Dynamic role of GlyT1 as glycine sink or source: pharmacological implications for the gain control of NMDA receptors

Glycine transporter 1 (GlyT1) mediates termination of inhibitory glycinergic receptors signaling in the spinal cord and brainstem, and is also diffusely present in the forebrain. Here, it regulates the ambient glycine concentration influencing the ‘glycine’-site occupancy of N -methyl-d-aspartate (NMDARs). GlyT1 is a reversible transporter with a substantial, but not excessive, sodium-motive force for uphill transport. This study examines its potential role as a glycine source, either by reversed-uptake or by heteroexchange. I explored how glycine accumulation triggers its release, facilitating the activation of NMDARs by glutamate applied alone. Indeed, glutamate evokes no current in “naive” oocytes coexpressing GluN1/GluN2A and GlyT1, a previously characterized cellular model, but now using GlyT1 as the only potential source of coagonist for NMDAR activation. After glycine uptake, however, glutamate evokes large currents, blocked by ALX-5407 and potentiated by sarcosine, a specific inhibitor and substrate of GlyT1, respectively. These results suggest higher occupancy of the co-agonist site when GlyT1 functions as a glycine source either by reversed-uptake or by heteroexchange. A difference between these two glycine-release mechanisms occurs at hyperpolarized potentials, which induce an apparent voltage-dependent block of NMDAR currents, whereas heteroexchange preserves NMDAR activation at these potentials. Together, these results confirm GlyT1-mediated efflux as a positive regulator of NMDAR co-agonist site occupancy, and demonstrate sarcosine heteroexchange effectiveness in enhancing coagonist site occupancy. Depending on its actual mode of transport, GlyT1-inhibitors and sarcosine may have distinct effects on ambient glycine and NMDAR facilitation, and be a source of variation in reversing NMDAR hypofunction in schizophrenia. Graphical Abstract


INTRODUCTION
Glycine acts as a high-affinity co-agonist for NMDA receptors at excitatory synapses (Johnson and Ascher, 1987).Its binding to the GluN1 subunit enables their activation by glutamate in a heterologous expression system (Kleckner and Dingledine (1988), see recent reviews in Hansen et al. (2021); Mony and Paoletti (2023)).
This seminal discovery by Johnson and Ascher revealed that glycine is spontaneously released by primary cultures of cortical and diencephalic neurons (Johnson and Ascher, 1987), suggesting that a glycine transporter, probably the glial isoform GlyT1, operates in reversed uptake mode in the absence of glycine (Attwell et al., 1993;Shibasaki et al., 2017).
In agreement with a 2:1:1 Na + :Cl − :glycine stoichiometry, glycine uptake is electrogenic, carrying 1 elementary charge per glycine; therefore the time integral of the transport current is directly proportional to glycine accumulation (Roux and Supplisson, 2000).

Diffusion-limited unstirred layer in Xenopus oocyte
Xenopus oocytes are large cells, with 1.0 to 1.2 mm diameter, mechanically protected by a fibrous vitelline membrane (Wolf et al., 1976).Their plasma membrane forms numerous invaginations and microvilli that expand their surface by ∼7-fold, creating an unstirred layer that limlits the exchange time even in high-flow chambers (Barry and Diamond, 1984;Costa et al., 1994;Supplisson and Bergman, 1997).The reduction of passive diffusion in this extracellular unconvected fluid compartment (Occhipinti et al., 2014), is advantagenous for studying how transporters deplete or accumulate their substrate in a small, not well stirred, juxtamembrane environment.Ultimately, creating a stopped-flow condition that limits diffusion of the bath solution to the membrane enhances the juxtamembrane depletion of the substrate resulting from its uptake (Supplisson and Bergman, 1997;Zuo and Fang, 2005;Sun et al., 2014;Sheipouri et al., 2020).

