Functional characterization of SGLT1 using SSM-based electrophysiology: Kinetics of sugar binding and translocation

Beside the ongoing efforts to determine structural information, detailed functional studies on transporters are essential to entirely understand the underlying transport mechanisms. We recently found that solid supported membrane-based electrophysiology (SSME) enables the measurement of both sugar binding and transport in the Na+/sugar cotransporter SGLT1 (Bazzone et al, 2022a). Here, we continued with a detailed kinetic characterization of SGLT1 using SSME, determining KM and KD app for different sugars, kobs values for sugar-induced conformational transitions and the effects of Na+, Li+, H+ and Cl− on sugar binding and transport. We found that the sugar-induced pre-steady-state (PSS) charge translocation varies with the bound ion (Na+, Li+, H+ or Cl−), but not with the sugar species, indicating that the conformational state upon sugar binding depends on the ion. Rate constants for the sugar-induced conformational transitions upon binding to the Na+-bound carrier range from 208 s−1 for D-glucose to 95 s−1 for 3-OMG. In the absence of Na+, rate constants are decreased, but all sugars bind to the empty carrier. From the steady-state transport current, we found a sequence for sugar specificity (Vmax/KM): D-glucose > MDG > D-galactose > 3-OMG > D-xylose. While KM differs 160-fold across tested substrates and plays a major role in substrate specificity, Vmax only varies by a factor of 1.9. Interestingly, D-glucose has the lowest Vmax across all tested substrates, indicating a rate limiting step in the sugar translocation pathway following the fast sugar-induced electrogenic conformational transition. SGLT1 specificity for D-glucose is achieved by optimizing two ratios: the sugar affinity of the empty carrier for D-glucose is similarly low as for all tested sugars (KD,K app = 210 mM). Affinity for D-glucose increases 14-fold (KD,Na app = 15 mM) in the presence of sodium as a result of cooperativity. Apparent affinity for D-glucose during transport increases 8-fold (KM = 1.9 mM) compared to KD,Na app due to optimized kinetics. In contrast, KM and KD app values for 3-OMG and D-xylose are of similar magnitude. Based on our findings we propose an 11-state kinetic model, introducing a random binding order and intermediate states corresponding to the electrogenic transitions detected via SSME upon substrate binding.

References -p. 21 2 S1. Supplementary Methods S1.1. Sensor preparation and electrophysiological measurements SSME was performed using the SURFE 2 R N1 device (Nanion Technologies GmbH) as described in detail previously (Bazzone et al., 2017a;Bazzone and Barthmes, 2020). SURFE 2 R N1 sensors of different diameters (1 mm or 3 mm) were coated first with an alkanethiol layer, then with 1,2-diphytanoyl-sn-glycero-3-phosphocholine (DPhPC) to form the SSM in the presence of resting solution (R). The composition of R depends on the assay (see below). Aliquots of frozen membrane vesicles are thawed and diluted 1:10 in R. Following sonication using a tip sonicator (UP 50 H, Dr. Hielscher, equipped with MS 1 tip; 10 bursts, 20% amplitude, 0.5 s cycle time), 10 µl are added to each sensor. The membrane vesicles attach to the SSM on the sensor and form a coupled membrane system. After centrifugation of the sensors (2,000 g, 30 min), they are ready-to-use.
The SURFE 2 R N1 performs a fast solution exchange providing the substrate to activate the transporters immobilized on the sensors. Measurements take place under continuous flow of solutions. In our standard protocol, three solutions are flushed across the sensor, each for 1 s and at a speed of 200 µl/s: (1) During the flow of non-activating solution (NA) a current baseline is recorded; (2) When switching to the sugar containing activating solution (A), SGLT1 is activated and a capacitive current is recorded. (3) At the end of the experiment, NA flow restores initial conditions. Multiple solution exchange experiments are carried out on the same sensor. The solution exchange protocol is provided in each figure as a perfusion scheme that contains the compositions of R, NA and A. In case multiple conditions are measured using the same sensor, the measurement sequence is indicated within the perfusion scheme. All currents shown are recorded during the flow of activating solution, starting with the solution exchange from NA to A.

S1.2. Solutions for electrophysiological measurements
The main buffer used to prepare the measurement solutions contained 30 mM Tris/HCl, 3 mM EDTA, 1 mM EGTA and 120 mM NMDG/SO 4 . If not stated otherwise, the pH was titrated to 7.4 using KOH. When H + /sugar cotransport was investigated (Figure 7), titration was performed to achieve the acidic pH values as indicated.
