Current Progress on Equilibrative Nucleoside Transporter Function and Inhibitor Design

Structure–Function Relationship of hENT1

Investigation of the 3D structure of ENTs has been hampered by low endogenous expression levels in many host organisms and difficulty in isolating them in a functional form. However, a wealth of information about the structure–function relationships of hENT1 has been produced using site-directed mutagenesis and generation of chimeras between human and rat NTs, with ligand binding assays ( Table 6 ). These studies have provided many insights into key interactions used by different ligands during ENT inhibition or substrate transport. Here we discuss the mutations in structural context of a new homology model of hENT1.

OPEN IN VIEWER

Table 6.

Summary of Mutagenesis Studies in hENT1 Protein.

Residue	Location	WT Effect	Substitution	Mutation Effect
G24	TMD1	Involved in substrate/inhibitor recognition	R	Incapable of binding with NBMPR due to disruption of substrate binding site
W29	TMD1	Creates aromatic ring stacking for NBMPR interaction	C, A, G	Impaired interactions with inhibitorsDecreased adenosine transport due to small or absence of side chain
M33	TMD1	Important binding site for dilazep, dipyridamole, and nucleosides	I	Decreased sensitivities to dilazep and dipyridamoleDecreased uridine transportNo effect on NBMPR
N48	1st extracellular loop	Glycosylation	Q	Lower expression levels, decreased nucleoside transport activities
P71	1st extracellular loop	Mitochondrial targeting	L	Reduced mitochondrial localization and toxicity
E72	1st extracellular loop	Mitochondrial targeting	G	Reduced mitochondrial localization and toxicity
N74	1st extracellular loop	Mitochondrial targeting	P	Reduced mitochondrial localization and toxicity
M89	TMD2	Interacts with the purine ring of adenosine, guanosine, and NBMPR	C, T	Reduction of NBMPR affinity due to hydroxyl group of theronine
L92	TMD2	Either carbonyl group in inosine or guanosine or proton at the N-1 position and nitrobenzylthiol at C-6 in NBMPR interact directly/indirectly with Leu92 but not the amino group of adenosine	Q, P	Steric or indirect interactions between Leu92 and nucleoside substrates and inhibitorsMay involve structural change
G154	TMD4	Nucleoside 3′-hydroxyl group of sugar ring involved in interactions	C, S	Binding affinities with cytidine and adenosine decreasedLess sensitive to NBMPR inhibitionAlteration in tertiary structure of transporters
S160	TMD4	Interacts with nitrobenzylthiol group of NBMPR	C	Decreased permeant affinities and increased sensitivity to dipyridamoleMight alter the interaction of nitrobenzylthiol group of NBMPR with transporters
G179	TMD5	Hydrophobic interactions with nucleosides or NBMPR	L, V, C	Impaired transport activity; causes indirect steric effects on binding of uridine or NBMPR or with flanking region residue S178 or Q180
G184	TMD5	Hydrophobic interactions with nucleosides or NBMPR	L, V, C	Impaired transport activity; causes indirect steric effects on binding of uridine or NBMPR or with flanking region residue S178 or Q180
F334	TMD8	Hydrophobic interactions with dilazep and dipyridamoleCatalytic turnoverProtein folding	C, V, I	Decreased adenosine transport activityAlso reduced sensitivities to dilazep and dipyridamoleNo effect for NBMPR, soluflazine, and draflazineAltered membrane distribution
N338	TMD8	Hydrogen bond between draflazine and soluflazine	Q	Decreased adenosine transport activity
L442	TMD11	Forming part of the dipyridamole binding siteAlso involved in helix–helix interaction with residue 33 of TM1	T	Increased uridine and adenosine transport activityMay affect conformations of TM1 and TM11 helices

