Forty Four Years With Baruch Kanner and The Chloride Ion

Abstract

Baruch Kanner and this author have had parallel careers investigating neurotransmitter transporters. At multiple times during their careers, they have found themselves collaborating or competing, but always learning from each other. This commentary elaborates on the interactions between the Kanner and Rudnick laboratories, with a focus on transporters in the Neurotransmitter: Sodium Symporter (NSS) family of amino acid and amine transporters. A key focus of these interactions is the mechanism by which chloride ions activate and drive transport.

In 1977, Shimon Schuldiner and I, both having recently finished our postdoctoral training in Ron Kaback’s lab, were starting to get our own laboratories off the ground. I wanted to get away from bacterial transporters and explore the mechanism of biogenic amine transporters, which are targets for psychoactive drugs. Very little was known at that time about how these transporters worked, aside from their requirement for sodium. My first paper on serotonin transport into platelet membrane vesicles was published that year [1] and Shimon had been talking about it with Baruch Kanner, then a new investigator in the Medical School of the Hebrew University in Jerusalem. Baruch had arrived from a postdoctoral stint in Ephraim Racker’s lab, with valuable experience in purification and reconstitution of membrane proteins. He was trying to measure GABA (γ-aminobutyric acid) receptor activity in membrane vesicles from brain. Shimon suggested that he might want to look at GABA transport in the preparation, to see if it would work similarly to the serotonin system in platelets. We now know how that turned out; Baruch established the GABA transporter as one of the best characterized mammalian transport proteins.

At that time, one of the key questions for transport scientists concerned the driving force responsible for accumulation of neurotransmitters in neurons. Our work had shown that an inwardly directed gradient of sodium ions and an outwardly directed potassium ion gradient were driving forces for serotonin uptake, and that chloride was also required [1, 2]. In 1978, Baruch published two papers on GABA transport and another two on glutamate transport in plasma membrane vesicles from rat brain. The first GABA paper [3] showed that GABA accumulation was driven by a gradient of Na⁺ (out > in), that Cl⁻ or another monovalent anion was also required, and that the process was enhanced by a transmembrane electrical potential (interior negative). The second paper showed that GABA transport could be reconstituted into proteoliposomes using a detergent extract from rat brain membranes [4]. Just as with the membrane vesicles, [³H]GABA accumulation in the proteoliposomes was dependent on external Na⁺ and Cl⁻ and stimulated by the ionophore valinomycin, which would generate an interior negative diffusion potential in these K⁺-loaded vesicles. Nigericin, in contrast, dissipated Na⁺ and K⁺ gradients by exchanging the two cations and thereby abolished transport. Adding unlabeled GABA or cholate to the external medium, after accumulation, caused efflux of the labeled GABA by exchange or disruption of the vesicles, respectively. These findings set the stage for many subsequent discoveries relating to GABA transport. In particular, reconstitution of transporters was a step toward purification of the proteins responsible for neurotransmitter transport and their eventual molecular characterization.

At that point, the transporters were known as activities, rather than molecular entities. Examining their characteristics might allow characterization of the similarities and differences between them. Baruch published another pair of papers in 1978 about glutamate transport into rat brain vesicles and its reconstitution in proteoliposomes [5, 6]. There were interesting parallels and differences between the two systems. Both GABA and glutamate transport were Na⁺ dependent, but only GABA transport required Cl⁻. Both systems were stimulated by an internal negative membrane potential, created by a K⁺ gradient (in > out) and valinomycin, but glutamate transport and not GABA transport required the presence of cytoplasmic K⁺. Clearly, these two transporters behaved differently, and as we would learn later, they were members of two distinct transporter families with different structures. They also each had similarities to serotonin transport in membrane vesicles isolated from platelets. Serotonin transport required both Na⁺ and Cl⁻, like GABA transport [1], but, like glutamate transport, was stimulated by internal K⁺, although not through generation of an internal negative membrane potential [2]. Baruch continued to do outstanding research on glutamate transport. However, the remainder of this commentary will focus on his work with GABA transport, which more closely paralleled my own studies on serotonin transport.

In that same year of 1978, I spent a month in Jerusalem, visiting with Shimon and Baruch and their Israeli colleagues. It was a memorable visit, not only for the scientific exchanges, but also because I met my future and present wife working side by side with her in Shimon’s lab. Baruch, Shimon and I had many opportunities to discuss our interests and the many possibilities and challenges that membrane transport offered. It set the stage for lifelong friendships and the friendly competition that developed as our scientific interests diverged and re-converged through the years (Fig. 1).

Fig. 1.

