Extracellular betaine perfusion on Xenopus laevis oocytes heterologously expressing rGAT1 induces inward transport currents

Fully grown Xenopus laevis oocytes have a highly efficient biosynthetic apparatus that can perform all the post-translational modifications needed for correct protein targeting and function. We use the micro-injection technique to gain the heterologous expression of the protein of interest in Xenopus laevis oocytes [25]. BGT-1 is reported to transport betaine and GABA in a concentration-dependent manner, with a higher affinity for GABA than betaine [26, 27]. The perfusion of betaine (1, 3, 10, 30, 50 mM) on the oocytes expressing rat GAT1 (rGAT1) resulted in concentration-dependent inward transport currents, like canine BGT-1 (cBGT-1) (Fig. 1A–C). The representative traces of these betaine-induced currents, in Fig. 1 A, are comparatively shown with the currents induced by GABA 300 µM, which is the saturating concentration for GAT1 in X. laevis expression system [28,29,30].

Fig. 1

Betaine induces concentration- and sodium-dependent inward currents in Xenopus laevis oocytes expressing rGAT1, which can be blocked by SKF89976a. A Representative traces of current recorded for GABA 300 µM and increasing betaine concentration (0.3–10 mM) in non-injected control oocytes, oocytes expressing cBGT-1, and rGAT1 (top to down). B A violin scattered plot shows the concentration-dependent response of betaine (0.1–50 mM) in rGAT1. The current values are shown as mean ± SEM of 21/5 n (number of oocytes)/N (number of batches). C The kinetic analysis of the betaine transport in rGAT1 yielded Imax = −76.16 ± 1.12 nA, K0.5 = 11.57 ± 1.28 mM, (see Table). In the inset, for cBGT-1 the parameters were Imax = −23.01 ± 1.21 nA, K0.5 = 1.69 ± 0.21 mM. All data were fitted using logistic fit model, with current values shown as mean ± SEM of 8/3 n/N. D A representative trace of currents recorded for GABA 300 µM and betaine 10 mM in the presence of ND98, TMA98, and ND98 + SKF89976a 30 µM. E The histogram shows the mean values of the currents recorded as reported in D, the Na+ dependence, and the blocking effect of SKF89976a 30 µM on the inward induced currents by GABA 300 µM and betaine 10 mM. Current values are shown as mean ± SEM of 8/3 n/N. All recordings were performed at holding potential Vh = -60 mV

At the holding potential Vh = −60 mV, the kinetic analysis of these inward transport currents provided a half-maximal transport coefficient K0.5 = 11.57 ± 1.78 mM, the maximal transport current Imax = −76.15 ± 3.46 nA, and the transport efficiency Imax/K0.5 = −6.58 ± 0.80 nA/mM (Table in Fig. 1). As a negative control, the non-injected oocytes were perfused with the same concentrations of betaine. The oocytes expressing rGAT1 and the non-injected control oocytes were also exposed to glycine (from 1 to 50 mM), and in both cases no inward current was observed (Fig. S1A). Additionally, also in Chinese hamster ovary (CHO) cells overexpressing human GAT1 (hGAT1) we recorded a concentration-dependent betaine transport current using the automated whole-cell patch clamp technique (Patchliner™ from Nanion GmbH, Germany) (Fig. S2).

Betaine transport by rGAT1 is sodium-dependent and can be blocked by the inhibitors of GAT1

GAT1 is a sodium-dependent symporter, meaning its transport activity requires a sodium gradient [31]. To examine the sodium dependence of betaine induced currents in oocytes expressing rGAT1, after testing them in sodium chloride buffer solution (ND98) we replaced sodium chloride with trimethylammonium chloride (TMA-Cl). The experiments with this buffer solution (TMA98) resulted in the almost complete loss of the GABA- and betaine-induced inward transport currents (Fig. 1D, E). This confirms that like GABA, the betaine transport by rGAT1 also requires sodium. We also tested the effects of SKF89976a (30 µM), a GAT1 inhibitor [32], on betaine transport in the oocytes expressing rGAT1. The presence of SKF89976a in the perfused solutions caused a strong inhibition of the transport current elicited by GABA 300 µM and betaine 10 mM (Fig. 1D, E). Similar effects were also observed with other GAT1 inhibitors tiagabine and NO-711 (Fig. S1B), these data confirm that the betaine- currents are mediated by rGAT1.

