Glutamate is the predominant excitatory neurotransmitter in the brain (Curtis and Johnston, 1974; Fonnum, 1984) and is critical to numerous central nervous system (CNS) functions (Tang et al., 1999; Matsuzaki et al., 2004; Chantranupong and Sabatini, 2018; Boender et al., 2020; Stephenson-Jones et al., 2020). Anomalously elevated glutamate in the synapse results in CNS neurotoxicity; removal of glutamate occurs via glutamate transporters (excitatory amino acid transporters; EAATs), as enzymes that metabolize glutamate in the synaptic cleft are scarce. Glutamate transporters transfer the neurotransmitters over a transmembrane glutamate gradient by harnessing preexisting ion electrochemical gradients (Danbolt, 2001). These secondary active transporters belong to the solute carrier family 1 and are mainly expressed in neurons and glial cells. Among five subtypes of EAATs (EAAT1-5), EAAT2 plays an important role in catalyzing the metabolism of this excitatory transmitter. The EAAT2 subtype is expressed predominately in astroglia and is responsible for nearly 90% of glutamate transport in the forebrain (Tanaka et al., 1997; Vandenberg and Ryan, 2013; Martinez-Lozada et al., 2016). Researchers have reported that genetic mutations cause changes in the properties of important amino acid residues, leading to abnormalities in the structure and function of the transporter (Brocke et al., 2002; Zhang and Qu, 2011; Rong et al., 2015, 2016). Also, dysregulation of EAAT2 has been implicated in some neurodegenerative diseases including but not limited to amyotrophic lateral sclerosis (Takahashi et al., 2015; Jiang et al., 2019; Malik and Willnow, 2019), Alzheimer’s disease (Takahashi et al., 2015; Ugbode et al., 2017; Malik and Willnow, 2019; Sharma et al., 2019), and epilepsy (Takahashi et al., 2015; Guella et al., 2017; Malik and Willnow, 2019). For this reason, functional studies of EAAT2 are critical and may reveal potential genetically based therapeutic targets.

In addition to transporting glutamate, mammalian EAATs cotransport 3 sodium cations and 1 proton, and antiport 1 potassium cation (Zerangue and Kavanaugh, 1996; Levy et al., 1998). Indeed, when glutamate molecules are transported, an electrogenic net current with two positive charges is generated. This current depends on chloride anion conductance (activated by substrate transport and thermodynamically uncoupled) to curb the accumulation of positive charges and maintain continuous transfer (Billups et al., 1996; Ryan and Mindell, 2007).

In 2004, researchers identified and described the crystal structure of Glt_Ph, a prokaryotic homolog of EAATs that originated from archaeal ortholog Pyrococcus horikoshii (Yernool et al., 2004). This homolog shares ∼37% sequence identity with human EAAT2 although is more conservative in the C-terminal (critical for facilitating the transport core to bind substrates and cations; Matin et al., 2020). Notably, Glt_Ph is a homotrimer, and each subunit with an independent permeation pathway has eight transmembrane segments (TM1-8) and two re-entrant hairpin loops (1 and 2; HP1-2; Yernool et al., 2004). The carboxy terminus of Glt_Ph (particularly TM7-8 and HP1-2) is actively involved in substrate transport (Yernool et al., 2004; Matin et al., 2020). Furthermore, the amino-terminal “cylinder” formed by TM1-6 secures these functional segments. Hence, each monomer “owns” an independent translocation pathway (Yernool et al., 2004), which corroborates the functions of its mammalian counterparts: a trimer with three identical and independent monomers (Gendreau et al., 2004; Koch and Larsson, 2005; Leary et al., 2007). By coupling with the ion gradient, the transporter catalyzes the movement of the substrate through the lipid bilayer, thereby facilitating the completion of the substrate transport (Gouaux, 2009). In 2017, researchers published the crystal structure of EAAT1 in both a substrate binding state and an outward facing state. This research highlighted the organization of diverse functional domains in transporters: each protomer consists of a scaffold domain containing TM1, 2, 4, and 5 as well as a conserved transport domain containing TM3, 6, 7, 8, HP1 and HP2 (Canul-Tec et al., 2017).