Depending on the experimental conditions, different cations and sugars were used in NA and A solutions. In our standard assay 300 mM NaCl is added to both, NA and A solutions. In addition, the NA solution contains 250 mM Mannitol and the A solution contains 250 mM D-glucose, i.e. SGLT1 was always activated via sugar concentration jumps. In sugar specificity experiments, D-glucose was replaced by either MDG, D-galactose, OMG, D-xylose, DOG or fructose. When different sugar concentrations are used, the A solution (250 mM sugar) is diluted using NA solution (250 mM Mannitol) to keep a total osmolarity of ~250 mOsm/L. For the sugar concentration dependence, sensors were prepared in solution containing 250 mM Mannitol and measurements were carried out in sequence from high to low sugar concentrations.
When different Na + concentrations are used, Na + was supplemented with K + to achieve a constant cation concentration of 300 mM. If not stated otherwise, measurement solutions without Na + contain 300 mM K + instead. In some experiments, Na + is replaced by Li + to investigate Li + /sugar cotransport (Figure 7). When Clconcentrations are altered, Clwas supplemented by gluconate to achieve a constant anion concentration of 300 mM. Measurements in absence of Clcontain 300 mM gluconate.
In our standard assays, the R solution used for sample dilution and sensor preparation matches the NA solution, hence the internal composition of the vesicles matches the external composition before SGLT1 is activated via sugar concentration jump. For measurements in the presence of cation gradients (Figure 7), the resting solution contains 300 mM K + instead of Na + or Li + . The same is true when a cation concentration sequence is measured. Here, measurements were carried out in sequence from low to high cation concentrations. Before the measurement, the sensor was incubated for 3 minutes at the given cation concentration, assuming that the internal cation concentration equilibrates as found previously (Bazzone et al., 2022a).

S1.3. Data normalization
Since current amplitudes vary between sample batches and sensors, usually every dataset recorded from a single sensor was normalized before averaging across sensors to calculate the mean and standard deviation.
When concentration dependent peak currents or integrals are analyzed, the same normalization procedure was followed. First, all values recorded from one sensor are normalized to the value obtained for the highest substrate concentration. Then all normalized datasets recorded from different sensors are averaged yielding mean values and standard deviations. In addition, the average of the absolute values for the highest substrate concentration was determined. In order to get an estimate of I max and Q max , the normalized averaged dataset was multiplied by this average value. As a consequence of the normalization, standard deviation of the average data point for the highest substrate concentration is zero and all hyperbolic fits were performed without weighting.
In order to compare I max and Q max between different sugar substrates, we have measured 500 mM concentration jumps of all substrates on the same sensor ( Figure 6). This yields accurate relative values for the different substrates. Normalized average datasets for each substrate then were normalized to reflect the relative peak currents for the different sugar substrates at 500 mM concentration. Therefore, I max and Q max values provided in Table 1 may be directly compared across substrates. S1.4. Separation of PSS and transport currents during data processing In SSME, usually the peak current recorded during the A solution flow is used for analysis. In most cases, it correlates with the steady-state transport rate of the transporter. However, for some transporters also substrate-induced PSS currents were observed, potentially overlaying with the transport current phase (Garcia-Celma et al., 2010;Bazzone et al., 2022b;Mikušević et al., 2019;Bazzone et al., 2016;Bazzone et al., 2017b).
As shown previously, in case of SGLT1 and depending on the experimental conditions, the peak current may be affected by both, transport and PSS current phases (Bazzone et al., 2022a). Hence, we did not simply use the peak currents to analyze the transport properties, but different methods to separate the phases representing PSS current and transport current phases.
To analyze transport currents, we used three different approaches: (1) the total integral during A solution flow reflecting the charge translocation. Usually, the overall charge translocation is dominated by transport since PSS charge translocation often only transfers a fraction of an elementary charge per transporter. The integration method may be used for datasets recorded from 3 mm and 1 mm sensors.
(2) the current obtained from 1 mm sensors about 50 ms after sugar addition. At this time point PSS currents already decayed to zero; (3) the steady-state current which was obtained via reconstruction of the transporter current (Tadini-Buoninsegni and Fendler, 2015).