The voltage-dependent transport of betaine by GAT1 shows that it is a slower substrate than GABA

GAT1 is a voltage-dependent transporter, and the voltage-steps experiments on X. laevis oocytes expressing rGAT1 reveal pre-steady state, steady-state, and leak currents [33,34,35]. The voltage jumps in the presence of betaine (1, 3, 10, 30, 50 mM) performed on the oocytes expressing rGAT1 elicited voltage-dependent transport currents in rGAT1, similar to that induced by GABA [36] (Fig. 2B).

Fig. 2

The pre-steady state analysis of betaine transport by rGAT1. The current response for each condition was collected by giving 0.8 s long squared pulse at –20 mV of from the holding potential of −60 mV, starting from −140 mV to + 40 mV. A The representative traces of the voltage-step response for ND98, GABA 300 µM, SKF89976a 30 µM, and different indicated betaine concentrations. The dashed red line indicates the holding current for the oocyte at the holding potential. B The current and voltage (I–V) relationship from −140 mV to + 40 mV. C The total charge dislocation and voltage (Q–V) relationship. D The decay time constant and voltage (τ–V) relationship. E The relationships of unidirectional rate constants outrate (α) and inrate (β, shown as dashed line) with voltage. All the reported values were collected in the presence of ND98 alone and with betaine 0.1, 1, 10, 50 mM. In C–E the voltage levels tested were from −120 mV to + 20 mV. For B–E, all values are shown as mean ± SEM of 3/1 n/N

The pre-steady state (PSS) currents in GAT1 are due to the binding and unbinding of Na+ to the transporter inside the membrane electric field [37] and usually they disappear in the presence of a saturating concentration of substrate [34]. Hence expectedly, at saturating GABA concentration (300 µM), we observed the vanishing of PSS currents as they became too fast to be detected and only the steady state currents were detected (Fig. 2 A, top row). In the presence of SKF89976a (30 µM), the PSS and steady state currents both disappeared (Fig. 2A top row) as this compound blocks the access of Na+ to the GAT1 vestibule limiting the translocation of substrate and ions [38, 39]. The capacitive currents recorded in the presence of the inhibitors are the fast relaxation component due to the membrane capacitance of the oocytes.

Interestingly, even in the presence of increasing betaine concentrations, we observed the persistence of a slow transient component. These relaxations are larger than expected (Fig. 2 A, middle and last row) when compared to that at low GABA concentrations [30]. The PSS currents in the presence and absence of betaine were analysed after subtraction from the currents in the presence of SKF89976a 30 µM [25, 31] The remainders were integrated to obtain the dislocated charge (Q) and fitted with single exponential to determine the relaxation time constant (τ). The Q–V relationship reflects the charge moved into the membrane electric field at each tested transmembrane potential. Whereas the τ-V relationship provides the rate of the charge dislocation in the transporter vestibule. These data were further analysed to calculate outrate α and inrate β that indicate the rate of charge moving to and from the transporter cavity respectively (Fig. 2C–E) [30, 40]. All of the Q–V, τ–V, and rate constant analysis were done for betaine 0.1, 0.3, 1, 3, 10, 30, and 50 mM, in the figures the data for betaine 0.1, 1, 10, and 50 mM are reported (for the remaining concentrations Fig. S3).

The Q–V relationship showed that low concentrations of betaine (0.1–1 mM) induced charge displacements similar to ND98 alone (Fig. 2C). With further increase in extracellular betaine (10 and 50 mM) the charge displacement decreased but less than expected and did not disappear like for saturating GABA [30]. The τ–V relationship provided a similar result, where the decay time constant decreased with the increase in betaine concentration (Fig. 2D). This indicates an accelerated transport rate in the presence of betaine but again differently from what is known for GABA [30]. Further analysis of α and β showed that at low betaine concentrations, the charge entered the transporter cavities faster than it could leave (β > α), similar to the substrate free Na+-buffer solution. Interestingly, β in the presence of betaine 0.1 mM is lower than in the presence of Na+-buffer alone, in particular at voltages lower than −80 mV (Fig. 2E).

