Handling of intracellular K+ determines voltage dependence of plasmalemmal monoamine transporter function

The concentrative power of the transporters for dopamine (DAT), norepinephrine (NET), and serotonin (SERT) is thought to be fueled by the transmembrane Na+ gradient, but it is conceivable that they can also tap other energy sources, for example, membrane voltage and/or the transmembrane K+ gradient. We have addressed this by recording uptake of endogenous substrates or the fluorescent substrate APP+(4-(4-dimethylamino)phenyl-1-methylpyridinium) under voltage control in cells expressing DAT, NET, or SERT. We have shown that DAT and NET differ from SERT in intracellular handling of K+. In DAT and NET, substrate uptake was voltage-dependent due to the transient nature of intracellular K+ binding, which precluded K+ antiport. SERT, however, antiports K+ and achieves voltage-independent transport. Thus, there is a trade-off between maintaining constant uptake and harvesting membrane potential for concentrative power, which we conclude to occur due to subtle differences in the kinetics of co-substrate ion binding in closely related transporters.

Despite the similarity in structure and function, the three transporters differ in many more 60 aspects than just ligand recognition: the transport stoichiometry of SERT and DAT/NET is 61 considered to be electroneutral and electrogenic, respectively. It has long been known that 62 SERT antiports Kin + , i.e., intracellular K + (Rudnick & Nelson, 1978); for DAT and NET, Kin + 63 is thought to be immaterial (Gu et  If true, only SERT can utilize the chemical potential of the cellular K + gradient to did bind to the inward facing state of DAT and NET but, in contrast to SERT, Kin + was released 79 prior the return step from the substrate free inward-to the substrate free outward-facing 80 conformations. We also found that substrate uptake by DAT and NET, unlike SERT, was 81 voltage-dependent under physiological ionic gradients. Moreover, the absence of Kin + had no 82 appreciable effect on the transport rate of DAT and NET. The transient nature of Kin + binding 83 was incorporated into a refined kinetic model, which highlighted the differences between SERT 84 and DAT/NET. Notably, this model allows for a unifying description, which attributes all 85 existing functional differences between DAT, NET and SERT to the difference in the handling 86 of Kin + .