2
J o u r n a l P r e -p r o o f Journal Pre-proof

Electrophysiology
Three to eight days post-injection, two-electrode voltageclamp (TECV) recordings were performed using an OC-725D amplifier (Warner Instruments) with oocytes coexpressing GlyT1 and NMDAR.Before experiments, oocytes were preincubated in 30 µM BAPTA-AM for 20 to 30 min to help buffer intracellular Ca 2+ during long recordings.Glass electrodes filled with 3 M KCl solution had a typical resistance of 1 MΩ.Oocytes were held at −40 mV, and perfused continuously with a solution containing (in mM) 100 NaCl, 1.8 CaCl 2 , 1 MgCl 2 , 5 Hepes, pH 7.2 (with KOH) at 25 C. Glutamate was applied in Ca 2+ -and Mg 2+free Ringer solution (replaced by 0.3 mM BaCl 2 ).As low divalent increases the leakage current, Ba 2+ Ringer was applied only 10 s before and maintained 15 s after each glutamate application.Glycine uptake was performed in normal CaCl 2 and MgCl 2 Ringer containing 10 µM d-APV.Currents were typically filtered at 100 Hz and acquired at 1 kHz using a Digidata 1440A and the pCLAMP 10 software suite (Molecular Devices).Weighted time constants (τ w = A1 τ1+A2 τ2 A1+A2 ) were calculated from the amplitude and time constant of the exponential decay time course during washout.Glycine accumulation was derived from the time-integral of the uptake current as: IGly dt z F , with z represents GlyT1 charge coupling (1.01 e/gly, Roux and Supplisson (2000); Vandenberg et al. (2007)), and F is the Faraday's constant (96 485 C mol −1 ).

GlyT1 functions as an ambient glycine sink controlling NMDAR activation
In GlyT1 − oocytes, co-application of glutamate and glycine evoked NMDAR currents with amplitude comparable to or slightly greater than those recorded with dserine as a co-agonist (2 µM each, Figure 1A).In contrast, GlyT1 + -oocytes showed a marked reduction in ambient glycine levels, as the current amplitude decreased by 54 % compared to the response to d-serine (from 112 ± 9 % (n = 7) to 51.6 ± 6.3 % (n = 9), Figures 1B, 1C).This effect is even greater when the membrane is stepped at a more negative potential, which increases the driving force for both, NMDAR currents and glycine uptake.Increased uptake results in further glycine depletion, leading to an immediate increase in current amplitude that is followed by relaxation to a steady state, but surprisingly with little or no change in amplitude (Figure 1B).One minute application of ALX-5407 (5 µM) inhibited GlyT1 and restored the glycine current to full amplitude (120 ± 12 %, n = 4), higher than that induced by d-serine (Figures 1B,1C).In particular, GlyT1 appears to accelerate the rate of glycine clearance during its washout, as indicated by a reduced weighted decay time constant (τ w , inset Figures 1B).In GlyT1 + oocytes, the mean τ w was 487 ± 58 ms (n = 8), which is 44 % lower than the τ w calculated for oocytes lacking transporter binding sites for buffering and capturing glycine (1105 ± 111 ms, n = 22, P<.001 Wilcoxon rank sum test; this second group includes values from GlyT1 −oocytes, application of d-serine, and inhibition by ALX-5407, as their means were not statistically different, P = 0.78).

