Bidirectional transport of 2-chloroadenosine by equilibrative nucleoside transporter 4 (hENT4): Evidence for allosteric kinetics at acidic pH

Adenosine has been reported to be transported by equilibrative nucleoside transporter 4 (ENT4), encoded by the SLC29A4 gene, in an acidic pH-dependent manner. This makes hENT4 of interest as a therapeutic target in acidic pathologies where adenosine is protective (e.g. vascular ischaemia). We examined the pH-sensitivity of nucleoside influx and efflux by hENT4 using a recombinant transfection model that lacks the confounding influences of other nucleoside transporters (PK15-NTD). We established that [3H]2-chloroadenosine, which is resistant to metabolism by adenosine deaminase, is a substrate for hENT4. Transport of [3H]2-chloroadenosine at a pH of 6.0 in PK15-NTD cells stably transfected with SLC29A4 was biphasic, with a low capacity (Vmax ~ 30 pmol/mg/min) high-affinity component (Km ~ 50 µM) apparent at low substrate concentrations, which shifted to a high capacity (Vmax ~ 500 pmol/mg/min) low affinity system (Km > 600 µM) displaying positive cooperativity at concentrations above 200 µM. Only the low affinity component was observed at a neutral pH of 7.5 (Km ~ 2 mM). Efflux of [3H]2-chloroadenosine from these cells was also enhanced by more than 4-fold at an acidic pH. Enhanced influx and efflux of nucleosides by hENT4 under acidic conditions supports its potential as a therapeutic target in pathologies such as ischaemia-reperfusion injury.

[ 3 H]Substrate uptake. SLC29A4-transfected PK15-NTD cells (PK15-hENT4), and the un-transfected PK15-NTD cells, were seeded in 6-well plates and allowed to grow for 2 days. Upon reaching 80-90% confluency, culture media was removed and plates washed with PBS (137 mM NaCl, 2.7 mM KCl, 10 mM Na 2 HPO 4 , 1.8 mM KH 2 PO 4 , pH 7.4) twice and then incubated in room temperature (~22 °C) transport buffer (120 mM NaCl, 20 mM Tris, 3 mM K 2 HPO 4 , 10 mM glucose, 1 mM CaCl 2 , 1 mM MgCl 2 , pH 7.5) for 15 min. Time courses of substrate uptake were constructed by exposing the cells to the substrate ([ 3 H]adenosine or [ 3 H]2-chloroadenosine) in room temperature transport buffer, pH 6.0 or pH 7.5, for times ranging from 1 to 15 min. 30 µM was chosen as the substrate concentration to use for the initial time course studies, as this is within the range of local adenosine concentrations achieved pathologically in ischaemic tissues. Uptake was terminated by rapidly aspirating the reaction mixture from the plates and washing the cells thrice with ice-cold PBS. Cells were digested in 1 N sodium hydroxide overnight, and radioactive content of each sample was measured using standard liquid scintillation techniques. Samples of the cell digest from each well were also taken for assessment of protein content for intra-and inter-assay normalisation. hENT4-mediated uptake was defined as the difference between total uptake by the PK15-hENT4 cells (at pH 6.0 or 7.5) and that measured in parallel using the PK15-NTD cells (at the corresponding pH). Uptake by the un-transfected PK15-NTD cells, which are devoid of ENT1 and ENT2 transport activity 21,25 , showed no acidic pH-dependent accumulation of [ 3 H]nucleosides and was thus considered to represent "background" non-transporter mediated accumulation. Based on initial time course profiles, it was determined that a time point of 5-6 min provided a reasonable estimate of the initial rate of [ 3 H]substrate influx. Therefore, this time frame was used to obtain initial rates of uptake over a range of concentrations of substrate to define transporter kinetics.
