Cooperative mechanism of neurotransmitter recognition and transport by the human vesicular polyamine transporter

Abstract

Human vesicular polyamine transporter (hVPAT) from the SLC18 antiporter family sequesters polyamine neuromodulators into secretory vesicles in exchange for H⁺ in the brain, thereby sustaining learning and memory formation. As a potential therapeutic target for combatting psychostimulant use disorder, hVPAT (or SLC18B1) regulates the homeostasis of monoamine neurotransmitters via an unknown mechanism. Here we report the cryo-electron microscopy structures of hVPAT in complexes with histamine and serotonin, revealing up to three distinct neurotransmitter-binding sites within the lumen-facing SLC18 antiporter. Complementary proteo-liposome-based transport and ligand-binding assays suggest that hVPAT transports multiple monoamine neurotransmitters simultaneously, exhibiting variable H⁺/substrate stoichiometry and positive cooperativity. Unexpectedly, this cooperativity appears to arise from the direct interactions between the bound histamine or serotonin molecules, which also contributes to substrate selectivity. In this work, our findings uncover a hitherto unidentified modality for monoamine neurotransmitter recognition and offer unprecedented insights into the general principles governing membrane transport.

Vesicular transporters are essential for storing and releasing neurotransmitters. Here, authors report structures of the human vesicular polyamine transporter bound to histamine and serotonin, revealing a distinctive transport mechanism that enables its broad functional repertoire.

Introduction

Vesicular polyamine transporter (VPAT or SLC18B1), a member of the SLC18 antiporter family, plays a crucial role in chemical neurotransmission by using the preexisting proton gradient to load polyamines into secretory vesicles^1–3. The SLC18 proteins belong to the Major Facilitator Superfamily (MFS), the largest known family of solute transporters⁴, which typically consist of twelve membrane-spanning helices (TM1-TM12) organized into the structurally similar N-domain (TM1-TM6) and C-domain (TM7-TM12). Despite considerable variations in substrate specificity, coupling mechanism, and physiological function across and within the subdivisions, most MFS transporters share a common protein fold and operate via a rocker-switch-type alternating access mechanism^5,6.

VPAT in particular underpins learning and memory formation by enabling the quantal release of polyamines such as spermine and spermidine, modulators of key players in chemical neurotransmission, which include the NMDA and AMPA receptors³. Additionally, VPAT has been associated with cocaine addiction, implicating the antiporter in the regulation of monoamine neurotransmitter homeostasis, since cocaine disrupts neurotransmitter balance by blocking the reuptake of serotonin and dopamine, among others⁷. Thus, despite the lack of knowledge on whether or how VPAT transports monoamine neurotransmitters, human VPAT (hVPAT) has emerged as a potential target for combatting psychostimulant use disorder, which inflicts a serious public health burden⁷. Indeed, knockout of the VPAT gene in mice increased their sensitivity to the psychostimulant amphetamine². Furthermore, the malfunction of hVPAT has been linked to the pathophysiology of Parkinson’s disease, highlighting its critical role in maintaining the balance of polyamine neuromodulators and/or monoamine neurotransmitters⁸.

Besides the brain, VPAT is also expressed in many other tissues in mammals¹. Of note, VPAT is highly expressed in the bone marrow-derived mast cells as well as activated macrophages, contributing to the secretion of polyamines and immune function^9,10. Importantly, polyamines regulate the storage and release of histamine and serotonin from the mast cells, although the underlying mechanism remains elusive¹¹. Nonetheless, previous studies showed that the VPAT gene knockdown in the mast cells not only abolished the transport of spermine or spermidine, but also inhibited the antigen-evoked release of histamine, despite the presence of VMAT2, an SLC18 antiporter known to transport histamine and other monoamine neurotransmitters⁹. These findings implied that VPAT in the mast cells controls or regulates the secretion of histamine, a key signaling molecule involved in inflammation and allergic reactions, pointing to a broader functional repertoire of VPAT beyond its established role in neurotransmission³.

It may be relevant to note that histamine or serotonin can inhibit the uptake of spermine or spermidine mediated by the detergent-solubilized hVPAT, which can catalyze the uptake of serotonin almost as efficiently as polyamines¹. Besides chemical neurotransmission, serotonin plays a vital role in the physiological function of platelets, in which VPAT is also highly expressed¹². Thus, VPAT likely contributes to hemostasis or blood clotting by promoting the secretion of serotonin during platelet activation. Other than the mast cells and platelets, VPAT has also been identified in the primary human melanocytes as a highly expressed transporter for spermidine, which can increase melanogenesis and reduce premature hair graying or skin hypopigmentation¹³. Notably, the knockout of a pain-sensing gene Slc45a4, which encodes a neuronal polyamine transporter, also gave rise to white-speckled hair in mice¹⁴. This phenotypic overlap suggested that VPAT shares functional similarities with a pain-sensing neuronal transporter and represents a promising therapeutic target for treating hypopigmentation disorders and/or premature hair graying³.

To date, significant progress has been made in elucidating the molecular mechanism of hVPAT, fueled by the structural determination of its lumen-facing state bound to spermine, spermidine, protons, or tetrabenazine, a non-competitive inhibitor¹⁵. These cryo-EM structures revealed the polyamine-binding site in hVPAT and suggested a rather unusual antiporting mechanism that involves both direct and indirect competition between polyamine and protons¹⁵. However, these structures shed no light on the interactions between hVPAT and any monoamine neurotransmitter. Nonetheless, the cryo-EM structures of histamine- and/or serotonin-bound human VMAT1 or VMAT2, the monoamine-specific members of the SLC18 family, have been reported^16,17. However, due to the particularly low protein sequence similarity (<19% identical) between hVPAT and VMAT1 or VMAT2 (Supplementary Fig. 1), their mechanisms of neurotransmitter recognition and transport are likely to differ¹⁵.

In this work, to elucidate the interactions between hVPAT and monoamine neurotransmitters, which underlie important aspects of psychostimulant action, immune function, or blood clotting³, we functionally characterize hVPAT by utilizing the proteo-liposome-based transport and ligand binding assays^1,15. We find that hVPAT and hVPAT^EM, a hVPAT variant utilized in the reported cryo-EM study¹⁵, bind and transport multiple monoamine neurotransmitters simultaneously. Furthermore, we present the histamine- and serotonin-bound structures of hVPAT^EM, identifying a distinctive molecular mechanism of neurotransmitter recognition and transport. This work offers a structural and conceptual framework to understand how a secondary active transporter concurrently moves multiple substrate molecules across the membrane, broadening the spectrum of hVPAT activity and enlightening our understanding of neurotransmitter transport in general.

Results

hVPAT-mediated transport of histamine

We expressed full-length hVPAT in insect cells and purified the detergent-solubilized antiporter by a combination of affinity and gel-filtration chromatography¹⁵. We then reconstituted the purified hVPAT into proteo-liposomes and measured the uptake kinetics of [³H] histamine (Supplementary Fig. 2a). The dose-response curve, i.e., the graph of transport rates plotted against histamine concentrations, in the 0–200 µM range appeared rather sigmoidal, so we fitted our data to the allosteric sigmoidal enzyme kinetics¹⁸, which is essentially the Michaelis–Menten kinetics model with a variable Hill slope. Data analysis subsequently revealed that the K_1/2 (the concentration of histamine that produces a half-maximal uptake velocity) and V_max (the maximum transport velocity) are 151 µM and 23.2 nmole/min/mg protein, respectively, with an apparent Hill coefficient of 2.4.

Although the Hill coefficient does not correlate with the number of substrate-binding sites, it is a measure of cooperativity, and the next integer above the apparent Hill coefficient reports the minimal value for the number of ligand-binding sites¹⁸. Thus, our data implied that the monomeric¹⁵ hVPAT transports multiple (≥3) histamine molecules concurrently with positive cooperativity, in sharp contrast to human VMAT1 or VMAT2, which interact with a single histamine molecule per transport cycle^4,16. To exclude the possibility that the observed cooperativity was due to the mixed orientations of hVPAT within liposomes, we analyzed the polyamine uptake kinetic data obtained from the same batch of purified protein¹⁵, and found that the spermine transport rates could be fitted to the allosteric sigmoidal enzyme kinetics model with a Hill slope of 0.99, in agreement with single-site binding. Altogether, our results suggested that positive cooperativity is an intrinsic property of hVPAT-mediated transport of histamine under the tested conditions.