The author (left) with Shimon Schuldiner (center) and Baruch Kanner (right) in Erice, Sicily in 2005

In those days, one of our main interests was to understand the stoichiometry of ion-coupling to transport. One approach to this problem was to determine whether transport was affected by the transmembrane electrical potential. If a transport process is stimulated by an inside negative potential, we suspect that positive charge may enter the cell or vesicle with substrate. However, stimulation of transport rate could mean simply that the rate determining step in transport is influenced by the potential. However, if the steady state, or equilibrium, level of accumulation is increased by the inside negative potential, we can conclude that positive charge enters with substrate. Baruch understood this and measured the steady state accumulation of GABA under conditions where the Na⁺ or Cl⁻ gradient varied [7]. His results showed that, although a gradient of either ion could drive GABA accumulation, coupling to the Na⁺ gradient was twice as strong as to the Cl⁻ gradient, indicating a 2:1 ratio of Na⁺ to Cl⁻ entry with GABA. These results were consistent with Baruch’s observation that GABA accumulation was stimulated by a membrane potential (inside negative). GABA enters as a neutral zwitterion and the excess of one positive charge on two Na⁺ ions over the compensating negative charge on one Cl⁻ ion rendered GABA accumulation electrogenic, and thereby sensitive to a transmembrane electrical potential.

The simplest stoichiometry consistent with Baruch’s results was that two Na⁺ ions and one Cl⁻ ion were transported into the cell or vesicle with each GABA molecule (Fig. 2a). Much later, detection of two Na⁺ sites and one Cl⁻ site in structures of the homologous dopamine and serotonin transporters (DAT and SERT) [8, 9] validated Baruch’s findings. Results from my lab indicated a different stoichiometry for serotonin transport. Using similar methods, we found that only one Na⁺ ion was transported with serotonin [10] and that one K⁺ ion was transported in the opposite direction [1, 2] (Fig. 2b). At that point, mechanisms of serotonin and GABA transport seemed quite different, with different ions being used and different Na⁺ stoichiometries. However, without molecular characterization of the proteins carrying out these processes, the relationship between these transporters remained a mystery.

Fig. 2.

Reaction cycles for GAT-1 (a) and SERT (b). These diagrams are meant to show the differences between GAT-1 and SERT stoichiometry. GABA is transported together with 2 Na⁺ ions and one Cl⁻ ion while serotonin (5-HT) is transported together with one Na⁺ ion and in exchange for one K⁺ ion, which is transported in the opposite direction. Both transporters require Cl⁻ for function. Transporters, represented by T, are shown in outward- (T_o) and inward-facing (T_i) conformations with substrates (GABA or 5-HT), and ions Na⁺, K⁺ and Cl⁻ bound as indicated. For SERT (B) Na⁺ is shown bound at the Na2 site. The Na1 site (like the Cl⁻ site) is apparently always occupied unless Na⁺ (or Cl⁻) is removed from the extracellular medium

All this changed in the early 1990s. Baruch had managed to purify the GABA transporter from rat brain to a high degree and showed that it was still functional [11]. In collaboration with Nathan Nelson’s and Henry Lester’s labs, he determined enough protein sequence to generate oligonucleotide probes. These probes were used to screen a rat brain cDNA library and isolate a clone that coded for a functional GABA transporter [12]. mRNA transcripts from this clone, when injected into Xenopus oocytes, conferred the ability of the oocytes to take up GABA with the same requirements for Na⁺ and Cl⁻ and inhibitor sensitivity of the initial membrane vesicles from rat brain. GAT-1, as it was now called, was one of the first mammalian transporters to be cloned, and this work represented a major milestone in neurotransmitter transporter research. But GAT-1 was only the beginning. To establish relationships between different neurotransmitter transporters, more needed to be cloned.

The next year, Susan Amara’s lab used expression cloning to isolate a cDNA encoding a norepinephrine transporter from a library derived from a human cell line that accumulated norepinephrine [13]. NET, as the protein encoded by this clone was named, had many similarities (ion requirements, inhibitor sensitivity) to the serotonin transporter and its sequence was homologous to that of GAT-1. It now seemed like there was a family of neurotransmitter transporters that included GAT-1, NET and likely the serotonin transporter. This was confirmed by Randy Blakely, collaborating with Marc Caron, who used the two existing sequences to identify a cDNA clone by PCR using primers based on NET and GAT-1 [14]. At the same time, Beth Hoffman used an expression cloning strategy to identify a similar cDNA clone from RBL cells, which accumulate serotonin [15]. The serotonin transporter, now called SERT, was the third member of the family, which became known as the Neurotransmitter: Sodium Symporter (NSS) family of transporters. Subsequent identification of glycine and dopamine transporters [16–19] established this family as being responsible for the reuptake of almost all small neurotransmitters in the central nervous system. Notably absent from this group of transmitters were glutamate (taken up by a different transporter family) and acetylcholine, which is inactivated by hydrolysis, not reuptake. Baruch was also instrumental in cloning GLT-1, one of the first transporters responsible for synaptic reuptake of glutamate [20], demonstrating that it was unrelated to the NSS family.