GlyT1 functions as a glycine source facilitating NMDAR activation Glycine supply by GlyT1-mediated reversed-uptake
Stopped-flow experiments were conducted to provide initial evidence for NMDAR activation facilitated by GlyT1reversal.In GlyT1 + oocytes with low transporter expression, stopped-flow condition accentuates the reduction of [Gly] jm , thereby decreasing co-agonist site occupancy and NMDAR current, whereas stopped-flow has no effect when d-serine serves as co-agonist (Figure 2A).
Figure 2B shows the current trace for a low expressing GlyT1 + oocyte after glycine uptake followed by an extended washout period (I Gly =−33 nA for 200 µM glycine in normal Mg 2+ -and Ca 2+ -Ringer with 10 µM d-APV; the time integral of the glycine-evoked current was 13.9 µC for 3 min application).Application of glutamate alone evoked a modest NMDAR current (−17.7 nA), but its amplitude increases 28-fold (to −496 nA) in the stopped-flow condition, suggesting that a small glycine efflux by GlyT1 reversed-uptake generates a larger accumulation of ambient glycine in the stopped-flow condition, enhancing coagonist site occupancy (Figure 2B).
The use of high-expressing GlyT1 + oocytes increases glycine efflux and achieves steady-state potentiation of the NMDA current during continuous flow with a glycine-free solution (Figure 3).For each oocyte tested, applications of glutamate (2 µM), alone or with sarcosine (50, 100, 200 µM) were repeated for three different and consecutive conditions defined as: 1) naive, as initially the oocytes have a low glycine content because they are maintained for several days in single wells containing 1 ml of Barth solution supplemented with gentamycin and d-APV; 2) after glycine uptake (application of 200 µM glycine for 3 to 15 minutes); 3) after ALX-5407 (one minute application of 5 µM ALX-5407 induces an "irreversible" inhibition of GlyT1).Current amplitudes were normalized to the re- In naive GlyT1 + oocytes, glutamate alone failed to evoke a current, confirming that GlyT1 is not a glycine source per se under basal conditions (Figures 3Aa, 3B,  3C).However, after uptake, GlyT1 reversal provides a continuous supply of glycine that modulates the co-agonist site occupancy, allowing glutamate-only signaling, with current amplitudes comparable to those evoked by 2 µM d-serine (Figures 3Ae,3B).Greater glycine accumulations were achieved in high-expressing oocytes, as shown by a mean time integral of the uptake current of 80.4±5.7 µC (n = 7), corresponding to 833 ± 58 pmol of glycine (mean uptake current: −186 ± 39 nA (n = 7, from −322 to −79 nA at −40 mV); glycine application was 200 µM for 8.1 ± 1.5 min at an average Vm of −80 mV).
A hyperpolarizing voltage step during glutamate application increased the driving force for NMDAR current while decreasing glycine efflux, resulting in a higher initial current that relaxed to a lower steady-state amplitude (Figures 3Ae, 3B, and 3D), demonstrating that GlyT1 indirectly controls NMDAR activation.Inhibiting GlyT1 with ALX-5407 (5 µM) blocked the only source of extracellular glycine and terminated the facilitation of NMDAR activation by glutamate alone (Figures 3Ai, 3B).Together, these results demonstrate that glycine efflux, via GlyT1 reversed-uptake in glycine-loaded oocytes, effectively facilitates glutamate gating of low-affinity GluN1/GluN2A receptors (Figures 3Ae, 3B, and 3C).

Glycine supply by sarcosine heteroexchange facilitates NM-DAR activation
Sarcosine is a weak NMDAR co-agonist (McBain et al. (1989), but see Zhang et al. (2009)).In GlyT1 − oocytes, co-application of 2 µM glutamate and 200 µM sarcosine induced small currents, with an amplitude 41.5 ± 9.1 % (n = 7) of the response with 2 µM d-serine as a co-agonist.In naive GlyT1 + oocytes, however, the glutamate + sarcosine current represents 84.2 ± 7.4 % (n = 8, P =0.0033, Welch Two Sample t-test) of the glutamate + d-serine response.This potentiation suggests that sarcosine may already facilitate glycine supply by heteroexchange, whereas glycine efflux by reverse uptake was not detected in this condition (see above).It should be noted, however, that part of the current induced by 200 µM sarcosine is due to its electrogenic uptake as a GlyT1 substrate, although the precise contribution of this transport current was not quantified.Indeed, after glycine uptake, the potentiation of the NMDAR current by sarcosine far exceeds this initial difference, indicating that GlyT1-mediated heteroexchange greatly increases glycine efflux and NMDAR co-agonist site occupancy compared to the reverse uptake condition.All traces in Figure 3A are plotted on the same scale, resulting in the clipping of the three sarcosine traces after uptake.These three traces are shown below with a different current scale for clarity.
The potentiation of the NMDAR current by 50 µM sarcosine (896 %, from 0.229 ± 0.032 (n = 9) to 2.05 ± 0.23 (n = 8), P<0.001, Welch Two Sample t-test) appears to saturate at higher concentrations (Figure 3C), reflecting either saturation of glycine efflux and/or high occupancy of the co-agonist site.As expected, GlyT1 inhibition by ALX-5407 reduces the glutamate-gated current to levels comparable to those recorded in GlyT1 − oocytes (Figure 3C).In addition, membrane hyperpolarization with a voltage step does not affect heteroexchange as the glutamate current amplitude increases, in contrast to the reduction observed with reverse uptake, a contrast highlighted in the summary Figure 3D.