Furthermore, we found that valinomycin-invoked K⁺ diffusion potential could substantially alter the histamine uptake rate (Supplementary Fig. 2a), suggesting that the hVPAT-mediated transport is an electrogenic process, which gave rise to an extrusion of one or more net positive charges per cycle. This observation stood in stark contrast to the hVPAT-mediated transport of spermine or spermidine^1,15, which is electroneutral, highlighting the mechanistic promiscuity of this SLC18 antiporter. Notably, this result is reminiscent of other SLC18 members, VMAT1, VMAT2, and VAChT, all of which catalyze an electrogenic exchange with an H⁺/substrate stoichiometry⁴ of 2:1.

In addition, we measured the binding of [³H] histamine to hVPAT at pH 8.0 (Supplementary Fig. 2b), using a method that we had developed for the binding of spermine¹⁵. Since the ligand binding curve also appeared sigmoidal, we fitted the data to specific binding with Hill slope, revealing a K_d (dissociation constant) of 14.0 µM, a B_max (histamine/protein stoichiometry) of 2.7, and an apparent Hill coefficient of 2.5. Together, these data implied that the detergent-purified hVPAT binds to multiple histamine molecules cooperatively, with the number of binding sites most probably being 3, as defined by the B_max. In comparison, spermine binds to hVPAT with a B_max of 0.98, further validating our ligand binding assay and highlighting the differences between polyamine and monoamine in their interactions with hVPAT¹⁵. Collectively, our data support that hVPAT transports three histamine molecules simultaneously with positive cooperativity, a feature not previously reported for any SLC18 antiporter^3,4.

Furthermore, hVPAT appears to exchange four or more protons for three histamine molecules, each of which predominantly carries a (+1) charge under our experimental conditions (at pH ~8.0). Given that there are only four membrane-embedded acidic protein residues in hVPAT^3,15, our data suggested that the H⁺/histamine stoichiometry is 4:3. Moreover, the physiological concentration of histamine in the human blood is normally <1 µM¹⁹, our data thus implied that hVPAT safeguards against neurotransmitter toxicity in the brain by sequestering the excessive and potentially harmful histamines into intracellular vesicles. On the other hand, the histamine concentrations in the mast cells can reach mM level¹⁹, in line with the pivotal role of hVPAT in mediating histamine secretion from these cells⁹.

In parallel, we characterized the fusion protein hVPAT^EM, which had been used in our prior cryo-EM study¹⁵. Based on our histamine uptake and binding assays as described above, we found that hVPAT^EM most probably binds and transports three histamine molecules simultaneously (Supplementary Fig. 2), recapitulating the function of hVPAT. Specifically, our data analysis revealed that the K_1/2 and V_max for hVPAT^EM are 319 µM and 5.3 nmole/min/mg protein, respectively, with a Hill coefficient of 2.5. Furthermore, the K_d, B_max, and Hill coefficient for histamine binding are 30.4 µM, 2.5, and 2.1, respectively. As such, we concluded that hVPAT^EM can be utilized for investigating the interactions with histamine. The reduced transporter activity of hVPAT^EM relative to that of hVPAT, on the other hand, may result from the structural restraints imposed by the rigidly attached soluble protein domains and/or the alteration of protein conformational equilibrium, as observed for the similarly designed fusion protein of VMAT2¹⁷.

Thereafter, we collected cryo-EM data on hVPAT^EM, which was purified in the presence of added histamine and exchanged into amphipols (Supplementary Fig. 3), using an experimental set-up reported previously¹⁵. We processed the cryo-EM images by using cryoSPARC²⁰ and improved the non-uniform refined map further by using spIsoNet²¹, cryoFEM²², and EMReady2²³, which are machine-learning-based software for correcting data anisotropy and/or enhancing density features (Supplementary Table 1). In particular, the best cryo-EM map produced from spIsoNet was calculated to a global resolution of 3.5 Å (Supplementary Fig. 3), based on the FSC threshold of 0.143 (or FSC_0.143) between the two independently calculated half maps.

The histamine-binding sites in hVPAT

Our cryo-EM map (Supplementary Figs. 3 and 4) revealed a lumen-facing conformation of the SLC18 antiporter (Fig. 1a, b). Since the pH of the protein sample is 8.0, we assert that this structure portrays a lumen-facing, deprotonated state, given the cytosolic pH of ~7.4 under a physiological setting. At the interface between the pseudo-symmetric N- and C-domains, we observed three conspicuous, lollipop-shaped density features located approximately halfway into the membrane (Fig. 1c and Supplementary Fig. 5). Since the size and shape of these density features are consistent with those of histamine, with calculated Q scores²⁴ being ≥0.76 (Fig. 1c), we assigned them as three histamine molecules, HSM1-HSM3. We made this assignment also because these ligands are anchored within the polyamine-binding pocket¹⁵, which is encapsulated by numerous aromatic and hydrophobic amino acids (Fig. 2a), in addition to the membrane-embedded E83^TM2, E171^TM5, and D255^TM7 (Fig. 2b). Since their side-chain carboxylates have calculated pK_a values²⁵ of 6.8, 6.7, and 6.4, respectively, we concluded that these three protein residues all carry a (−1) charge. Nonetheless, to conduct the pK_a predictions, we used the software PROPKA3, which, despite its wide popularity and reported accuracy²⁵, provides empirical estimates of the pK_a values outside the native membrane environment. Thus, these predictions may deviate from the in vivo pK_a values and warrant a cautious interpretation.

Fig. 1. Cryo-EM density for the histamine-bound hVPAT^EM.

a, b The cryo-EM density for the histamine-bound hVPAT^EM as viewed from the lipid bilayer, with the membrane simplified as a grey rectangle. Cryo-EM density for the N- and C-domains of hVPAT is colored blue and orange, respectively; whereas the density for histamine and water molecules is shown in magenta and red, respectively. The cryo-EM density is contoured at 8.0 in ChimeraX v.1.8 to highlight the α-helical pitch. c Two zoomed-in views displaying the fitting of the three histamine molecules, HSM1-HSM3 (magenta sticks), into the cryo-EM density (magenta), which is contoured at 8.0 in ChimeraX v.1.8. The calculated Q scores, a measure of resolvability, are indicated for the histamine molecules.

Fig. 2. The histamine-bound structure of hVPAT.

a Ribbon diagram of the histamine-bound structure as viewed from the membrane bilayer. The N- and C-domains are colored blue and orange, respectively. The bound histamine (magenta) and relevant protein residues are drawn as sticks. b Close-up view of the binding sites for histamine molecules, HSM1-HSM3, with the histamine molecules (magenta) and relevant protein residues shown as stick models, and water molecules (W1–W2) as red spheres. Salt bridges, H-bonds, and cation–π interactions are highlighted as blue dashed lines (distances in angstroms are also indicated when applicable), whereas the aromatic stacking or van der Waals interactions are indicated by dashed lines in gray. c The overlay of the histamine- and spermine-binding sites in hVPAT. The bound histamine molecules (magenta), spermine (gray), and the relevant protein residues are shown as sticks. Red arrows highlight the structural differences discussed in the text. d Superimposition of the histamine-binding sites onto the low-pH structure (light green). The histamine molecules (magenta) and the relevant protein residues are illustrated as stick models. Red arrows highlight the structural differences, and red crosses indicate the steric clash between Y51/F251 at low pH and the otherwise bound substrates.

Moreover, we identified direct interactions between the bound histamine molecules and the antiporter, both electrostatic and hydrophobic in nature (Fig. 2b). Specifically, the positively charged ethylamine in HSM1, approximately 4.3 Å from the side-chain carboxylate of E83, makes cation-π interactions with the side chain of F144^TM4. Moreover, the aromatic imidazole group from HSM1 makes hydrophobic interactions with the side chains of M48^TM1 and Y51^TM1. Intriguingly, the imidazole moiety of HSM1 also makes favorable cation-π interactions with the ethylamine group from the adjoining HSM2 and forms edge-to-face aromatic stacking interactions with the imidazole moiety in the neighboring HSM3. As such, the direct contacts made among the histamine trio are poised to stabilize the binding of HSM1 to the antiporter.