[ 3 H]2-Chloroadenosine efflux. PK15-hENT4 cells in 6-well plates were loaded with 30 μM [ 3 H]2-chloroadenosine in transport buffer, at pH 6.0, for 12 min, in the presence of 50 nM ABT-702 to minimize metabolism by adenosine kinase (concentrations and optimum loading times determined from the substrate uptake experiments). After loading, cells were washed thrice with ice-cold PBS to remove remaining extracellular [ 3 H]substrate. To estimate the amount of extracellular [ 3 H] remaining on the plates after this wash procedure, the washed cells were exposed to 800 µl of ice-cold buffer for ~10 s and the [ 3 H] content of a 500 µl aliquot of this wash media assessed; this 'background' was subtracted from all efflux data. For the efflux time courses, efflux was initiated by adding room temperature (~22 °C) substrate-free buffer (800 µl), pH 6.0, 7.5 or 8.2 (±test inhibitors) to the plates. After a defined incubation time (1-18 min), 500 µl of the extracellular media was removed and assessed for [ 3 H] content. Cell protein content of each well was also measured and used for assay normalisation as described for the [ 3 H]substrate uptake assays. Intracellular pH (pH i ) measurement. pH i was measured using a PTI Deltascan spectrofluorometer 26 .
Cells were grown on glass coverslips (Thomas ® RedLabel ® Micro Cover Glasses) to ~90% confluency. Cells were loaded with BCECF-AM (20,70-bis(2-carboxyethyl)-5(6) carboxyfluorescein-acetoxymethyl ester) (Molecular Probes Inc., Eugene, OR, USA) at a final concentration of 2.5 µg/ml (1 µl of 1 mg/ml stock in DMSO in 400 µl serum-free culture media at 37 °C for approximately 20 min). The AM group is de-esterified during this incubation, trapping BECEF within the cell. A dual excitation (440 and 502.5 nm), single emission (528.7 nm) ratio ensures that measurement is independent of cell number or amount of BECEF inside the cells. After loading, the coverslip was transferred to a cuvette holder in transport buffer, pH 7.5, with constant stirring by a magnetic stir bar at room temperature for 15 min. Following which, incubation times and conditions used for [ 3 H]2-chloroadenosine uptake assays were applied to determine how pH i changes during uptake at the different pH conditions used (6.0, 7.5, 8.2). For each experiment, a three-point pH calibration curve was made with a high KCl transport buffer (120 mM NaCl was replaced by 120 mM KCl) and 10 μM nigericin. Statistical analysis. Experiments were repeated at least five times in duplicate (actual N values reported in the Figure legends). Data are expressed as mean ± S.E.M. Statistical significance was determined by Student's t test or one-way analysis of variance (ANOVA) plus an appropriate multiple comparisons post-test, as indicated in the Figure legends, with P < 0.05 denoting significance. GraphPad Prism v8.01 was used to define the best fit (F-test, P < 0.05) nonlinear relationship represented by the data, and for all statistical analyses.

Results
SLC29A4/ENT4 expression. PK15-NTD cells do not have any detectable transcript for SLC29A4, whereas SLC29A4 expression was evident using RNA isolated from PK15-NTD cells stably transfected with SLC29A4 (Fig. 1a). Myc-ENT4 protein was detected in the SLC29A4 transfected cells by immunoblotting with an anti-myc antibody, and had a molecular mass of approximately 61 kDa (Fig. 1b). The predicted molecular mass of hENT4 is ~55 kDa, suggesting that the recombinant hENT4 protein expressed in these cells may be glycosylated (hENT4 is predicted to be glycosylated at N523 near the C-terminus).