To analyze PSS currents, peak currents from 1 mm sensors were used. In contrast to 3 mm sensors, time resolution is increased (Bazzone et al., 2017a). Since most of the recorded PSS currents are fast, peak currents are dominated by PSS currents when 1 mm sensors are used. However, there may still be a remaining impact of the transport current phase. Alternatively, we also used the charge translocation within the first 30 ms upon sugar concentration jump, mainly being affected by PSS charge translocation. In absence of Na + , we used both, peak currents and integrals to analyze PSS currents, since transport is not observed under these conditions. S1.5. Data processing: EC 50 values, including K M and K D app Despite different read-outs, the data processing to derive EC 50 values for PSS and transport currents is the same. Measurements were performed on at least 3 to 5 different sensors. All different concentrations used for the analysis were measured on the same sensor.
Due to variances across sensors, values obtained from the same sensor are normalized to the value obtained for the highest concentration. Datasets then were averaged across sensors to obtain a standard deviation. In order to retain information about relative I max or Q max values the averaged, normalized datasets may be further processed (compare section S1.3).
Average values and standard deviation for each concentration are plotted and fitted using a standard Hill equation I=I max *c n /(EC 50 n +c n ) to obtain EC 50 and I max values. If not stated otherwise, the Hill coefficient n was set free for the fit. In some cases the modified Hill equation I = I max + (I max -I min )*c n /(EC 50 n +c n ) was used to include I min at 0 mM concentration. This is the case for the cation EC 50 of the PSS current, since the presence of the cation alters the sugar-induced PSS currents but is not a requirement to detect sugar binding induced currents ( Figure 3, Figure 7). S1.6. Data processing: rate constants The current decay of the transient currents may be fitted assuming an exponential decay. Depending if only PSS or transport and PSS current phases are visible, the mono-exponential equation I = A1*exp(-t/τ PSS ) ( Figure 2B) or the bi-exponential equation I = A1*exp(-t/τ PSS )+ A2*exp(-t/τ SS ) ( Figure 4B) was used. In case of the bi-exponential equation, the slow time constant τ SS was always assigned to the transport current phase and the fast time constant τ PSS to the PSS current phase. The rate constant of the PSS reaction is derived from k obs,PSS = 1/τ PSS . k obs values were determined from at least 3 to 5 different sensors and averaged to determine standard deviation. We found k obs being sugar concentration dependent in the presence of Na + (Figure 4). Assuming that the electrogenic reaction directly follows sugar binding, the concentration dependent k obs can be used to determine k on and k off of the electrogenic reaction and the K D for sugar binding using a fit with the model equation k obs,PSS = k on *c n /(K D n +c n ) + k off (Garcia-Celma et al., 2010). For this fit we fixed n=1, since only one sugar molecule binds to SGLT1.

S2.1. Symmetry of transport
We showed before that PSS currents dominate only in influx mode, while the peak currents in the efflux mode are dominated by the slow transport component (Bazzone et al., 2022a). Here we wanted to test if the transporter has similar steady-state properties in both transport directions by determining the K M for D-glucose during Na + /D-glucose efflux.
To perform efflux experiments with precisely defined internal substrate concentrations, vesicles are preloaded with the substrate via dilution and sonication, before adding them to the sensors (section S1.1). Consequently, for each substrate concentration, a different sensor was used. The standard deviation is increased upon averaging, due to missing normalization. However, we found K M = 2.2 ± 1.6 mM and I max = 761 ± 117 pA for D-glucose during Na + /D-glucose efflux, using the current 50 ms after the sugar concentration jump (Figure SI-1A). As a control, we repeated the influx assay with the same substrate concentrations and reproduced K M = 1.7 ± 0.1 mM and I max = 885 ± 125 pA using the same analysis procedure (Figure SI-1B), indicating that there may be only a slight asymmetry between steady-state influx and efflux properties regarding D-glucose transport in SSME experiments.
For the efflux assay, we used 300 mM Na + in both, internal and external solutions at 0 mV and only the sugar concentration gradients as the driving force. Apparent affinities for MDG in Na + /sugar efflux mode found in literature are usually higher and between 7 and 56 mM (Sauer et al., 2000;Eskandari et al., 2005;Quick et al., 2003), indicating a more prominent asymmetry. Since we found that the K M for MDG in SSME experiments is higher than for D-glucose (Table 1) our data approaches the lower end of the spectrum of literature values.