GlyT1-block of NMDAR-facilitation by reversed uptake at negative potentials
To examine these differences in the voltage dependence of glycine supply by GlyT1 reversed-uptake or heteroexchange, current-voltage (I-V) relationships were constructed for different conditions as shown in Figure 4A.
The I-V curves, generated in response to 2 µM glutamate either applied alone (glycine provided by reverse uptake, left panel) or coapplied with 50 µM sarcosine (glycine provided by heteroexchange, right panel), were constructed using slow voltage ramps under the three conditions previously defined for the GlyT1 + oocyte (naive, after glycine uptake, and after ALX-5407 inhibition).In particular, the I-V curve for glycine supply via reversed uptake (shown as the blue curve in the left panel of Figure 4A) exhibits a pronounced voltage-dependent block with a negative slope, in apparent similarity with the Mg 2+ -block of NMDAR (Nowak et al., 1984).However, these recordings were conducted in a 0.3 mM Ba 2+ Ringer solution, without Mg 2+ and Ca 2+ , suggesting that this voltage-dependent inhibition must be indirect, probably caused by a decrease in co-agonist site occupancy, indicating reduced glycine release at negative potentials.Indeed, sodium ions are transported against their electrochemical gradients during a complete reverse uptake cycle, as shown in the schema of Figure 4B (left panel), thus limiting the rate of glycine efflux at negative voltages.In contrast, the I-V curve for sarcosine heteroexchange after glycine uptake (blue curve in the right panel of Figure 4A) appears nearly linear, with only small deviation s at very negative potentials as NMDAR becomes sensitive to traces of Mg 2+ (Kuner and Schoepfer, 1996;Retchless et al., 2012).This linear I-V supports that the sarcosine/glycine heteroexchange is electroneutral, as it does not involve a a net transfer of the cosubstrate ions (Figure 4B).These I-Vs are consistent with the observed difference in current amplitude during hyperpolarized voltage steps, as described in Figure 3D.In naive oocytes, the flat I-Vs (orange curves) confirm negligible glycine efflux via reverse uptake, except for Vm>+20 mV.Furthermore, the presence of an NMDAR outward current in sarcosine supports that heteroexchange with glycine occurs even at low intracellular glycine levels in naive oocytes.Finally, the absence of glutamate-evoked 4 J o u r n a l P r e -p r o o f Journal Pre-proof current after incubation with 5 µM ALX-5407 confirms the lack of glycine release by endogenous amino acid transporters of Xenopus oocytes.