Similarly, the ethylamine group in HSM2, about 3.9 Å from the side-chain carboxylate of E171, makes cation-π interactions with the side chain of W384^TM11 and also forms a direct H-bond with a water molecule W1 (Fig. 2b). Moreover, the imidazole group from HSM2 makes extensive aromatic stacking interactions with the side chain of F251^TM7. Once again, HSM2 makes direct contacts with both HSM1 and HSM3: the ethylamine groups in HSM2 and HSM3 make cation-π interactions with the imidazole moieties in HSM1 and HSM2, respectively, further cementing the binding of HSM2 to the antiporter. Furthermore, the imidazole groups in HSM2 and HSM3 make aromatic stacking interactions with each other.

Finally, the ethylamine group in HSM3 forms a salt bridge with the side chain of D255^TM7, whereas its imidazole moiety makes extensive aromatic stacking interactions with the side chain of Y51^TM1 (Fig. 2b). Besides the direct contacts between HSM3 and the protein, both HSM1 and HSM2 also interact with HSM3. Overall, in accord with our transport and binding assays, there are three disparate histamine-binding sites formed within the substrate-binding pocket, and the direct interactions among the histamine molecules seem to account for the observed positive cooperativity. Furthermore, our data points to a previously unidentified neurotransmitter recognition mechanism (Fig. 2c): the antiporter selects its substrate on the basis of not only the structural and chemical complementarity between the protein and ligands, but also the inherent ability of the histamine molecules to interact with each other, i.e., via cation–π and aromatic stacking interactions (Fig. 2b). Both of these features confer substrate selectivity, despite the moderate interactions between the antiporter and individual ligands.

Notably, the three positively charged histamine molecules are in proximity to E83, E171, and D255 (Fig. 2b), all of which are protonation sites in hVPAT¹⁵, implying that the monoamine substrates, similar to polyamines, can directly compete against H⁺ for the binding of these three acidic amino acids. The fourth protonation site¹⁵, D140^TM4 (predicted pK_a ~ 5.4), on the other hand, lies outside the histamine-binding sites and forms a direct salt bridge with the side-chain guanidinium group in R137^TM4 (Fig. 2b), consistent with a deprotonated state (Fig. 2d). Perhaps surprisingly, the side chain of R137 also forms cation-π interactions with the aromatic side chain of F79^TM2, which was not observed under the same pH in the polyamine-bound structure of hVPAT¹⁵ (Fig. 2c).

Mutagenesis study of the histamine-binding sites

To validate the functional importance of the interactions observed between hVPAT^EM and histamine molecules, we replaced M48, Y51, E83, F144, E171, F251, D255, and W384 individually with alanine in hVPAT and reconstituted the detergent-solubilized mutants into proteo-liposomes^1,15. Based on the radio-labeled histamine uptake assay, the substitution of Y51, E83, F144, E171, F251, D255, or W384 exerted pronounced deleterious effects on the transport function (Fig. 3a, b), by markedly decreasing (>7.5-fold) the V_max and increasing (>3.8-fold) the K_1/2 (Supplementary Table 2). The transporter activity of all these mutants was almost reduced to the background level, indicating that these histamine-contacting protein residues are important for the binding of substrate as well as the conformational changes required for the hVPAT-mediated uptake of histamine. By contrast, the mutant M48A exhibited higher transporter activity than all the other mutants (Fig. 3a), suggesting that M48 plays a less important role in stabilizing the antiporter-histamine interactions.

Fig. 3. Functional study of the histamine-binding sites.

a, b Radiolabeled histamine uptake kinetics of the liposome-reconstituted hVPAT variants. The transporter-mediated uptake of histamine here is driven by an outwardly directed proton gradient. Data are presented as mean values ± SD. Error bars represent SD (n = 3, i.e., three biological replicates were examined). The measured K_1/2, V_max, and Hill coefficient are included in Supplementary Table 2. c, d should be c,d Radiolabeled histamine binding to the purified hVPAT variants at pH 8.0. Data are presented as mean values ± SD. Error bars represent SD (n = 3, i.e., three biological replicates were examined). The measured K_d, B_max, and Hill coefficient are listed in Supplementary Table 3.

Notably, within experimental error, all mutants retained Hill coefficients greater than 1.0 (Supplementary Table 2), indicating preserved positive cooperativity. This observation suggests that while mutations of histamine-contacting residues in hVPAT may have altered specific protein-ligand interactions, they are insufficient to obliterate the inter-substrate interactions or cooperativity. Overall, the transport assay corroborated the functional importance of the histamine-contacting protein residues, implying that the cooperativity stems from the interactions among the substrates, rather than being solely mediated through the histamine-contacting amino acids.

Additionally, we measured histamine uptake activity of the single mutants F79A, R137A, and D140A (Fig. 3a, b), revealing that these mutations all markedly reduced the V_max (>12-fold). Moreover, the mutations F79A and D140A increased the K_1/2 > 8-fold, whereas R137A augmented the K_1/2 by merely ~2-fold. These observations suggested that F79, R137, and D140, although not directly involved in the binding of histamine in the lumen-facing state, all play pivotal roles in the hVPAT-mediated transport of histamine. It is conceivable that the side chains of F79 and D140 neutralize R137 in the deprotonated antiporter and preclude the positively charged side-chain guanidinium from protruding towards the substrate-binding site, which may avoid premature dissociation of the cationic substrates during transport. Additionally, D140 is one helical turn from the histamine-contacting F144 (Fig. 2b), thus the salt bridge between D140 and R137 may stabilize the binding of histamine indirectly through this spatial proximity. As such, the mutation of D140 can also suppress the binding of histamine to hVPAT.

To examine the interactions between histamine and the hVPAT mutants directly, we also measured the binding of radiolabeled histamine to the detergent-solubilized hVPAT variants at pH 8.0 (Fig. 3c, d). We found that all the mutations, except for M48A and R137A, markedly augmented (>11.8 fold) the K_d (Supplementary Table 3) while preserving a B_max or an apparent Hill coefficient greater than 1.0 (Supplementary Table 3). These results lent further credence to the importance of the observed histamine-binding protein residues in substrate binding and suggested that F79 and D140, although not in direct contact with any bound substrate molecule, play crucial roles in stabilizing the binding of histamine to the antiporter. Lastly, these data helped to rationalize why histamine affects the VPAT-mediated accumulation of polyamines in the mast cells⁹, i.e., likely by directly competing for the substrate-binding site, since the histamine-contacting Y51, D255, or W384 also interacts with polyamines in the deprotonated antiporter¹⁵ (Fig. 2c).

Comparison of the ligand-bound structures of hVPAT

The histamine-bound structure of hVPAT^EM could be superimposed onto that of the spermine-bound form (PDB 9D7X), yielding an RMS deviation of 0.8 Å for 384 Cα positions (Fig. 2c), which indicated overall structural similarity. Nevertheless, the histamine-contacting Y51, E83, F251, D255, and W384 all assume distinct side-chain conformations from their spermine-binding counterparts, apparently enabling hVPAT to accommodate a trio of monoamine neurotransmitters by utilizing the poly-specific binding pocket. The interactions among the histamine molecules effectively act as a “linchpin” to stabilize the binding of three neurotransmitter molecules, despite fewer interactions between the antiporter and each histamine molecule, as opposed to those made with spermine. Another structural difference centers on R137, which rotates toward the aromatic side chain of F79 upon the binding of histamine to form parallel cation-π interactions. It is not immediately clear, however, why R137 makes such an adjustment upon binding histamine, although the side-chain guanidium of R137 is only 4.6 Å from the phenolic hydroxyl in histamine-contacting Y51, likely forging favorable charge-dipole interactions.