88
Single cell uptake of APP + . We combined advantages of transporter-targeted radiotracer 89 assays and electrophysiology by setting up a system wherein APP + (Fig. 1A, left)   properties. Center, Schematics of the setup used in this study. The setup included: 1) a dichroic glass 112 that specifically reflects light of 440 nm but allows transmission of those with wavelength of 550 nm, 113 2) an inverted microscope with a 100x objective (for single cell fluorescence recordings), 3) an electrode 114 and patch clamp amplifier that allows for voltage control, 4) a photomultiplier tube (PMT) that converts 115 emitted fluorescence into electrical signal and 5) and an acquisition system that allows for filtering, 116 digitizing and two-channel recordings of APP + induced fluorescence and currents. Right, A theoretical 117 trace for a two-channel recording that displays simultaneous APP + -induced currents (red trace) and 118 fluorescence (orange trace) in real time. (B) Representative traces of the APP + -induced currents and 119 fluorescence in empty HEK293 cells or in HEK293 cells expressing DAT, NET and SERT patched 120 under normal physiological ionic conditions. In all traces, the rapid rise and decline in fluorescence on 121 applying and washing off APP + , respectively, is probably indicative of APP + adherence to the plasma 122 membrane and non-specific in nature. All three transporters show linear increase in fluorescence as a 123 function of time, indicative of APP + accumulation in cells expressing DAT, SERT and NET. APP + 124 induces robust DAT-mediated currents that comprise both peak and steady state currents. Only peak 125 currents (represented as magnified inset) were seen on APP + application to cells expressing NET and 126 SERT. AUarbitrary units. 127 Concentration dependence of APP + -induced currents and fluorescence. The slope of the 128 linear increase in fluorescence has the dimension of a rate (i.e. fluorescence*s -1 ) and is hence a 129 suitable readout for the uptake rate of APP + . We therefore determined the concentrations 130 required for achieving half-maximal uptake rates and measured the concomitant currents at -60 131 mV; original representative traces from single cells expressing DAT, NET and SERT are shown 132 in panels A&D, B&E, and C&F, respectively of Fig.2. In DAT, the APP + -induced currents 133 increase over the same concentration range as the rise in the rate of fluorescence. Accordingly, 134 the KM-values, which were estimated from fitting the data to a hyperbola (KM = 27.7 ± 7.1 µM 135 and 21.5 ± 10 µM, respectively), were indistinguishable within experimental error (Fig. 2J). 136 Dopamine induced transporter mediated currents (Fig. 2G) with a KM of 4.4 ± 1.4 µM. Thus, 137 when compared to dopamine, APP + is a low-affinity substrate of DAT (Fig.2J). In SERT, APP + 138 did not elicit any appreciable steady state currents (even at the highest concentration tested, i.e., 139 600 µM), but a robust concentration-dependent increase in peak current amplitudes (Fig. 2F). 140 APP + , nonetheless, accumulated intracellularly in SERT-expressing cells (Fig. 2C) indicating 141 that APP + was a substrate of SERT, which was translocated inefficiently. The KM for uptake of 142 APP + (32.0 ± 13 µM) was about an order of magnitude higher than the KM of 5-HT (3.6 ± 1.4 143 µM) estimated from 5HT-induced steady state currents (Fig. 2L). This indicates that APP + 144 uptake is also a low-affinity substrate of SERT. In cells expressing NET, APP + accumulated in 145 a concentration-dependent manner (Fig. 2B). Electrophysiological resolution of NET 146 associated currents revealed that neither norepinephrine (Fig. 2H) nor APP + (Fig. 2E) elicited 147 any steady state currents in NET on rapid application. In addition, APP + -induced peak currents 148 were considerably smaller than peak currents elicited by norepinephrine (cf. Fig. 2E and 2H).  37.3 ± 17 µM (n = 9)) in HEK293 cells expressing NET. Neither norepinephrine nor APP + induce any 164 steady state currents in NET (even at the highest concentration tested, i.e., 600 µM). (L) Normalized 165 concentration response of APP + -induced fluorescence (KM = 32.0 ± 13 µM (n = 6)) and serotonin-166 induced steady state currents (KM = 3.6 ± 1.4 µM (n = 9)) in SERT-expressing HEK293 cells. APP + did 167 not induce any steady state currents in SERT (even at the highest concentration tested, i.e., 600 µM).

168
All experiments were performed under physiological ionic conditions. All fluorescence and current 169 amplitudes were normalized to those obtained at 600 µM (which was set to 1) and the data points were 170 fitted with a rectangular hyperbola. All datasets are represented as means + S.D. AUarbitrary units, 171 norm.normalized. 172 Voltage dependence of APP + -induced currents and uptake. The data summarized in Fig. 2 173 indicate that APP + is a substrate, which is taken up with comparable KM by DAT, NET and 174 SERT. Accordingly, we applied APP + for 15 s at a concentration corresponding to the KM range  and Fig.3G, respectively). The voltage-dependence of these currents overlaps with that of DAT-190 mediated APP + uptake (Fig. 3J). In SERT, APP + did not induce sufficiently large steady state 191 currents to determine any current-voltage relationship (cf. Fig. 3F). However, serotonin induced 192 robust steady state currents (Fig. 3I), the amplitude of which was reduced by ~50% reduction 193 at + 30 mV (diamond symbols, Fig. 3L). It was not possible to do this comparison in NET (Fig. 194 3K), because neither norepinephrine nor APP + (cf. Fig. 3E and 3H) elicited steady state currents. of APP + -induced fluorescence (circle symbols, n = 6) and serotonin-induced steady state currents 206 (diamond symbols, n = 11) in SERT-expressing HEK293 cells. APP + did not induce any steady state 207 currents in SERT. All experiments were performed under physiological ionic conditions. All 208 fluorescence and current amplitudes were normalized to those obtained at -90 mV (which was set to 1) 209 and the data points were fitted with to the Boltzmann equation (except APP + uptake by SERT, which 210 was fit to a line). All datasets are represented as means ± S.D. AUarbitrary units, norm.normalized.