To analyze the positions of hENT1 predicted to have an active role in ligand binding and/or substrate transport, we created a homology model of the hENT1 structure (see Supplemental Information). Since homology searches using the hENT1 sequence (UniProtKB Q99808) against the Protein Data Bank (PDB) entries at the National Center for Biotechnology Information (NCBI) (http://blast.ncbi.nlm.nih.gov/) failed to reveal suitable structural templates for homology modeling, we used I-TASSER structure prediction to produce a 3D model for hENT1. In this approach,^21,22 threading using Local Meta-Threading-Server (LOMETS) ²³ is first used to identify structural templates from PDB, which are then used to assemble template fragments into a complete 3D model of the full-length target protein. The results of the protein homology/analogy recognition engine Phyre2 ²⁴ and the homology detection and structure prediction server HHPred^25,26 showed a good correlation with the model of hENT1 obtained using I-TASSER. The templates used for modeling by each server are listed in Supplemental Table S1 . The secondary structure elements for the I-TASSER model selected for the 3D structural analysis were assigned by DSSP.^27,28 The hENT1 model had a C score of −1.87 providing confidence for the quality of the 3D model created by I-TASSER (ranges between −5 and +2, with a higher value being better ²¹ ) and a TM score of 0.49 ± 0.15, which shows the topological similarity between the template and the model (TM score > 0.5 approximate cutoff value). ²³ The quality of the 3D model for hENT1 ( Suppl. Fig. S3 ) was also assessed to be consistent with a TM structure with a good quality for the TM regions by QMEANBrane, which is a local model quality estimation method for membrane proteins combining statistical potential strained on membrane protein structures with a per residue weighting scheme. ²⁹ Furthermore, ProQ optimized to find correct models ³⁰ gave an LG score of 5.076 (>4 for an extremely good model) and a MaxSub score of 0.166 (>0.1 for a correct model) for the hENT1 model when the atomic coordinates and the secondary structure predicted by PSIPRED v3.3 ³¹ were used in the quality assessment. Furthermore, the TMs in the 3D model of hENT1 are consistent with the secondary structure predictions. In conclusion, the 3D homology model of hENT1 is of high quality and, combined with experimental site-directed mutagenesis data, can be used to draw mechanistic conclusions about the roles of individual amino acid residues in ENT1 ligand binding and substrate transport.

Implications of Different Mutations on hENT1 Structure and Function

Many site-specific mutational studies have addressed the contribution of individual residues for ligand binding and transport within ENT1.^32–35 The putative locations for these residues are depicted in the 3D homology model of hENT1 in an inward-facing conformation ( Fig. 4A ). The residues that have been mutated in the various studies mainly point toward the cavity between the TM helices ( Fig. 4B ) with the exception of G24, G179^(5.50), and S160. The two glycines face toward the membrane, whereas S160 points toward TM3 and is closer to the intracellular surface than other mutated residues ( Fig. 4B ).

OPEN IN VIEWER

Figure 4.

3D model for hENT1 showing key residues involved in protein function. (A) Side view of an inward open hENT1 model showing the key TMs highlighted in colors. (B) Side view of TM1, TM2, TM4, TM5, TM8, and TM11 with the residues suggested to be important for the hENT1 function. Despite S160 and G179^(5.50), the residues are located near each other in the middle of the helices. During the review process of this article a crystal structure of a truncated hENT1 transporter was published. ³⁶

ENT1 contains a highly conserved tryptophan residue at position 29^(1.57) of TM1 facing the ENT1 central cavity, which likely plays an important role in all ENTs. Several substitutions at this position (W29C, W29T, W29A, and W29G) result in impaired binding of NBMPR and other inhibitors, whereas the conservative W29F mutation exhibits wild-type-like affinity, suggesting that the identity of the side chain at this position is critical for inhibitor binding. W29^(1.57) may facilitate interactions with the nitrobenzyl group of NBMPR by aromatic ring stacking. Substitution of W29 with a smaller amino acid (such as glycine, alanine, cysteine, or valine) resulted in decreased nucleoside binding. Replacement with tyrosine caused a small increase in the transport of adenosine and other purine nucleosides. Strikingly, replacing W29^(1.57) with a threonine showed a drastic decrease in the transport of uridine but not adenosine. It has been suggested that W29T may form a hydrogen bond with another residue that is specific and critical for pyrimidine nucleoside interactions without dramatically influencing purine nucleoside transport, which is altered by mutation of the nearby L442 mutant (L442I). ³² Together, these results suggest that W29^(1.57) is involved in permeant selection and translocation. ³⁷ Another highly conserved residue among ENTs is G24^(1.52) in helix TM1. Arginine replacement of G24 has been shown to affect substrate binding, protein folding, and correct targeting of hENT1 to the plasma membrane. Furthermore, G24R impairs binding of NBMPR, even when the protein is localized to the plasma membrane, suggesting a key role for this residue in substrate binding, in addition to promoting protein stability and correct plasma membrane trafficking. ³⁸

