In Vitro and In Vivo Models for Drug Transport Across the Blood-Testis Barrier

Abstract

The blood-testis barrier (BTB) is a selectively permeable membrane barrier formed by adjacent Sertoli cells (SCs) in the seminiferous tubules of the testes that develops intercellular junctional complexes to protect developing germ cells from external pressures. However, due to this inherent defense mechanism, the seminiferous tubule lumen can act as a pharmacological sanctuary site for latent viruses (e.g., Ebola, Zika) and cancers (e.g., leukemia). Therefore, it is critical to identify and evaluate BTB carrier-mediated drug delivery pathways to successfully treat these viruses and cancers. Many drugs are unable to effectively cross cell membranes without assistance from carrier proteins like transporters because they are large, polar, and often carry a charge at physiologic pH. SCs express transporters that selectively permit endogenous compounds, such as carnitine or nucleosides, across the BTB to support normal physiologic activity, although reproductive toxicants can also use these pathways, thereby circumventing the BTB. Certain xenobiotics, including select cancer therapeutics, antivirals, contraceptives, and environmental toxicants, are known to accumulate within the male genital tract and cause testicular toxicity; however, the transport pathways by which these compounds circumvent the BTB are largely unknown. Consequently, there is a need to identify the clinically relevant BTB transport pathways in in vitro and in vivo BTB models that recapitulate human pharmacokinetics and pharmacodynamics for these xenobiotics. This review summarizes the various in vitro and in vivo models of the BTB reported in the literature and highlights the strengths and weaknesses of certain models for drug disposition studies.

SIGNIFICANCE STATEMENT

Drug disposition to the testes is influenced by the physical, physiological, and immunological components of the blood-testis barrier (BTB). But many compounds are known to cross the BTB by transporters, resulting in pharmacological and/or toxicological effects in the testes. Therefore, models that assess drug transport across the human BTB must adequately account for these confounding factors. This review identifies and discusses the benefits and limitations of various in vitro and in vivo BTB models for preclinical drug disposition studies.

Introduction

The blood-testis barrier (BTB) is a critical component of the male reproductive tract that safeguards the normal spermatogenic process from external threats. It is a selectively permeable cell barrier formed by adjacent Sertoli cells (SCs) in the seminiferous tubules that express intercellular junctional proteins to limit the accumulation of reproductive toxicants in the male genital tract (Dym and Fawcett, 1970). Except for spermatogonia, all developing germ cells are protected within the adluminal compartment of the seminiferous tubules. In addition to the network of membranes that physically blocks access to the male genital tract, SCs also express various transporters along their basal membranes to regulate the disposition of compounds across the BTB (Mital et al., 2011; Mruk et al., 2011; Mruk and Cheng, 2015). Moreover, SCs express and secrete immunomodulatory factors to achieve testicular immune privilege (Fijak and Meinhardt, 2006; Li et al., 2012; Zhao et al., 2014). Together, these three factors actively protect developing germ cells residing on either side of the BTB. However, the seminiferous tubule lumen can act as a sanctuary site for some viruses (e.g., Ebola, HIV, Zika) and cancers (e.g., leukemias) because certain xenobiotics cannot effectively cross the tight SC epithelia to achieve pharmacologically relevant concentrations in the adluminal compartment. There are many reports of testicular persistence and/or relapse of viruses and cancers, despite eradication from the blood and other tissues (Nesbit et al., 1980; Ritzén, 1990; Byrn and Kiessling, 1998; Kiessling et al., 1998; Politch et al., 2012; Houzet et al., 2014; Deen et al., 2017; Robinson et al., 2018; Ma et al., 2021). This phenomenon highlights a need to identify the mechanisms that permit antiviral or antineoplastic therapies to successfully treat these diseases in the testes.

Uptake transporters present at the basal membrane of SCs are vital for the delivery of small molecule therapeutics across the BTB. Pharmacologically relevant uptake transporters include the organic cation transporters (OCTs), novel organic cation transporters (OCTNs), organic anion transporters (OATs), organic anion transporting polypeptides (OATPs), and equilibrative nucleoside transporters (ENTs). Efflux transporters, such as P-glycoprotein (P-gp), breast cancer resistance protein (BCRP), and the multidrug resistance-associated proteins (MRPs), act in opposition to prevent the accumulation of toxic molecules in SCs, although not all uptake transporter substrates are (good) efflux transporter substrates. There are also dozens of other understudied BTB transporters that can influence the disposition of drugs to treat testicular indications, although they will not be discussed in this review. Collectively, the listed transporters are involved in the flux of many clinically relevant drugs and environmental toxicants, but many of these BTB pathways have not been adequately defined in rodents, humans, and other species.

There are in vitro and in vivo BTB models across different species reported in the literature, but there is a significant lack of information on their relevance for drug transport studies. As a result, there is a need to evaluate the benefits and limitations of these in vitro and in vivo BTB models to assess testicular drug disposition. Furthermore, the information that is known about some efflux transport pathways, such as MRP4 and BCRP, indicate that rodent models may not recapitulate events in humans (Hau et al., 2021, 2023). Therefore, a standardized array of clinically relevant BTB models and assays to study drug disposition across the BTB is warranted. The purpose of this review is to provide a comprehensive overview of the mammalian BTB models described in the literature and to outline their potential in preclinical drug disposition studies.

