in humans, with the exception of MATE2-K being absent in the kidney of mice and rats, while Mate1 is additionally absent in the liver of rats [15,47].

3. Dose-limiting toxicities associated with chemotherapeutic agents

3.1. Nephrotoxicity

Cisplatin is a standard of care drug for the treatment of various malignant solid tumors in both pediatric and adult populations [48–51]. The clinical utility of cisplatin is limited by the onset of debilitating toxicities, including nephrotoxicity [52–54], which can cause severe renal impairment. Approximately one-third of patients receiving cisplatin develop acute kidney injury (AKI) during the course of treatment, and in severe cases, patients display a permanent incomplete recovery of renal function despite discontinuation of treatment. More than 50 percent of the drug is excreted into the urine within the first 24 hours [55,56], and renal cortical concentrations of platinum are orders of magnitude greater than that observed in the systemic circulation and compared to other organs [57], suggesting a key renal-mediated mechanism of accumulation. The pathophysiology of cisplatin-induced nephrotoxicity involves renal tubular apoptosis and necrosis, particularly in the S3 segment of proximal tubule cells and of the distal nephron, manifesting clinically as imbalances in salt and essential nutrients, tubular acidosis and a reduced glomerular filtration rate (GFR) [58–60]. Although proper hydration of patients and prophylactic treatment with diuretics can manage the symptoms of nephrotoxicity, the prevalence of this side effect remains high and greatly diminishes quality of life [59,61].

Renal concentrations of platinum are highest in the S2 and S3 segment of proximal tubular cells [52,62], highlighting a key cell-specific and transporter-mediated process. In rodents, the deficiency of both Oct1 and Oct2 conferred protection against cisplatin-induced nephrotoxicity [63] while the OCT2 808G>T (rs316019) genetic variant in humans was associated with reduced changes in serum creatinine, a hallmark of acute kidney injury, after one cycle of cisplatin treatment as compared to baseline and to patients lacking this variant [64–66]. Furthermore, sexually dimorphic expression of Oct2 in proximal tubular cells of male rodents correlated with a greater propensity and susceptibility to cisplatin-mediated tubular injury compared to female rodents [12,67]. Cisplatin is also a transported substrate of MATE1, and rodent deficiency for Mate1 was associated with diminished urinary excretion of total platinum, potentiating nephrotoxicity [68]. The pharmacologic targeting of these transport mechanisms using OCT2 and MATE1 inhibitors protected or exacerbated nephrotoxicity, respectively [69–71]. The translational significance of OCT2-dependent intervention strategies, in which cisplatin is administered after an OCT2 inhibitor to restrict access of the chemotherapy to tubular cells, remains somewhat disappointing. This may be explained by the partial reduction in biomarkers of cisplatin-induced nephrotoxicity (e.g., serum creatinine and blood urea nitrogen), the use of non-specific OCT2 inhibitors lacking potency at clinically-achievable concentrations, the incomplete recovery of pathological damage in the kidneys that occurs despite the OCT2 inhibition [63], and/or the failure of the intervention strategy to modulate uptake mechanisms that mediate cisplatin-induced AKI in an OCT2-independent manner. Regarding the latter, it should be pointed out that several reports have suggested the existence of a negatively charged N-acetylcysteine S-conjugate (NAC-1) derived from cisplatin that can act as an intermediate, highly reactive nephrotoxic thiol. NAC-1 is generated by cisplatin conjugation with glutathione and cysteine before accumulation in proximal tubular cells [72]. NAC-1 was recently found to be a transported substrate of the organic anion transporters OAT1 [SLC22A6] and OAT3 [SLC22A8], which are also expressed on the basolateral membrane of proximal tubules, and can contribute to cisplatin nephrotoxicity independently of Oct2 in a signaling pathway that is upstream of p53 [73]. However, similar to Oct2-deficient rodents, genetic deficiency of Oat1 and Oat3 or pharmacological inhibition with probenecid only offers partial reduction in serum markers and pathological damage [74]. This study further demonstrated that concomitant administration of the Bcr-Abl tyrosine kinase inhibitor nilotinib, which can simultaneously inhibit both Oct2-mediated uptake of cisplatin [75] and Oat1/Oat3-mediated uptake of NAC-1 through non-competitive mechanisms, afforded complete protection against cisplatin-induced proximal tubular injury in vivo [74].