[ 3 H]Adenosine uptake. Initial studies were conducted to confirm that hENT4 mediated the uptake of [ 3 H] adenosine in an acidic pH-sensitive manner, as described previously 3,18 . These assays were done in the presence of 50 nM ABT-702 to inhibit adenosine kinase and 100 nM erythro-9-(2-hydroxy-3-nonyl)adenine (EHNA) to inhibit adenosine deaminase. In all cases, time course data fit best to a one-phase (versus two-phase) association profile (Equation: Y = Y0 + (Plateau-Y0)*(1-exp(-K*x), with Y0 fixed to zero for fitting mediated uptake data, but left variable when fitting total uptake data with no background subtracted). The uptake of 30 µM adenosine by the PK15-hENT4 cells was significantly greater at pH 6.0 versus pH 7.5, whereas no significant difference with pH was seen for adenosine uptake by the un-transfected PK15-NTD cells (Fig. 2a). In addition, there was no significant difference in adenosine uptake between PK15-hENT4 cells and PK15-NTD cells at pH 7.5. The near linear accumulation of [ 3 H] by the cells at pH 6.0 even at 15 min of incubation suggests that there continues to be some trapping of [ 3 H]adenosine metabolites (adenine nucleotides) by the cells over this time frame even in the presence of the adenosine kinase inhibitor ABT-702. However, given that caveat, when the adenosine uptake by the PK15-NTD cells at pH 6.0 was subtracted from uptake by the PK15-hENT4 cells at pH 6.0, an initial rate of hENT4-mediated uptake of 9 pmol/mg/min could be estimated based on the maximum accumulation (187 ± 63 pmol/mg) and t 1/2 (22 ± 11 min) derived from the fitted curve (Fig. 2b). Inhibitors of other SLC29A family members (ENT1/ENT2) were then tested for their ability to inhibit the uptake (6 min) of 1 μM [ 3 H]adenosine at pH 6.0 (Fig. 3a). NBMPR, dilazep, draflazine, and soluflazine all produced a significant inhibition of [ 3 H] adenosine uptake by hENT4 when tested at 10 µM, but had no effect at 1 µM. These inhibitors would be expected to inhibit ENT1-and/or ENT2-mediated nucleoside uptake to some extent at the concentrations used 21,25,27,28 .
Dipyridamole was the only compound tested in this set with significant inhibitory activity against hENT4 at 1 µM (Fig. 3a). In comparison, the hENT4 inhibitor decynium 22 (D22; 1 μM) 29 completely blocked the acidic pH-dependent component of [ 3 H]adenosine uptake (Fig. 3b). Higher concentrations of D22 (10 µM) appeared to affect uptake at pH 7.5, but it was not statistically significant in this data set. Finally, if 2-chloroadenosine is a substrate for hENT4, one would expect it to compete with [ 3 H]adenosine for the substrate binding site on the hENT4 transporter. As shown in Fig. 3c  course fit best to a one-phase (versus two-phase) association profile. There was no significant difference in [ 3 H]2-chloroadenosine uptake by the PK15-NTD cells between pH 7.5 and 6.0 (Fig. 4a), in either the presence or absence of the hENT4 blocker D22. These findings, along with the lack of acidic pH-sensitive [ 3 H]adenosine uptake (Fig. 2a) by PK15-NTD cells, and no detectable mRNA transcript in the PK15-NTD cells (Fig. 1a), indicates that there is no endogenous ENT4 activity in this cell line. Therefore, since the PK15-NTD cells are also devoid of ENT1, ENT2, and concentrative Na-dependent nucleoside transporters 21,25 , enhanced 2-chloroadensine accumulation upon transfection of the PK15-NTD cells with SLC29A4 may be considered hENT4-mediated. A potential complicating factor, however, is the reported presence of an equilibrative nucleobase transporter (ENBT1) in the PK15-NTD cells for which 2-chloroadenosine may have some affinity 30,31 . To address this, we assessed the ability of adenine, a substrate for ENBT1 but not ENT4 18 , to inhibit 2-chloroadenosine uptake by PK15-NTD cells at both pH 7.5 and 6.0. Adenine, at concentrations sufficient to completely block ENBT1 activity (10 mM) had a slight suppressant effect on the uptake of 30 µM 2-chloroadenosine at both pH 6.0 and pH 7.5, with the difference at pH 6.0 achieving statistical significance (Fig. 4b). The acidic pH sensitivity of www.nature.com/scientificreports www.nature.com/scientificreports/ [ 3 H]2-chloroadenosine uptake was then assessed using the SLC29A4-transfected PK15-NTD cells (Fig. 4c). There was no significant difference in 2-chloroadenosine (30 µM) uptake between PK15-hENT4 and PK15-NTD cells at pH 7.5. In contrast, at pH 6.0, PK15-hENT4 cells accumulated 30 µM [ 3 H]2-chloroadenosine at a