Detection of GABA and betaine using LC–MS/MS in X. laevis oocytes expressing rGAT1

While the electrophysiological characterization of the betaine-GAT1 interaction has shown that betaine is a slow substrate of GAT1, we also wanted to detect a direct and voltage-free uptake of betaine by GAT1. In this work, we developed a radiolabelled-free method using a tandem LC–MS/MS based approach to detect the substrates taken up by rGAT1 heterologously expressed in X. laevis oocytes (Fig. 3A). To verify the sensitivity and linearity of the LC–MS/MS response to betaine and GABA, standard solutions were created, and a calibration curve was obtained (Fig. S4). The detected retention time for GABA was 1′ 55″ and for betaine 2′ 30″ (Fig. 3).

Fig. 3

Detection of GABA and betaine transport by rGAT1 using the LC–MS/MS protocol on X. laevis oocytes. A A cartoon of the protocol developed to extract the cytosol contents of the oocytes and detect the presence of the substrate of interest. B A representative trace showing the presence of GABA in rGAT1 expressing oocytes incubated in GABA 1 mM. The histogram on the right shows a qualitative measurement of a concentration-dependent GABA uptake by rGAT1 expressing oocytes incubated in different GABA concentrations (1–300 µM). C A representative trace showing the presence of betaine in rGAT1 expressing oocytes incubated in betaine 30 mM. The histogram on the right shows the qualitative measurement of a concentration-dependent betaine uptake by rGAT1 expressing oocytes incubated in different betaine concentrations (0.1–30 mM). All values are shown as arbitrary units per minute per oocyte ± SEM of 15/2 n/N. The table at the bottom shows the collision energy required to obtain a unique production ion and the retention time (in the 15-min-long protocol) for the detection of the peak correlated to GABA and betaine

The oocytes heterologously expressing rGAT1 were incubated in different concentrations of GABA (1, 3, 10, 30, 100, 300 µM) and betaine (0.1, 0.3, 1, 3, 10, 30 mM) for 25 min in groups of five oocytes per sample. For both GABA and betaine, a concentration-dependent uptake by rGAT1 was observed (Fig. 3B, C histogram on right). As a negative control, the non-injected oocytes were also incubated in the same concentrations of GABA and betaine, and their analysis showed an absence of uptake for both substrates.

Both GABA and betaine induce efflux of [3H]-GABA in rGAT1 expressing HEK293 cells

The human embryonic kidney cell line (HEK293) is widely utilized as the platform to express membrane proteins. In the absence of radiolabelled betaine, it was difficult to detect its inward transport in HEK293-rGAT1 cells using uptake assays. In our previous work [41], we demonstrated that GAT1 has a bi-directional transport characteristic. It can function in inward and outward modes with different affinity for GABA influx and efflux. By utilizing this characteristic of GAT1 we performed the release assay experiment with betaine outside on HEK293-rGAT1 cells preloaded with radiolabelled GABA ([3H]-GABA).

We tested the effects of betaine on HEK293-rGAT1 cells pre-loaded with 0.01 µM [3H]GABA at 37 °C for 20 min. The experiment was initiated by replacing the pre-loading buffer with plain Kerbs Ringer HEPES buffer (KHB) and starting the experiment with the collection of baseline efflux values (Fig. 4). Basal [3H]GABA efflux was 0.12 ± 0.02% min−1. Addition of betaine induced a time-, concentration-, and rGAT1-dependent efflux of [3H]GABA, in the presence and absence of the ionophore monensin (10 µM) (Fig. 4D, E). Monensin (mon) is a sodium ionophore that selectively collapses the Na+ and H+ gradients, reducing the electrochemical driving force for uptake and favouring the transporter efflux by an increase of sodium inside the cell [42]. As a positive control, pre-loaded HEK293-rGAT1 cells were exposed to GABA, and the results matched with the time-, concentration-, and rGAT1-dependent efflux (Fig. 4A, B), similar to our previous report [41]. For kinetic analysis, the drug-induced efflux was calculated as the mean efflux of the fraction where the value started plateauing divided by the length of the fraction (= 2 min). The fitting of the data using non-linear regression by the logistic fitting model provided the parameters as K0.5,GABA = 36.28 ± 11.27 µM (with mon 10 µM K0.5,GABA = 36.48 ± 2.57 µM) and K0.5,betaine = 6.73 ± 2.21 mM (with mon 10 µM K0.5,betaine = 7.71 ± 1.13 mM) (Fig. 4C, F). We also performed a negative control experiment with tiagabine, a GAT1 inhibitor, that resulted in the absence of [3H]-GABA efflux (Fig. 4G).