Compared to cisplatin, the related drug oxaliplatin lacks significant nephrotoxic activity, despite being a transported substrate of the rodent and human OCT2, MATE1 and MATE2-K [76]. While Oct2 transports both platinum agents with similar kinetics, the extent of total platinum accumulation in the kidney was far greater in cisplatin-treated animals compared to oxaliplatin [77]. The differences observed in the accumulation and elimination of these platinum agents is potentially related to differences in transport kinetics for Mate1 in mice, and MATE1 and MATE2-K in humans [77], suggesting higher luminal efflux of oxaliplatin into the urine compared to cisplatin. However, this explanation of differential tubular secretion rates for cisplatin and oxaliplatin as a driving discriminatory mechanism for their different nephrotoxic properties is not completely satisfactory, and is inconsistent with comparative in vivo studies in mice indicating that the renal excretion of both agents is strongly dependent on Mate1. It also largely ignores intrinsic differences in renal metabolic activation pathways that results in the formation of potent nephrotoxins in the case of cisplatin but not oxaliplatin, and fails to recognize the differing contribution of other potentially critical renal ABC efflux transporters in the process, such as MRP2 and MRP4.

Similar to oxaliplatin, the novel platinum derivative, phenanthriplatin, is a high affinity substrate for both OCT2 and MATE1/2-K, and it has been argued that these are kinetically favorable attributes that would be efficacious in the treatment of OCT-expressing tumors, while efficient MATE-mediated efflux may decrease the renal tubular residence time and thereby prevent the debilitating toxicity profiles observed in the kidneys following cisplatin treatment [78]. Similar observations have been documented with the alkylating agent, ifosfamide, which is associated uniquely with nephrotoxic activity and the development of renal Fanconi syndrome [79]. Bioactivation of ifosfamide mediated by OCT2 transport into proximal tubular cells has been proposed as a differentiating mechanism between ifosfamide on the one hand and cyclophosphamide’s lack of nephrotoxic activity on the other [80]. Although the nephrotoxic properties of these chemotherapeutic agents is predominantly attributed to originate with basolateral OCT2-mediated uptake and subsequent apical MATE1- and MATE2-K-mediated efflux, it should be pointed out that significant differences in transport kinetics and transporter-mediated interactions involving MATEs could potentially contribute to lower or higher incidences of nephrotoxicity in patients, depending on relative interindividual differences in functional expression of these proteins. While interindividual differences in expression may predict susceptibility to drug-induced organ damage in tissues with a clear functional role of uptake transporters, it is important to note that these expression differences also contribute as mechanisms of chemoresistance in tumor cells. Consistent with this notion, several studies have demonstrated that the down-regulation of OCT6 in lung and colon cancer tissues is associated with cellular resistance to cisplatin and oxaliplatin [81,82]. Thus, intervention strategies aimed at targeting uptake transporters to ameliorate drug-induced toxicity must be evaluated for antagonized therapeutic efficacy due to unintended inhibition of transport mechanisms regulating entry into tumor cells.

3.2. Neurotoxicity

In addition to their varying degrees to which cisplatin and oxaliplatin induce AKI, platinum agents can also cause severe and debilitating, dose-limiting peripheral neurotoxicity, and this has resulted in extensive preventative efforts as evidenced by the increasing emergence of large numbers of neuroprotective clinical trials [83]. Clinical manifestation of neurotoxicity occurs immediately after infusion and is characterized by paresthesia, ataxia and dysesthesia in the extremities of the body [84,85]. Within the nervous system, platinum agents are unable to penetrate the blood-brain barrier, but rather preferentially accumulate in peripheral sensory neurons present along dorsal root ganglia (DRG) of the spinal cord [86]. In peripheral sensory neurons, induction of nuclear and mitochondrial DNA damage, disruption of ion channels while glial activation of astrocytes and microglial cells are mechanisms of pharmacological interest in preventing mechanical and thermal neuropathic pain [87]. At higher cumulative doses, incompletely recovery and permanent dysfunction of sensory neurons represents a major clinical problem [86].

Somewhat unexpectedly, several studies have demonstrated that OCT2 transcripts are detectable in rodent and human DRG samples, suggesting a carrier-mediated mechanism of accumulation into these structures [88]. Consistently, the genetic deletion of Oct2 in mice prevented the onset of acute oxaliplatin-induced thermal and mechanical allodynia, while this phenotypic preservation occurred in the absence of detectable changes across structurally similar and putative SLC or ABC transporters of relevance to oxaliplatin [88]. This protection can also be recapitulated in wild-type animals using pretreatment with dasatinib, a pharmacological SRC-family kinase inhibitor, which inhibits both mouse and human OCT2 through a noncompetitive mechanism that regulates the post-translational modification of OCT2 through tyrosine phosphorylation by the kinase YES1 [89]. Interestingly, other reports have suggested oxaliplatin is also a transported substrate of rat Octn1 and the use of genetic targeting constructs or pharmacologic inhibition with ergothioneine, an Octn1-specific substrate, protected rats from chronic forms of oxaliplatin neurotoxicity [90–92]. However, due to the reported expression of Octn1 on the mitochondrial membrane and intrinsic antioxidant activity of ergothioneine, the observed neuroprotection may be independent of a direct contribution to the transmembrane transport of oxaliplatin into target cells. This hypothesis would be consistent with data suggesting that ergothioneine can reduce cellular concentrations of oxaliplatin in Octn1-deficient DRG neurons of mice [44]. Furthermore, it is unknown whether ergothioneine competes or exhibit inhibitory activity against putative transporters of relevance to oxaliplatin in neurons. The discrepant findings reported for the involvement of transporters involved in oxaliplatin-induced peripheral neurotoxicity in rodents may also be attributed to differences in transport kinetics or expression between the two species, and further studies are required to delineate the translational contribution of human OCT2, OCTN1, and other SLC family members to this side effect.