significantly greater rate than seen at pH 7.5 in either the PK15-hENT4 or PK15-NTD cells. However, as observed for [ 3 H] adenosine, the cellular accumulation of [ 3 H] using [ 3 H]2-chloroadenosine as the substrate did not saturate even at the longest incubation time of 15 min. To determine whether this represented trapping of radiolabel due to the formation of impermeable [ 3 H]2-chloroadenosine metabolites, the adenosine kinase inhibitor ABT-702 was added. ABT-702 (50 nM) reduced the total accumulation of [ 3 H] from the [ 3 H]2-chloroadenosine leading to saturation of the cellular uptake profile by 15 min (Fig. 4b). This suggests, that over this extended time of incubation, 2-chloroadenosine was being metabolised by adenosine kinase resulting in the trapping of [ 3 H] as phosphorylated [ 3 H]2-chloroadenosine derivatives. ABT-702 was therefore used in all subsequent experiments to minimise metabolism of 2-chloroadenosine by adenosine kinase. Figure 4d shows that 10 mM adenine also decreased 2-chloroadenosine uptake in the PK15-hENT4 cells to a degree similar to that seen for the PK15-NTD cells (Fig. 4b), likely due to the activity of ENBT1 in these cells. When [ 3 H]2-chloroadenosine uptake at pH 6.0 in the PK15-NTD cells (which includes the ENBT1-mediated uptake component) was subtracted from uptake at pH 6.0 in the PK15-hENT4 cells, the initial rate of hENT4-mediated [ 3 H]2-chloroadenosine (30 µM) uptake was estimated as 8 pmol/mg/min with a maximum accumulation of 124 ± 65 pmol/mg and a t 1/2 of 16 ± 11 min. This is similar to the initial rate of hENT4-mediated uptake defined above for 30 µM [ 3 H]adenosine. The uptake of a range of concentrations of [ 3 H]2-chloroadenosine up to 3 mM (solubility limit), in the presence of ABT-702, was then assessed using a 5 min incubation time (approximates initial rate of influx) in PK15-NTD cells and PK15-hENT4 cells at pH 7.5 and 6.0 ( Fig. 5a,b). Uptake by the PK15-NTD cells was linear with concentration, indicative of a non-mediated process, and there was no significant difference in uptake by PK15-NTD cells at pH 6.0 versus pH 7.5. However, in addition to the enhanced uptake at pH 6.0 in the PK15-hENT4 cells, there was also significantly more uptake in the PK15-hENT4 cells at pH 7.5 when compared with uptake by the PK15-NTD cells at pH 7.5, particularly at the higher concentrations of substrate (>750 µM). These data sets were then analyzed to obtain the acidic pH-dependent transport component (Fig. 5c, www.nature.com/scientificreports www.nature.com/scientificreports/ minus PK15-NTD, pH 7.5). Analyses of these data sets revealed that the acidic-pH-dependent transport component ( Fig. 5c) was best described (F-test, P < 0.05) by a allosteric sigmoidal relationship (Y = V max *X^h/ (K half^h + X^h) where V max is the maximum transport velocity for the substrate, K half is the concentration of substrate that produces half-maximal transport velocity (EC 50 ), and h is the Hill coefficient. The acidic pH-dependent component described in Fig. 5b had a V max of 324 ± 18 nmol/µg/min, a K half of 573 ± 47 µM, and a Hill coefficient of 2.5 ± 0.4 which is indicative of positive cooperativity. The Hill Plot derivation of these data show a significantly biphasic relationship (Fig. 5g), with a Hill coefficient of 0.68 ± 0.14 at substrate concentrations less than 200 µM and a Hill coefficient of 1.89 ± 0.27 at higher concentrations. The latter Hill coefficient is significantly greater than one, which is also indicative of positive cooperativity. When only the low concentrations of 2-chloroadenosine (1-200 µM) were considered, representing the portion of the acidic pH-dependent substrate/ rate profile more reflective of endogenous adenosine concentrations, the acidic-pH dependent transport data fit best to a simple non-cooperative Michaelis-Menten relationship (Fig. 5d) with a V max of ~30 pmol/mg/min and a K m ~50 µM. Note that due to the large variability that results from the data subtraction process used to obtain this sub-component, only approximations of the kinetic constants can be derived. In contrast, the hENT4-mediated transport component that remained at neutral pH ( Fig. 5e) was best described by a simple non-cooperative Michaelis-Menten relationship (Y = V max *X/(K m + X)), giving a V max of ~500 pmol/mg/min, a K m of ~2 mM (with no evidence of a high affinity component (Fig. 5f), and a Hill coefficient not significantly different from 1 (1.15 ± 0.12, Fig. 5h).