In Vitro Models

Rodent Sertoli Cell Models

Rodent SC models are commonly used for a variety of applications to study the physiology of the testes and the BTB. This is due to the comparatively easy access to fresh rodent testes for cell isolation and the mandatory use of rodent models in preclinical studies for assessing reproductive toxicity. Historically, BTB drug transport has been studied with primary SCs isolated from immature and mature rats or mice. Immature rodents are used due to their ease of isolation and the lack of contaminating germ cells for SC isolation; however, mature rodents with an established BTB are more physiologically relevant. In vivo, the establishment of the functional and complete BTB occurs after 15 to 18 days postpartum in mice (Moroi et al., 1998; Hogarth, 2015; Willems et al., 2010; Xiong et al., 2018) and 15 to 21 days postpartum in rats (Bergmann and Dierichs, 1983; Russell et al., 1989; Toyama et al., 2001). This is important because data collected using immature SCs may not be representative of the findings collected in adult animals due to ontogenetic changes in transporter expression (Brouwer et al., 2015; Cheung et al., 2019; van Groen et al., 2021; Felmlee and Zhang, 2022). These transporter expression changes may be, in part, associated with changing hormone or immunomodulatory factor levels as animals mature and spermatogenesis begins. Interestingly, one study observed downregulation of P-gp in rat SCs after exposure to estradiol-17β, testosterone, and transforming growth factor-β3 but no change after exposure to interleukin-1α and tumor necrosis factor-α (Su et al., 2012). On the other hand, estradiol-17β and interleukin-1α also upregulated rat OATP1A5, but the other hormones or cytokines had no effect. Hormone and cytokine levels can fluctuate significantly over time and are also sensitive to various environmental factors (e.g., infections) (Larsson et al., 2015; Ter Horst et al., 2016; Bell, 2018). For example, testosterone levels are generally low after birth and gradually rise during peripuberty due to the onset of spermatogenesis, but its downregulatory effect on P-gp is counterintuitive. It is possible that hormone and cytokine levels indirectly regulate the intracellular levels of various biomolecules (e.g., steroids, prostaglandins, and nucleosides) by modulating transporter expression. However, it is clear that the effect of hormones and cytokines on transporter expression warrants further evaluation because there are dozens of other physiologically and/or pharmacologically relevant transporters expressed in the testes.

It is also possible that transporter expression and localization can be altered during exposures to certain environmental toxicants, such as zearalenone or cadmium (Su et al., 2012; Koraïchi et al., 2013). Cadmium chloride exposure was shown to downregulate P-gp, MRP1, ABCG1, OATP1A5, PEPT1, and ZIP8 in cultured rat SCs, whereas neonatal exposure to zearalenone causes MRP4 to redistribute from the basal membrane to the apicolateral membrane of SCs. It is known that certain disease states affect transporter expression in other tissues (Evers et al., 2018). Therefore, it is conceivable that some testicular indications may cause changes in transporter expression and, ultimately, affect the disposition of drugs to the testes. Thus, it is critical to account for age, environmental exposures, and other confounding factors when assessing in vitro and in vivo drug disposition to the testes. This process may involve the use of more than one in vitro and in vivo model and/or the use of multiple animal replicates, but it is necessary to make an informed conclusion.

While drug transport is often performed on flat monolayers of nonpolarized cells plated on standard plastic cell culture vessels, differences in transporter expression between immature and mature SCs can affect drug disposition across these membranes. Additionally, drug transport on these nonpolarized monolayers is not representative of in vivo epithelial barriers due to cell polarization directing transporter localization (Cao et al., 2012). Junctional complexes formed between adjacent SCs separate the cell membrane into apical/apicolateral or basal/basolateral domains, leading to differential transporter expression on each side. There is no distinction between the basal and apicolateral transporters when SCs are cultured as flat monolayers. Consequently, it is possible that while transporter-mediated uptake of a compound may be observed in vitro, it would not cross the in vivo BTB because the transporter is expressed at the apicolateral membrane of SCs. Transporter expression and localization maps such as the one described by Hau et al. (Hau et al., 2023) for human testes could provide some context, but functional studies to validate these findings are still warranted.

The specificity, selectivity, expression, and localization of drug transporters at the BTB can vary between species (Chu et al., 2013; Hau et al., 2021). For example, a previous study showed that, whereas MRP4 localizes to the basal membrane of SCs in humans and rhesus macaques, expression is restricted to the apicolateral membrane of SCs in normal rodents (Klein et al., 2014). Several other studies reported basal membrane localization of MRP4 in humans (Morgan et al., 2012; Huang et al., 2016). Importantly, another study observed MRP4 at the basal membrane of SCs, Leydig cells, and peritubular myoid cells in rats but found that it redistributes to the apicolateral membrane of SCs, spermatocytes, and spermatids after neonatal treatment with zearalenone (Koraïchi et al., 2013). Interestingly, unlike in rats or humans, MRP4 in mice is exclusively localized to the Leydig cells (Morgan et al., 2012, 2015). These contradictory findings are not limited to MRP4, as they have been observed with another major efflux transporter, BCRP. In humans, BCRP is localized to the basal and apicolateral membranes of SCs, peritubular myoid cells, developing germ cells, and vascular endothelial cells (Melaine et al., 2002; Huang et al., 2016). Notably, although BCRP in rats is expressed in the same cells as humans, it is absent at the basal membrane of SCs (Qian et al., 2013). Furthermore, BCRP is exclusively localized to the vascular endothelial cells in mouse testes (Enokizono et al., 2007). These observations suggest that the use of rodent models for studying drug transport across the BTB may not always be representative of what is observed in humans. Nevertheless, studying drug transport with these in vitro rodent models can provide insight into at least some of the transporter mechanisms present in SCs across species. Indeed, flat monolayers of primary rodent SCs have been used to meticulously study the interaction between nucleosides and nucleoside analogs with the ENTs (Kato et al., 2005, 2006, 2009; Klein et al., 2013; Miller et al., 2021a). Unfortunately, this model only describes basic drug-transporter interactions and does not account for transporter interactions at specific membranes of a polarized SC.