3.3. Ototoxicity

Cisplatin is unique among platinum chemotherapeutics in its ability to frequently induce severe hearing-loss (ototoxicity), where the incidence ranges from 20 to 80% in pediatric patients, and this side effect may permanently affect early speech and ultimately hamper social development [93,94]. Excessive generation of reactive oxygen species (ROS), increases in cochlear inflammatory responses and induction of apoptosis mediated by platinum-DNA adduct formation have been proposed as mechanisms of cisplatin ototoxicity [95,96]. However, antioxidant or anti-inflammatory drugs used prophylactically do not prevent permanent damage, and only offer moderate protection in the management of cisplatin-induced cochlear injury [95,96]. It is possible that the initial accumulation of cisplatin into the cochlea cells represents a key trigger for the following pathophysiological changes in cisplatin-induced hearing loss. Indeed, both mouse Oct2 and human OCT2 are expressed in the organ of Corti and stria vascularis of the cochlea, and genetic deficiency or pharmacological targeting of Oct2 with cimetidine protected rodents against toxicity [71]. A retrospective and exploratory analysis of 11 SNPs in the human SLC22A2 gene identified a genetic variant, c.808G>T; Ser270Ala (rs316019), that conferred significant protection against cisplatin-induced ototoxicity in a pediatric cohort [97]. In vitro insertion of the 808G>T allele into OCT2, located within the transmembrane domain [98], reduced transport kinetics (e.g., V_m and K_m) of OCT2 substrates (e.g., MPP and dopamine) [98–100] and a moderate increase in serum creatinine [101], supporting the notion that the observed human variant functionally reduces transport affinity of cisplatin and cellular injury. Of note, pantoprazole, an inhibitor of OCT2 [102], did not ameliorate ototoxicity or nephrotoxicity in a pediatric cohort of osteosarcoma patients treated with cisplatin [103]. While prediction of the applied dosing schedules for pantoprazole-mediated OCT2 inhibition was performed using pharmacokinetic modeling and simulations, these strategies were not adequately validated and ultimately lack rationally designed preclinical and clinical studies to document the local (cochlea or kidney) and systemic (blood) changes in endogenous substrates of OCT2 as a pharmacodynamic biomarker of OCT2 function. While the reported intervention study with pantoprazole was negative, caution is warranted against the conclusion that the proposed OCT2 inhibition concept is intrinsically flawed, and different results could have been obtained with a more potent inhibitor given at a dose and schedule known to affect cochlear biomarkers of OCT2 function.

In connection with MATE1 function, it is worth pointing out that, in an adult cohort of patients with head and neck squamous cell carcinoma (HNSCC) receiving cisplatin-containing therapy, the presence of one or two copies of the SLC47A1 variant (rs2289669) significantly predisposed to cisplatin-induced ototoxicity [104]. However, it is uncertain whether this variant produces a gain or loss of function phenotype. Pharmacogenomic analysis of this same variant in patients with type 2 diabetes receiving metformin indicated a reduced function phenotype, as demonstrated by an increase in glucose lowering, a marker of pharmacodynamic response to metformin [105]. On the other hand, the protective effects observed among HNSCC patients suggested a gain of function since MATE1-mediated cisplatin efflux and treatment response was unaffected in the HNSCC cohort [104]. The unique bi-directional antiporter characteristics of MATE1, compared to other SLCs, suggest that the function of this variant may be dependent on tissue localization and cellular polarity of MATE1-expressing cells.

3.4. Cardiotoxicity

Anthracycline agents (e.g., doxorubicin, daunorubicin, epirubicin and idarubicin) are among standard of care chemotherapeutic agents in the treatment of various solid tumors and hematological malignancies in children and adults. Severe and debilitating cardiotoxicity associated with anthracyclines is of significant clinical concern due to the risk of detrimental and irreversible cardiac damage that is accompanied with a high incidence of cardiovascular mortality, especially in young adults and adolescent survivors [106,107]. Initial studies evaluating mechanisms of transmembrane transport of doxorubicin concluded that passive diffusion of the unionized drug is a dominant pathway of uptake [108]. However, the existence of saturation kinetics in vitro and the apparent cationic nature of the drug at physiological pH are supportive of the involvement of a carrier-mediated mechanism [109,110]. Extensive pharmacogenomics studies have identified the class of ABC transporters as important determinants of doxorubicin-induced cardiac injury. Polymophoric variants in MDR1, MRP2, and BCRP have been linked with altered risks of cardiotoxicity, presumably due to reduced function and increased intracellular levels of doxorubicin and its metabolite doxorubicinol in cardiomyocytes [111]. Furthermore, OCT6 polymorphisms have been associated in preliminary studies with pharmacokinetic variability of doxorubicin in humans [112,113]. Despite these observations, direct evidence supporting a role of OCT6 and other OCTs in anthracycline disposition, particularly in the myocardium, remains lacking.