Random mutagenesis and adenosine complementation assays have also identified the central roles of TM2 residues M89 and L92 and TM4 residue S160 in inhibitor binding and/or nucleoside transport. ³⁴ Consistent with the prediction from the helical wheel analysis in our 3D model, both M89 and L92 face the same side of TM2 and point to the ligand binding cavity. ENT1 M89 substitutions result in increased (M89L or M89Q) or severely diminished (M89C or M89T) NBMPR binding ³⁴ but have no effect on the ability of ENT1 to bind dipyridamole. S160 is located at the end of TM2 facing away from the ligand binding cavity ( Fig. 4B ). The mutation S160C selectively decreases the binding of dipyridamole, but not that of the other inhibitors and natural nucleosides, suggesting an important role for this hENT1 residue in the selective binding of dipyridamole. Since mutation of M33 and L442I also sensitizes hENT1 toward dipyridamole (vide infra) and these residues are distant from S160 in our homology model, it is likely that S160 does not contribute to ligand binding through direct interactions, but may instead have an important role for maintaining hENT1 in a dipyridamole-accessible conformation.

Another important residue at the extracellular end of hENT1 TM1 is M33^(1.61), which has been shown to be important for binding of dilazep, dipyridamole, and physiological nucleosides. ³⁹ In the 3D model of hENT1 M33 lies above W29^(1.57) in TM1 and interacts with TM2 residues M89 and L92I ( Fig. 5A,B ). When M33 was substituted to isoleucine (the corresponding residue in hENT2), the transporter became less sensitive to dilazep and dipyridamole, whereas the reciprocal mutation in hENT2 (I33M) increased the sensitivity to these inhibitors. Remarkably, neither mutation affected transport inhibition by NBMPR, which suggests that residue 33 plays a key role in the ability of dilazep and dipyridamole to inhibit hENT1 and hENT2 transport. The predicted topology model for hENT1 suggests that position 33 is the last residue in the first TM segment and may therefore be solvent accessible and/or in the plane of the extracellular bilayer–solvent interface. ⁴⁰ Moreover, these results suggest that I33 may be involved in uridine interaction with hENT2 but not hENT1. Both hENT1 and hENT2 are known to have different permeant binding properties because, unlike hENT1, hENT2 is capable of transporting nucleobases and antiviral dideoxynucleoside analogs. ⁴¹ These observations suggest that in addition to the main high-affinity dilazep and dipyridamole binding site, hENT1 may contain a second low-affinity allosteric site for these ligands.^42,43 Hence, it is assumed that hENT1 contains a high-affinity binding site for permeants, NBMPR, and other inhibitors, such as dilazep and dipyridamole, and an allosteric low-affinity site that binds nucleosides, nucleobases, and inhibitors when present at high concentrations. It is currently not clear whether residue 33 of hENT1 and hENT2 is directly involved in permeant or inhibitor binding or whether the effects observed with M33 mutations are due to changes in the ENT tertiary structure.

OPEN IN VIEWER

Figure 5.

Close-up views of the predicted ligand binding site in hENT1. (A) Side view. (B) Top view. (C) Putative interaction between M33 in TM1 and L442 in TM11. The closest distance between the residues is 8.7 Å in the hENT1 model, close enough to be involved for dipyridamole binding.