Although these basic rodent monolayer models are a valuable tool, the development of a two-compartment cell culture system to study transepithelial permeability and polarized chemical secretion across primary rodent SCs was a significant breakthrough (Janecki and Steinberger, 1986, 1987; Janecki et al., 1987; Steinberger and Klinefelter, 1993). These polarized cell models are much more representative of the in vivo BTB; however, they still neglect the potential physiologic importance of the peritubular myoid cells and developing germ cells that, along with the SCs, compose the seminiferous tubules. Initial studies with these primary rat SC layers reported transepithelial electrical resistance (TEER) values exceeding 100 Ohms•cm² (Janecki et al., 1991a,b; Steinberger and Klinefelter, 1993). More recent studies observed peak TEER values of 70 Ohms•cm² from primary rat SCs (Kaitu’u-Lino et al., 2007; Nicholls et al., 2009; Mruk and Cheng, 2011). Similarly, TEER values in primary mouse SCs are reported to reach up to 60 Ohms•cm² (Wu et al., 2017a,b; Hu et al., 2021; Liu et al., 2021). But these are comparatively low TEER values compared with those observed in Caco-2 and MDCK cells, which surpass 1000 Ohms•cm² when the tight junctions are fully developed (Prozialeck and Lamar, 1997; Srinivasan et al., 2015). It is possible that the unusual morphology and formation of tight junctions near the basal membrane of SCs contribute to the lack of a coherent, 3D epithelial monolayer on two compartment cell culture inserts. As a result, SCs plated in two-compartment inserts may permit paracellular diffusion of some chemicals that would not occur in vivo and, consequently, cannot accurately reflect the significantly tighter in vivo BTB. Although there is currently no way to directly measure the TEER of an intact epithelial barrier in vivo, a recent study showed that the membrane-impermeable protein biotinylation reagent sulfo-NHS-LC-biotin was unable to penetrate the intact in vivo BTB but was unimpeded when the barrier was disrupted by CdCl₂ (Chen et al., 2018). Furthermore, other studies have demonstrated that the intact in vivo BTB was impenetrable by various tracers including lanthanum and NHS-linked biotin (Neaves, 1973; Tarulli et al., 2008; Carette et al., 2010; McCabe et al., 2010; Pelletier, 2011). One study assessed the membrane permeability of primary mouse SCs; however, it was not relevant to the paracellular permeability of the BTB (Liu et al., 2020). These findings suggest that the disposition of some small molecules across the BTB is largely impossible without the help of carrier-mediated processes like transporters.

Previous studies have assessed the transepithelial transport of [³H]L-carnitine and [¹⁴C]TEA across primary rat SCs cultured on two-compartment inserts (Kobayashi et al., 2005; Maeda et al., 2007). Based on higher transporter affinity for L-carnitine, compared with TEA, OCTN2 was suggested to be responsible and to localize to the basal membrane of rat SCs and OCTN1 to the apicolateral membrane. L-carnitine is critical for maintaining normal testicular physiology to support spermatogenesis and is the endogenous substrate of OCTN2 (Tamai et al., 1998; Wang et al., 1999; Palmero et al., 2000; Agarwal and Said, 2004; Mongioi et al., 2016). Furthermore, OCTN2 is known to localize to the basal membrane of human SCs, where it may support the uptake of L-carnitine and other molecules across the BTB (Hau et al., 2022a, 2023). A similar study that evaluated transepithelial transport of [³H]adjudin across primary rat SCs demonstrated that it is a substrate for OCTN2, OATP1A5, OATP6B1, and OATP6C1 expressed at the basal membrane (Su et al., 2011). But, as described previously, these transepithelial transport experiments are not adequate models of the in vivo BTB due to the absence of tighter intercellular junctional complexes to prohibit paracellular diffusion and the lack of other testicular cells that can influence SC protein expression. For example, developing germ cells are known to stimulate the expression of genes such as transferrin in SCs (Djakiew and Dym, 1988; Le Magueresse et al., 1988; Stallard and Griswold, 1990). Furthermore, the peritubular myoid cells that interact with SCs are also partly responsible for generating and maintaining the extracellular matrices that affect junctional protein dynamics at the BTB (Skinner et al., 1985). Consequently, a simplified in vitro transepithelial transport model of the rodent BTB is insufficient to reproduce the complexities of the in vivo barrier. Similarly, transport assays with nonpolarized monolayers of rodent SCs can provide higher throughput compared with two-compartment models, but, as noted earlier, the absence of normal epithelial structure compromises the interpretation of the results. Nevertheless, despite these limitations, and with important caveats outlined here, drug transport assays conducted with primary rodent SCs cultured as a monolayer on standard cell cultureware or two-compartment inserts are important for exploratory in vitro studies.

To expedite studies on the male reproductive system, several immortalized rodent SC lines have been established. One of the most used rodent SC lines are TM4 cells, which were isolated from the testis of 11- to 13-day-old, immature BALB/c mice (Table 1) (Mather, 1980). Since these cells were isolated from an immature mouse, drug transport data generated from TM4 cells may not be accurately reflected in studies with mature SCs or animals. Nevertheless, TM4 cells have been used to study a variety of basic physiologic functions in the testis including cell signaling, metabolism, compound transport, and BTB integrity. Basic transport studies have been performed with flat monolayers of TM4 cells plated on standard plastic cell culture plates (Selva et al., 2004; Kato et al., 2005, 2009; Robillard et al., 2012; Huang et al., 2014; Hoque et al., 2015; Ge et al., 2018; Kubo et al., 2022). Studies assessing the transport of clinically relevant drugs in TM4 cells have been largely limited to nucleoside-based antivirals (Kato et al., 2005, 2009; Robillard et al., 2012; Hoque et al., 2015). However, there are many transporters that can transport other clinically relevant compounds across SCs that have not been analyzed in TM4 cells. Compared with primary rat SCs, TM4 cells had a lower capacity for uridine uptake over 10 minutes and were mediated by a saturable and nonsaturable component (Kato et al., 2005). TM4 cells also had negligible functional expression of the Na⁺-dependent concentrative nucleoside transporters, whereas the concentrative nucleoside transporters appeared to mediate at least 25% of the uridine transport in primary rat SCs (Kato et al., 2005). Furthermore, the affinity of various endogenous nucleosides and nucleobases appeared to vary between primary rat SCs and TM4 cells, which may be attributed to species differences in transporter expression and function (Kato et al., 2005, 2006, 2009). These observations differ when compared with primary human SCs and human TERT-immortalized SC (hT-SerC) line, which appear to only functionally express ENT1 (Hau et al., 2020).

TABLE 1.