With GAT-1, SERT and NET clones available, there was plenty to do. Although the sequences implied a transmembrane topology, it needed to be verified experimentally. Also, the field needed a platform for testing mutations in these transporters to find out where key residues were located. Baruch and I decided to join forces to express GAT-1 in mammalian cells. We had been using an expression system dependent on an engineered vaccinia virus that expressed the T7 RNA polymerase to amplify and transcribe plasmid cDNA into mRNA encoding SERT [21]. This system also worked well for GAT-1 expression [22], and both our labs used it extensively.

We also started a project to see if chimeras between GAT-1 and NET would be functional. The only chimeras that retained activity were those in which only the N- or C-terminal tails were swapped. Baruch found that even small deletions in the loops between predicted transmembrane segments would disrupt activity [23]. This result highlighted the importance of extramembrane loops in transporter function; the role of these loops has never been adequately explained. Progress in the field now depended on a structural model which could be used to ask how the transporters move their substrates through the membrane. Testing the functional effect of mutations provided local information on the mutated positions but didn’t relate those effects to the larger conformational changes. We were like the proverbial blind men with an elephant, and only a structural model would allow us to reach the next level of understanding.

Until the atomic structure of an NSS protein was available, biochemical methods were the only way to test the predictions of transmembrane topology that were inferred from the sequences. Both Baruch’s lab and mine attempted to address this question. Baruch deleted glycosylation sites that were native to GAT-1 and inserted novel glycosylation sites to identify extracellular segments [24]. My lab made specific mutations, inserting reactive cysteine and lysine residues in predicted loops of SERT and tested their ability to react with impermeant extracellular reagents [25]. Together, we arrived at a common topology that was largely verified in 2005 when Eric Gouaux and his colleagues published a structure of the bacterial amino acid transporter LeuT, the first structure of an NSS transporter [26]. This structure and later ones of SERT, DAT and GlyT1 [8, 9, 27] demonstrated that all NSS transporters adopted a similar fold. Baruch went on to determine the topology of the glutamate transporter GLT-1, which revealed an unusual topology [28] that was verified by the structure of a bacterial homolog [29].

The LeuT structure contained two Na⁺ sites, consistent with the two Na⁺ ions symported with GABA, but not with the one Na⁺ symported with serotonin. Moreover, LeuT didn’t require Cl⁻, and the location of the Cl⁻ site in NSS neurotransmitter transporters remained a mystery. This second question set up the most interesting interaction between my lab and Baruch’s – namely the role of Cl⁻ in transport. In the late 70 s, we demonstrated that serotonin transport required an external anion such as Cl⁻ [1]. Baruch’s lab showed that the same was true for GABA transport [3] and that a Cl⁻ gradient by itself could drive GABA accumulation [7]. Although both transporters required Cl⁻, a Cl⁻ gradient alone consistently failed to drive serotonin accumulation. Currently, the evidence suggests that Cl⁻ does not dissociate from SERT when 5-HT does, but instead stays bound through the transport cycle (Fig. 2B) [30].

Because both transporters required a monovalent anion like Cl⁻, there was intense interest in where Cl⁻ bound. LeuT does not require Cl⁻, and the only Cl⁻ found in the original LeuT X-ray structure was on the periphery of the protein [26]. However, analysis of the LeuT structure and comparison of its sequence to that of GAT-1, SERT and other Cl⁻-dependent NSS transporters suggested to Baruch and to Lucy Forrest (then a postdoc with Barry Honig) that Glu290 in LeuT provided a negative charge positioned at the location corresponding to where Cl⁻ bound in SERT and GAT (Fig. 3). Glu290 in LeuT corresponds to a serine in Cl⁻-dependent NSS transporters (Ser372 in SERT, Fig. 3 bottom), which coordinates the Cl⁻ ion. We were working with Lucy and Barry to provide some biochemical evidence in SERT to support their proposed Cl⁻ site when we learned that Baruch had already submitted a manuscript with similar evidence for the Cl⁻ site in GAT-1. Both labs showed that replacing the serine corresponding to LeuT Glu290 (Ser372 in SERT and Ser331 in GAT-1, Fig. 3 bottom) with glutamate or aspartate allowed transport in the absence of Cl⁻. After a fevered race to publish these studies, our papers appeared within weeks of each other [31, 32].

Fig. 3.