Inhibition of hENT4-mediated nucleoside uptake. Adenosine significantly inhibited the hENT4-mediated uptake of 30 µM [ 3 H]2-chloroadenosine at pH 6.0 but not at pH 7.5 (Fig. 6a). The lack of effect of adenosine on at pH 7.5 likely reflects the relatively low concentration of [ 3 H]2-chloroadenosine used for this particular assay. hENT4-mediated uptake at neutral pH is only apparent at higher substrate concentrations (see Fig. 5). Uridine, on the other hand, which can inhibit all of the other known purine nucleoside transporters (ENT1, ENT2, CNT2, CNT3) in the low-to mid-µM range 32,33 , did not inhibit hENT4 at concentrations up to 3 mM (Fig. 6b). This suggests www.nature.com/scientificreports www.nature.com/scientificreports/ that 3 mM uridine may be used to block mediated uptake of [ 3 H]2-chloroadenosine by other nucleoside transporters without inhibiting hENT4 in cell populations that express multiple nucleoside transporter subtypes. This lack of effect of uridine on [ 3 H]2-chloroadenosine uptake by PK15-hENT4 cells, along with the insensitivity of [ 3 H]adenosine uptake in the PK15-hENT4 cells to known inhibitors of ENT1 and ENT2 (Fig. 3a), also confirms that the base PK15-NTD cells are indeed deficient in all nucleoside transporter activity. The known hENT4 substrate MPP + inhibited 30 µM 2-chloroadenosine uptake with an IC 50 of 220 µM (log IC 50 = −3.66 ± 0.11). An unconstrained sigmoid curve fit to the MPP + data indicated that MPP + appeared to inhibit only the acidic pH-dependent component of the uptake (extrapolated lower limit of 6.5 ± 6.2% of control). This is in contrast to what was observed when using D22 or the more hENT4-selective analogue of dipyridamole, TC-T6000 17 . These agents appeared to inhibit both the acidic pH-dependent uptake and the hENT4-mediated uptake of 2-chloroadenosine at neutral pH (based on the observation that the inhibition profiles did not plateau at 0% in Fig. 6c). A trend in this regard was also seen for 10 µM D22 (Fig. 3b). While it is not possible to realistically determine the base of the inhibition profile from the data shown, it can be estimated from the results shown in Fig. 5a that about 55% of the uptake of 30 µM 2-chloroadenosine by the PK15-hENT4 cells is dependent upon an acidic pH. Therefore, inhibition of all hENT4-mediated uptake should theoretically plateau at about −82% on the graph shown in Fig. 6c. Thus, constraining a sigmoid curve fit to −82% allows a rough estimation of an IC 50 value of 2.3 µM for both D22 (logIC 50 = −5.64 ± 0.05) and TC-T6000 (logIC 50 = −5.63 ± 0.04). These IC 50 values for MPP + , D22, and TC-T6000 are comparable to those determined previously 17,18,34 , particularly when one considers the different expression models and substrate concentrations used.