List of immortalized rodent SCs published in the literature

Cell Line	Species (Strain)	Age	References
15P-1	Mus musculus (C57BL/6J x DBA/2 PyLT transgenic)	6 months postpartum	Paquis-Flucklinger et al., 1993
42GPA9	Mus musculus (Tg(PyLT) transgenic)	6 months postpartum	Bourdon et al., 1998
93RS2	Rattus norvegicus (CD)	16 days postpartum	Jiang et al., 1997
A3/C2#	Mus musculus (C57BL/6J)	6 weeks postpartum	Sato et al., 2013
ASC-17D	Rattus norvegicus (Sprague-Dawley)	120 days postpartum	Roberts et al., 1995
C3H	Mus musculus (C3H Heston)	16 months postpartum	Franks, 1968
G5/SG5/J5#	Mus musculus (BALB/c)	6 days postpartum	Hofmann et al., 2003
MSC-1	Mus musculus (hMIS-SV40 transgenic)	13 weeks postpartum	Peschon et al., 1992
PASC1	Rattus norvegicus (Sprague-Dawley)	adult*	Kabbesh et al., 2021
RTS3-3	Rattus norvegicus (Wistar SV40-tsA58 transgenic)	adult*	Tabuchi et al., 2003
S14-1	Mus musculus (BALB/c)	20 days postpartum	Boekelheide et al., 1993
SCIT-C8	Rattus norvegicus (Wistar)	19 days postpartum	Konrad et al., 2005
SerW3	Rattus norvegicus (Wistar)	17 days postpartum	Pognan et al., 1997
SMAT1	Mus musculus (B6.D2 x B6.CBA F2 AMH-SV40 transgenic)	6.5 days postpartum	Dutertre et al., 1997
TM4	Mus musculus (BALB/c)	11–13 days postpartum	Mather, 1980
TR-ST	Rattus norvegicus (Long-Evans)	2 years postpartum	Mather et al., 1982
TTE3	Mus musculus (C57BL/6J SV40-tsA58 transgenic)	8 weeks postpartum	Tabuchi et al., 2002

^#Multiple clones were generated in the study.

*Age was not reported at the time of cell isolation.

In addition to drug transport across flat monolayers, a few studies have assessed transepithelial transport across TM4 cells in a two-compartment system, but there are conflicting reports on TEER values for unperturbed TM4 cells. Whereas some studies report TEER values of ∼20 to 60 Ohms•cm² (Papadopoulos et al., 2016; Schrade et al., 2016; Wang et al., 2022; Zhang et al., 2022), others report significantly higher values (> 200 Ohms•cm²) (Hu et al., 2014; Zhou et al., 2018; Fisher et al., 2019). These conflicting values may be attributed to different culture systems and methods (note: some publications incorrectly reported TEER as Ohms/cm² rather than Ohms•cm², and these values have been recalculated here based on available information). While transepithelial transport studies may be adequately performed with TM4 cells achieving TEER values of > 200 Ohms•cm², the reproducibility and clinical relevance is questionable due to the TEER values observed in primary mouse SCs (Wu et al., 2017a,b; Hu et al., 2021; Liu et al., 2021) and differing expression, localization, and selectivity/kinetics of the mouse transporter orthologs compared with other species (Hau et al., 2022a, 2021, 2023).

There are many other immortalized rodent (rat or mouse) SC lines, albeit generated from animals of different ages, and these are listed in Table 1. Unlike primary rodent SCs or TM4 cells, fewer studies assessing transporter function have been performed with these cell lines. One study investigated the function of the H⁺-coupled monocarboxylate transporter 4 on lactate transport in 15P-1 cells following cadmium or lead exposure (Yu et al., 2019). Monocarboxylate transporter 4 predominantly exports lactate out of cells, although it can function as a bidirectional transporter for other endogenous monocarboxylates and interacts with monocarboxylate xenobiotics like indomethacin and diclofenac (Manning Fox et al., 2000; Sasaki et al., 2016; Contreras-Baeza et al., 2019). Nevertheless, the utility of mouse 15P-1 cells when directly investigating xenobiotic disposition remains understudied. Similarly, there is only one published report on transporters in mouse 42GPA9 cells; however, this study was limited to L-ascorbic acid transporters (Angulo et al., 2008). There have been several studies assessing the reproductive toxicity of dichlorodiphenyltrichloroethane, doxorubicin, and other toxicants in the SerW3 cell line (Fiorini et al., 2004; Bernard et al., 2007; Fiorini et al., 2008; Qiu et al., 2016; Tremblay and Delbes, 2018), although there are only two additional studies that were limited to investigating the interaction between flusilazole or zearalenone with ATP-binding cassette transporters (Koraïchi et al., 2013; Karacaoğlu, 2022). Unfortunately, the expression and function of pharmacologically-relevant uptake transporters have not been explored in any of these immortalized rodent cell lines, which is a major disadvantage for their informed use in drug disposition studies.

The use of rodent BTB models carries an important caveat; namely, they are not always representative of clinical findings. Some transporters, such as rodent and human OCT1, share similar selectivity and kinetics for at least some drugs; however, other OCT1 substrates/inhibitors may diverge from this trend (Floerl et al., 2020). Additionally, functional differences in P-gp transport between species have been shown to affect drug disposition across the blood-brain barrier (Syvanen et al., 2009). Consequently, assessing drug disposition across the BTB in different species may be subject to similar correlation issues. Drug transport with primary rodent SCs is straightforward because gene expression in immortalized rodent SC lines can be highly variable and fresh primary cell isolation is convenient. However, primary rodent SCs may not express orthologous transporters to human cells, and the selectivity, kinetics, and localization of the orthologous transporters that are expressed can be significantly different (Chu et al., 2013; Hau et al., 2021, 2023). Furthermore, the availability of human SC models and other heterologous expression systems for human transporters renders in vitro rodent models less useful. In cases where in vitro rodent models are used, it is extremely important to validate the generated drug transport data with more clinically relevant models for testicular drug disposition.

Human Sertoli Cell Models

Unlike rodent SCs, obtaining freshly isolated, healthy human SCs is difficult due to the relative rarity of healthy donor tissue and the associated regulatory and logistical hurdles. Furthermore, the human BTB is not completely established until puberty (around age 12–13), reflecting the absence of developing germ cells that require the protection afforded by the BTB (Furuya et al., 1978). This corresponds to observations made in the testes of rodents, where the BTB is not completely formed until later in life. Consequently, it is fundamentally important to use adult human SCs for BTB studies instead of those from a fetus or child due to major differences in gene expression, physiology, and morphology. Isolation of primary adult human SCs has been reported in many studies (Lipshultz et al., 1982; Holmes et al., 1983; Buch et al., 1988; Berensztein et al., 1992; Santiemma et al., 1992; Cudicini et al., 1997; Chandley et al., 1996; Cowan et al., 2010; He et al., 2010; Chui et al., 2011; Guo et al., 2015; Jesus et al., 2016; Wen et al., 2017), but the modest yield of these preparations limits the number of studies that can be conducted at one time. Gene expression profiles in primary cells are generally unstable after dissociation from the native tissue and these cells will eventually undergo cellular senescence (Hayflick, 1965; Olovnikov, 1996). Because of these practical difficulties and limitations associated with the use of primary human SCs, heterologous expression systems and/or nonhuman SCs have been the gold standard for drug transport studies.