Fig. 4

Betaine induces efflux of [3H]-GABA in pre-loaded HEK293 cells overexpressing rGAT1. A Time course of the efflux of [3H]-GABA in the presence of increasing GABA concentrations (1–300 µM). B Time course of the efflux of [3H]-GABA in the presence of monensin 10 µM and different GABA concentrations (1–300 µM). C The kinetic analysis of the [3H]-GABA-induced efflux with and without monensin 10 µM. D Time course of the efflux of [3H]-GABA in the presence of different betaine concentrations (1–100 mM). E Time course of the efflux of [3H]-GABA in the presence of monensin 10 µM and different betaine concentrations (1–100 mM). F The kinetic analysis of the [3H]-GABA-induced efflux with and without monensin 10 µM. G Time course of the efflux of [3H]-GABA in the presence of tiagabine 10 µM with and without monensin 10 µM. All GABA and betaine solutions were prepared using KHB as the buffer solution. Data were fitted using logistic fit model and values are shown in the table at the bottom. Data are mean ± SEM from three individual experiments, performed in duplicate

Molecular dynamics and docking experiments show that betaine binds in the same binding pocket of GAT1 as GABA

Having observed that betaine is a slower substrate of GAT1, with lower affinity than GABA (Figs. 1, 2), we were motivated to understand its binding mechanism with GAT1 using molecular dynamics. In the absence of a crystal or cryo-EM structure of hGAT1 in the outward-facing conformation, we selected the Alphafold homology model to analyse the stability of bound GABA and betaine [37, 43, 44]. The bound ions (two Na+ and one Cl−) were added to the hGAT1 model using as a reference the outward-open human SERT crystal structure (PDB ID: 5I71) [45]. The docking studies for betaine to the outward-open hGAT1 provided a successful docking with the best fitness score of 42.24 (Fig. 5A). As a positive control, docking simulations for GABA were also run, which resulted in fitness score of 50.52 (Fig. S5). The carboxyl head of betaine, which overlaps with the carboxyl head of docked GABA, interacts with Na1 (the sodium bound to the NA1 sodium binding site) and formed short-range (d < 3 Å) contacts with the backbone amide protons of L64 and G65, and with the side chain hydroxyl proton of Y140. The tail containing three methyl groups formed two medium-range (3 Å < d < 5 Å) contacts with residues S295 and S396.

Fig. 5

Molecular docking and MD simulation of betaine and GABA in hGAT1 show that betaine stably binds to GAT1 and forms less polar contacts than GABA. A The successful docking of betaine in hGAT1 with zoomed-in view of the binding site. B The overlapping of GABA-bound hGAT1 Alphafold in outward-open (in white) with cryo-Em structure of the hGAT1 in the inward-occluded (PDB: 7Y7W, in red) with zoomed-in view of the GABA binding site, in the presence of water molecule stabilized by T400 shown in yellow dashed line. C The same overlapping for betaine-bound structure shows the tail of betaine forming hydrogen-bond with water molecule that is stabilized by carboxyl head of T400 shown in yellow dashed line. D MD simulation results for hGAT1 in the outward open conformation bound to betaine is with a zoomed-in view as the S1-site. The average occupancy from three simulations is visualized using an isosurface, color-coded according to the legend. E The docking poses (in white) and the respective end structures (in light red) at 50 ns resulting from MD simulations. F The root mean square displacement of each replica smoothened with a running average over 2 ns. G. The root mean square fluctuation of hGAT1 residues by plotting a mean root mean square fluctuations value from the three replicas, emphasizing residues belonging to TM helices with a grey bar. In panel A, the short-range contacts (d < 3 Å) are indicated as red dashed lines and medium-range contacts (3 Å < d < 5 Å) as yellow dashed lines. The representation illustrated includes hGAT1 as ribbons (in A–D: outward-open in white. In C: inward-occluded in red), betaine and GABA shown as sticks, and Na+ (purple) and Cl− (green) as spheres