It was shown that TM1 and TM11 are critical for dipyridamole binding within hENT1, CeENT1, and hENT2. ³⁹ This is indicated by the finding that the mutation L442I in hENT1 and M33 and L442^(11.51) substitution in hENT2 make the protein sensitive to dipyridamole compared with wild-type protein. Based on the assumption that TM1 and TM11 are in close proximity, a mechanistic model of dipyridamole interaction was proposed: increased dipyridamole binding affinity is suggestive of a functional interaction between L442^(11.51) and M33, which may be favored by the proximity of TM1 and TM11 induced by enhanced interactions resulting from the introduction of the L442I mutation. ³⁹ Because the M33 of TM1 and leucine and isoleucine residues of TM11 contain the highly conserved GXXXG helix–helix interaction motif, it is likely that the mutation of these residues affected the conformations of both TMs. Studies have shown that the mutation L442T changes the conformation of TMs such that the helices move farther apart. ³⁹ In the 3D model of hENT1, M33 and L442^(11.51) are well positioned and far enough from each other to be involved in dipyridamole binding ( Fig. 5C ). Based on these studies, and our homology model, we propose that dipyridamole interactions are complex and involve contributions from several residues on different parts of the protein.

Mutations in TM1 and conservative substitutions of the hydrophobic leucine and isoleucine residues at the TM11 position in hENT1 had only minor effects on ENT1 uridine transport. Analysis of uridine and adenosine transport by the TM11 mutants revealed that L442T at the TM11 position resulted in greatly increased uridine and adenosine transport by hENT1, whereas the L442I mutation showed impaired adenosine transport without effects on uridine transport, suggesting specific alterations to the shape of the permeant binding pocket. ³⁹ Similar substrate-selective effects affecting the binding of NBMPR, dilazep, and inosine, but not dipyridamole, adenosine, or pyrimidines, were observed with mutations of L92. ⁴⁴ Because of the structural differences between the purines adenosine, inosine, and guanosine, it was suggested that L92 would directly interact with the C-6 and N-1 positions of the purine group, ⁴⁴ but not the amino group of adenosine. Substituting L92 with polar (glutamine) or nonpolar (proline) residues may hinder hENT1 interactions with inhibitors or substrates due to structural changes influencing neighboring or other residues directly involved in substrate and inhibitor interaction. As L92 is adjacent to P91, this position of the α-helix may be sterically stressed, and sensitive to neighboring mutations. ⁴⁴ However, the observed selective effects of L92 mutations on the affinity of inosine and guanosine, but not adenosine, suggest that these mutations do not cause gross alterations to the transporter structure. In line with the notion that L92 together with M89 and W29 would promote the structural integrity of the ligand binding site, all of these residues cluster together in the central cavity of the hENT1 homology model ( Fig. 4B ).

The conserved residue G154 in hENT1 is equivalent to S141 in hENT2, and the corresponding hENT1 mutation G154S significantly decreases hENT1 sensitivity to NBMPR and reduces its affinity for other nucleosides. ⁴⁵ In the 3D model of hENT1 ( Fig. 4D ) G154 is located in the ligand binding cavity proximal to and below TM1 W29^(1.57). NBMPR and other nucleosides interact with G154 via the ribose ring, and the G154S mutation diminishes the dependency of hENT1 on the presence of the ribose ring, thus converting it into an hENT2-like transporter. ⁴⁶ Additionally, the ribose ring alone is not enough for a compound to act as an hENT1 substrate; it must possess a 3′-hydroxyl group to be recognized.^46,47 Highlighting the differences in substrate selectivity between hENT isoforms, only hENT2 can transport anti-HIV drugs, such as 3′-azido-3′-deoxythymidine or azidothymidine (AZT), dideoxycytidine (ddC), and dideoxyinosine (ddI), along with nucleobases, such as adenine and hypoxanthine. ⁴⁶