Although there are many published reports recognizing the importance of drug transport at the BTB, there are few studies that directly assess the function of pharmacologically relevant drug transporters in human SCs (Klein et al., 2013; Hau et al., 2020; Miller et al., 2021a; Hau et al., 2022b). Many of these studies have primarily focused on the function of Na⁺-independent ENT1 and ENT2. Other common uptake and efflux transporters are expressed in human testes (Kullak-Ublick et al., 1995; Melaine et al., 2002; Pizzagalli et al., 2002; Suzuki et al., 2003; Bart et al., 2004; Bleasby et al., 2006; Huber et al., 2007; Klaassen and Aleksunes, 2010; Robillard et al., 2012; Fietz et al., 2013; Klein et al., 2014; Huang et al., 2016; Hau et al., 2022a), but little is known about how these other transporters collectively affect drug disposition across the BTB. Due to the importance of identifying reproductive toxicity during drug development, understanding transporter-mediated disposition across the human BTB and which models are appropriate is vital for successful preclinical and clinical studies.

Currently, there are two immortalized human SC lines that have been reported in the literature (Wen et al., 2017; Hau et al., 2020). Both cell lines were generated by transducing human telomerase reverse transcriptase from the primary SCs of adult patients. The hS1 cell line was generated from the primary SCs of an adult male of unknown age with obstructive azoospermia (Wen et al., 2017). However, these cells are not publicly available, and drug transport studies with them have not been reported. Conversely, the mRNA expression of common drug transporters was described in hT-SerCs, which were derived from a recently deceased, healthy 36-year-old adult male (Chui et al., 2011; Gaur et al., 2018; Hau et al., 2020). Functional transport studies with hT-SerCs have been reported, although the protein expression levels of many transporters are unknown (Hau et al., 2020; Miller et al., 2021a; Hau et al., 2022b). The novel male contraceptives, H2-gamendazole and gamendazole, are known to exert reversible infertility in several species and are taken up by human and rat SCs by one or more transporters, given its large structure and negative charge at physiologic pH (Tash et al., 2008a; Tash et al., 2008b; Gupta et al., 2011; Hau et al., 2022b). It is speculated that the transporter(s) responsible for H2-gamendazole uptake is (are) shared between species; however, additional work is necessary to identify each specific transporter. However, species differences in testicular transporter expression, localization, and function have been reported and is a major concern for drug disposition to the testes (Melaine et al., 2002; Enokizono et al., 2007; Morgan et al., 2012; Klein et al., 2013; Koraïchi et al., 2013; Klein et al., 2014; Morgan et al., 2015; Huang et al., 2016; Hau et al., 2020, 2021, 2023).

It is important to note that these transport studies were conducted using monolayers of hT-SerCs on standard plastic cell culture plates. Unfortunately, hT-SerCs and primary human SCs do not form a tight, polarized epithelial monolayer on two-compartment systems. Compared with rodent SCs, MDCK cells, and Caco-2 cells, human SCs form a leakier in vitro barrier, as is evident by low TEER values ranging from ∼10 to 30 Ohms•cm² (Chui et al., 2011; Chen et al., 2017; Siemann et al., 2017; Gaur et al., 2018; Hau et al., 2020). Consequently, human SCs cultured on two-compartment inserts permit significant paracellular diffusion of test compounds, and transporter-mediated transepithelial flux of these compounds is difficult to resolve above this background signal. Overexpression of tight junction proteins in hT-SerCs to promote intercellular interactions has been attempted (unpublished data); however, there was no difference in TEER values compared with untransfected cells. This limits their utility and relevance in preclinical drug disposition studies; however, like rodent SCs, human SCs are one of many predictive tools to study drug transport across the BTB. Human SCs may be a better predictive tool of drug disposition across the BTB since they express human transporters that may exhibit differing substrate selectivity or kinetics, although further work is needed to provide a more comprehensive comparison.

Other Mammalian Sertoli Cell Models

Although isolation of SCs from other mammalian species is feasible and has been done for pigs, bulls, sheep, canines, nonhuman primates, and other animals (Vazeille and Chevalier, 1980; De Martino et al., 1985; Renier et al., 1986; Tung et al., 1987; Coombs and Jenkins, 1988; Monet-Kuntz and Fontaine, 1989; Merhi et al., 2001; Davidson et al., 2007; Guibert et al., 2011; Zhang et al., 2013), the clinical relevance of these models is ambiguous. In addition, functional drug transporter studies have not been performed with any of these mammalian SCs. Although there have been reports on BTB transporter expression and localization in nonhuman primates and horses (Klein et al., 2014; Merkl et al., 2016), it is restricted to only the BTB localization of MRP1, MRP4, MRP5, and MRP8 in mature rhesus macaques and to the ATP-binding cassette transporter A1 in horses. The expression and localization of common pharmacologically relevant uptake transporters at the BTB are unknown in these species; however, nonhuman primate models are generally regarded as the closest representation of human physiology. An extensive map of transporter localization, selectivity, and kinetics for the relevant species must be generated first before widespread adoption of these other models in BTB drug transport studies can ensue. Currently, the use of primary human SCs, SC lines, or heterologous expression systems for simple drug transporter studies across the BTB is preferable to cells from other mammalian species.

Complex 3D Culture Models

Although current SC in vitro models are largely limited to conventional flat monolayers of cells plated on standard cell cultureware or on two-compartment inserts, some studies have used co-culture methods to improve the genotypical or phenotypical characteristics of two-compartment cell systems (Hilgendorf et al., 2000; Megard et al., 2002; Zhang et al., 2011). However, reconstructing a tight SC barrier to study active transport across the BTB remains a major objective. As previously explained, SCs tend to form weak intercellular interactions in vitro on two-compartment inserts, even with cell culture additives, matrices, and co-cultures. Furthermore, assessing transepithelial transport of, or inhibition of transport by, tens or hundreds of different compounds with these systems is unrealistic. Most compounds interact with more than one transporter and the transporters expressed by peritubular myoid cells and interstitial Leydig cells may affect drug disposition across the BTB, since some of the same common drug transporters are also expressed by these cells as in SCs (Klein et al., 2014; Hau et al., 2022a, 2023). Therefore, more complex in vitro BTB models that more effectively recapitulate in vivo physiology are warranted.