Interestingly, the recent work of Zhu et al. assigned five water molecules in GABA binding pocket in hGAT1. They showed in the inward occluded state, the amino acid group of GABA is stabilized by the hydroxyl group of T400 and a water molecule [46]. When we overlapped our hGAT1 model with GABA docked in the outward-open state with the inward-occluded cryo-EM structure of hGAT1 with GABA from Zhu et. al. (Fig. 5B), we observed similar stabilization of docked GABA in our model. Moreover, we also performed the same overlapping of the cryo-EM structure by Zhu et. al. with our outward-open hGAT1 with betaine docked (Fig. 5C). This overlapping showed that betaine molecule is not long enough to have a stable polar-contact with T400, but in the presence of a water molecule (stabilized by T400) it can form a stable hydrogen-bond and could allow the conformational change to the inward-open state.

The MD simulations were performed using the three docking poses of betaine with the best fitting scores (Fig. 5D, E). The simulations were run for 50 ns (the simulation parameters are reported in supplementary information as Code S1) and resulted in betaine stably bound to hGAT1, with root mean square displacement lower than 0.2 nm (Fig. 5F). The simulated GAT1 also resulted in a stable conformation, showing root mean square fluctuations greater than 0.2 nm only in residues belonging to the extracellular and intracellular loops (Fig. 5G).

The relationship of GABA and betaine in rGAT1 depends heavily on their extracellular concentrations

The neuroprotective properties of betaine have been reported in many publications, with a positive correlation with the GABAergic pathway. To determine the mechanism of action betaine with GAT1 and the possible impacts on the GABAergic pathway, we explored its relationship with GABA, the primary substrate of GAT1. We investigated the GABA-betaine relationship using TEVC and LC–MS/MS. The competitive assay experiments on TEVC were performed using six GABA concentrations ranging from 1 to 300 µM with eleven betaine concentrations from 0.001 to 50 mM. The GABA concentrations chosen to study the GABA betaine relationship are based on the kinetic parameters GABA transport by rGAT1 and hGAT1 obtained for and reported in the literature [29, 30, 47, 48]. The representative traces of the currents collected in the competitive assay at the holding potential of −60 mV are reported in Fig. 6A. With GABA concentration below 10 µM and betaine below 10 mM, we observed a reduction of the total transport current indicating an inhibition of the GABA transport. With betaine 10 mM and above, this blocking effect disappeared, and a collective larger inward transport current was observed. The mean values of the currents for all 84 conditions are reported as a heat map of GABA-betaine competition (Data available in Table S3). The data showed that the GABA-betaine relationship starkly depends on their individual extracellular concentration (Fig. 6B).

Fig. 6

Betaine has a dual role in rGAT1, a GABA inhibitor at low concentrations and a secondary substrate at higher concentrations. A Representative traces of GABA betaine assay in X. laevis oocytes expressing rGAT1 at holding potential Vh =  − 60 mV, where the oocyte was perfused with different GABA concentrations (1, 3, 10, 30, 100 μM) along with betaine 0.1, 1, and 10 mM. B The heatmap analysis of the GABA betaine competitive assay shows their concentration-dependent relationship, using the combination of GABA (1–300 μM) with betaine (0.001–50 mM). Data are shown as mean ± SEM of 6/2 n/N. C Detection of GABA and betaine in the oocytes incubated in GABA 3, 10, 30 µM with betaine 0.1 mM, using LC–MS/MS protocol. The qualitative analysis of GABA and betaine uptake by the oocytes is represented in this bar plot with the uptake values, as arbitrary units, of each oocyte per minute, data shown with SEM and obtained from n = 3 with five oocytes in each sample. The p values were obtained by ordinary one-way ANOVA method followed by Bonferroni’s multiple comparisons test, with a single pooled variance with statistical significance of p < 0.05. D The kinetic analysis of different betaine concentrations (0.001–50 mM) with GABA 10 μM shows the dual behaviour of betaine in rGAT1, as at the lower concentrations (left) the GABA transport current is reduced with an increase in betaine, and at the higher concentration (right) the total transport current increases. All data were fitted using the logistic fitting model, fitting values shown in the inset, and current values shown as mean ± SEM of 6/2 n/N