TM5 of hENT1 contains two highly conserved residues, G179^(5.50) and G184^(5.55), which may contribute to helix packing in ENTs from different organisms since bulky amino acid substitutions (G179L, G179V, G179C) impair nucleoside transport, whereas smaller substitutions (G179A, G179S) do not affect nucleoside binding. ⁴⁸ Earlier work suggested that G179^(5.50) would be located in a hydrophobic region of TM5, along with other nonpolar residues (A183, F186, A190). ³⁵ However, in our hENT1 model, G179(5.50) (Fig. 4B and 5A) and the nearby non-polar residues clearly face the membrane–TM1 interface, a notion that is also supported by further mutational studies. In contrast to G179^(5.50), mutations in G184^(5.55) appear to prevent correct plasma membrane localization of hENT1, indicating that this residue may be important for correct folding or trafficking. ³⁵ Substitution of G184^(5.55) with a cysteine revealed six TM5 residues on one face of an amphipathic helix, where the G184C residue is accessible for sulfhydryl modification, indicating that this residue in hENT1 is solvent exposed and may form part of the nucleoside translocation pathway. In the hENT1 model, G184^(5.55) is located next to N338 in TM8 (see below) and N30^(1.58) in TM1, and a larger residue in the mutants might sterically hinder ligands interacting with N338.

Sequence alignment of ENTs from different organisms shows high evolutionary conservation at residues F334 and N338 ( Fig. 5A,B ). Also, evolutionarily distant ENT orthologs from organisms as diverse as plants to protozoa contain bulky aliphatic side chains at the corresponding site, suggesting a high requirement for hydrophobic residues at this position. ⁴⁹ The introduction of several mutations at these sites (F334C, F334V, F334I, N338M, N338S, N338A) significantly decreased the ability of hENT1 to transport adenosine. ³³ In contrast, polar mutations N338D and N338Q were well tolerated, and a particular mutation (F334Y) even resulted in an increase in adenosine transport. Similar to the effect on adenosine transport, bulky hydrophobic F334 mutations reduced ENT1 sensitivity to dipyridamole and dilazep. ³³ Conversely, the same F334 mutations had little to no effect for the binding of NBMPR, soluflazine, and draflazine. Although the nearby residue N338 is highly conserved, both hydrophilic (N338S, N338D) and hydrophobic (N338M) substitutions were well tolerated, suggesting that asparagine is not required for ENT1 stability or the ability to transport adenosine despite a high degree of conservation.

Collectively, these results suggest that large hydrophobic substitutions at F334 favor the binding of dipyridamole and dilazep, whereas hydrophilic mutations at N338 promote hENT1 interactions with soluflazine and draflazine. Our structural model indicates that both F334 and N338 are located close to one another on the same side of TM8 ( Fig. 5A,B ), implicating this hENT1 region within the substrate cavity in ligand interactions and transport activity. ⁵⁰

Proposed Mechanism of Ligand Binding and Translocation

To date, no experimental structures for any ENT orthologs are available. The closest homolog for which structural data exist, a CNT from Vibrio cholerae, ⁵¹ shares only moderate sequence identity with ENT1 (9.5%), has a different membrane topology, and couples nucleoside transport with proton transfer, and is therefore not useful for investigating ENT family members. Computational models such as ligand-based quantitative structure–activity relationships (QSARs) built on in vitro affinity data have provided information about ligand requirements for interactions with ENT. Most of these studies have focused on the analysis of the NBMPR class of compounds and have revealed a critical role for the NBMPR nitrobenzyl moiety required for high-affinity ENT binding. In particular, NBMPR analogs with electron-withdrawing groups such as a nitro group at the 6-position of the benzyl substituent contribute strongly to ligand binding affinity. ⁵² The size of the NBMPR molecule and enhanced affinity with a nitro group at the 6-position suggest that the ENT binding pocket is quite large and can accommodate the inhibitor in a region that tolerates the presence of a bulky charged moiety. Further studies revealed that substituents in the NBMPR C2-purine position negatively affect hENT1 binding affinity. ⁵³ Also, the C-3 hydroxyl of the NBMPR sugar moiety was revealed to be essential for interaction with hENT1. ⁵⁴