S2.2. Ordered binding models
The apparent affinity for the sugar depends on the Na + concentration, indicating binding cooperativity. Na + enhances the accessibility of the sugar binding site, increasing the sugar affinity and indicating an ordered binding model as proposed in literature (Sala-Rabanal et al., 2012;Loo et al., 2013;Gorraitz et al., 2017;Adelman et al., 2016;Peerce and Wright, 1984;Hirayama et al., 2007). Assuming that Na + binds before the sugar, the Na + concentration dependent K M values for D-glucose may be used to derive K M,Na # (S), the K M value for D-glucose when the Na + binding sites of SGLT1 are saturated (Figure SI-2A), as previously shown for H + /sugar cotransport (Bazzone et al., 2016). The determined K M,Na # (S) for D-glucose is 0.68 ± 0.07 mM and the Hill coefficient for Na + binding is n = 2.02 ± 0.02, as expected for the binding of two Na + ions. The quality of the model fit (R² > 0.9999) supports the ordered binding model.
On the other hand, the apparent affinity for Na + also depends on the sugar, indicating that binding cooperativity works in both directions. We applied the same model equation assuming that the sugar binds before Na + . We determined K M,S # (Na + ) = 48 ± 12 mM and a Hill coefficient for sugar binding of n = 0.52 ± 0.13 (Figure SI-2B). The fit is not exceptionally good with R² = 0.98. However, the data demonstrates that the sugar is not only able to bind in absence of Na + (Figure 2), but also able to bind before Na + under transport conditions, which in response increases the apparent affinity for Na + . 6 S2.3. The role of chloride in sugar binding and Na + /sugar cotransport Chloride is known to bind to SGLT1, increasing the K M for Na + , while not being transported (Loo et al., 2000). On the other hand, Loo et al. found that I max for Na + /sugar transport and K M for MDG are relatively unaffected by Cl -. Using SSME, we aimed to expand this knowledge by investigating the effect of Clon sugar binding and on Na + /sugar translocation triggered by sugar concentration jumps under 0 mV conditions. S2.3.1 Chloride affects all kinetic parameters during Na + /D-glucose cotransport We measured Na + /MDG cotransport using different MDG and Na + concentrations in the presence of ~300 mM Clin both, internal (R) and external (NA,A) solutions. The internal solution is defined during sensor preparation (section S1.1). We then compared the results with the same experiment, but partially replacing Clby gluconate -(internal solution: 15 mM Cl -/ 285 mM gluconate -; external solution: 0 mM Cl -/ 300 mM gluconate -) ( Figure SI-7).
From the MDG concentration sequence we found that chloride affected almost all kinetic parameters: (1) The K D,Na app and K M for MDG both double when Clconcentration is reduced, from 96 mM to 189 mM and from 2.9 mM to 6.8 mM, respectively. Opposingly, the effect of Clon K M for MDG found by Loo et al. was only ~10% at -50 mV (Loo et al., 2000). Interestingly, the K D,Na app /K M ratio is almost independent of chloride, being 33 and 28 in this set of experiments and at high and low Clconcentrations, respectively. The effect on K M likely represents a direct consequence of Claffecting K D,Na app . (2) The transport I max decreases from 950 pA to 450 pA when Clconcentration is reduced. Hence, Clnot only enhances sugar affinity, but also the transport rate. This is roughly in agreement with the findings of Loo et al., who observed a decrease in transport current of about 30 % when Clwas removed at 0 mV (Loo et al., 2000). (3) Finally, I max of the PSS peak current increases from 3 nA to 5.3 nA at high Clconcentrations, indicating a different sugar-induced charge translocation when Clhas bound. It is rather unlikely that chloride contributes with its negative charge towards the overall charge displacement measured upon sugar binding, since this would lead to an overall reduced charge translocation at high chloride concentrations -the opposite of what we observed. Chloride rather affects the conformational state of SGLT1 upon sugar binding. This is similar to the observation that the charge of the cation (Na + , Li + , H + ) has no direct impact on I max of the PSS current (section 3.3.5 of the main manuscript). Cations and Cldefine the overall conformational state of SGLT1 after sugar binding.
Following the MDG concentration sequence, we checked the impact of Clon K M and EC 50 PSS for Na + . Similar as for the sugar, the values decrease from EC 50 PSS > 300 mM and K M = 87.5 mM at low chloride concentrations to EC 50 PSS = 252 mM and K M = 49.5 mM at high Clconcentrations. This is also in strong agreement with Loo et al. who observed an increase in K M for Na + from 41 mM at 106 mM Clto 62 mM at 20 mM Clconcentration (Loo et al., 2000). Interestingly the PSS peak current increases in the presence of Clby a factor of 1.7 ± 0.1, independent of Na + concentration, indicating the lack of binding cooperativity between Na + and Cl -.