Among the class of OCTs, mRNA transcript and immuno-reactivity is present in the vasculature, endothelial cells, and cardiomyocytes [34,42]. OCTN2 is the highest OCT expressed in the myocardium, and this is not surprising given the cardio-protective role of L-carnitine against oxidative stress and inflammation, and in regulating cardiac energy metabolism [114]. Following OCTN2, expression levels were in decreasing order observed for OCT3 > OCTN1 > OCT1, although their precise localization and expression pattern in these cell types remain controversial [42]. In vitro overexpression of OCTN1, OCT1, and OCT6 indicate a role of cationic-type transporters in facilitating doxorubicin and daunorubicin accumulation in leukemic and ovarian cancer cells [17,115]. The contribution of these transporters to doxorubicin-induced cardiotoxicity remains to be elucidated, though preliminary studies have indicated an important independent contribution of Oct3 in mice [116]. Since OCT3 is known to transport important cardiac signaling hormones (e.g., catecholamines) and given that genetic variants in the SLC22A3 gene locus are associated with the development of coronary artery disease [35], further studies are required to substantiate these findings.

4. Tyrosine kinase inhibitors and OCT/MATEs: Implications for drug interaction studies

In the past decade, the development and approval of small molecule targeted therapies, especially tyrosine kinase inhibitors (TKIs) for the treatment of various solid tumors and hematological malignancies has revolutionized the conventional landscape of cancer treatment [117]. TKIs largely act by competing with adenosine 5’- triphosphate (ATP) on binding sites of tyrosine kinases, such that the displacement of ATP results in the diminished activity of mutated or overexpressed protein kinases [117]. Drug-drug interactions and the high degree of interindividual pharmacokinetic variability observed with TKIs present a great risk for the development of toxicity or suboptimal therapeutic efficacy, especially for drugs with narrow therapeutic windows [118,119].

Mechanistic insights into the contribution of OCTs to the systemic pharmacokinetic profile of TKIs remain largely unstudied. Recently, several reports proposed that tumoral expression of OCT1 mRNA may be a prognostic biomarker for sorafenib treatment and response in patients with hepatocellular carcinoma (HCC) or cholangiocarcinoma [120]. Consistent with this observation, inherited polymorphic variants in OCT1 or pharmacological targeting by quinidine, an OCT1 inhibitor, diminished sorafenib transport activity in tumor cells and overexpressed model systems [120–123]. This notion is further supported by evidence demonstrating that localization of OCT1 protein at the plasma membrane, rather than transcript levels, correlated with improved outcomes in HCC patients receiving sorafenib [122]. Although these reports conclude that OCT1 expression may influence sorafenib response, others studies have suggested that sorafenib transport may be facilitated through a transporter-independent mechanism or mediated by transporters other than OCT1 [124–126]. For example, a recent study demonstrated that genetic deletion of Oct1 in mice did neither diminish liver accumulation of sorafenib nor influence sorafenib plasma pharmacokinetics, and was not associated with compensatory gene expression changes [124]. The possibility that OCT1 may not play an important role in the uptake of sorafenib in cancer cells is further supported by the findings that sorafenib can extensively accumulate in liver cancer cell lines with low levels of OCT1 transcripts or in cells with impaired OCT1 functional activity [125,126]. It is conceivable that clinical associations of OCT1 transcript and protein levels only represent a prognostic tool rather than a predictive biomarker for efficacy in patients receiving sorafenib treatment. In addition, genetic variants in OCT1 could be a mere surrogate biomarker of poor response due to linkage disequilibrium with variants in genes that are causally connected with the pharmacologic action of sorafenib. It is also possible that differences in OCT1 expression observed in HCC patient populations and sorafenib response is an indirect result of sorafenib-mediated OCT1 inhibition, resulting in reduced accumulation of endogenous OCT1 substrates or metabolites that may be essential for liver function (e.g., thiamine) [22]. Similarly, functional activity of OCT1 has been suggested as a prognostic marker for long-term resistance in patients with chronic myeloid leukemia receiving imatinib [127]. However, follow-up preclinical studies have demonstrated an OCT1-independent mechanism of imatinib accumulation in hepatocytes and CD34+ mononuclear cells [128–130]. Similar observations have been made for other putative OCT1-TKI interactions, including those with erlotinib, gefitinib, and pazopanib [131,132]. The lack of consistent preclinical and clinical findings on the role of OCT1-mediated liver clearance of TKIs warrant further investigation using properly validated in vivo models.