Evolutionarily, NTs share common features with the major facilitator superfamily (MFS) membrane transporters. ⁵⁵ The MFS transporters function by an alternating access mechanism, ⁵⁶ and hENT1 is believed to act by a similar ligand translocation mechanism because of shared topology with the MFS transporters. This model would indicate that hENT1 could exist in at least two primary conformations exposing either extracellular or intracellular substrate binding sites. In this model, these conformations alternate regardless of ligand binding in a way that only exposes a single ligand binding site at a time ( Fig. 1B ). Interestingly, NBMPR can only engage an open outward conformation of the transporter ⁵⁷ and binding of NBMPR may lock the transporter in an inward occluded conformation. Despite the apparently passive alternating access mechanism, it appears that the rate of associated conformational changes can vary in a manner dependent on the particular transported permeant, suggesting that some ligands have an ability to promote ENT conformational changes required for permeant transport. For example, the binding of pyrimidine induces a rapid conformational change, whereas 2-chloroadenosine binding reduces the rate of conformation changes in the transporter. ⁵⁸ Finally, it remains unclear whether hENT1 functions entirely in a monomeric form or whether it can oligomerize and contains several allosteric ligand binding sites. Supporting the presence of additional low-affinity allosteric ligand binding sites, [ ³ H]NBMPR competition binding experiments with dipyridamole revealed a binding mode that suggests cooperativity or multiple binding sites. ⁵⁸ A similar two-site binding mode for NBMR, dilazep, and dipyridamole was also observed in our studies ⁵⁹ at 37 °C. Finally, the variable rates of NBMPR dissociation in the presence of excess nucleosides and inhibitors such as dilazep and dipyridamole further suggests the presence of multiple binding sites or cooperative binding mechanisms. ⁵⁸ However, further structural studies are required to conclusively settle the number of binding sites or possible hENT1 multimerization mechanisms.

Recent Advances in Discovery of New ENT Modulators

Because of the well-characterized protective role of extracellular adenosine, ⁶⁰ preventing adenosine uptake by pharmacological ENT inhibition has been postulated to provide beneficial effects in pathological conditions such as stroke, ischemia-reperfusion injury, and diabetes. In this context, classical ENT inhibitors such as NBMPR, and coronary vasodilators dilazep and dipyridamole and their congeners have demonstrated the feasibility of inhibiting ENTs to promote protective adenosine signaling. Historically, NBMPR has been used as a probe compound to study adenosine transporter function and biology and to quantify transport elements of various cell types. However, these classical inhibitors lack selectivity for different ENT isoforms, which may limit the potential for their further therapeutic development. ¹⁶ Some ENT isoforms such as ENT4 are expressed in a highly restricted tissue pattern, suggesting that small molecules targeting only selected ENT isoforms could have less adverse effects arising from inhibition of unwanted target proteins. Additionally, highly isoform-specific inhibitors would serve as valuable probe compounds for studying the roles of individual ENT isoforms in the regulation of nucleoside transport in complex physiological environments and different animal models of human disease. Many kinase inhibitors also share structural features with adenosine and have been shown to inhibit ENT1 transport, raising the possibility of off-target effects of classical ENT modulators. ⁶¹