To gain additional insights into the coupling between H⁺ and histamine, we superimposed the histamine-bound structure onto the structure of hVPAT^EM obtained at pH~5 (PDB 9D7U)¹⁵, which gave rise to a rms deviation of 1.4 Å over 384 Cα atoms (Fig. 2d). Upon inspection, we observed discernible differences in Y51, F79, R137, D140, F144, F251, D255, and W384, all of which surround the bound histamine molecules (Fig. 2b, d). In particular, the side chains of Y51 and W384 rotate away upon the lowering of pH and disengage from the interactions with HSM3 and HSM2, respectively, thus disfavoring the rebinding of substrates. In parallel, F144 and F251 shift away their side chains at low pH to rupture the cation–π or aromatic stacking interactions with HSM1 and HSM2, respectively, further destabilizing the binding of histamine. Lastly, the side chain of R137 dissociates from F79 or D140 at low pH, which consequently points toward the histamine-binding sites, discouraging the cationic neurotransmitters from interacting with the antiporter.

All these structural differences, on the other hand, are likely to be triggered or accompanied by the protonation of the membrane-embedded E83/E171/D255 and D140, which would break the charge-charge interactions formed between the antiporter and HSM1-HSM3 and R137, respectively. Of particular importance, HSM1-contacting F144 is only one helical turn from D140, and hence the structural changes in one position are likely to propagate to the other, adding yet another element to the histamine/H⁺ coupling mechanism. Taken together, all the above-described differences between the low pH and histamine-bound structures form the molecular basis for the antiporting mechanism, by which the binding of H⁺ to hVPAT evokes the dissociation of three histamine molecules, either directly (in the case of E83, E171, or D255) or indirectly (for D140). Since the calculated pK_a values of the side-chain carboxylates of E83, E171, and D255 are comparable (Fig. 2b), we posited that the protonation of these acidic residues occurs simultaneously, and the synchronous release of three histamine molecules from the antiporter ensues. On the other hand, although the predicted pK_a values are suggestive, this mechanistic aspect deserves further experimental validation.

hVPAT-catalyzed transport of serotonin

Previous studies demonstrated that hVPAT also transports serotonin, or 5HT, via an unknown mechanism^1,3. Since serotonin is substantially bulkier than histamine, we reasoned that serotonin may interact with hVPAT differently from histamine. Indeed, after we measured the kinetics of hVPAT-mediated uptake of [³H] serotonin (Supplementary Fig. 6a), we found that the K_1/2, V_max, and Hill coefficient are 167 µM, 11.2 nmole/min/mg protein, and 1.6, respectively. Moreover, we examined the binding of [³H] serotonin to hVPAT (Supplementary Fig. 6b), revealing a K_d of 31.3 µM, a B_max of 1.8, and a Hill coefficient of 1.8, respectively. Therefore, our data converged to suggest that hVPAT transports two, as evidenced by the B_max, rather than three serotonin molecules concurrently. Furthermore, we found that hVPAT-mediated uptake of serotonin is also electrogenic and results in an egress of one or more net positive charges per transport cycle (Supplementary Fig. 6a), since the uptake rates were markedly augmented by the valinomycin-imposed, inside-positive membrane potential. As such, our data implied that the H⁺/serotonin stoichiometry for hVPAT is 3:2 or 4:2.

Since the physiological concentrations of serotonin in the human blood are normally below 1 µM²⁶, our findings imply that hVPAT loads excessive and potentially toxic serotonin into membrane vesicles to protect the brain. Nonetheless, the serotonin concentrations in the platelets usually amount to the mM level²⁶, our data thus raised the possibility that hVPAT is involved in the secretion of serotonin from the platelets, thereby regulating hematosis (blood clotting). Because abnormal platelet serotonin levels have been associated with stroke or cardiovascular diseases, and dysregulation of spermine or spermidine levels has been linked to ischemic stroke^3,14, our findings supported hVPAT as a potential target for battling stroke or associated cardiovascular diseases.

Moreover, we found that hVPAT^EM retained the ability to bind and transport serotonin (Supplementary Fig. 6), with the measured K_1/2, V_max, and Hill coefficient being 340 µM, 2.9 nmole/min/mg protein, and 1.6, respectively. The measured K_d, B_max, and Hill coefficient (for serotonin binding), on the other hand, are 63.6 µM, 1.7, and 1.7, respectively. Overall, our data indicated that VPAT^EM could be used for studying the interactions between hVPAT and serotonin. Subsequently, we performed single-particle cryo-EM analysis on hVPAT^EM purified in the presence of added serotonin at pH 8.0 (Supplementary Table 1). We processed the cryo-EM images by using cryoSPARC, and improved the density map further by using spIsoNet, cryoFEM, and EMReady2, with the best spIsoNet map calculated to a nominal resolution of 3.6 Å at FSC_0.143.

Serotonin-binding sites in hVPAT

Our cryo-EM map revealed two prominent, tadpole-shaped density features within the substrate-binding pocket of the lumen-facing antiporter (Fig. 4a and Supplementary Fig. 7). These two density features (Fig. 4a), which we modeled as serotonin molecules, 5HT1 and 5HT2, are congruent with the size and shape of the monoamine neurotransmitter (Q score ≥ 0.52), and also in close proximity to numerous aromatic and hydrophobic amino acids, as well as the negatively charged E171^TM5 (calculated pK_a 6.7) and D255^TM7 (pK_a 5.9), while acknowledging that these pK_a predictions carry inherent limitations²⁵. In the deprotonated, serotonin-bound antiporter, the ethylamine group in 5HT1 forms a direct salt bridge with the side-chain carboxylate of E171 (Fig. 4b). In addition, the positively charged ethylamine in 5HT1 makes cation–π interactions with the side chain of W384, whereas the indole group from 5HT1 makes van der Waals interactions with the side chains of M48 and Y51.

Fig. 4. The serotonin-bound structure of hVPAT^EM.

a Two close-up views demonstrating the fitting of the two bound serotonin molecules (green sticks) into the cryo-EM density (green), contoured at 8.0 in ChimeraX v.1.8. The calculated Q scores are indicated for the two serotonin molecules, 5HT1 and 5HT2. b Detailed view of the binding sites for 5HT1 and 5HT2, with the serotonin molecules (green) and relevant protein residues shown as sticks, and a water molecule (W3) as a red sphere. Salt bridges, H-bonds, and cation–π interactions are highlighted as dashed lines in blue (distances in angstroms are specified if applicable), while the aromatic stacking or van der Waals interactions are highlighted by dashed lines in gray. c The overlay of the histamine- and serotonin-binding sites in hVPAT. The histamine (pink) and serotonin molecules (green), as well as the relevant protein residues, are shown as stick models. Red arrows indicated the structural differences discussed in the text. d Superimposition of the serotonin-binding sites onto the low-pH structure (gray). The serotonin molecules (green) and the relevant protein residues are illustrated as stick models. Red arrows indicate the structural differences, and red crosses highlight the steric clash between Y51/F251 at low pH and the otherwise bound serotonin molecules.

Furthermore, the indole groups from both 5HT1 and 5HT2 form edge-to-face aromatic stacking interactions between each other (Fig. 4b), stabilizing the binding of two substrates to the antiporter and rationalizing the cooperativity in serotonin uptake and binding. Besides 5HT1, 5HT2 also makes direct contact with the antiporter. In particular, the positively charged ethylamine group in 5HT2 forms a direct salt bridge with the side chain of D255^TM7, makes cation–π interactions with the aromatic side chain in F277^TM8, and interacts with the side chain of M280^TM8 through van der Waals interactions. Moreover, the side chain carboxylate in D255 is within H-bonding distance with the indole group from 5HT2, which is further buttressed by the side chain of F251^TM7 by virtue of van der Waals interactions.

The negatively charged side-chain carboxylate of E83^TM2 (predicted pK_a 6.4), by contrast, is too far to directly interact with any of the serotonin molecule in the lumen-facing state, with the closest approach (between E83 and 5HT1) being 7.6 Å (Fig. 4b). Notably, D140^TM4 (pK_a ~4.0), which lies outside the serotonin-binding sites, makes ionic interactions with the side chain of R137^TM4, in line with a fully deprotonated state. In addition, the side-chain guanidium group of R137 makes cation-π interactions with the side chain in F79^TM2, akin to their counterparts in the histamine-bound structure (Fig. 4c, d), depicting a shared feature of the monoamine-interacting hVPAT. Although R137 is not in direct contact with either histamine or 5HT, the removal of the side-chain guanidinium group from R137 moderately reduced the binding of monoamines (Fig. 3a, c and below), which may partly result from the loss of stabilizing cation–π interactions with F79 that is just one helical turn from the histamine-contacting E83 (Fig. 2b). Of note, although E83 makes no contact with 5HT in the lumen-facing structure (Fig. 4b), it may interact with the bound monoamine in cytosol-facing hVPAT¹⁵.