S2.3.2 Clonly binds to SGLT1 from the intracellular site
To determine K M and EC 50 PSS values for Clvia a Clconcentration sequence, we used different Cl -/gluconate ratios to vary the Clconcentration and activated SGLT1 via 250 mM MDG concentration jumps in the presence of 300 mM Na + . In SSME, concentration sequences are usually performed sequentially on the same sensor. When attempting the Clsequence, starting with vesicles equilibrated in 15 mM Cl -, we could not see any effect on the currents upon stepwise Claddition by changing the Clconcentration of the external solution from 0 mM to 300 mM ( Figure SI-8A). Clcould not affect SGLT1 activity when provided with the external solutions (NA,A). Recently, we showed that the vesicles used in SSME measurements contain SGLT1 in right-site-out orientation (Bazzone et al., 2022a). Hence, the results indicate, that Clonly binds to SGLT1 from the intracellular site. However, Loo et al. found effects of Clwhen removing it from the external solution in TEVC experiments on SGLT1 expressing oocytes (Loo et al., 2000). This conflict might be a result of different Clpermeabilities between the oocytes plasma membrane in the presence of membrane voltage and the plasma membrane vesicles derived from CHO cells at 0 mV which were used in our SSME assays.

S2.3.3 K M and EC 50 PSS for Cl -
Given that external Cldoes not affect SGLT1 activity (section S2.3.2), we preloaded vesicles with different Clconcentrations via sonication before sensor preparation and performed only one measurement using a single Clconcentration per sensor to determine K M and EC 50 PSS values for Cl -. The PSS and transport current phases change with Clconcentration ( Figure SI-8B), as expected from MDG and Na + dependent measurements at high and low Cl -(section S2.3.1). From this dataset we determined K M and EC 50 PSS values for Cl -. From the current 50 ms after substrate jumps we determined a K M value for Clof 6.8 ± 2.8 mM ( Figure SI-8C). The EC 50 PSS value concluded from the PSS peak current is 21 ± 2 mM ( Figure SI-8D). Loo et al. found K M = 21 mM (Loo et al., 2000). The hyperbolic fits of the Clconcentration dependent transport and PSS currents suggest that both currents are massively decreased at 0 mM Cl -, indicating that the residual transport activity and PSS electrogenicity of SGLT1 at 0 mM internal Clis close to zero.

S2.4. Model Simulations
The following paragraphs aim to critically analyze the 11-state model and what aspects of the experimental data it is able to describe. Clearly, it does not describe all kinetic data available on SGLT1, which would require additional intermediate states within the empty carrier along with additional electrogenic steps observed in conventional electrophysiology and as discussed by Loo et al. and Longpré et al. (Loo et al., 2006;Longpré et al., 2012).
S2.4.1 The 11-state kinetic model explains substrate binding and steady-state kinetics found with SSME For the simulations we aimed to explain substrate binding and steady-state kinetics within an allowed discrepancy of a factor of ~2 between the model simulation and the experimental data. The proposed set of rate constants for the 11-state model (Table SI-3) is able to explain the following experimental data found using SSME (compare also Table SI-2): 1. Steady-state kinetics for Na + and sugar o K M for Na + under saturating sugar conditions is between 39 mM and 59 mM in SSME and 31 mM in the model simulations. o K M for the sugar under saturating Na + conditions is between 0.8 and 1.9 mM in SSME and 2 mM in the model simulations. o Cooperativity between Na + and Sugar represented by relative changes of K M values with cosubstrate concentration is essentially the same within the model and experimental data: our model simulations show that a random binding order using the given rate and equilibrium constants explain the dependence of co-substrate concentrations and K M values very well (Table  SI-2). The relative change of the K M for D-glucose at 300 mM Na + compared to 100 mM, 50 mM 8 and 20 mM Na + is 2.9 ± 0.8, 9.4 ± 6 and 55 ± 39 ( Figure 5) and 3.5, 9.9 and 34 for the experimental dataset and the model simulations, respectively. The relative change of the K M for Na + at 250 mM D-glucose compared to 20 mM, 4 mM and 1 mM D-glucose is similarly well represented by the model: In the experiment we obtained ratios of 1.3 ± 0.1, 2.3 ± 0.3 and 3.6 ± 0.9 ( Figure 5); the model generates ratios of 1.8, 2.5 and 2.9 respectively. o Impact of Na + gradients: At 300 mM external and internal Na + , removing Na + from the internal solution in model simulations increases the steady-state current 2.1-fold. In the experiment, steady-state currents are increased by a factor of 3.48 ± 0.58. o Sugar Efflux: The K M value for the sugar in efflux mode at saturating Na + conditions is slightly increased compared to the influx mode. We observed a K M of 2.2 ± 1.7 mM in SSME and 5.5 mM in model simulations. 2. EC 50 of the sugar-induced PSS (referred to as K D app for the sugar) o K D,Na app for D-glucose: The sugar EC 50 of the sugar-induced PSS in the presence of Na + is between 14.5 mM and 38 mM in SSME and 20 mM in the model simulations. o K D,K app for D-glucose: The sugar EC 50 of the sugar-induced PSS in absence of Na + is 210 ± 52 mM in SSME and 182 mM in the model simulations. o EC 50 PSS for Na + : The Na + EC 50 of the sugar-induced PSS is between 231 mM and 285 mM in SSME and 102 mM in the model simulations.