Testicular organoids, ex vivo models, and other 3D cultures have gained substantial popularity in recent years due to their closer approximation to in vivo physiology. Many advanced testicular models have been reported in the literature with different approaches to generate a more physiologically relevant model, as reported in Table 2. However, some of these systems are derived from fetal tissues and may be less representative of an established BTB in adult animals. Furthermore, some of these models rely on extracellular matrices to maintain the desired cellular aggregate morphology, which can adversely affect drug transport studies. Although some of these complex models are also grown on two-compartment inserts, they aggregate and adopt a spheroid-like morphology unlike the flat monolayers in the previously described models. Drug transporter function has also been studied in organoid models, but these experiments rely on fluorescent substrates that are not available for some transporters (Achilli et al., 2014; Zhang et al., 2019). Recently, one study demonstrated that measuring transport of radiolabeled glucose, fructose, or dipeptide glycyl-sarcosin across human intestinal organoids was feasible (Zietek et al., 2020). Unfortunately, these techniques have not been applied to testicular organoid models. An ex vivo assay to assess radiolabeled uridine transport across intact rat seminiferous tubules has been described (Klein et al., 2013), but this method is technically challenging, time-intensive, and does not adequately define the role of apicolateral transporters in SC transcellular transport, which may be important for the disposition of certain compounds into the adluminal compartment of the seminiferous tubule. It is feasible that drug transport assays can be adapted for use in some of the cultured models listed in Table 2, but they would need to be individually optimized for radiolabeled or fluorescent substrates and for unlabeled compounds detected by liquid chromatography-tandem mass spectrometry. It is important to note that these complex models can be used for other applications when studying reproductive physiology, but high-throughput drug transport studies with these models are impractical and are best suited for the most promising drug candidates.

TABLE 2.

List of complex rodent or human 3D culture models that may be adapted for use in low-throughput drug transport studies

Model	Species	Age	References
Air-liquid interface	Mus musculus (male pCXN-GFP, Acr-GFP, or Gsg2-GFP transgenic x female ICR, C57BL/6J, or ICR x C57BL/6F1 F2)	0.5–14.5 days postpartum	Gohbara et al., 2010; Komeya et al., 2016; Sato et al., 2011a; Sato et al., 2011b; Yokonishi et al., 2013; Yokonishi et al., 2014
	Mus musculus (Acr-GFP, Gsg2-GFP, and WBB6F1-Sl/Sld transgenic)	2.5 days–28 weeks postpartum	Sato et al., 2015
	Mus musculus (CD-1)	2.5–37 days postpartum	Arkoun et al., 2015; Dumont et al., 2015; Dumont et al., 2016; Rondanino et al., 2017
	Mus musculus (C57BL/6J or CBAB6F1)	3–5 days postpartum	Reda et al., 2017; Richer et al., 2021
	Mus musculus (ddY)	5 days postpartum	Suzuki and Sato, 2003
	Rattus norvegicus (Sprague-Dawley)	5–7 days postpartum	Liu et al., 2016; Reda et al., 2016
	Rattus norvegicus (Long-Evans)	12–14 days postpartum	Steinberger and Steinberger, 1965; Steinberger et al., 1964
	Homo sapiens	6–12 weeks gestation fetus	Lambrot et al., 2006
		2–12 years prepubertal	de Michele et al., 2017
		77.94 ± 6.84 years	Roulet et al., 2006
		Adult*	Baert et al., 2020; Vermeulen et al., 2018
Hanging drop	Mus musculus (CD-1)	Gestation fetus* (adult females aged 2–3 months)	Potter and DeFalco, 2015
	Homo sapiens	7–9 weeks gestation fetus	Jørgensen et al., 2015
		Adult*	Jørgensen et al., 2014; Vermeulen et al., 2018
Microfluidic device	Mus musculus (male Acr-GFP transgenic x female ICR or B6D2F1 F2)	0.5–5.5 days postpartum	Komeya et al., 2016
	Mus musculus (CD-1)	7–8 days postpartum	AbuMadighem et al., 2022
Organoid-based cellular aggregate or seminiferous tubule cord (grown on different extracellular matrices)	Mus musculus (unknown strain)	8–10 days postpartum	Sakib et al., 2019
	Mus musculus (C57BL/6J)	6 days postpartum	Zhang et al., 2014
	Mus musculus (BALB/c)	1 2, 4, or 8 weeks postpartum	Abu Elhija et al., 2012
	Mus musculus (CD-1)	5 days–16 weeks postpartum	AbuMadighem et al., 2018; Edmonds and Woodruff, 2020; Enders et al., 1986; Stukenborg et al., 2008
	Rattus norvegicus (Sprague-Dawley)	5–60 days postpartum	Alves-Lopes et al., 2017; Hadley et al., 1985; Hadley et al., 1990; Harris et al., 2015; Harris et al., 2016; Lee et al., 2006b; Lee et al., 2011; Legendre et al., 2010; Reda et al., 2014; Reuter et al., 2014; Zhang et al., 2017
	Rattus norvegicus (CD)	7 days postpartum	Gassei et al., 2008; Gassei et al., 2010; Pan et al., 2013
	Rattus norvegicus (SIV-1)	Newborn, 8–10, 18–25, 35–40, and 90 days postpartum	Zenzes and Engel, 1981
	Rattus norvegicus (unknown albino strain)	25 days postpartum	Enders et al., 1986
	Homo sapiens	5–9.5 weeks gestation fetus	Oliver et al., 2021
		6 months–75 years	Baert et al., 2017; Gholami et al., 2018; Jabari et al., 2020; Mincheva et al., 2018; Pendergraft et al., 2017; Sakib et al., 2019; von Kopylow et al., 2018
		Adult*	Lee et al., 2006a; Lee et al., 2007
Whole seminiferous tubule	Rattus norvegicus (Sprague-Dawley)	8 days–15 weeks postpartum	Klein et al., 2013; Mao et al., 2021; Perrard et al., 2016
	Rattus norvegicus (Brown Norway)	3–5 months postpartum	Li et al., 2016
	Homo sapiens	25–83 years	Perrard et al., 2016; Seidl and Holstein, 1990

*Age was not reported at the time of isolation.