Substrate binding and important conformational changes are also related to the opening and closing of the gate, namely the non-helical tip of HP2 of the transporter (Garaeva et al., 2019; Wang and Boudker, 2020). Indeed, the binding of substrates and ions outside the cell is related to the opening of the tip of HP2. Movement of the tip of HP2 also allows the substrate and ions to dissociate from the inward-facing state. In the outward-facing state, when the tip of HP2 opens to expose the substrate and ion-binding sites to the extracellular solution, HP1 remains fixed. This also applies to the inward facing state where the sole gate HP2 becomes open and HP1 is locked, accompanied by the movement of the rest of the transport domain (Garaeva et al., 2019; Arkhipova et al., 2020; Wang and Boudker, 2020; Qiu et al., 2021).

The loop between TM3 and TM4 undergoes multiple conformational changes during substrate transport, as confirmed by a protease cleavage experiment in 2010 (Compton et al., 2010). The crystal structure of Glt_Ph makes apparent that this long loop is situated in close proximity to transport domains. Some researchers have even shown that the TM3-4 loop rich in proline undergoes substrate-dependent conformational changes, as confirmed by limited trypsin hydrolysis analysis and fluorescein 5-maleimide accessibility experiments (Figure 1A; Compton et al., 2010). In addition, although the protein maintains its structural integrity, cleavage of the protein backbone within the TM3-4 loop through the engineered factor X protease recognition site leads to an almost complete loss of transport activity (Compton et al., 2010; Mulligan and Mindell, 2013). These findings indicate that the TM3-4 loop plays a vital role in the transportation cycle. Studies have also performed cysteine-scanning mutations on the TM3-4 loop and identified many single amino acid substitutions that cause significant changes in the activity of the transporter (Compton et al., 2010). In the prokaryotic crystal structure, for example, the TM3-4 loop undergoes a substrate-sensitive configuration change during transportation that is an important part of the transportation mechanism (Compton et al., 2010; Mulligan and Mindell, 2013). However, there is only about 12.5% sequence homology in TM3-4 loop between the eukaryotic EAAT2 and the prokaryotic Glt_Ph; the conformation of EAATs TM 3–4 loop and its biological functions have not yet been fully elucidated. Studies on the crystal structure of prokaryotes Glt_Ph indicate that HP1, HP2, TM7, and TM8 may form a substrate-binding pocket during the transportation process (Yernool et al., 2004; Boudker et al., 2007; Reyes et al., 2009). In particular, when the transporter is in an inward-facing conformation, the substrate-binding region (including TM7) moves 18 Å toward the cytoplasm (Qu and Kanner, 2008; Reyes et al., 2009; Zhang et al., 2019), which indicates that TM7 has a conformation-dependent spatial position for transport and actively maintains transport functions (Figure 1B). In the eukaryotic EAAT1 heat-resistant mutant EAAT1_cryst, the equivalent region in alanine-serine-cysteine transporter 2 (ASCT2) replaced TM3-4c of EAAT1 to obtain a stable crystalline EAAT1 construct. To improve crystallinity, two predicted N-glycosylation sites of the transporter, N155T and N204T, were introduced through specific mutations. Therefore, the undefined regions in the crystal structure include the TM3-TM4a (residues 153–173) and TM4b-c (residues 200–208) loops and the N (residues 1–28) and C (residues 490–522) termini. Because of the partial region replacement by ASCT2, the identity of TM3-4 loop between the EAAT1_cryst and the wild-type EAAT1 sequence is 16.1% (Canul-Tec et al., 2017), and the spatial relationship between TM3-4 loop residues in EAAT2 and TM7 is still unknown.