Functional study of the serotonin-binding sites

To confirm the functional significance of the protein residues described above, we replaced M48, Y51, F79, E83, R137, D140, E171, F251, D255, F277, M280, and W384 individually with alanine in hVPAT and examined their impact on serotonin uptake and binding activity. We found that all these mutations, barring M48A and M280A, had prominent adverse effects on the antiport function (Supplementary Tables 4 and 5), by markedly decreasing (>5-fold) the V_max (Fig. 5a, b) and concomitantly increasing the K_1/2 (≥5-fold) or K_d (≥9-fold, Fig. 5c, d). One notable exception is the mutant R137A, the K_d of which is ~2.4-fold that of the wild-type antiporter (Supplementary Table 5). These results thus validated the importance of the observed serotonin-antiporter interactions and further suggested that E83 and D140, although not in direct contact with the serotonin molecules (Fig. 4b), play indispensable roles in stabilizing the binding of serotonin. Furthermore, all tested mutants exhibited apparent Hill coefficients greater than 1.0, reinforcing our assertion that the cooperativity is mediated by the interactions between 5HT1 and 5HT2, rather than the protein residues flanking the ligand-binding sites.

Fig. 5. Mutagenesis study of the serotonin-binding sites.

a, b Radiolabeled serotonin transport kinetics of the liposome-reconstituted hVPAT variants. The transporter-mediated uptake of serotonin is driven by an outwardly directed proton gradient. Data are presented as mean values ± SD. Error bars represent SD (n = 3, i.e., three biological replicates were examined). The measured K_1/2, V_max, and Hill coefficient are listed in Supplementary Table 4. c, d Radiolabeled serotonin binding to the hVPAT variants at pH 8.0. Data are presented as mean values ± SD. Error bars represent SD (n = 3, i.e., three biological replicates were examined). The measured K_d, B_max, and Hill coefficient are listed in Supplementary Table 5.

The serotonin-bound structure of hVPAT^EM could be overlaid onto that of the histamine-bound form, resulting in an RMS deviation of 0.9 Å for 384 Cα positions (Fig. 4c) and highlighting structural similarity. Notably, the serotonin-contacting Y51, F251, F277, and W384 all assume different side-chain conformations from their histamine-binding counterparts, perceivably adapting the histamine-binding pocket to shelter a duo of bulkier serotonin molecules. Furthermore, unlike the antiporter-bound histamine molecules, which interact with each other via both aromatic stacking and cation-π interactions, the serotonin molecules contact each other without engaging the positively charged ethylamine groups, portraying a distinct mode of neurotransmitter binding.

In addition, we superimposed the serotonin-bound structure of hVPAT^EM onto the low pH form (PDB 9D7U)¹⁵, which led to an RMS deviation of 1.5 Å over 384 Cα atoms (Fig. 4d). We found that the serotonin-contacting Y51, F251, F277, and W384 make the largest structural changes as the lumen-facing antiporter becomes protonated. These changes would cause the loss of favorable antiporter-serotonin interactions (for F277 and W384) or steric clash between the particular protein residues (for Y51 and F251) and the bound substrates, explaining why a protonated antiporter disfavors the binding of serotonin. Furthermore, these structural changes likely follow or accompany the protonation of the membrane-embedded E171, D255, and D140, which would sever their favorable electrostatic interactions with 5HT1, 5HT2, and R137, respectively, showcasing a common theme in the coupling between H⁺ and monoamine neurotransmitters.

Proposed mechanism of neurotransmitter transport

Based on our structural and biochemical data, we deduced a mechanism for the hVPAT-mediated transport of histamine (Fig. 6). Specifically, in the lumen-facing, deprotonated hVPAT, E83, E171, and D255 make favorable medium- (3.9–4.3 Å) or short-range (2.8 Å) electrostatic interactions with the three histamine molecules, respectively. In addition, D140 and R137 form a salt bridge. Upon the protonation of E83, D140, E171, and D255, however, the interactions between hVPAT and histamines, as well as those between D140 and R137, are ruptured, releasing the histamine molecules from the antiporter as well as repositioning R137 toward the histamine-binding site, which readies the protonated, substrate-free antiporter to switch to its cytosol-facing conformation.

Fig. 6. Proposed mechanism of hVPAT-mediated transport of histamine.

The N- and C-domains of hVPAT are drawn as cyan and yellow ovals, the histamine- and H⁺-binding protein residues as yellow (deprotonated) or green (protonated) pentagons. R137, histamine, H⁺, and lipid bilayer are simplified as a red pentagon, a magenta triangle, a green droplet, and a grey rectangle, respectively. As the histamine-bound hVPAT (state I) becomes protonated, the neutralization of E83, E171, and D255 induces the dissociation of three histamine molecules, whereas R137 disengages from the protonated D140. Release of R137 toward the histamine-binding sites facilitates the release of substrates from the protonated hVPAT (state II) and discourages substrate rebinding due to charge-charge repulsion. The lumen-facing, protonated hVPAT subsequently switches to the cytosol-facing conformation (state III), and the binding of three histamine molecules to E83, E171, and D255 triggers the release of three protons. The binding of substrates to hVPAT (state IV) pushes R137 away from the substrate-binding site and promotes the formation of a salt bridge between R137 and D140, which triggers the release of a fourth proton into the cytosol and completes the transport cycle. Notably, the histamine-bound and low pH structures, which were obtained at pH 8.0 and 4.9, respectively, emulate states I and II, respectively. Importantly, to achieve a stoichiometric H⁺/monoamine coupling, hVPAT is either bound to the substrate or H⁺ during the transport cycle, making the deprotonated, substrate-free state functionally irrelevant.

Subsequently, the binding of three histamine molecules to E83, E171, and D255 triggers the deprotonation of their side-chain carboxylates, which is accompanied or followed by the deprotonation of D140 and the re-establishment of its ionic interactions with R137. In so doing, hVPAT exchanges four protons for a trio of histamine molecules, giving rise to a H⁺/substrate stoichiometry of 4:3. For the transport of serotonin (Supplementary Fig. 8), by contrast, only two substrate molecules are transported (Fig. 4b), yielding a H⁺/substrate ratio of 4:2. Notably, the competition between H⁺ and histamine or serotonin is both direct (for E83, E171, or D255), and indirect (for D140), with the transport cooperativity attributed to the interactions within the bound substrate molecules.

To obtain further support for this working mechanism, we examined the H⁺/substrate stoichiometry by measuring the reversal, or equilibrium membrane potential, which is defined as the membrane potential at which there is no net substrate flux^27–29. Specifically, for a transport process simplified as $n{H}_{in}^{+}+m{S}_{out}^{+}{\iff }_{VPAT}n{H}_{out}^{+}+m{S}_{in}^{+}$, i.e, hVPAT exchanges m molecules of monoamine neurotransmitters (S⁺) for n number of protons, the reversal potential can be calculated from Eq. (1) as follows^27,28.

$$E_{rev}=\frac{-E_0}{\frac{n}{m}-1}\left\{\frac{n}{m}log\frac{{\left[H^{+}\right]}_{in}}{{\left[H^{+}\right]}_{out}}-log\frac{{\left[S^{+}\right]}_{in}}{{\left[S^{+}\right]}_{out}}\right\}$$ 1

The derivation of Eq. (1) can be found in the “Methods” section. In essence, the reversal potential (E_rev) depends on the transport stoichiometry (n/m) as well as the intra- and extra-liposomal H⁺ and neurotransmitter concentrations, i.e., [H⁺] and [S⁺], respectively, whereas E₀ is ~58 millivolts at 20 °C (the experimental temperature). Thus, we analyzed the liposome-reconstituted hVPAT, clamped the membrane potentials²⁹ by using valinomycin and a number of transmembrane K⁺ gradients (Supplementary Table 6), and for each potential, we measured the hVPAT-mediated histamine (Supplementary Figs. 9 and 10) or serotonin (Supplementary Figs.11 and 12) transport rates under known H⁺ and substrate concentrations^27,28. Overall, our data were consistent with an H⁺/substrate stoichiometry of 4:3 for mono-cationic histamine³⁰ (Supplementary Fig. 9), or 4:2 for serotonin (Supplementary Fig. 11), further substantiating our formulated mechanism.