The recently updated FDA guidance for industry in January 2020 recommends the evaluation of renal OAT1, OAT3, OCT2, MATE1 and MATE2-K, and hepatic OATP1B1, OATP1B3, MDR1, and BCRP for transporter-mediated drug interactions. If preclinical ADME data indicate active renal secretion (CL_r – [f_{unbound, plasma} * GFR]) is responsible for ≥25% of systemic clearance, in vitro transporter-mediated drug interaction studies are required to be performed for the renal transporters OAT1, OAT3, OCT2, MATE1 and MATE2-K. In this context, a drug is considered a substrate if (1) accumulation in a cell-based overexpression system is ≥2-fold compared to accumulation observed in cells transfected with an empty vector; and (2) a known inhibitor decreases uptake by ≥50% at a concentration that is 10-times the predicted inhibition constant (K_i) value. Similarly, a drug is considered an inhibitor if the ratio of IC₅₀ or K_i is 10-fold greater than the average maximum unbound inhibitor concentration (I_max,u) observed in the systemic circulation in patients receiving approved doses [21]. Additionally, the need for in vivo clinical drug interaction as a victim or perpetrator is determined based on the drug’s putative site of action, concomitant use within the patient population, or is dependent on a drug’s narrow therapeutic index [133]. Furthermore, supportive evidence from a pharmacodynamic marker of OAT2, OCT2 and MATE function (e.g., serum creatinine) is required in mechanistic studies [134–136]. However, in the regulatory guidance documents, OCT1 or MATE1-mediated hepatic interactions are not required for evaluation.

Structural and physiochemical analysis of TKIs suggest that many of these agents are cationic and highly lipophilic in nature, and are thus themselves potential transported substrates or inhibitors for OCTs and MATEs. Comprehensive interrogation of the available prescribing information and package inserts showed that 7 of the 48 FDA-approved TKIs are claimed to be inhibitors of OCT2 or MATE1 transport function, respectively [137]. Of note, the presence or absence of an OCT2- or MATE1-mediated drug-drug interaction is not mentioned in the prescribing information for a majority of FDA-approved TKIs. An additional search of relevant literature evidence on TKIs identified 15/48 as confirmed inhibitors of OCT2 (K_i range: 0.02 – 10 μM) and 17/48 as inhibitors of MATE1 (K_i range: 0.09 – 10 μM) (Table 1), while a subset of TKIs also potently inhibit OAT1 and OAT3 function [74,89,137,138]. According to the prescribing information, clinical elevations of increases in serum creatinine (all grades) are observed with the majority of TKIs; however, whether this observation is a consequence of the TKI’s direct influence on cationic-type transport or related to intrinsic nephrotoxic activity associated with TKIs, as documented for example with vemurafenib, remains to be elucidated. With the emerging and ever increasing number of TKIs approved, it is plausible that a large subset of TKIs in Table 1 are actually undocumented substrates or inhibitors of OCT2 and MATE1, and remain to be verified as such.

Table 1. FDA-approved tyrosine kinase inhibitors.

List of FDA-approved tyrosine kinase inhibitors and their indications. ✓ represents TKI as either an OCT2 or MATE1 inhibitor described in the prescribing information. X represents TKI is not an inhibitor of OCT2 or MATE1 at clinically-relevant concentrations, as provided by the prescribing information. Fecal/urine recovery and incidence of creatinine elevations are also provided by the prescribing information. IC₅₀ or K_i values are either exact values or estimates after an exhaustive literature search. The list also contains the reference number and citations that provided all table data or values.