Here, we aimed to reveal the conformational change of brain transporter EAAT2 by way of transport assay on paired cysteine mutant L149C/M414C. After engineering multiple paired cysteine mutants between the TM3-4 loop and TM7, we found that oxidation severely inhibits the transport activity of L149C/M414C (Figure 1C). In oxidative conditions, when pretreated with different mediums, the transport activity of L149C/M414C varies because of the change in the distance between the two residues. Indeed, we confirmed that the residues at positions L149 and M414 in the single mutant have relatively high accessibility. These results indicate that during the transport cycle of EAAT2, distance between two residues on the TM3-4 loop and TM7 physically changed. This finding may provide evidence of the spatial structural change between these two key segments when EAAT2 ferries cargo.

The TM3-TM4 loop has been previously characterized by the crystal structure model, and is connected to both the anchored TM4 and the ambulatory TM3; the loop spans across critical segments for transport including HP2, TM7, and TM8 (Yernool et al., 2004; Canul-Tec et al., 2017). However, the role of this loop in the transport of glutamate and the relative relationship of the loop to these segments remain largely unknown (Compton et al., 2010; Mulligan and Mindell, 2013; Wang and Qu, 2021). Here, based on the crystal structure of three different conformations of Glt_Ph, we designed the double mutant L149C/M414C located on the TM3-TM4 loop and TM7 to probe how this loop participates in the translocation and conformation change relative to TM7. In the crystal structure of substrate binding state and outward facing state of EAAT1, the introduction of substitutions or mutations necessary to complete stable crystallization results in an unresolved TM3-TM4 loop, which may disrupt the crystallization (Canul-Tec et al., 2017). Hence, we merely refer to the model of Glt_Ph for designation.

The transport activity of double mutant L149C/M414C is lower than cys-less EAAT2, although much higher than other double mutants in the absence of CuPh (Figure 2A). We demonstrate that this difference occurs because of the reduced apparent affinity or surface expression of L149C/M414C. Indeed, the kinetic experiment of L149C/M414C shows that the substrate affinity and maximal transport rate decrease following the treatment of CuPh; this decrease may be a result of the formation of disulfide bond (Table 1). However, western blot assays in L149C/M414C show impaired membrane targeting, as the mutants appear to aggregate in cytoplasm. The impaired ability of membrane targeting in double cysteine mutants may be due to its retention in the endoplasmic reticulum in an immature form. Indeed, an experiment published in 2004 reports that some conserved residues between TM7 and TM8 play a vital role in the forward trafficking of EAAT2 (Kalandadze et al., 2004). In contrast, we observed no relationship between our western blot assays and the reduced activity of L149C, likely because the accumulated effect results in reduced activity, despite no significant difference between L149C and CL-EAAT2 (Supplementary Figures 1F,G).

Here, we demonstrate that, compared to the double mutant, co-transfected single mutants are resistant to the inhibition of CuPh (Figure 2D). Hence, intramolecular cross-linking results in inhibition of transport activity. We also assayed the activity of L149C/M414C in the presence of DTT and demonstrate that pre-incubation with CuPh and subsequent treatment with DTT results in comparable activity of L149C/M414C and double mutants that were not exposed to any treatment. This activity was also significantly different from the activity of double mutants that were treated with CuPh alone (Figure 3), suggesting that incubation with CuPh results in the formation of disulfide bonds.