Discussion

Numerous clinically relevant transporters from the ABC, CDF, MFS, NSS, RND, and ZIP families contain two or more functionally important binding sites for substrates, i.e., mutations in any of the binding sites can abolish the transport function^31–36. Unlike the canonical, single-substrate transporters⁵, however, the mechanism by which a transporter recognizes or transports multiple substrate molecules remains mysterious. Our data suggested that hVPAT translocates two or three monoamine neurotransmitters simultaneously, via a mechanism fundamentally different from that of the NSS neurotransmitter transporters with a LeuT-fold, in which a second substrate allosterically regulates the transport of the first³⁴, or the I239T/G354E variant of an MFS multidrug transporter MdfA, which concurrently transports two substrates but there is no direct interaction between the two³⁷. Significantly, our cryo-EM structures revealed that there are extensive interactions between the hVPAT-bound monoamine substrates, leading to positive cooperativity. Such a cooperative mechanism, to our knowledge, has never been structurally characterized for any membrane transporter before.

This cooperativity enables the hVPAT-catalyzed transport rates to rise rapidly once the substrate concentrations surpass a threshold (~50 µM for monoamines). In so doing, hVPAT is well-positioned to protect the brain against the detrimental effects caused by excessive monoamine neurotransmitters, without interfering with the regular chemical neurotransmission, since at low (≤10 µM) neurotransmitter concentrations, the hVPAT-mediated transport of histamine or serotonin is minute and negligible. On the other hand, in the mast cells or platelets, where the histamine or serotonin concentrations are in the mM range, the ability of hVPAT to bind and transport monoamine neurotransmitters in addition to polyamine neuromodulators allows for a functional cross-talk between different signaling pathways, i.e., histamine or serotonin can regulate the storage and secretion of polyamines from the same cell, and vice versa⁹. To mediate such a cross-talk or signal integration, poly-specific transporters such as VPAT exhibit a mechanistic advantage over more substrate-selective transporters, including VMAT2, despite their colocalization within the same cells.

Moreover, hVPAT displays striking mechanistic flexibility, which is manifested in a wide range of H⁺/substrate stoichiometric ratios, i.e., 4:1 for spermine, 3:1 for spermidine¹⁵, 4:3 for histamine, and 4:2 for serotonin. This H⁺/substrate stoichiometry appears to be dictated by the number of substrate molecules that can spatially occupy the substrate-binding pocket, the ability of substrate to form favorable ionic contacts with the membrane-embedded acidic protein residues, and the capability of substrate molecules to interact with each other, which collectively place restraints and limitations on the size, shape, and chemical structures of the ligands, conferring substrate poly-specificity. As such, hVPAT lies at the intersection between its monoamine-specific SLC18 relatives^3,4 and the distantly related but rather promiscuous, bacterial MFS transporter MdfA³⁷.

One of the shared structural features between the SLC18 antiporters and MdfA is signified by R137 in hVPAT and its equivalent residues in others, although their mechanistic roles have diverged substantially during evolution, even among the human SLC18 members. Indeed, although VMAT1 and VMAT2 transport monoamine neurotransmitters, they differ from hVPAT in their substrate recognition or transport mechanism. Firstly, VMAT1 and VMAT2 bind and transport only a single histamine or serotonin molecule per cycle, with a fixed H⁺/substrate stoichiometry of 2:1. This difference can be partly attributed to the physical size of the substrate-binding pocket, with those of VMAT1 and VMAT2 being substantially less voluminous than that of hVPAT, although all the SLC18 antiporters contain no fewer than four membrane-embedded acidic residues. Secondly, the substrate specificity of VMAT1 and VMAT2 stems from the interactions between the antiporter and neurotransmitter, whereas that of hVPAT also arises from the ability of the monoamine substrates to interact with each other. Thirdly, in VMAT1 or VMAT2, two acidic residues (VMAT1^E320 and VMAT1^D407) make direct contacts with the bound serotonin, whereas only one (E171 or D255) interacts with each substrate molecule in hVPAT, which partly explains why VMAT1 or VMAT2 binds monoamine neurotransmitters with higher affinity and specificity than hVPAT. Of note, D255 in VPAT is equivalent to VMAT1^E320 or VMAT2^E312 (Supplementary Fig. 1), although previous studies showed that the protonation state of VMAT2^E312 exerts little effect on the dissociation of serotonin¹⁷.

As such, the human SLC18 antiporters exhibit substantial mechanistic differences in their neurotransmitter recognition or transport mechanism, highlighting the multitasking capacity and substrate poly-specificity of hVPAT as a dual monoamine- and polyamine- transporter. The structural characteristics outlined above evidently equip hVPAT to underpin not only chemical neurotransmission but also immune crosstalk or platelet function, supporting hVPAT as a broad regulator of polyamine and monoamine homeostasis rather than solely a mediator of neurotransmission. Accordingly, our findings expand the functional scope of hVPAT, reinforcing its physiological role beyond the central nervous system, particularly in modulating immune response or coagulation. In the mast cells, for instance, hVPAT can function as a transporter of histamine, polyamine, or even serotonin, which may allow this antiporter to sense, integrate, and respond to complex environmental and physiological cues during host defense.

Methods

Membrane protein expression

The codon-optimized genes encoding human VPAT (hVPAT) and its variants were synthesized by GenScript (NJ, USA) and cloned into the expression vector pFastBac1 at the restriction sites EcoR1 and XhoI¹⁵. For functional studies, the hVPAT variants were fused with a thrombin-cleavable, C-terminal FLAG (i.e., DYKDDDDK) and deca-histidine dual affinity tags. For cryo-EM studies, the mVenus-hVPAT-nanobody fusion protein was attached to thrombin-cleavable, C-terminal affinity tags (FLAG and deca-histidine).

Bacmids were produced in the DH10Bac cells, and baculoviruses were generated and propagated in Sf9 cells (Fisher Scientific, #A35243) by using the Bac-to-Bac expression system (Fisher Scientific)¹⁵. The P0 (Passage 0) virus was generated by using 5−10 µg bacmid DNA with 20 µl ExpiFectamine in 200 µl Grace’s medium for ~2 million cells plated in 3 ml of Grace’s medium in a cell culture flask. After incubation for 4 days at 27 °C, the supernatant was collected and supplemented with 10% (v/v) fetal bovine serum (FBS, Cytiva HyClone). The P1 baculovirus was produced by adding 1 ml P0 virus to 30 ml (~1 million/ml) Sf9 cells in Grace’s medium (Gibco) supplemented with 10% (v/v) FBS, followed by incubation at 27 °C for 3–4 days.

The P1 baculovirus was used for membrane protein expression¹⁵, and the amount of virus used was optimized based on small-scale protein expression with 40-ml Tni (High Five, Fisher Scientific, #B85502) cells cultured in the Express Five medium (Fisher Scientific). The large-scale protein expression was usually performed by infecting Tni cells (at ~1 million/ml) with a 1:500–1:1000 (v/v) ratio (virus vs. insect cells). After 48 h of incubation at 27 °C, the Tni cells were harvested by centrifugation at 6000 × g for 15 min, 4 °C, washed with fresh medium, centrifuged again, and stored at −80 °C before use. Cells were not tested for mycoplasma or authenticated, which were typically purchased <1 month before protein expression.