211
We note that the sigmoidal Boltzmann and the line function are both arbitrary fits to the data. Neither 212 one of them, is suitable to model the processes, which underlie the depicted voltage dependence. The 213 decision to use one or the other was based on the fidelity of the resulting fit. 214 Impact of intracellular cations on APP + uptake. It is safe to conclude from the observations 215 summarized in Fig. 3 that NET and DAT differ from SERT in their susceptibility to voltage: 216 transport of APP + by NET and DAT is voltage-dependent; in contrast, influx of APP + mediated 217 by SERT is essentially independent of voltage. Previous studies showed that Kin + was antiported  Fig. 5J-L). This is consistent with the idea that Nain + and Kin + bind to 274 overlapping sites in SERT and DAT. Nain + /Kin + , diamond symbols) or an intracellular environment of high Na + (163 mM Nain + , square 282 symbols). In the case of DAT (J), we also included data obtained with high intracellular Li + (163 mM 283 Liin + , triangle symbols). Peak current amplitudes were normalized to those obtained at -60 mV (which 284 was set to 1) and the individual datasets were fit with to the line equation. The slope of the voltage 285 dependence in the presence and absence of 163 mM Kin + was significantly different for all three 286 transporters (DAT (p < 0.0001), NET (p = 0.019) and SERT (p <0.0001); F-test). All data points are 287 represented as means ± S.D. 288 Effect of Kin + on uncoupled conductance and catalytic rate of monoamine transporters. 289 Because internal potassium did not affect DAT-mediated uptake (Fig. 4A), we examined the 290 role of Kin + in DAT by determining its effect on transport-associated currents. The presence 291 (square symbol in Fig. 6A) and absence of Kin + (circle symbol in Fig. 6A) (Fig. 6B). Currents through SERT (amplitudes of which are represented as 298 diamond symbol in Fig. 3L and square symbol in Fig.6C) are completely uncoupled from the 299 catalytic transporter cycle; in spite of its electroneutral stoichiometry, SERT mediates an 300 inwardly directed current, which is eliminated by removal of Kin + (circle symbol in Fig. 6C).  Fig. 6E and 6F, the catalytic 314 rates of DAT and NET in the presence or absence of Kin + is very similar. In fact, Kin + fails to 315 render the peak current recovery by DAT voltage-independent (see supplementary Fig. 1). In 316 contrast, SERT shows a ~2 fold deceleration in the catalytic rate in the absence of Kin + when 317 compared to its recovery rate in the presence of high Kin + (Fig. 6G). These observations are in 318 stark contrast to the indistinguishable uptake by SERT of APP + observed in the presence or 319 absence of Kin + (Fig. 4C). This discrepancy can be accounted for by APP + being a poor 320 substrate, an explanation, which is supported by our observations that APP + did not induce any 321 detectable steady state currents in SERT (Fig. 2F and 3F). All three monoamine transporters 322 can also operate in the exchange mode, which is the basis for the actions of amphetamines (Sitte 323 and Freissmuth, 2015). As a control, we examined the recovery in the presence of high Nain + , 324 which precludes cycling in the forward transport mode and thus forces the transporters into 325 exchange: as predicted, high Nain + accelerated the recovery of all three transporters (square 326 symbols in Fig. 6E-G). symbols), devoid of