	Indication	OCT2	IC50	MATE1	IC50	Fecal/Urine (% recovered)	Creatinine (% incidence)	Reference/ Citation
Acalabrutinib	BTK	X		X		84/12	↑ SCr; 5%	4523127
Afatinib	EGFR HER2/4				< 10 μM	85/4	↑ SCr; 49%	4207081, [89]
Alectinib	ALK	X				98/0.5	↑ SCr; 28 – 38%	4273596
Avapratinib	PDGFRa KIT			✓		70/18	↑ SCr; 29%	4544122
Axitinib	C-KIT PDGFR		< 10 μM		< 10 μM	41/23	↑ SCr; 55%	3078397, [89]
Baricitinib	JAK			✓		20/75	↑ SCr	4271150
Binimetinib	MEK1 BRAF					62/31	↑ SCr; 93%	4283608
Bosutinib	BCR-ABL SRC		≤ 50 nM		< 10 μM	91/3	↑Renal Impairment; 13%	4083804, [89]
Brigatinib	ALK			✓		65/25		4090797
Cabozantinib	C-MET		< 10 μM		< 10 μM	54/21	↑ SCr; 58%	4375294, [89]
Ceritinib	ALK					92/1	↑ SCr; 58%	4104052
Cobimetinib	MEK1/2 BRAF					76/18	↑ SCr; 99%	3845167
Crizotinib	ALK ROS1 MET	✓	0.34 μM		0.34 μM	63/22	↑ SCr; 98%	4127887, [89,137]
Dabrafenib	BRAF	✓				71/23	↑ SCr; 21%	4255750
Dacomitinib	EGFR HER2/4					79/3	↑ SCr; 24%	4327054
Dasatinib	BCR-ABL SRC		0.02 μM		0.84 μM	85/4	↑ SCr observed	4179887, [89,138]
Encorafenib	BRAF	✓				47/47	↑ SCr; 93%	4283598
Entrectinib	TRKs ROS1 ALK					83/3	↑ SCr; 73%	4477931
Erdafitinib	FGFRs	✓		X		69/19	↑ SCr; 52%	4418725
Erlotinib	EGFR		4.21 μM		7.93 μM	83/8		3308430, [89,138]
Fedratinib	JAK2	✓		✓		77/5	↑ SCr; 10%	4478261
Fostamatinib	SYK					80/20		4249943
Gefitinib	EGFR		24.4 μM		1.92 μM	86/4	↑ SCr; 3%	4310729, [89,137,138]
Gilteritinib	FLT3			✓		64/16	↑ SCr; 3%	4440577
Ibrutinib	BTK					80/10	↑ SCr; 9%	4222705
Imatinib	BCR-ABL		2.37 μM		0.05 μM	68/13	↑ SCr	3993828, [89,137,138]
Lapatinib	EGFR HER2					27/2		4359049
Larotrectinib	TRKs	X		X		58/39		4354331
Lenvatinib	VEGFRs					64/25	↑Renal impairment; 18%	4274001
Lorlatinib	ALK ROS1	X		✓		41/48		4344940
Midostaurin	FLT3					95/5	↑ SCr; 25%	4090671
Neratinib	EGFR HER2/4					97/1	↑Renal impairment; 0.4%	4566456
Nilotinib	BCR-ABL		0.1 μM		3.38 μM	93/	↑ SCr; <1%	4200046, [89,138]
Nintedanib	PDGFRs FGFRs VEGFRs SRC					93/1		4487848
Osimertinib	EGFR	X		X		68/14		4174859
Pazopanib	VEGFRs PDGFRs KIT		≤ 1 μM		< 0.47 μM	90/4		3823458, [89,137,152]
Pexidartinib	CSFR1 KIT FLT3	X		✓		65/27		4471832
Ponatinib	ABL	X	< 10 μM		< 10 μM	85/5	↑ SCr; 21%	4019587, [89]
Regorafenib	VEGFR				< 10 μM	71/19		4090114, [89]
Ruxolitinib	JAK1/2	X	< 10 μM		< 10 μM	22/74		4191667, [89]
Sorafenib	BRAF FLT3		40.8 μM		1.43 μM	77/19		4357655, [137,138]
Sunitinib	PDFGRs VEGFRs FLT3		1.73 μM		0.28 μM	61/16	↑ SCr; 27%	4182165, [89,137,138]
Tofacitinib	JAK1/3					70/30	↑ SCr observed	4269956
Trametinib	MEK1/2 BRAF	X		X		80/20	↑ SCr; 21%	4255758
Upadacitinib	JAKs	X		X		38/24		4478363
Vandetinib	EGFR VEGFR	✓	≤ 1 μM		< 10 μM	44/25	↑ SCr; 16%	3480537, [89]
Vemurafenib	BRAF					94/1	↑ SCr; 27%	4084937
Zanubrutinib	BTK	X				87/8		4520008