The decay time-course of NMDARs facilitation by GlyT1 reversed-uptake depends on their sensitivity to glycine
After several minutes of uptake and subsequent glycine washout, a steady outward current is recorded in GlyT1 + oocytes, indicating an electrogenic glycine efflux by GlyT1mediated reverse uptake (see Figure 4B).
To examine the decay time course of NMDAR facilitation, glutamate-evoked currents were recorded at different time points before and after glycine uptake in the GlyT1 + oocyte (Figure 5A).While facilitation of NMDAR activation greatly enhances the signal generated by glycine efflux, it also distorts the decay kinetics due to the narrow range of sensitivity to glycine.In the experiment shown in Figure 5A, a decay time t 1/2 of 9.8 min was measured, indicating that tonic glycine supply allows glutamate-gating of NMDAR for a long period post-uptake.This slow decay time course is even more pronounced in GlyT1 + oocytes that co-express high affinity GluN1/GluN2B receptors, as shown previously (Figure 5B is adapted from Supplementary Figure 16 in Guellec et al. (2022)).Remarkably, the amplitude of the glutamate-evoked current remained at 6.5 µA, i.e. 80% of the initial value, one hour after glycine washout.

GlyT1 reversed-uptake as a glycine source for NMDAR co-agonist site
This study demonstrates, in a coexpression model, that GlyT1 can function as a steady source of glycine supply to modulate NMDAR co-agonist site occupancy.
It is posited that GlyT1 can deplete or accumulate glycine within a small juxtamembrane compartment ([Gly] jm ) that is sensed by the NMDAR co-agonist site (Supplisson and Bergman, 1997).Under constant perfusion, [Gly] jm dynamics depend on two fluxes: the GlyT1-mediated net flux across the membrane (J Gly ) and passive diffusion (J dif ) through an unstirred layer between the bath and the juxtamembrane compartment.
J Gly is directly proportional to the transport current (I Gly ): where z and F have already been defined and S is the spherical surface of the oocyte (0.038 cm 2 for a diameter of 1.1 mm).
According to Fick's first law of diffusion, the glycine flux (J dif ), between the bath compartment ([Gly] b ) and the juxtamembrane compartment is given by : where D is the effective diffusion coefficient for glycine (5 × 10 −6 cm 2 /s), and h is the thickness of the unstirred layer.When oocytes are perfused in a glycine-free solution ([Gly] b = 0), the diffusion flux simplifies to: A steady state condition is reached when J dif = −J Gly , for which [Gly] jm detected by the NMDAR co-agonist site corresponds to: This equation predicts stationary [Gly] jm between 1 and 1.9 µM for an average outward current of I Gly = +17 nA, assuming a range of 11 to 20 µm for h (Costa et al., 1994;Supplisson and Bergman, 1997).These values confirm the ability of GlyT1 to modulate ambient glycine concentrations within a range suitable for GluN1/GluN2A receptors (Figure 3), comparable to the response evoked by 2 µM d-serine (or glycine) as a co-agonist.
Although it may seem counterintuitive that GlyT1, a tightly coupled cotransporter, would exhibit this futile cycle of exchange process, in which the influx of one substrate facilitates the efflux of another, evidence under voltageclamp conditions confirms this trans-stimulation.Specifically, external glycine triggers an inward current in GlyT1expressing oocytes -indicating a net glycine influxwhile simultaneously increasing the unidirectional efflux of 14 C-glycine, a trans-stimulation not observed in GlyT2expressing oocytes (Guellec et al., 2022).
In the absence of external glycine, the ion-motive force provided to GlyT1 by the coupling of 2 Na + and 1 Cl − remains intact and directed inward.Reversed uptake however, is a complete backward transport cycle requiring dissociation of all three ions from the outward-facing conformation and their reassociation in the inward-facing conformation (Figure 4B).Both steps are unfavorable at the 5