The crystal structure of bacterial homolog Glt_Ph and eukaryotic EAAT1 could aid in the exploration of the relative motion of different segments in glutamate transporters. To determine the relative conformational change between the TM3-TM4 loop and TM7 during the transport cycle, we measured the transport activity of double mutant L149C/M414C in different extracellular mediums. TBOA, a substrate analog with aspartate moiety and a bulky benzyl group, increases the extracellular facing state of the transporter (Qu and Kanner, 2008). After pre-incubation with TBOA, the activity of L149C/M414C was lower than those in NaCl solution, which represents the substrate binding state or “occluded” state (Shlaifer and Kanner, 2007; Reyes et al., 2009; Figure 4). Notably, when the transporter increased its outward ratio, these two residues became closer. Extracellular potassium ions or L-glutamate also reversed the orientation of substrate influx and increased the ratio of cytoplasm facing conformation of glutamate transporters (Shlaifer and Kanner, 2007; Qu and Kanner, 2008; Crisman et al., 2009). We found that pretreatment with potassium ions or L-glutamate protects L149C/M414C from inhibition by CuPh, as activity was restored compared with the activity of L149C/M414C in NaCl solution (Figure 4). This suggests that two residues are likely distanced from each other. We also compared our findings with the published crystal structure of Glt_Ph by measuring the Cα–Cα distance between His114 and Ala328, which are equivalent to Leu149 and Met414 in EAAT2 in models representative of different conformations of this homolog. When Glt_Ph transitioned to the inward facing state, the distance between two residues does not change substantially (Figures 6A,C). However, when Glt_Ph transitioned to the outward facing state, the distance between the two residues increased (Figures 6A,B). The distance we observed in our experiments is different from that shown by the Glt_Ph crystal structure. The reasons for this difference are as follows. First, prokaryotic Glt_Ph and eukaryotic EAATs originate from different species that are evolutionarily different (sequence homology ∼12.5% in TM3-4 ring between eukaryotic EAAT2 and prokaryotic Glt_Ph), although important functional segments are highly conserved (Vandenberg and Ryan, 2013). Second, mutations were made in the published crystal structure model of Glt_Ph [i.e., seven His residues in non-conserved site of Glt_Ph (Yernool et al., 2004)] that may have caused the crystal structure models to differ from the structure of the transporters in their native membrane environment. The TM3-4 loop was not completely resolved in the structure of EAAT1 and EAAT3; however, in the Cryo-electron microscopy structure of EAAT3 in the outward facing state (PDB ID code 6 × 2Z), Lys123 (namely the Leu149 in EAAT2) was resolved, and we measured the Cα–Cα distance between this site and Leu384 (Met414 in EAAT2). The distance between the two sites is 4 Å, and this proximity might support our finding that the two residues become closer together in an outward facing state (Figure 4). Furthermore, the distance is shorter compared with its counterpart in the model of Glt_Ph in the outward facing state (Figure 6B; Canul-Tec et al., 2017; Qiu et al., 2021).

We also considered the alternation of accessibility of single mutants L149C and M414C. As mentioned, TBOA, L-glutamate and potassium ions influence the conformation of glutamate transporters. Using HeLa cells pre-incubated with MTSET, we show that application of analog TBOA does not protect the single mutant from the inhibition of MTSET in L149C, although treatment with extracellular potassium ions and L-glutamate was capable of doing so (Figure 5B). These three agents were incapable of reversing or reducing the transport activity of M414C (Figure 5C). Combined, these data suggest that the aqueous accessibility of L149C increased when EAAT2 shifted to the inward facing conformation. We also show that application of CuPh decreases the apparent affinity of L149C/M414C and we suggest that in this conformation, where two cysteine residues are in close proximity, the formation of the disulfide bond disrupts the movement of the two residues and hinders transport of the substrate. We also speculate that this may occur because M414C, located in the TM7, is proximal to several residues that participate in the sodium binding in the homolog Glt_Ph (Boudker et al., 2007; Qiu et al., 2021). Furthermore, the residues at positions L149 and M414 in the single mutant have relatively high accessibility, and we suggest that the TM3-TM4 loop may be supported by the L149C to function and ultimately aids in substrate translocation.

Our findings indicate that when EAAT2 ferries its cargo, the TM3-TM4 loop undergoes a proximal conformational change to TM7. Here, we report the significant role of the TM3-4 loop in the transport cycle and outline the interaction with key transport segment TM7.

The original contributions presented in the study are included in the article/Supplementary Material; further inquiries can be directed to the corresponding author/s.

SQ and RC designed the research. QQ and JW performed the experiments. SQ, QQ, JW, and GL analyzed the data. SQ, QQ, and JW wrote the manuscript. All authors: finalization and approval of the content of the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.