Nevertheless, the adaptation of these technologies and techniques for measuring transport of select compounds across complex testicular culture models is promising. The widespread use of these models is still limited by cost, time, and throughput compared with conventional models. Furthermore, the use of heterologous expression systems to study the selectivity and kinetics of one or more human transporters may be appropriate if the transporter(s) that interact with a drug of interest have been defined. Transport in heterologous expression systems is the same as transport in SCs, although some drugs can interact with multiple efflux transporters expressed by SCs that may not be present in the former model. Despite these limitations, the information gained through each of these in vitro models (e.g., primary/immortalized rodent and human SC monolayers, heterologous expression systems, complex 3D culture models) for BTB drug disposition is highly valuable prior to applying it to in vivo models.

In Vivo Models

Rodent Models

Current U.S. Food and Drug Administration (FDA) guidelines recommend the use of a rodent or nonrodent animal model to assess the pharmacokinetic and toxicokinetic characteristics of an investigational new drug (https://www.fda.gov/regulatory-information/search-fda-guidance-documents/m3r2-nonclinical-safety-studies-conduct-human-clinical-trials-and-marketing-authorization). The use of rats and mice, among other preclinical species, for assessing testicular disposition of xenobiotics is standard practice in biomedical research. There are many different laboratory strains of rats and mice, and although it may be unlikely that different strains will exhibit significant transporter gene expression differences that affect the pharmacokinetics of most drugs, caution should still be noted. It may be an important factor because differences in gene expression and function (e.g., efflux transporters, cytochrome P450s, and macrophage antibodies) between rodent strains have been identified (Turk et al., 2004; Lash et al., 2014; Jin et al., 2017; Li et al., 2019; Schaffenrath et al., 2021). Unlike in vitro models, there are many more parameters to consider when using in vivo models to evaluate drug penetration across a physiologically intact BTB. For example, the expression and localization of certain transporters at the BTB (e.g., MRP4 and BCRP) can generate results that cannot to translate to humans, as described previously (Koraïchi et al., 2013; Klein et al., 2014; Hau et al., 2021, 2023). Due to differences in MRP4 localization at the BTB between rats, mice, and humans, a promising drug candidate that is a substrate for MRP4 can generate contradictory results in a rodent disposition study. Moreover, there is no known rodent ortholog for some transporters, such as human OAT4, which limits the use of rodent models when studying this transporter or others without a relevant ortholog. In addition, differences in transporter ortholog selectivity and kinetics can affect how well the testicular disposition data correlates between rodents and humans or nonhuman primates. Therefore, it is critical to determine the expression and localization of rodent BTB transporters and functionally characterize them to compare with data from human tissues.

A previous study showed that ENT1 and ENT2 are localized to the basal and apicolateral membranes of SCs in rodents and humans, respectively (Klein et al., 2013). These transporters were shown to be functional in rodent and human SCs (Kato et al., 2005, 2009; Klein et al., 2013; Hau et al., 2020), which led to a recent study that assessed ENT-mediated testicular disposition of clofarabine in Sprague-Dawley rats (Miller et al., 2021a). Clofarabine is a reported ENT substrate (King et al., 2006; Miller et al., 2021b; https://www.accessdata.fda.gov/drugsatfda_docs/nda/2004/21-673_Clolar_biopharmr.pdf), whereas lamivudine is not (Miller et al., 2020, 2021b), which is why they were chosen as probe compounds by Miller et al. In that study, clofarabine was further confirmed to be a substrate of ENT1 and could be inhibited by an ENT-specific inhibitor, NBMPR, in primary rat SCs and hT-SerCs (Miller et al., 2021a). Rats were dosed with 10 mg/kg NBMPR-P, the phosphorylated prodrug of NBMPR, or vehicle 20 minutes before 10 mg/kg clofarabine, 10 mg/kg lamivudine, or vehicle dosing. These clofarabine and lamivudine doses were sufficient to detect within the testes and were well within the dose ranges of previously published studies (Bonate et al., 2005; Arthaud and Seals, 2007; Li et al., 2020; https://www.accessdata.fda.gov/drugsatfda_docs/nda/2004/21-673_Clolar_biopharmr.pdf; https://www.accessdata.fda.gov/drugsatfda_docs/label/2004/021673lbl.pdf; https://www.accessdata.fda.gov/drugsatfda_docs/label/2015/206510lbl.pdf). Testicular accumulation of clofarabine in Sprague-Dawley rats in the presence of NBMPR compared with the vehicle control, only approached statistical significance. On the other hand, lamivudine was not a substrate of ENT1 or ENT2 and testicular disposition was completely unaffected by NBMPR. Interestingly, clofarabine causes testicular toxicity in rats, mice, and canines even at a fraction of the clinically recommended dose, according to the FDA package insert (https://www.accessdata.fda.gov/drugsatfda_docs/label/2004/021673lbl.pdf). Despite the lack of statistical significance in the study by Miller et al., these cumulative observations suggest that ENT-mediated disposition across the BTB can be an important physiologic and pharmacological pathway for some nucleosides and nucleoside analogs in rodents and humans (Miller et al., 2021a). However, further studies are necessary to determine if ENTs (and other transporters) mediate the transport of other clinically relevant xenobiotics across the in vivo BTB.

Many other xenobiotics are known to induce a pharmacological or toxicological response in the testes of rodents, including adjudin (Su et al., 2011), cisplatin (Vawda, 1994; Sherif et al., 2014; Soni et al., 2016), cyclophosphamide (Trasler et al., 1988; Matsui et al., 1995), doxorubicin (Imahie et al., 1995), (H2-)gamendazole (Tash et al., 2008a,b; Gupta et al., 2011), imatinib (https://www.accessdata.fda.gov/drugsatfda_docs/nda/2001/21-335_Gleevec_pharmr_P1.pdf), and methotrexate (Padmanabhan et al., 2009; Vardi et al., 2009). The major transporters that accept some of these xenobiotics as substrates have been defined using in vitro heterologous expression systems, although the transporters that accept other compounds as substrates, like H2-gamendazole and gamendazole, are unknown. Unfortunately, comparatively few in vitro and in vivo studies have attempted to identify the BTB transport processes for these compounds. Furthermore, there is a critical lack of information on the expression and localization of transporters at the rodent BTB, which is an issue when studying drug disposition across this barrier. Due to the complexity of in vivo pharmacokinetics, it is prudent to first identify major BTB drug transporters with immunolocalization studies and in vitro models then confirm those findings in vivo like with ENT1 and ENT2.