J o u r n a l P r e -p r o o f
Journal Pre-proof holding potential of −40 mV due to the inwardly-directed electrochemical gradients for Na + .Consequently, more hyperpolarized potentials which reduce glycine efflux exponentially (Roux and Supplisson, 2000), indirectly cause voltage block of NMDAR current due to reduced glycine occupancy of the co-agonist site.
In contrast, sarcosine-driven heteroexchange is electroneutral because it bypass the requirement to form an empty transporter in the Inward Occluded conformation (Figure 4B).Therefore the heteroexchange transport cycle is not blocked at negative voltage and benefits from two favorable substrate gradients: inward for sarcosine and outward for glycine.
Since the unidirectional glycine efflux triggers by sarcosine heteroexchange is not proportional to the transport current, a kinetic model of the GlyT1 transport cycle is needed to estimate glycine release by this mechanism, although, it may be difficult to infer transporter kinetics from an indirect readout of glycine efflux as detected here with NMDAR activation, due to the narrow sensitivity range of the coagonist site.
Sequential and random order of binding models have been proposed for the GlyT1 alternative access cycle, but exchange remains largely unexplored (Aubrey et al., 2005;Erdem et al., 2019).Results from these models differ as they predict either a higher rate of reverse translocation (Aubrey et al., 2005), or alternatively, high allosteric cooperativity between cosubstratres sites, which is specific for GlyT1, facilitating the reverse mode and glycine release in a symmetric model (Erdem et al., 2019).
In contrast, the heteroexchange of amphetamines with dopamine or serotonine has been studied in detail for the monoamines transporters DAT and SERT, with different mechanisms, models, and transporter structures (Jones et al., 1999;Schicker et al., 2012;Wang et al., 2015;Sitte and Freissmuth, 2015;Hasenhuetl et al., 2018;Navratna and Gouaux, 2019).Fast reverse mode and exchange have also been characterized for excitatory amino acid transporters (EAATs) -albeit coupled to a symport of 3 Na + and 1 H + , and antiport of 1 K + , and with a different transport mechanism than GlyT1 -, and structures of several intermediate states of the EAAT symport/antiport cycle are now documented (Kavanaugh et al., 1996;Zhang et al., 2007;Canul-Tec et al., 2022;Qiu and Boudker, 2023).

Post-uptake, glycine accumulation triggers a steady overflow and NMDAR facilitation
To turn on glutamate signaling in GlyT1 + oocytes, glycine must first accumulate significantly by uptake.This intracellular accumulation leads to a tonic overflow of glycine by GlyT1 reversed-uptake after external clearance, increasing occupancy of the coagonist site and facilitating glutamategated activation of NMDAR.The mean time integral of the uptake current (80 µC) predicts glycine accumulation of 833 pmol, a value that exceeds by 9 to 14 times the amount of glycine measured in freshly dissociated oocytes (ranging from 59 to 92 pmol (Taylor and Smith, 1987;Meier et al., 2002)).Naive oocytes that have been deprived of external amino acids for days, may have even lower basal glycine levels.Assuming an oocyte (stage 5-6) with a water volume of 0.4 ml (Taylor and Smith, 1987), the estimated cytosolic glycine concentration would increase from 147 to 230 µM for naive oocytes to ∼ 2.1 to 2.3 mM after uptake.These concentration ranges are in good agreement with the EC 50 (4.3mM) estimated for the GlyT1 outward current in mammalian CHO cells (Aubrey et al., 2005), suggesting an efficient but non-saturating glycine supply after uptake.In contrast, negligible glycine efflux is expected in naive (untreated) oocytes, consitent with the absence of response to glutamate.Interestingly, the mean glycine concentration after uptake, corresponds to the ∼ 2 mM estimate for the concentration of glycine in astrocytes proposed by Attwell et al. (1993), based on Berger et al. (1977).