Rate constants for PSS currents
o Rates in the presence of Na + : For D-glucose we determined forward and reverse rates for the electrogenic induced fit of 208 s -1 and 56 s -1 , respectively ( Figure 4C). Accordingly, we defined k on and k off values for the transition following sugar binding to the Na + bound carrier in the 11state model with 200 s -1 and 60 s -1 , respectively. Simulation of the PSS currents yield rate constants matching with the experimentally observed constants (Table SI-2): At 250 mM and 8 mM D-glucose we measured k obs =244 s -1 and 103 s -1 , respectively. From the simulations we obtained 191 s -1 and 92 s -1 . o Rates in absence of Na + : k obs of sugar-induced PSS current in absence of Na + is independent of sugar concentration within the tested concentration range (5-500 mM). In SSME k obs for 500 mM and 15 mM D-glucose are 95 ± 18 s -1 and 85 ± 17 s -1 , respectively. Accordingly, we defined k on = 100 s -1 for the transition following sugar binding to the empty carrier. We used a k off of 57 s -1 , similar to the k off observed in the presence of Na + . The model correctly simulates that k obs in absence of Na + is almost independent of sugar concentration within the tested concentration range, yielding a k obs of 53 s -1 at 500 mM D-glucose and a k obs of 44 s -1 at 15 mM D-glucose concentration. o Na + induced PSS currents: We do not observe PSS currents upon Na + jumps in absence of sugar in both, SSME experiments (Bazzone et al., 2022a) and model simulations. k obs upon 60 mM Na + jumps in the presence of 250 mM sugar is 139 ± 20 s -1 (Bazzone et al., 2022a) and 167 s -1 in model simulations.

S2.4.2 What the model does not consider
Some kinetic properties of SGLT1 found via SSME are not fully described by the proposed model. Further modifications are required to take the following experimental results into account. This may be fine tuning of rate constants or additional intermediate states.
1. PSS peak currents are overestimated compared to the steady-state current. We applied a 2 ms time filter representing the time resolution of the solution exchange to accommodate for this.
However, the PSS to SS current ratio obtained from the model is still increased 3-fold compared to the respective ratio found with SSME, which is 4.38 ± 0.27. 2. Simulations in the presence of a 300 mM Na + gradient show that the transport rate increase by a factor of 2.1. Experimentally we observed a factor of 3.5 (inset of Figure 7A). However, the reason for the Na + gradient affecting the steady-state transport rate observed in the model is different: the steady-state current decreases when Na + concentration increases above 120 mM in both, external and internal solutions.
One major flaw of the model is the missing electrogenicity of the empty carrier translocation. Loo and Longpré et al. showed that the voltage induced PSS currents may be explained by different electrogenic steps within the empty carrier translocation (Longpré et al., 2012;Loo et al., 2006). When charge translocation is added for the empty carrier translocation (Figure 8, 8->1), the ratio between PSS and SS currents in the model can be adjusted to match the ratio observed in SSME experiments: the SS current increases according to the charge translocation set for the empty carrier translocation, while sugar-induced PSS currents are unaffected. However, including electrogenic empty carrier translocation will automatically lead to a PSS response upon Na + binding which we clearly lack in SSME experiments (Bazzone et al., 2022a). Therefore, we decided to omit electrogenicity for empty carrier translocation. An alternative solution would involve an additional, electroneutral rate limiting step between electrogenic empty carrier translocation and Na + binding. We did not follow up on this for simplicity reasons.