Since in our LC–MS/MS protocol the detection time for GABA and betaine uptake by rGAT1 are different and have distinct production-ions, it was possible to use this method to perform GABA and betaine competition in the same oocyte. Based on the electrophysiological findings, the inhibitory effect of betaine on GABA transport was the most at 0.1 mM concentration. Therefore, we incubated the oocytes expressing rGAT1 with GABA 3, 10, and 30 µM with betaine 0.1 mM for 25 min (n = 3, five oocytes per sample). The cytosolic contents detection by LC–MS/MS for the uptake of all three GABA concentrations were compared with and without betaine 0.1 mM (Fig. 6C). The histogram showed that in the presence of betaine 0.1 mM: with GABA 3 µM no GABA was taken up, with GABA 10 µM the GABA uptake was significantly reduced, and with GABA 30 µM the GABA uptake didn’t alter significantly.

Taken together, these data showed that betaine has a dual effect on GABA transport by rGAT1. This duality can be best visualized at GABA 10 µM with different betaine concentrations (Fig. 6D). When GABA 10 µM is co-applied with betaine from 1 to 300 µM, concentration-dependent inhibition of GABA 10 µM transport current was observed. At the same GABA concentration, the perfusion with betaine from 1 to 50 mM yielded a concentration-dependent increase in transport current. This kind of dual effect of betaine on GABA transport vanished when the extracellular GABA concentration was 30, 100, and 300 µM i.e., larger than K0.5,GABA.

Betaine slows down the GAT1 transport cycle inhibiting GABA uptake

The competition experiment of GABA and betaine was also performed with the voltage-steps protocol to study the steady and PSS transport currents. By observing the representative traces in the presence of GABA 10 µM with and without betaine 0.1 mM, the peculiar inhibitory action of betaine is visible. We observed a reduction in the current amplitude (indicated by the red dotted line) and increment in the decay time of the PSS currents in response to the voltage jumps (Fig. 7A). The I–V relationship for GABA 10 µM in the absence and presence of betaine 0.1 mM revealed the apparent blocking effect of betaine (Fig. 7B). Also, the Q–V analysis showed that in the presence of betaine 0.1 mM, the total charge dislocated by GAT1 did not increase significantly than of GABA 10 µM alone (Fig. 7C). While the τ–V relationship for low betaine concentration (0.1 mM) (Fig. 2) showed an overall similar decay time constant as ND98, whereas for low GABA concentration (10 µM) the decay time constant was much faster (Fig. 7D). Interestingly, τ for GABA 10 µM with betaine 0.1 mM, slowed down significantly, especially at the positive transmembrane potentials (Fig. 7D), indicating slowing down of the transport cycle. Similarly, both decay constants, α and β for GABA 10 µM in the presence of betaine 0.1 mM, showed an overall reduction, significantly in α at the positive transmembrane potentials (Fig. 7E). Altogether, it is evident that in the presence of betaine the transport rate of GAT1 for GABA decreased.

Fig. 7

Betaine inhibits the GABA uptake by slowing down the transport cycle of rGAT1. A The representative traces of the voltage-step response, from −140 mV to + 40 mV of the oocyte expressing rGAT1, at holding potential Vh = −60 mV, with non-saturating GABA 10 μM (left) and GABA 10 μM + betaine 0.1 mM (right), the dashed red line indicates the holding current for the oocyte at the holding potential. B The current and voltage relationship of GABA 10 μM alone and with betaine 0.1 mM, from −120 mV to + 20 mV, shows the transport current reduction at all voltages. C The total charge dislocation and voltage relationship of GABA 10 μM alone and with betaine 0.1 mM, from −120 mV to + 20 mV shows more charge dislocation happening in the presence of betaine. D The total decay time constant and voltage relationship of GABA 10 μM alone and with betaine 0.1 mM, from −120 mV to  + 20 mV shows slowing down of the transport cycle in the presence of betaine. E The relationships of unidirectional rate constants outrate (α) and inrate (β, shown as dashed line) with voltage for GABA 10 μM alone and with betaine 0.1 mM, from −120 mV to + 20 mV. β in the presence of betaine does not decrease, but α decreases significantly at positive membrane potentials. All current responses were collected by giving 0.8 s long squared pulse at −20 mV of voltage jump. All current values shown as mean ± SEM of 3/1 n/N. The p values were obtained by the two-tailed p test with statistical significance of p < 0.05