Diminished efficacy and increase risk of toxicity are major potential consequences of drug-drug interactions, especially for agents that have a narrow therapeutic window. One example highlighting this risk is the commonly prescribed anti-arrhythmic drug, dofetilide, where a mere 1-ng/mL increase in plasma concentration is associated with an average increase in the QTc prolongation of 15 ms. Previous work demonstrated that concurrent administration of dofetilide with the OCT2 and MATE1 inhibitor, cimetidine, dose-dependently increased the area under the curve of dofetilide due to impaired renal tubular secretion [139], increased dofetilide-induced QTc prolongation, and increased the risk for Torsades de Pointes. The concurrent use of certain FDA-approved TKIs, such as vandetanib and vemurafenib, is contraindicated with dofetilide, presumably due to an unacceptable risk associated with anticipated increases in systemic exposure or local inhibition of efflux from cardiomyocytes [140]. Although the mechanistic underpinning of these potentially life-threatening, albeit hypothetical scenarios remains further study, it is likely that some off-target effects of TKIs, including inhibition of clinically-important transporters creates a liability by possibly potentiating drug-induced toxicities. In addition to the fact that OCT-inhibitory TKIs can influence the function of uptake transporters, several studies have also demonstrated that many TKIs also inhibit ABC transporters that are relevant to drug efflux in tumor cells and as well as excretory pathways of xenobiotics in the liver and kidney [19,20,141,142]. Although the strategic use of TKIs as modulators of transport function may reduce tissue injury, this intervention approach will only be feasible if there are no detrimental effects on the pharmacokinetics and no negative impact on the antitumor efficacy due to concurrent inhibition of relevant ABC transporters. Detailed preclinical studies evaluating the systemic and distribution changes to on- and off-target (tumor vs. healthy cells) must thus be conducted to ensure drug safety.

A particularly poorly studied area of potentially fruitful research is related to the extent to which the inhibitory potential of TKIs against drug transporters and the predictive ability of in vitro test systems is dependent on specific experimental conditions of the employed models. This not only refers to the type of prototypical transport substrate (e.g., TEA, MPP or metformin) and concentration range of the substrate(s) used, but also to culture media and incubation conditions. These details are not provided in the regulatory guidelines, but rather left at the discretion of investigators. TKI-transport inhibition studies have often remained unpublished and unverified, and this is an unsustainable situation in need of an appropriate and cost-effective solution. This is because implementation of dependable and reproducible approaches to unequivocally establish TKI-transporter interactions -or the lack thereof- in a statistically sound and unbiased manner is required to avoid the generation of conflicting results and should ultimately be utilized to inform the design of subsequent in vivo validation studies.

An additional concern pertains to a general lack of information in the prescribing information of TKIs on the mechanism by which these agents can inhibit transport function (non-competitive vs. competitive; reversible vs. irreversible). The presence or absence of either pre- and co-incubation of TKIs with probe substrates for the transporter(s) of interest could further contribute to their lack of reported inhibitory potential of a drug, and result in false negative observations. For example, dasatinib potently inhibits OCT2 function (IC₅₀ = 15.9 nM) through a non-competitive mechanism that is dependent on the phosphorylating kinase YES1 (K_d = 0.3 nM). However, co-incubation, based on a presumed competitive mechanism of inhibition, with the prototypical transport substrate TEA identified dasatinib as only a weak inhibitor of OCT2. In contrast, pre-incubation was sufficient to significantly influence OCT2 function [89]. Of the 48 TKIs investigated (Table 1), only for crizotinib and vandetanib is there strict concordance between the prescribing information and published literature. This is not surprising given the differential capacity for TKIs to inhibit OCT/MATE function and the discrepancy in published reports, prescribing information, as well as interindividual variations among different laboratory settings. Recently, it was suggested that previously reported conflicting findings are due, in part, to the fact that the potency of transport inhibitors can be highly dependent on the substrate used in the overexpressed model systems [143]. More specifically, this study demonstrated that the prototypical OCT/MATE substrate MPP is the least sensitive to inhibition compared to other probe substrates: metformin, TEA, NBD-MTMA and ASP, with an average of 6-fold higher IC₅₀ values, while metformin produced the lowest variability compared to non-MPP substrates [144–146]. Given that OCTs and MATEs share a high degree of substrate and inhibitor specificity, the choice of model substrates used in in vitro cationic-type transport assays should become an integral component of a rationally-designed strategy to make informative, translationally-relevant predictions.

The absence of regulatory guidelines regarding the design and interpretation of in vitro studies aimed at demonstrating an inhibitory potential of drugs with transporters has likely contributed to this dismal state of affairs. Since TKIs are most commonly prescribed on a chronic basis (e.g., once or twice daily) and given together with multiple other medications, it is hoped that investigators will be encouraged to re-evaluate the potential liability of TKIs as perpetrators of transporter-mediated drug-drug interactions to gain new mechanistic insights and to improve the safety of currently used polypharmacy regimens.