Figure SI-1: Comparison of K M values for D-glucose during Na + /D-glucose cotransport in influx and efflux modes.
All current traces were recorded at pH 7.4 in the presence of 300 mM Na + in all measurement solutions and upon D-glucose concentration jumps between 0.1 mM and 250 mM. For the efflux assay, vesicles were preloaded with the respective concentration of D-glucose, during sensor preparation. Efflux was triggered by removing D-glucose from the sensor, generating an outward directed D-glucose concentration gradient. The influx assay was performed as described in the main manuscript. A Representative current traces recorded from one 1 mm sensor in influx mode. At higher D-glucose concentrations, the peak current is dominated by the sugar-induced PSS current. B Representative current traces recorded from different 1 mm sensors in efflux mode. The transport current dominates the peak at all concentrations tested. C Efflux and influx currents 50 ms after the sugar concentration jump are fitted using a hyperbolic equation to derive K M values of 2.2 ± 1.6 mM and 1.7 ± 0.1 mM for D-glucose efflux and influx modes, respectively. In efflux mode, all concentration jumps are performed on an individual sensor and averaging occurs without normalization, leading to a somewhat higher standard deviation as shown in the plot. For the influx assay, measurements were performed on three different sensors, using the full concentration sequence each. As for the efflux data, averaging was performed without normalization, leading to higher standard deviation.

Figure SI-2: Fit of co-substrate concentration dependent K M values using ordered binding models.
For this analysis the dataset shown in Figure 5 was used. An ordered binding model was applied as demonstrated previously (Bazzone et al., 2016). The model equations were derived from the given ordered binding models, either assuming that the cation or the sugar binds first. The model equations consider the K D values, e.g. real affinities for the substrates. Instead, we used the co-substrate concentration dependent K M values due to the lack of K D app data for Na + . Hence, the model fit rather represents the effect of the ordered binding on K M , instead of K D . We replaced K D by K M for this analysis since K D and K M were found to be proportional to each other, i.e. increasing K D will also increase K M (Table 1). K M # (X) reflects the K M value for transport of substrate X, when 100% of the transporter population is bound to co-substrate Y. K M approaches K M # (X) when the co-substrate Y is provided at oversaturating concentration. K M~( Y) reflects the K M value for transport of substrate Y, assuming it only binds to the empty carrier. A Model equation assuming that Na + binds before the sugar. Na + concentration dependent K M values for D-glucose used for the fit were taken from Figure 5C. The model equation fits very well (R²> 0.9999) and yield a K M # for D-glucose of 0.68 ± 0.07 mM and a Hill coefficient of n = 2.02 ± 0.02, in agreement with two Na + ions binding to SGLT1 before D-glucose binds. B Model equation assuming that D-glucose binds before Na + . The D-glucose concentration dependent K M values for Na + used for the fit were taken from Figure 5D. The fit is not as good as for the model in A (R² = 0.98) but shows that D-glucose is able to bind before Na + , at least to some extent and even under physiological concentrations. The fit yields a K M # for Na + of 48 ± 12 mM and a Hill coefficient of n = 0.52 ± 0.13.

Figure SI-3: Comparison of current traces, reconstructed transporter currents and K M plots for different sugar species during Na + /D-glucose cotransport.
The data reflects typical influx experiments in the presence of 300 mM Na + as shown in Figure 1 for D-glucose. All concentrations of the same sugar were applied to the same sensor. Each concentration dependence was repeated at least 5 times using different sensors. A Representative current traces recorded on a 1 mm sensor. B Via circuit analysis, transporter currents were reconstructed from the current traces shown in A (Tadini-Buoninsegni and Fendler, 2015). Steady-state currents are revealed from the transport phase of the original current, while the PSS current is essentially unaltered. C-E To obtain K M values, datasets for the same sugar were normalized to the current or charge obtained for the highest sugar concentration, then averaged. Normalized, averaged datasets were then multiplied by the average current amplitude obtained for the highest concentration of the respective sugar within this set of experiments in order to include information about relative I max for the different substrates (section S1.3). Fits using hyperbolic equations to derive K M and relative I max values for each sugar species were performed by using three different types of data from the same datasets. For the fits the following data was used.   Figure 4B for D-glucose. D To decrease the number of variables, the K D value of the fitting equation was fixed to the one determined using the K D,Na app plot shown in B. The Hill coefficient was fixed to n=1 to reflect the binding of one sugar molecule. For MDG and D-galactose a negative k off resulted from the fit, hence k off was fixed to 0 for these sugars. E Identical fit using the same dataset as in D, but with the K D values as additional variables (not fixed). K D app values obtained from the fits are similar to those obtained from the plots shown in A-C. For D-glucose and MDG a better estimate of k on and k off is achieved compared to D. F Plots of the concentration dependent k obs determined from the transport current phase. The current decay is a consequence of the membrane being charged upon Na + /sugar cotransport.  Figure 2 for D-glucose. All concentrations of the same sugar were applied to the same sensor. Each concentration dependence was repeated at least 5 times using different sensors.