Sarcosine as a glycine releasing agent
One motivation for this study was to clarify the mode of action of sarcosine, often described as a GlyT1 inhibitor or glycine reuptake inhibitor, and used as a tool to effectively increase ambient glycine levels, such as in the amygdala (Bossi et al., 2022), while NFPS, a "true" GlyT1 inhibitor, produced a slow accumulation in the same brain structure (Li et al., 2013).However, the results obtained with this coexpression model show that in the absence of extracellular glycine, thus ruling out inhibition of uptake, sarcosine is indeed a potent glycine-releasing agent by GlyT1mediated heteroexchange when intracellular glycine levels are high.
It can be assumed that this limiting condition is met by astrocytes, which accumulate glycine and whose endfeet surround capillary endothelial cells, close to the largest and ultimate source of glycine.Therefore, sarcosine heteroexchange, which is fully functional at the negative resting potential of astrocytes, could locally scale up ambient glycine, potentially to higher levels than "true" GlyT1 inhibitors, which are dependent on glycine diffusion from an extracellular source, or leakage by Asc-1, a sodiumindependent amino acid transporter expressed in glia (Ehmsen et al., 2016).
The cytoplasmic accumulation of sarcosine by uptake and heteroexchange should lead to a decrease in glycine release over time.However, sarcosine is both a product and a precursor of glycine, as both amino acids can be interconverted by one-carbon metabolic enzymes (glycine-N-methyltransferase for sarcosine synthesis and sarcosine dehydrogenase for its conversion to glycine (Ducker and Rabinowitz, 2017;Pérez-Sala and Pajares, 2023)).Therefore, a sustained and steady release of glycine by heteroexchange is possible provided that sarcosine is converted to glycine by sarcosine dehydrogenase, a mitochondrial enzyme expressed in astrocytes (Pérez-Sala and Pajares, 2023).
These glycine-releasing properties, combined with competitive inhibition of glycine uptake, may contribute to the clinical effect of sarcosine used as an adjuvant therapy to 6 J o u r n a l P r e -p r o o f Journal Pre-proof improve negative and positive symptoms in schizophrenic patients (Tsai et al., 2004;Lane et al., 2005;Curtis, 2019).In addition, it should be noted that altered one-carbon metabolism has been associated with shizophrenia (Smythies et al., 1997;Krebs et al., 2009).

ACKNOWLEDGEMENTS
I would like to thank Jon Johnson and Pierre Paoletti for inviting me to contribute to this special issue in honor of Philippe Ascher.Philippe was an exceptional mentor and remained so long after I joined his laboratory.He was always excited to learn about new results, even on transporters, and discuss their potential implications.His help in editing a manuscript was invaluable.The oocyte coexpression model was not his way of elucidating the role of glycine as a coagonist, but he encouraged me and was surprised by the dynamic interaction between GlyT1 and NMDAR that this simple model revealed.I thank Annick Ayon for her help with molecular biology and Stéphane Dieudonné for generous discussions and support.This research did not receive any specific grant from funding agencies in the public, commercial, or notfor-profit sectors.I thank the National Institute of Health and Medical Research (INSERM) and the National Center for Scientific Research (CNRS) for their continuous support.

Declaration of generative AI and AI-assisted technologies in the writing process
During the preparation of this work, carried out without the help of Philippe Ascher, I used ChatGPT and DeepL to check grammar, spelling and English.After using these tools, I reviewed and edited the content as needed and takes full responsibility for the content of the publication.GlyT1 is represented in four conformations using the Rocking model (Drew and Boudker, 2015;Fan et al., 2021) as a template: Outward Facing (OFC), Inward Facing (IFC), Outward Occluded (OOC) and Inward Occluded (IOC).Each step is reversible, but not mentioned in the schema for simplicity.(A-B) Peak amplitude of the current evoked by 2 µM glutamate before (a) and after (c-e) glycine uptake in GlyT1 + oocytes coexpressing either (A) GluN1/GluN2A or (B) GluN1/GluN2B.The current traces below correspond to different time points before and after glycine uptake, as indicated by a letter.The GluN1/GluN2B panel is adapted from the Supplementary Figure 16 in Guellec et al. (2022).The uptake currents are plotted with the same current and time scales in both panels, so that the surface integrals are comparable.