Comparison of ion binding sites in outward occluded states of Cl⁻-independent LeuT (PDB 2A65) and Cl⁻-dependent SERT. The ionized γ-carboxyl of Glu290 in LeuT (top) is replaced in Cl⁻-dependent NSS transporters by the bound Cl⁻ ion, as shown in a model of SERT in an outward-occluded conformation [43] (bottom). In addition to Ser372, which corresponds to LeuT Glu290, the Cl ion is coordinated by residues conserved within the family, including Tyr121, Ser336 and Gln332. Gln332 in SERT, or Gln250 in LeuT is believed to interact with both the bound Cl⁻ in Cl⁻-dependent NSS transporters (or the negatively charged Glu290 in LeuT) and the ion pair represented by Arg30-Asp404 in LeuT and Arg104-Glu493 in SERT. Kanner first proposed that this interaction might influence transporter conformation [40] and this influence was recently demonstrated for the GlyT1 glycine transporter [42]. Transmembrane (TM) helices are color-coded as follows: TM1, raspberry; TM2, salmon; TM5, limon; TM6, forest (shown as ribbon for clarity); TM7, light teal; TM10, blue (ribbon)

There was some initial disagreement between the labs on the components of the Cl⁻ binding site. Baruch proposed, correctly, that a highly conserved glutamine residue corresponding to Gln250 in LeuT (Fig. 3, top) was part of the site (Gln332 in SERT, Fig. 3, bottom), but some of our data could be interpreted as refuting that proposal. The issue was settled when Poul Nissen, collaborating with Jonathan Javitch, solved the X-ray structure of LeuT E290S with Cl⁻ bound [33]. Although this mutant could not transport, it could bind substrate, and this binding required Cl⁻. The Javitch lab also showed that another bacterial NSS transporter, Tyt1, Cl⁻-independent like LeuT, could become Cl⁻-dependent by replacing an endogenous aspartate and inserting a serine, although the resulting activity was about 1% that of wild type Tyt1 [32]. Our lab was able to convert yet another bacterial NSS transporter, TnaT, to become completely Cl⁻-dependent with the same activity as wild type TnaT by replacing the aspartate corresponding to LeuT Glu290 with serine [34]. Structures of DAT and SERT verified that the Cl⁻ site in LeuT E290S corresponded to the true Cl⁻ site in neurotransmitter transporters [8, 9].

With the Cl⁻ site established, the next question was how it acted. By 2012, there were structures of LeuT in outward and inward-open states [35], and it was clear that two domains moved relative to one another as LeuT transitioned between these states during the reaction cycle [36]. But how did ion and substrate binding influence transport? Ultimately, these binding events must act to allow or prevent the specific conformational changes that constitute the transport process. Quick and Javitch showed that the cytoplasmic permeation pathway of Tyt1, a bacterial NSS transporter, was stabilized in the closed position by Na⁺ [37], and we later showed that in LeuT, this influence was due to Na⁺ binding at the Na2 site, where the two domains meet at the beginning of the cytoplasmic pathway [38]. Similarly, substrate bound at the central binding site interacts with the two domains to overcome the effect of Na⁺ by inducing the extracellular pathway to close [39]. However, unlike the sites at which Na⁺ and substrate bound, the Cl⁻ site proposed for GAT-1 and SERT was not at that domain interface. How could it influence the conformational changes required for transport?

In the structure of the LeuT E290S mutant, Cl⁻ was coordinated by Gln250 (Fig. 3, top) as Baruch had proposed [33]. Moreover, subsequent structures of DAT and SERT showed that the corresponding glutamine coordinated Cl⁻ in those proteins (Fig. 3, bottom) [8, 9]. Baruch noticed that in LeuT, this residue was close to an ion pair (Arg30-Asp404, Fig. 3, top) that had been proposed to function as a gate between the substrate site and the external medium [26]. Mutation of the corresponding glutamine in GAT-1 led to a loss of activity that could be partially restored by the S331E mutation that rendered GAT-1 Cl⁻-independent [40]. Until recently, this was the only evidence suggesting that Cl⁻ acted to strengthen the ion pair, and molecular dynamics simulations of such an action by Cl⁻ were inconclusive [33].

To resolve the issue we turned to GlyT1, a Cl⁻-dependent NSS glycine transporter. GlyT1, like LeuT, is an amino acid transporter. However, GlyT1 requires Cl⁻ and symports it with glycine [41] In GlyT1, Cl⁻ acts independently of substrate to close the extracellular pathway, consistent with favoring formation of the ion pair pinpointed by Baruch [42]. Mutation of the glutamine residue that coordinates Cl⁻ also led to dramatic changes in transporter conformation, confirming Baruch’s prediction.

My scientific relationship with Baruch Kanner has now spanned over 40 years and has been marked by our parallel progress using two important neurotransmitter transporters GAT-1 and SERT. It has been a journey of joint discovery in which each of us has benefited from the other. I consider myself lucky to have had such an open and collaborative relationship with Baruch and I hope it has enriched his scientific journey as much as it has mine.