A Representative current traces recorded on a 1 mm sensor. B-C Fits using hyperbolic equations to derive K D,K app and relative I max or Q max values for each sugar species were performed using the peak currents or the charge translocation (current integral), respectively. The data was averaged and normalized as described in Figure SI-3. D Plots of the concentration dependent k obs determined via mono-exponential fit of the current decay as shown in Figure 2B for D-glucose. k obs is independent from sugar concentration in the tested concentration range for all sugars. In the presence of Na + , Li + and at acidic pH Na + -, Li + -and H + -coupled MDG cotransport is measured. In the presence of K + at alkaline pH, only PSS currents are detected. Sensors were prepared in the presence of the respective cations and all MDG concentrations were applied to the same sensor. Data was averaged and normalized across at least 4 different sensors as described in Figure SI- Figure 1 for D-glucose. To determine K M and EC 50 PSS for Na + , 250 mM MDG concentration jumps were performed in the presence of different Na + concentrations as shown in Figure 3 for D-glucose. Data was averaged and normalized across at least 4 different sensors as described in Figure SI- Each current trace is recorded from a different sensor with vesicles equilibrated at different Clconcentrations before sensor preparation. Since the vesicles storage buffer contained 150 mM Cl -, the minimum internal Clconcentration is limited by the dilution ratio with the respective measurement solutions containing 0 -300 mM Cl -. The corrected internal Clconcentration range used was 11.25 mM to 289 mM as indicated. C Currents 50 ms after substrate concentration jump are fitted using a hyperbolic equation to derive a K M of 6.8 ± 2.8 mM for Cl -. Average currents from n=3 different sensors for each internal Clconcentration were used. No normalization was applied, since individual data points were derived from different sensors. D Same as C, but peak currents are fitted to derive an EC 50 PSS of 21 ± 2 mM for Cl -. In order to improve the fit, we fixed n = 1 and I min = 0.  parameters labeled as 'raw data' refer to I max or Q max values obtained from different sets of sensors for each sugar species as found from the raw data provided in Figure SI-3, SI-4 and SI-5. There was no normalization to reflect proper relative currents or charges for different sugars. Parameters labeled as 'norm.' reflect those provided in the main manuscript. Norm. values were recalculated using the raw data values by normalization to the datasets shown in Figure 6. Here, all sugar species were recorded on the same sensor, hence parameters may be compared directly. For comparison of I max and Q max across sugar substrates, only the normalized values are used. Second, K D app and K M values obtained from different analysis methods are shown to provide insights about how parameters compare, when different read-outs are used. To obtain K M values three methods are compared: we used the steady-state current from current reconstruction (K M (I rec )), the current 50 ms after substrate jump (K M (I t=50ms )) and the current integral (K M (Q 0-1.0 )). To determine the sugar affinity for the Na + bound carrier (K D,Na app ), we have used the peak currents from the raw data (K D,Na app (I peak )), the peak currents from current reconstruction (K D,Na app (I rec,peak )), the current integral within the first 30 ms after substrate jump (K D,Na app (Q 0-0.03s )) and a model equation fit based on the k obs values determined via bi-exponential fit of the current decays (K D,Na app (k obs )). To determine the sugar affinity for the empty carrier (K D,K app ), we either used the peak currents (K D,K app (I)) or the current integral (K D,K app (Q)).
Table SI-3: Parameters for the 11-state kinetic model describing Na + /D-glucose cotransport. Rate constants used to simulate the kinetic model shown in Figure 8 are listed. The last column explains the reasoning behind the value. Rate constants for the substrate induced conformational transitions recorded as PSS currents in SSME and K D values for sugar binding were experimentally determined. They were combined with information from the literature as explained. Some rate constants are modified to account for the law of detailed balance (Alberty, 2004), others are modified to match the experimental data.