Although transport data collected from in vivo models can closely correlate with clinical observations, conclusions concerning the role(s) of individual transporters require an extensive array of in vitro and in vivo studies. Since most compounds interact with more than one transporter, in vitro transport studies can inform in vivo experiments, and vice versa, to improve model correlation. The most important consideration is that caution must be taken when using rodents to study drug disposition and toxicity in the testes due to the issues outlined here. Another recent concern for in vivo studies is the push by the FDA to reduce, refine, and replace animals in research, meaning alternative methods such as in vitro and in silico modeling to predict the absorption, distribution, metabolism, and excretion (ADME) characteristics of drugs are becoming increasingly popular.

Nonhuman Primate and Other Mammalian Models

Direct drug disposition studies with human patients are limited to highly regulated clinical studies; however, nonhuman primate models are the closest approximation to human genetics and physiology. Therefore, it is expected that most, if not all, BTB transporters will be expressed in the same locations in humans and nonhuman primates. One study found that MRP1 (basal membrane of SCs), MRP4 (basal membrane of SCs), and MRP8 (round spermatids) are expressed in the same location in humans and rhesus macaques (Klein et al., 2014); however, the localization of many other transporters remains undetermined. Like human studies, nonhuman primate studies are also highly regulated, and FDA recommendations to reduce, refine, and replace these animal studies are warranted. Testicular disposition studies to verify transporter function in nonhuman primates would require physical castration or outright sacrifice, which is unsustainable for drug discovery programs that require routine testing. Confirmational testicular disposition studies in nonhuman primate models may be viable for exceptional drug candidates as a penultimate step for an investigational new drug application. Some drugs, such as H2-gamendazole, have been shown to be effective as a contraceptive agent in nonhuman primates, which confirm observations in other species including mice, rats, and rabbits (Gupta et al., 2011). These findings suggest that BTB pharmacokinetics and pharmacodynamics of some xenobiotics can be adequately modeled in different species, which can then be used to develop in silico models for predictive drug discovery, although an extensive array of data is necessary to achieve accurate and precise predictions for different drug scaffolds.

Like other mammalian in vitro models, alternative species such as hamsters, canines, and rabbits have been used to assess testicular toxicity (Heywood and James, 1978; Rao et al., 1982; Morton, 1988; McEuen and Miller, 1991; Foote and Carney, 2000; Shimomura et al., 2004; Noritake et al., 2011; Vidal and Whitney, 2014). However, there is limited information on transporter expression, localization, and selectivity in these species and how their BTB physiology compares relative to humans, nonhuman primates, and rodents. More recently, pharmaceutical companies have begun exploring the idea of performing ADME studies directly in humans using a microtracer approach by dosing patients with subpharmacologically active ¹⁴C-labeled drugs and assessing their pharmacokinetic profiles (Young et al., 2023). The human ADME microtracer study is a powerful strategy; however, it does not help in identifying tissue concentrations, which still require mass balance studies in other species. Therefore, rodent, nonhuman primate, and other mammalian models remain important tools to study drug disposition across the BTB, and further investigation into the expression, localization, and function of transporters in these species is warranted.

Conclusions

The BTB limits the penetration and accumulation of many endogenous and exogenous compounds in the testes due to the presence of a physical membrane barrier and the expression and function of various transporters at the basal membrane of SCs. The expression, selectivity, and common localization of BTB uptake transporters, including the OATs, OCTs, OCTNs, and OATPs, is an indirect indicator for testicular drug disposition. Additionally, more promiscuous efflux transporters such as P-gp, BCRP, and the MRPs actively reduce compound accumulation in and through SCs. Several in vitro and in vivo models are available to study the collective role of these transporters as outlined here. However, there are advantages and disadvantages that must be considered with each model as discussed in this review. Multiple models to assess drug transport across the BTB must be used to account for the biologic complexity of this physiologic barrier. For example, assessing drug transport of a test compound in human, nonhuman primate, and rodent SCs can be supplemented with in vivo rodent studies to generate a holistic picture of BTB disposition.

One potential workflow is to first identify if the uptake of a compound into cultured rodent SCs is transporter-mediated, which can then inform a rodent in vivo disposition study like in Miller et al. (Miller et al., 2021a). An in vitro transport study with human SCs can also be performed at the same time to correlate the findings with rodent SC studies. Potential species differences in transporter expression and localization will also be very important to consider at this stage, because transporter-mediated uptake by transporters expressed at the basal or apicolateral membranes when SCs are plated as flat monolayers are indistinguishable. If possible, transport studies with nonhuman primate SCs may also provide additional data to support future disposition studies in nonhuman primates or humans. Another approach is to identify compounds that exert an in vivo pharmacological or toxicological effect in the testes, then assess SC uptake in various in vitro models. This would be an informed approach, in which some compounds may have well-defined transporter pathways that could help identify the transporters expressed in SCs. The latter approach may be preferable until a quality data set is established. In summary, comprehensive preclinical drug transport studies with multiple in vitro and in vivo BTB models are more likely to accurately reflect and predict observations in clinical studies and should become standard practice.

Abbreviations

ADME

absorption, distribution, metabolism, and excretion

BCRP

breast cancer resistance protein

BTB

blood-testis barrier

ENT

equilibrative nucleoside transporter

FDA

Food and Drug Administration

hT-SerC

human TERT-immortalized Sertoli cell

MRP

multidrug resistance-associated protein

OAT

organic anion transporter

OATP

organic anion transporting polypeptide

OCT

organic cation transporter

OCTN

organic cation transporter, novel

P-gp

P-glycoprotein

SC

Sertoli cell

TEER

transepithelial electrical resistance

Authorship Contributions

Wrote or contributed to the writing of the manuscript: Hau, Wright, Cherrington.