Measurement of intracellular pH. For the transport studies described above, cells were incubated in solutions of different pH for up to 30 min (15 min pre-incubation for uptake studies +15 min substrate uptake/efflux time course), with the intent being to establish a proton gradient across the cell membrane. However, measurement of intracellular pH during these incubation conditions established that the initial pH gradient across the cell membrane declined rapidly when cells were incubated in either pH 6.0 or pH 8.2 buffer, with the intracellular pH equilibrating with the extracellular pH within 5 min (Fig. 8a), which was the time point used to assess transporter kinetics at pH 6.0 versus 7.5 in this study (Fig. 5). The requirement of a proton gradient for the acidic pH-dependent uptake by hENT4 was also examined by assessing the rate of influx of hENT4-mediated [ 3 H]2-chloroadenosine (30 µM) in the presence and absence of the proton ionophore nigericin (10 µM). It was confirmed via intracellular pH analyses that this concentration of nigericin completely collapsed the proton gradient in these cells. Nigericin had no effect on the uptake of [ 3 H]2-chloroadenosine at a pH of 7.5 (Fig. 8b). With a pH of 6.0, nigericin treatment did result in a significant decrease in 2-chloroadenosine influx at the longer time points (affecting maximum accumulation but not influx rate), but it was still significantly higher than that observed at pH 7.5 (Fig. 8b) indicating that a complete collapse of the proton gradient did not eliminate the acidic pH effect on hENT4-mediated transport.

Discussion
Our results show that ENT4 mediates both the influx and efflux of 2-chloroadenosine, confirming that hENT4 is a bidirectional transporter for nucleosides like other members of the ENT family [35][36][37][38] . While adenosine uptake by hENT4 has been reported previously 18 , this is the first study to show that hENT4 can also transport adenosine analogues. We also confirm previous reports that substrate uptake by hENT4 is enhanced by acidic pH 18 . A novel finding, though, is that the acidic pH-dependent component of 2-chloroadenosine uptake by hENT4 displays positive cooperativity (Hill coefficient of ~2) at substrate concentrations in excess of 200 µM. Furthermore, at lower substrate concentrations, in the more physiological range (<200 µM), hENT4, at acidic pH, is similar to other ENT subtypes with a K m for adenosine of ~50 µM 24,39 . This supports the relevance of hENT4 in mediating the transmembrane flux of nucleosides in the acidic environments associated with conditions such as ischaemia-reperfusion injury [40][41][42] . Given that ENT1 and ENT2 function is suppressed in hypoxic tissue [43][44][45] , hENT4 may assume an even more dominant role in mediating adenosine flux in these conditions. In contrast, the component of hENT4-mediated transport that is not dependent on an acidic pH does not show cooperativity and follows simple Michaelis Menten kinetics with a much lower affinity for 2-chloroadenosine (K m ~ 2 mM, Hill coefficient = 1.15). Therefore, acidification of the environment of hENT4 enhances the affinity of 2-chloroadenosine for the transporter. The appearance of this high affinity component at acidic pH also explains why most of the uptake observed in the initial time course experiments using 30 µM 2-chloroadenosine appeared predominantly acidic pH-dependent (Figs 3, 4 and 6). Whether this also occurs for the endogenous substrate adenosine has yet to be determined, but the attainment of such detailed transporter kinetics for adenosine would be problematic due to its rapid metabolism. The mechanism(s) underlying the change in substrate affinity and the positive cooperativity in this system at the higher substrate concentrations is unknown at present. Further analysis will require a model that has a higher level of hENT4 expression and lower background uptake (improving 'signal:noise' ratio) than the current PK15-ENT4 model provides. One can speculate that it may reflect allosteric changes in transporter protein structure due to multiple substrate binding sites with different affinities, or it may involve substrate driven changes in transporter oligomerization. Other nucleoside transporters such as ENT1 and ENT2 have been shown to form dimers 46 , and allosteric modification of inhibitor binding to ENT1 by high concentrations of adenosine has been reported 47,48 . Likewise, sigmoidal transport kinetics have been described for L-adenosine uptake by chromaffin cells 49 . Similar positive cooperativity in substrate-transporter interactions has been noted for other solute carriers such as the citrate transporter SLC13A5 50 .