5. Expert opinion

Over the past few decades, the contribution of transporters to the accumulation and efflux of therapeutic drugs has been a focus for many academic institutions and pharmaceutical industries [1]. The preclinical utility of heterologous overexpression systems and genetically engineered model systems has substantially contributed to the in vitro and in vivo understanding of the initial pathophysiology of drug-induced organ damage [18]. Many conventional chemotherapeutic agents are now known to hitchhike on the tissue-specific drug transporters and this process can contribute to the incidence and severity of debilitating cellular injuries that could result in partial or complete tissue dysfunction. Many currently ongoing efforts in the field are focused on the treatment and management of symptoms by targeting an intracellular signaling cascade that influences this cellular dysfunction and injury [147], and the targeting of these intracellular mechanisms may antagonize antitumor properties due to overlapping signaling pathways between normal and cancer cells. In contrast, the targeting of transmembrane transport mechanisms unique to healthy tissues may offer preventative approaches that can be exploited strategically (Figure 1), without negatively influencing antitumor properties and disease management [148]. The class of TKIs is of particular interest in this context due to their inhibitory potency towards members of the SLC and ABC family and because of their already approved use in the treatment of a wide array of solid tumors and hematological malignancies [117]. The success of these approaches is also dependent on the lack of unintended inhibition in off-target and healthy cells. The strategy of integrating transport inhibitors to prevent drug-induced toxicity can be optimized to inhibit local exposure without altering the systemic pharmacokinetics of the chemotherapeutic agent. Rationally-designed preclinical and clinical studies that document the local and systemic changes in endogenous substrates of OCT/MATE should also be integrated to serve as a pharmacodynamic biomarker of transport function to achieve an optimal degree of temporary inhibition.

Figure 1.

Proposed mechanism of chemotherapy-induced cellular injury and clinical intervention. Expression and localization of cationic transporters in cochlear and proximal tubular cells, and dorsal root ganglion as mediators of established toxicity profiles associated with chemotherapeutic agents. Putative transport of doxorubicin in cardiomyocytes.

The recently updated FDA guidelines for in vitro drug interaction studies do not necessarily predict the probability of transporter-mediated drug interactions, and further revision of these guidelines is warranted. The absence of detailed guidance on experimental methodology and study design, which are now at the discretion of investigators, can contribute to the generation of false negative data, as evidenced by the noted discrepancies between data contained in package inserts and those available in the published literature (Table 1). It is our contention that the number of currently FDA-approved TKIs that is able to substantially affect the function of OCTs and/or MATEs is underestimated [116,140], and in that context the following recommendations are offered:

1). Background Uptake.

The identification of transporters involved in TKI uptake with the use of currently available in vitro model systems is difficult due to high background accumulation in most mammalian cells, such as HEK293, HeLa and MDCK-II cells [124]. Oocytes from Xenopus laevis are a potentially useful alternative overexpression system for such studies due to their ability to translate foreign genetic material (including post-translational modifications), correct localization to subcellular compartments, and have a low number of endogenous transport systems expressed on their plasma membranes that may differentiate between active uptake and signal noise [149]. Many TKIs also preferentially accumulate at high concentrations within lysosomes and/or bind extracellularly, and these properties may further artificially increase the background reading and be mistaken for mediated uptake. Manipulation of proton concentrations in incubation buffers and known inhibitors that are absent of lysosomal properties can potentially delineate between lysosomal trapping and transporter-mediated uptake [150,151].

2). Identification of inhibitors.

Transport of prototypical substrates (e.g., TEA, MPP and metformin) is dependent on their kinetic affinity for OCTs and MATEs. Although the guidelines outline I_{max, unbound} /IC₅₀ or K_i ≥ 0.1 as the criterion for consideration, a drug’s putative inhibition potential could also be dependent on the duration of uptake studies, and this is rarely considered in the experimental design. A requirement should be outlined regarding different inhibitor incubation methods, as transport function might be sensitive to the presence or absence of pre-incubation conditions. This also helps delineate between non-competitive and competitive mechanisms of inhibition. If a drug is identified as an inhibitor of a transporter, reversibility of the property should also be verified in order to allow translationally-relevant predictions to be made, depending on the chosen dose and frequency of administration of the agent to humans. Indeed, irreversible inhibition or prolonged times required for recovery of transport function are important considerations in order to arrive at meaningful predictions regarding the long-term functional consequences of TKI-mediated transport inhibition and to properly predict interactions with concomitant medications.

More comprehensive evaluation of drug-transporter interactions should become an integrated approach to human studies aimed at understanding basic pharmacokinetic properties of approved and investigational oncology drugs, and it is anticipated that such studies will continue to make significant contributions to the design of clinical studies aimed at improving the safety of patients afflicted with cancer.

Article highlights.

• Organic cation-type transporters are integral membrane proteins that facilitate the translocation of endogenous substrates and xenobiotics across a biological membrane.

• Debilitating toxicities associated with several chemotherapeutic agents are dependent on the tissue-specific expression of organic cation transporters that governs their cellular accumulation.

• Modulating transport function to prevent toxicity without antagonizing therapeutic benefits of anticancer agents represents a novel intervention strategy that can be translationally exploited.

• Tyrosine kinase inhibitors are an emerging class of anticancer therapeutics that may influence transport function to a greater extent than held presently, and these agents are thus liable to perpetrate drug-drug interactions and affect drug-induced toxicity.

• Recently updated regulatory guidelines on the conduct of in vitro drug interaction studies remain ambiguous and warrant further revision.