Despite problems associated with excess toxicity or poor availability of existing ENT1 inhibitors, progress in the design of new chemically distinct isoform-selective ENT inhibitors has been lacking in recent years. A recent study coauthored by us now suggests a potential new strategy for discovering potent macrocyclic inhibitors of different ENT isoforms. This method relies on the reengineering of rapamycin, a mechanistically unique inhibitor of mTOR. ⁶² In this study, a series of structural rapamycin variants were developed with an altered mTOR binding domain in order to discover new rapamycin-like macrocyclic inhibitors with the ability to expand the repertoire of inhibited protein targets. Screening of this library of “rapafucins” revealed a single compound that binds to and inhibits hENT1 with single-digit nanomolar potency in a manner dependent on simultaneous binding to FKBP12, mimicking the mode of action of rapamycin itself. ¹⁹ In demonstration of the therapeutic potential for ENT-targeting rapafucins, the authors demonstrated the beneficial efficacy of inhibiting ENT1 in an animal model of kidney reperfusion injury. The development of new ENT-targeting rapafucins will demonstrate the applicability of enhancing adenosine receptor signaling by inhibiting adenosine uptake in other diverse models of human disease. Interestingly, the identified rapafucin molecule, named rapadocin, inhibited ENT1 with high specificity over the related ENT2 isoform, suggesting the possibility to develop new highly isoform-selective ENT inhibitors. Importantly, because of their unique chemical structure, these macrocyclic inhibitors are unlikely to suffer from undesired pleiotropic off-target effects arising from the inhibition of nucleoside binding cellular targets. Similar to rapamycin, the ENT-targeting rapafucins will likely benefit from the unique properties offered by their unique mode of action. These properties arise from complexation with proteins of the FKBP family (in particular FKBP12) and include increasing the effective size of the inhibitor, allowing allosteric blockade of ENT substrate molecules and enhanced stability and efficient in vivo delivery. Future work will demonstrate the applicability and therapeutic benefits of isoform-selective ENT inhibition by rapafucin macrocycles.

Isolation of ENTs for Structural Studies

Despite their importance for the import of therapeutic NAs and as direct drug targets,^11,12 practically no structural information exists for hENT1 and other ENTs. Mostly this lack of structural information reflects the inherent difficulty to produce different ENTs in heterologous systems in an active and stable form and in sufficient quantities for purification and structural studies. Several groups have successfully expressed hENT1 in yeast systems,^39,63,64 and this has enabled detailed functional studies, of which many are summarized in this review; however, purifying ENTs from yeast has not allowed structural studies. Likewise, our attempts to express hENT1 in Escherichia coli, Saccharomyces cerevisiae, and Pichia pastoris failed to produce properly folded and targeted functional protein (unpublished data). In our experience, robust heterologous expression of hENT1 is most readily achieved in baculovirus-infected insect cells, which allows the isolation of milligram amounts of functional hENT1 when solubilized with lauryl maltose neopentyl glycol (LMNG) detergent. Using the guidelines we established, ⁴⁹ several other groups were also able to isolate functional hENT1.^38,65 Studies for the heterologous expression of other hENT family members are ongoing and will allow study of their biochemical differences, such as differences in transported substrates,^16,66–68 and structure-guided development of inhibitors, in particular for ENTs, from disease-relevant parasites.

Perspectives

Our current structural understanding of nucleoside transport and the regulation of ENT family transporters still remains at a juvenile stage. Most of the current molecular understanding of cellular nucleoside transport is derived from experiments using cell lines and animal models with contributing background activity from several ENT and CNT subtypes. To circumvent this problem, heterologous expression of individual ENTs and functional studies with isolated and reconstituted proteins is required to gain a better understanding of substrate selectivities of physiological nucleosides and their analog drugs with their corresponding ENT isoforms. Additionally, overexpressed proteins may be used to identify chemical probes by high-content screening approaches. One of the most challenging yet exciting areas of future research will be unraveling the complex network of functional interactions between ENTs, CNTs, and adenosine receptors (and possibly other unidentified purinergic receptors), metabolic enzymes, and signaling pathways. In addition, 3D structures for all ENT transporter types, complexed with their substrates and/or known inhibitors, would be desirable for creating more reliable models of their biological functions. With new tools in the form of FKBP12–rapadocin–ENT complexes, cryo-EM may provide a suitable route for future structural investigations. Approaches to stabilize distinct ENT conformations by use of small molecules, camelid antibodies, ⁶⁹ or recently introduced synthetic nanobodies ⁷⁰ may also be needed for the successful preparation of conformationally homogenous ENT samples suitable for structure determination by crystallization or single-particle electron microscopy approaches.