A goal of this study was to determine if the increased metabolic stability of 2-chloroadenosine would allow more effective analysis of hENT4 nucleoside transport activity. While 2-chloroadenosine is not metabolized by adenosine deaminase, it is still susceptible to phosphorylation by adenosine kinase 20 . This is not an issue when 2-chloroadenosine is used as a substrate for studies of other nucleoside transporters such as ENT1 with uptake time courses on the order of seconds 22,30 , but the slower kinetics of hENT4 raises the possibility that adenosine kinase activity may confound transporter kinetic analysis even when using more stable substrates such as 2-chloroadenosine. This was confirmed by conducting studies in the presence and absence of the cell permeable adenosine kinase inhibitor ABT-702. The inclusion of ABT-702 decreased the maximum accumulation of [ 3 H] (prevents the intracellular trapping of [ 3 H] as 2-chloroadenosine phosphates), but did not affect the initial rate of [ 3 H]2-chloroadenosine influx, (indicating that ABT-702 was not directly inhibiting the transporter). Therefore, in the presence of adenosine kinase inhibitors such as ABT-702, [ 3 H]2-chloroadenosine is a suitable substrate for assessing the kinetics of nucleoside transport by hENT4.
As noted by others, D22 34 and TC-T6000 17 , can inhibit hENT4-mediated substrate influx. The affinity of D22 for blocking 2-chloroadenosine uptake by hENT4 in SLC29A4-transfected PK15-NTD cells was consistent with previous studies where 10 μM D22 was able to completely prevent adenosine uptake by hENT4 in SLC29A4-transfected MDCK cells 18 and Xenopus oocytes 3 . The ENT1 inhibitors nitrobenzylthioinosine (NBMPR), dilazep, and dipyridamole, data for the concentration range of 1-200 µM in Panel e. These data were best described by a linear relationship. Panels g and h are the Hill Plot derivations of the data shown in Panels c and e, respectively. The slopes of the fitted lines, which represent the Hill coefficients, are indicated.  25 , were relatively weak inhibitors of hENT4 as expressed in this model. We also show that the lidoflazine analogues soluflazine and draflazine, which are potent ENT1 inhibitors in other systems 28 , are likewise relatively ineffective at inhibiting hENT4. The dipyridamole analogue, TC-T6000, which has been proposed as a more selective hENT4 inhibitor than dipyridamole 17 had an IC 50 of 2.3 µM for inhibiting 2-chloroadenosine uptake by hENT4 in SLC29A4-transfected PK15-NTD cells. This is more than 30-fold higher than that reported previously for its inhibition of [ 3 H]adenosine uptake by hENT4 (74 nM), suggesting that TC-T6000 affinity for hENT4 may be substrate or model dependent. Interestingly, MPP + , an established substrate for hENT4 34   www.nature.com/scientificreports www.nature.com/scientificreports/ also inhibited the uptake at neutral pH). This difference may be related to the fact that MPP + is a substrate for the transporter whereas D22 and TC-T6000 are not.