12
J o u r n a l P r e -p r o o f Journal Pre-proof

3
J o u r n a l P r e -p r o o f Journal Pre-proof sponse evoked by glutamate and d-serine in naive oocytes (2 µM each).

Figure 1 :Figure 2 :
Figure1: GlyT1 decreases ambient glycine sensed by the NMDAR co-agonist site in oocytes coexpressing GluN1/GluN2A subunits (A) In control (GlyT1 − ) GluN1/GluN2A expressing oocytes, currents evoked by glutamate (Glu) in combination with d-serine ( d-Ser) or glycine (Gly) show similar amplitudes that increase with a negative voltage step.All co-agonists were applied at 2 µM.(B) In GlyT1 + oocytes, glycine uptake reduces NMDAR activation compared to d-serine.In addition, a negative voltage step does not increase the current amplitude at steady-state.Inhibition of GlyT1 after one minute application of 5 µM ALX-5407 restored normal glycine response.Inset: GlyT1 inhibition slows the rate of glycine clearance: semi-log plot of the normalized off-response fitted by the sum of two exponentials.Amplitude-weighted time constants are given).(C) Summary of GluN1/GluN2A currents evoked by 2 µM glutamate + glycine in GlyT1 − , and GlyT1 + oocytes in control and after one minute incubation with 5 µM ALX-5407.Current amplitudes were normalized to the absolute amplitude of the current evoked with 2 µM d-serine.Error bars indicate SEM; P values calculated for the Wilcoxon signed rank test with Bonferroni adjustment.

Figure 3 :Figure 4 :
Figure3: Glycine supply by GlyT1, via reversed-uptake or heteroexchange with sarcosine, facilitates glutamate-gated activation of NMDARs.GlyT1 + -oocytes co-expressing GluN1/GluN2A were selected for high transporter expression, with an uptake current amplitude greater than −50 nA for 200 µM glycine at −40 mV.(A) Array of 15 current traces [a-l] recorded from the same oocyte for applications of 2 µM glutamate alone [a,e,i] or with sarcosine (50 µM [b,f,j], 100 µM[c,g,k]  or 200 µM[d,h,l]).Each application was repeated for 3 different conditions which are organized in columns for: #1) naive oocytes (left column, [a-d]); #2) after glycine uptake (middle column, [e-h]; time integral of the uptake current: 86.5 µC ); #3) after 5 µM ALX-5407 (right column, [i-l]; one minute incubation).The bottom panel shows the unclipped traces for sarcosine after glycine uptake [traces f-h].(B) Summary of the current amplitude change for the three conditions (naive, after Gly uptake, after ALX-5407) for each application (no coagonist, 50 µM Sar, 100 µM Sar, 200 µM Sar).P values from Wilcoxon rank-sum test, with Bonferroni adjustment for multiple comparisons; n = 8-9 oocytes for each concentration.(C) Sarcosine dose-response for application of 2 µM Glu as a function of sarcosine concentration for the three conditions (naive (green), after uptake (purple), after ALX-5407 (red)) for GlyT1 + oocytes The black curve is the same protocol for naive GlyT1 − oocytes.Currents were normalized to the amplitude of the response with 2 µM d-Ser as co-agonist, as indicated by the dotted line.(D) Opposite voltage dependence of NMDA current for hyperpolarized potential for glycine supply by reversed uptake (x=0, orange) or heteroexchange with sarcosine (50, 100 and 200 µM, purple).Two different voltage steps (−40 mV and −60 mV) were applied from different holding potentials, as indicated in the axis legend.

Figure 5 :
Figure 5: After uptake, GlyT1 maintains glycine supply and high occupancy of the NMDAR co-agonist site; the decay time of the potentiation depends on the sensitivity of the GluN2 subunit.(A-B)Peak amplitude of the current evoked by 2 µM glutamate before (a) and after (c-e) glycine uptake in GlyT1 + oocytes coexpressing either (A) GluN1/GluN2A or (B) GluN1/GluN2B.The current traces below correspond to different time points before and after glycine uptake, as indicated by a letter.The GluN1/GluN2B panel is adapted from the Supplementary Figure16inGuellec et al. (2022).The uptake currents are plotted with the same current and time scales in both panels, so that the surface integrals are comparable.