Membrane transporter purification

All the chemicals were purchased from Sigma-Aldrich, Millipore, and all the protein purification steps were at 4 °C unless stated otherwise¹⁵. Before protein purification, the Tni cells were resuspended into the buffer containing 100 mM TRIS-HCl, pH 8.0, 10 mM MgCl₂, 150 mM KCl, 1 mM TCEP, 1 µg/ml DNase I and protease inhibitors, which included 500 µM AEBSF, 2 µM E-64, 2 µM leupeptin, 2 µM pepstatin A, 1 µM aprotinin, and 0.5 mM benzamidine. The insect cells were ruptured by ≥2 passages through a cell disruptor (Avestin Emulsiflex-C3), and the membranes were obtained by centrifugation at 100,000 × g for 2 h. The membranes were resuspended into the buffer containing 50 mM TRIS-HCl, pH 8.0, 10 mM MgCl₂, 150 mM KCl, 1 mM TCEP, 500 µM AEBSF, 2 µM E-64, 2 µM leupeptin, 2 µM pepstatin A, 1 µM aprotinin, and 0.5 mM benzamidine. Detergent glyco-diosgenin (GDN, Anatrace) and cholesteryl hemisuccinate (CHS, Anatrace) were subsequently added to 0.5% (w/v) and 0.1 (w/v), respectively, and the sample was thoroughly mixed for 1 h. For cryo-EM studies, freshly prepared histamine or serotonin was added to 1 mM at pH ~8.0 during membrane solubilization and protein purification. At pH 8.0, both histamine³⁰ and serotonin⁴ are mono-cationic.

The insolubilized material was removed by centrifugation at 100,000 × g for 1 h, and the supernatant was mixed with Ni-NTA resin (1 ml for ~20 g of cell membranes) for 1 h, which was equilibrated in 20 mM TRIS-HCl, pH 8.0, 150 mM NaCl, 1 mM TCEP, 500 µM AEBSF, 2 µM E-64, 2 µM leupeptin, 2 µM pepstatin A, 1 µM aprotinin, 0.025% (w/v) GDN, and 0.005 (w/v) CHS. The Ni-NTA resin was washed with 20 column volumes of wash buffer containing 20 mM TRIS-HCl, pH 8.0, 500 mM NaCl, 1 mM TCEP, 0.025% (w/v) GDN, and 0.005 (w/v) CHS. For the transport assay¹⁵, thrombin in one column volume of equilibration buffer containing 20 mM TRIS-HCl, pH 8.0, 150 mM NaCl, 1 mM TCEP, 0.025% (w/v) GDN, and 0.005 (w/v) CHS (10 units for ~20 g of cell membrane), was mixed with the protein-bound Ni-NTA resin, which was incubated for 1 h. The Ni-NTA resin and buffer mixture was centrifuged at 5000 × g for 30 min, and the supernatant was passed through a column packed with anti-FLAG G1 resin (Genscript) in the equilibration buffer. The flow-through was concentrated (MWCO 100 kd) to ~100 µl and centrifuged at 50,000 × g for 1 h. The supernatant was subsequently run over a 24-ml Superdex 200 Increase column (Cytiva) at 0.4 ml/min in the equilibration buffer. The peak fractions were pooled and used for proteo-liposome reconstitution.

For the ligand binding assay¹⁵, the protein was eluted from the protein-bound Ni-NTA resin with two column volumes of 20 mM TRIS-HCl, pH 8.0, 100 mM NaCl, 450 mM imidazole, 1 mM TCEP, 0.025% (w/v) GDN, and 0.005 (w/v) CHS. The eluate was passed through a 120-ml Sephadex G50 (crude) column in the equilibration buffer. The sample was concentrated to ~100 µl, centrifuged at 50,000 × g for 1 h, and run over a 24-ml Superdex 200 Increase column at 0.4 ml/min in the equilibration buffer. The peak fractions were pooled and used for the radio-ligand binding assay. For the cryo-EM study¹⁵, the tag-free hVPAT^EM eluted from the Ni-NTA resin was incubated with amphipol PMAL-C8 (Anatarce) at a 1:30 (w/w) ratio for 2 h at 4 °C ¹⁵. Then BioBeads SM2 (BioRad) at a 1:50 (w/w) ratio was added, and the protein sample was incubated for 8 h at 4 °C. BioBeads SM2 were removed by centrifugation at 10,000 × g for 15 min, 4 °C. The sample was concentrated to ~100 µl and purified by using a 24-ml Superdex 200 Increase column (Cytiva) in 20 mM TRIS-HCl, pH 8.0, 150 mM NaCl, 1 mM TCEP, and 1 mM histamine or serotonin (pH ~8.0).

Radiolabeled histamine and serotonin transport assay

The detergent-solubilized hVPAT variants were reconstituted into liposomes^1,15. Briefly, the purified proteins (~100 µg) were mixed with n-decyl-β-D-maltoside, or DM (Anatrace, final concentration 1%, w/v) and brain polar lipid extract (porcine, Avanti Polar Lipids) at a final concentration of 0.5 mg/ml, and incubated for 30 min at 4 °C. The sample was centrifuged at 20,000 × g for 15 min, 4 °C, and the supernatant was mixed with 1 ml reconstitution mixture containing 20 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1 % (w/v) DM, 10 mg/ml mammalian brain polar lipids, and 3 mg/ml asolectin. The sample was sonicated in a bath sonicator (Misonix) to clarity, and then dialyzed (MWCO 50 kd) four times at 4 °C, each time against 1 L buffer containing 20 mM Tris-HCl, pH 8.0, 150 mM NaCl. The proteo-liposome mixture was diluted 25 times in intra-liposomal buffer typically containing 50 mM MES-NaOH, pH 5.0, 150 mM NaCl, unless otherwise specified (for experiments with valinomycin), and centrifuged at 100,000 × g for 1 h, 4 °C. The proteo-liposome pellet was resuspended into 200 µl intra-liposomal buffer, frozen in liquid nitrogen, and kept at −80 °C before use.

Histamine or serotonin transport assay was performed as follows^1,15. In brief, each proteo-liposome sample was thawed, sonicated to clarity, and purified by using a 20-ml Sephadex G50 (fine) column in 50 mM MES-NaOH, pH 5.0, 150 mM NaCl, unless otherwise specified. The peak fractions containing the proteo-liposomes were collected and concentrated (MWCO 100kd) to ~1 mg protein/ml. For experiments performed in the absence of valinomycin-imposed K⁺ diffusion potential, the histamine or serotonin uptake assay was performed in reaction buffer containing 50 mM Tris-HCl, pH 8.0, 150 mM NaCl. For experiments involving the valinomycin-imposed K⁺ diffusion potential²⁹, the intra-liposomal and reaction buffers were specified in Supplementary Table 6 and also the legends for Supplementary Figs. 9 and 11. Of note, prior to the reversal potential measurement, the proteo-liposomes were centrifuged at 100,000 × g for 1 h at 4 °C, resuspended in the corresponding buffer, subjected to three cycles of freeze-thaw for buffer exchange^27,28, and finally extruded through a 400-nm polycarbonate membrane (Whatman). Proteo-liposomes (~5 µl per reaction) were diluted into 150 µl of reaction buffer, and the transport was initiated by the addition of various concentrations of unlabeled histamine or serotonin, [³H] histamine (25 Ci/mmol, American Radiolabeled Chemicals), or [³H] serotonin (80 Ci/mmol, American Radiolabeled Chemicals), at 20 °C.

The histamine or serotonin concentrations were varied for the transport kinetic study, and 30-µl aliquots of the reaction mixture were taken 2 min after the reactions started at 20 °C, which was within the linear range of transport rates. The samples were centrifuged through a 1.5-ml Sephadex G50 (fine) micro-spin column at 1000 × g for 1 min, 20 °C. The eluate was transferred to a scintillation vial, to which 5 ml scintillation liquid (Perkin Elmer) was added. Radioactive decay of the isotope-labeled histamine or serotonin was recorded by using a liquid scintillation counter (Perkin Elmer). For each substrate concentration, radioactive decay values from the protein-free liposomes were subtracted from the respective proteo-liposome values. The resulting values were fitted by using nonlinear regression to the “allosteric sigmoidal enzyme kinetics”, which is basically the Michaelis–Menten kinetics model with a variable Hill slope to account for potential cooperativity¹⁸; and the kinetic parameters were calculated by using the GraphPad Prism v10. All experiments were repeated three times, and the error bars represent standard deviations of the mean. We selected the allosteric sigmoidal enzyme kinetics model over the standard Michaelis–Menten kinetics with a fixed Hill slope of 1.0, based on its superior visual agreement between the data points and fitted curve, improved goodness of fit statistics, and better P values calculated from the replicate test. Furthermore, a statistical comparison of the two analysis methods yielded a P value that favored the allosteric sigmoidal kinetics, i.e., Michaelis–Menten kinetics with a Hill slope greater than 1.0.