intracellular Na + or K + (0 mM Nain + /Kin + , diamond symbols) or contain high 342 intracellular Na + (163 mM Nain + , square symbols). Peak current amplitudes obtained at each test pulse 343 were normalized to that of the reference peak (which was set to 1) and fitted with a mono-exponential 344 function. The catalytic rates obtained were as follows: DAT (163 mM Kin + -3.74 ± 0.76 s -1 ; 163 mM 345 Nain + -10.68 ± 2.14 s -1 ; 0 mM Kin + /Nain + -4.45 ± 0.93 s -1 ), NET (163 mM Kin + -0.15 ± 0.013 s -1 ; 163 346 mM Nain + -0.22 ± 0.016 s -1 ; 0 mM Kin + /Nain + -0.15 ± 0.014 s -1 ) and SERT (163 mM Kin + -1.61 ± 0.12 347 s -1 ; 163 mM Nain + -2.57 ± 0.28 s -1 ; 0 mM Kin + /Nain + -0.74± 0.058 s -1 ). The data in E, F and G are 348 represented as means ± S.D (n = 5 for each condition). 349 A kinetic model for the transport cycle of monoamine transporters. The data, represented 350 in Fig. 6, can be explained by a model, which posits that all monoamine transporters can bind 351 Kin + , but that the bound Kin + is released on the intracellular side prior to the return step by DAT 352 and NET. In contrast, Kin + is released on the extracellular side after being antiported by SERT. 353 We tested the plausibility of this hypothesis by resorting to kinetic modeling. As a starting point 354 for modeling DAT and NET, we used the previously proposed kinetic model for DAT by 355 Fig. 7A), which is nested within 356 our proposed model. For NET, we posited a much slower dissociation-rate for the substrate 357 (indicated as green text) to account for the small substrate turnover rate and the absence of the 358 steady current component (cf. Fig.2H and Fig.6E). The model was expanded to account for  Notably, voltage independent substrate transport by SERT in the absence of Kin + (Fig. 4C) Fig.2). This 374 finding is consistent with the concept that only SERT can utilize protons in the return step. 375 We interrogated the kinetic models to generate synthetic data for APP + transport by DAT (Fig.   376 7C), NET (Fig. 7D) and SERT (Fig. 7E) at different membrane voltages. The synthetic data 377 generated through the respective kinetic models could faithfully reproduce our experimental 378 findings: i) only APP + uptake by SERT was voltage-independent (cf. Fig.7C-7E and Fig. 3A-379   3C); ii) the removal of Kin + abrogated the steady-state current only in SERT but not in DAT 380 (cf. Fig.7F/7G and Fig. 6A/6C); iii) the removal of Kin + did not slow down the return of DAT 381 and NET from the inward-to the outward-facing conformation, while it reduced this rate in 382 SERT by two-fold (cf. Fig. 6E -6G to Supplementary Fig.3C). For other simulated datasets, 383 please refer to Supplementary Fig.3. This indicates that the underlying assumptions are valid 384 and allow for a reasonable approximation, which has explanatory power: the differences in 385 handling of Kin + incorporated into the model are necessary and sufficient to account for the 386 differences in the forward transport mode of DAT, NET and SERT. here that all major differences can be accounted for by the distinct handling of Kin + : (i) in SERT, 422 physiological Kin + concentrations accelerated the rate of substrate uptake: it was 2-fold faster 423 than in the absence Kin + (Fig. 6G). In contrast, DAT and NET return to the outward-facing state 424 with the same rate regardless of whether Kin + is present or not ( Fig. 6E and 6F). Accordingly,