A complication in studying hENT4 activity in endogenous cell models is that it is often co-expressed with other ENT subtypes (specifically ENT1 and ENT2). ENT1 can be potently inhibited with NBMPR, but ENT2 is less sensitive to inhibition. The best inhibitor for ENT2, dipyridamole, inhibits hENT4 to some degree at the concentrations used to maximally inhibit ENT2 (>1 μM). Dipyridamole has also been shown to have non-selective actions on other systems including affecting cAMP and cGMP metabolism 51,52 . Furthermore, the influence of the sodium-dependent concentrative nucleoside transporters (CNTs) in heterogeneous systems is often negated by conducting nucleoside flux studies in the absence of sodium. This is problematic in the study of hENT4, as hENT4 activity has been shown to be sensitive to changes in membrane potential 18,53 . Therefore, we investigated if the pyrimidine nucleoside uridine can be used as an inhibitor for other ENTs and CNTs without affecting hENT4. We showed that 3 mM uridine, which is sufficient to completely block ENT1 and ENT2 54 , and CNT2 and CNT3 33,55 , does not affect hENT4 function. This means that 3 mM uridine can be used alone or in conjunction with NBMPR as an alternative to dipyridamole to suppress ENT1 and ENT2, the main nucleoside transporters in most endogenous cell lines, as well as CNT-mediated transport activity, for study of hENT4. .0 transport buffer. Cells were washed in ice-cold buffer to remove extracellular substrate, and then efflux was initiated by adding transport buffer at the indicated pH. Aliquots of extracellular solution were taken at the specified times and assessed for [ 3 H]content. Data were fitted by a one-phase association curve. Each point represents the mean ± S.E.M. from 5 independent experiments. *Significant difference between efflux at pH 6.0 and 7.5 or 8.2. # Significant difference between pH 6.0 and 7.5 (Two-way ANOVA with Tukey's multiple comparisons test, P < 0.05). (b) Inhibition of efflux by D22 and TC-T6000. [ 3 H]2-chloroadenosine efflux was measured at pH 6.0, as described for Panel a, in the presence of D22 (2 μM) or TC-T6000 (1 μM). Each bar represents the mean ± S.E.M. from 5 experiments. The dashed line represents the uptake at pH 7.5 (from Panel a), shown for comparison. *Significant difference from efflux at pH 6.0 in the absence of inhibitor (Two way ANOVA with Tukey's multiple comparisons test, P < 0.05).
Other SLC29A family members have been shown to be bidirectional with respect to substrate flux, mediating the flow of substrate across the plasma membrane in either direction down the substrate concentration gradient [35][36][37][38] . Under conditions of cellular stress, such as that which occurs in ischaemia-reperfusion events, adenosine can be generated intracellularly from the breakdown of adenine nucleotides and released from cells via equilibrative nucleoside transporters 15,16 . This loss of intracellular adenosine contributes to cellular dysfunction upon reperfusion by limiting restoration of cellular adenine nucleotide pools. It has been shown previously that hENT4 is able to mediate efflux of MPP + 34 , but the dependence of efflux on pH was not assessed, nor were other substrates besides MPP + tested. It is not feasible to study the efflux of preloaded adenosine as it is rapidly metabolized intracellularly into impermeable adenine nucleotides. However, the finding that PK15-hENT4 cells can accumulate 2-chloroadenosine in a saturable manner upon blockade of its metabolism by adenosine kinase with ABT-702, enabled analysis of [ 3 H]2-chloroadenosine efflux by hENT4. Our results show that hENT4 is indeed able to mediate 2-chloroadenosine efflux, and that this efflux could be inhibited by the hENT4 blockers D22 and TC-T6000. Like hENT4-mediated influx, 2-chloroadenosine efflux was enhanced by acidification of the media. The acidic pH-dependent component of hENT4-mediated nucleoside flux does not appear to be proton gradient driven, as we show that collapse of the proton gradient with nigericin did not eliminate the effect of acidic pH on 2-chloroadenosine uptake. Furthermore, we established that an artificially generated proton gradient dissipated within 5 minutes in this cell model, and all of the kinetic influx data shown in Fig. 5 was derived using a 5 minute time point to represent initial rates. Therefore, the acid effect may be due to protonation of amino acid residues of hENT4 affecting transporter function. Alternatively, it may be the protonation of the substrate itself that leads to an enhancement of hENT4-mediated uptake, as has been described for the organic anion transporter OCT2 56,57 . Interestingly, a recent study on the molecular determinants of the pH sensitivity of another SLC29A family member, ENT3, showed that specific amino acids on the face of ENT3 exposed to the acidic environment were important for its acidic pH-dependence 58 . An analogous mechanism may be operating for hENT4.
In summary, our data show that, like ENT1 and ENT2, hENT4 is a bi-directional transporter with respect to 2-chloroadenosine (and presumably adenosine) flux. In addition, we have delineated a positively cooperative relationship for 2-chloroadenosine-ENT4 transporter interactions at high substrate concentrations at acidic pH. We also show that hENT4, in an acidic pH environment, transports 2-chloroadenosine with high affinity at