425
Kin + did not affect rate of substrate uptake by DAT and NET ( Fig. 4A and 4B).  Fig. 5L). This was not the case for DAT and NET (Fig. 3J and 3K and Fig. 5J and 5K, Na2 site was proposed earlier by a study that employed extensive molecular dynamic 446 simulations to understand intracellular Na + dissociation from DAT (Razavi et al., 2017). Our 447 results highlight the fact that the voltage dependence of peak amplitudes is identical in the 448 presence of high Nain + and high Kin + (Fig. 5J) and thus support this conjecture.

449
The most parsimonious explanation for all differences between SERT, NET and DAT was to 450 posit that Kin + is antiported by SERT but not by DAT and NET. All three transporters carry a 451 negative charge through the membrane on return from the substrate-free inward-to the 452 substrate-free outward-facing conformation (presumably a negatively charged amino-acid). In 453 the case of SERT, however, the charge on the transporter is neutralized by the counter-454 transported Kin + . Because the return step is slow and therefore rate-limiting, it determines the 455 voltage dependence of substrate uptake. The Kin + -binding site in SERT, alternatively, can also 456 accept protons (Keyes and Rudnick, 1982). Hence protons -as alternative co-substrate that is 457 antiported -support the return step from the inward-to the outward facing substrate-free 458 conformation (Hasenhuetl et al., 2016). The alternative is to postulate, based on recent 459 evidence in LeuT (Billesbølle et al., 2016), that antiport of Kin + is a general feature of all SLC6 460 transporters. However, this can be refuted for DAT and NET for the following reasons: the 461 presence or absence of Kin + , did not change their catalytic rates ( Fig.6E and 6F). Thus, in the 462 absence of Kin + , the transporters seem to return from the substrate-free inward-to the substrate-463 free outward-facing conformation. In this case, however, the transporters carry one positive 464 charge less through the membrane. This change in ion translocation must, therefore, translate 465 into a concomitant, substantial change in the voltage dependence of substrate transport, i.e., the 466 voltage dependence of transport ought to be much steeper in the absence than in the presence 467 of Kin + . This was not observed (Fig.4A and 4B). Additionally, H + failed to accelerate the 468 catalytic rate of DAT (data not shown) and the slope of the IV-curve for the peak current 469 remained steep (Supplementary Fig.2). These observations indicate that protons (like Kin + ) was equivalent to that of the cognate substrate in DAT, but substantially lower in SERT. 486 Originally, NET expressed in HEK293 cells was reported to support both, a peak and a steady 487 current, when challenged with substrate (Galli et al., 1995). However, in the present study, we 488 only observed the peak current with our superfusion system, which allowed for rapid exchange 489 of solutions: neither APP + nor the cognate substrate norepinephrine elicited a steady current.

490
The absence of the steady current can be attributed to the very slow catalytic rate of NET: it is 491 evident that, by contrast with DAT and SERT, NET returns on a timescale of seconds from the 492 inward-to the outward-facing state: NET dwells in the inward-facing state with a lifetime of τ 493 = ~7 s (cf. Fig.6F), which is >10-20 times longer than the dwell time of DAT and SERT (DAT, 494 τ = ~0.3s; SERT, τ = ~0.6s, Fig. 6E & 6G). Thus, this very slow turnover explains the absence 495 of coupled or uncoupled NET-mediated currents in spite of the proposed electrogenic 496 stoichiometry (Gu et al., 1996). 497 Our analysis provides a unifying concept of substrate transport through all three monoamine to reconcile with the concept of transport by fixed stoichiometry. We, therefore, surmise that 518 DAT, NET and SERT operate with a mixed stoichiometry. Based on our data we conclude that 519 DAT and NET are less likely than SERT to antiport Kin + , because we cannot rule out that they 520 can occasionally carry the Kin + ion through the membrane. Conversely, SERT antiports Kin + in 521 the majority of its cycles but may return empty in some instances. We thus believe that the 522 differences between these three transporters with respect to their handling of Kin + represents a 523 continuum, as opposed to divergence, in ionic coupling and kinetic decision points during 524 substrate transport. The difference between SERT and DAT/NET represent different 525 approaches to an inherent trade-off and may reflect an adaptation to physiological requirements: 526 because of electrogenic binding and subsequent counter-transport of Kin + , SERT operates in the 527 forward transport mode with a constant rate regardless of membrane potential, but it cannot 528 exploit the membrane potential to fuel its concentrative power. In contrast, DAT and NET can 529 harvest the energy of the transmembrane potential to fuel its concentrative power. As a trade-530 off, the substrate uptake rate of DAT and NET is voltage-dependent and strongly reduced or 531 increased upon membrane depolarization or hyperpolarization, respectively.  Dopamine, norepinephrine, serotonin or APP + was applied using a DAD-12 superfusion system 562 and a 4-tube perfusion manifold (ALA Scientific Instruments), which allowed for rapid solution 563 exchange. Current traces were filtered at 1 kHz and digitized at 10 kHz using a Digidata 1550 564 (MDS Analytical Technologies). Current amplitudes and accompanying kinetics in response to 565 substrate application were quantified using Clampfit 10.2 software (Molecular Devices).

566
Passive holding currents were subtracted, and the traces were filtered using a 100-Hz digital 567 Gaussian low-pass filter.  We thank Verena Burtscher for discussion and comments on the data.