Histamine and serotonin binding assay

Radioligand binding assay was performed¹⁵ on the detergent-purified, His-tagged hVPAT variants (~0.5 mg), which were mixed with Ni-NTA resin (~500 µl) in 5 ml binding buffer containing 50 mM TRIS-HCl, pH 8.0, 0.5 M NaCl, 0.5 mM TCEP, 0.025 % GDN, and 0.005% CHS. The samples were incubated for 30 min at 4 °C and the unbound material was discarded after centrifugation at 1000 × g for 3 min, 4 °C. The Ni-NTA resin was subsequently washed with the binding buffer and divided into 200 µl aliquots, which were incubated for 30 min at 4 °C with increasing concentrations of unlabeled histamine or serotonin, and [³H] histamine (25 Ci/mmol, American Radiolabeled Chemicals), or [³H] serotonin (80 Ci/mmol, American Radiolabeled Chemicals).

An aliquot of 180 µl of the Ni-NTA resin with histamine or serotonin was transferred to a micro-spin column and centrifuged at 10,000 × g for 1 min, 4 °C. Unbound material was discarded, and the resin was resuspended in 100 µl of buffer containing 50 mM TRIS-HCl, pH 8.0, 100 mM NaCl, 0.5 mM TCEP, 450 mM imidazole, 0.025 % GDN, and 0.005% CHS. The Ni-NTA resin was centrifuged at 10,000 × g for 1 min, 4 °C, and the radioactivity of the flow-through was measured by using liquid scintillation. The amount of [³H] histamine or serotonin associated with the Ni-NTA resin at each ligand concentration in the absence of added protein was subtracted from the measurements made in the presence of the hVPAT variant. The resulting values were fitted by using nonlinear regression to “specific binding with Hill slope”, employing the GraphPad Prism v10. All experiments were repeated three times, and the error bars represent standard deviations of the mean.

Derivation of the reversal potential equation for hVPAT

hVPAT, an H⁺-coupled antiporter (exchanger), transports histamine or serotonin in the (+1) charge state (see below for a discussion on histamine). We describe the hVPAT-mediated transport as shown below:

$$\mathrm{n}{\mathrm{H}}_{\mathrm{in}}^{+}+\mathrm{m}{\mathrm{S}}_{\mathrm{out}}^{+}{\iff }_{\mathrm{VPAT}}\mathrm{n}{\mathrm{H}}_{\mathrm{out}}^{+}+\mathrm{m}{\mathrm{S}}_{\mathrm{in}}^{+}$$

where S⁺ is the positively charged histamine or serotonin, the subscripts out and in indicate H⁺ or histamine/serotonin outside or inside the proteo-liposomes, n is the number of protons transported per cycle and m is the corresponding number of histamine or serotonin molecules. We aim to determine n/m, the H⁺/substrate stoichiometric ratio for hVPAT. For this transport process:

$${\mathrm{\mu }}_{H^{+}in}={\mathrm{\mu }}_{H^{+}}^0+RTln{\left[H^{+}\right]}_{in}+z_{H^{+}}F{\Psi }_{in}$$

$${\mathrm{\mu }}_{H^{+}out}={\mathrm{\mu }}_{H^{+}}^0+RTln{\left[H^{+}\right]}_{out}+z_{H^{+}}F{\Psi }_{out}$$

$$\mathrm{\Delta }{\mathrm{\mu }}_{H^{+}}={\mathrm{\mu }}_{H^{+}in}-{\mathrm{\mu }}_H^{+}out$$

$$\mathrm{\Delta }{\mathrm{\mu }}_{H^{+}}=RTln\frac{{\left[{\mathrm{H}}^{+}\right]}_{in}}{{\left[{\mathrm{H}}^{+}\right]}_{out}}+z_{H^{+}}F\mathrm{\Delta }\Psi$$

Similarly,

$$\mathrm{\Delta }{\mathrm{\mu }}_{S^{+}}=RTln\frac{{\left[S^{+}\right]}_{in}}{{\left[S^{+}\right]}_{out}}+z_{S^{+}}F\mathrm{\Delta }\Psi$$

Among the above equations, µ is the chemical potential of H⁺ or the monoamine neurotransmitter, µ⁰ is the corresponding standard state chemical potential, R is the universal gas constant, T is the absolute temperature, F is the Faraday constant, z is the ion’s valence or charge (z_H⁺ = z_S⁺ = 1), and $\mathit{△}\mathit{\Psi }$ is the voltage difference across the liposomal membrane, i.e., $\mathit{△}\mathit{\Psi }={\mathit{\Psi }}_{\mathit{in}}-{\mathit{\Psi }}_{\mathit{out}}$.

When there is no net flux, i.e., the net antiport rate would become zero, then

$$\left(n{\mathrm{\mu }}_{H^{+in}}-n{\mathrm{\mu }}_{H^{+}out}\right)-\left(m{\mathrm{\mu }}_{S^{+}in}-m{\mathrm{\mu }}_{S^{+}out}\right)=n\mathrm{\Delta }{\mathrm{\mu }}_{H^{+}}-m\mathrm{\Delta }{\mathrm{\mu }}_{S^{+}}=0$$

or,

$$n\mathrm{\Delta }{\mathrm{\mu }}_{H^{+}}=m\mathrm{\Delta }{\mathrm{\mu }}_{S^{+}}$$

Therefore,

$$n\left\{RTln\frac{{\left[H^{+}\right]}_{in}}{{\left[H^{+}\right]}_{out}}+z_{H^{+}}F\mathrm{\Delta }\Psi \right\}=m\left\{RTln\frac{{\left[S^{+}\right]}_{in}}{{\left[S^{+}\right]}_{out}}+z_{S^{+}}F\mathrm{\Delta }\Psi \right\}$$

After we set z_H⁺ =1 and z_S⁺ =1, we can rearrange the above equation as follows,

$$\left(m-n\right)F\mathrm{\Delta }\Psi =nRTln\frac{{\left[H^{+}\right]}_{in}}{{\left[H^{+}\right]}_{out}}-mRTln\frac{{\left[S^{+}\right]}_{in}}{{\left[S^{+}\right]}_{out}}$$

Thus, the equilibrium or reversal potential E_rev can be calculated as follows,

$$E_{rev}=\mathrm{\Delta }\Psi =\left[\frac{mRT}{\left(m-n\right)F}\right]\left\{\frac{n}{m}ln\frac{{\left[H^{+}\right]}_{in}}{{\left[H^{+}\right]}_{out}}-ln\frac{{\left[S^{+}\right]}_{in}}{{\left[S^{+}\right]}_{out}}\right\}$$

Or

$$E_{rev}=\mathrm{\Delta }\Psi =\left[\frac{2.303mRT}{\left(m-n\right)F}\right]\left\{\frac{n}{m}log\frac{{\left[H^{+}\right]}_{in}}{{\left[H^{+}\right]}_{out}}-log\frac{{\left[S^{+}\right]}_{in}}{{\left[S^{+}\right]}_{out}}\right\}$$

We define E₀ as 2.303RT/F, which is ~58 millivolts (or mV) at 20^oC, and the above equation can be simplified as follows,

$$E_{rev}=\frac{m{E}_0}{m-n}\left\{\frac{n}{m}log\frac{{\left[H^{+}\right]}_{in}}{{\left[H^{+}\right]}_{out}}-log\frac{{\left[S^{+}\right]}_{in}}{{\left[S^{+}\right]}_{out}}\right\}$$

Therefore,

$$E_{rev}=\frac{-E_0}{\frac{n}{m}-1}\left\{\frac{n}{m}log\frac{{\left[H^{+}\right]}_{in}}{{\left[H^{+}\right]}_{out}}-log\frac{{\left[S^{+}\right]}_{in}}{{\left[S^{+}\right]}_{out}}\right\}$$ 1

Histamine may carry a (+1) or (+2) charge over the pH range used in this study. Specifically, the experimentally determined pK_a values for the imidazole and ethylamine moieties in histamine at 20 °C (the experimental temperature) are 6.14 and 9.93, respectively³⁰. Based on these pK_a values and our experimental conditions specified in Supplementary Fig. 9, the concentrations of extra-liposomal, mono- and di-cationic histamine are 10.5 and 1.5 µM, respectively, and those of the intra-liposomal forms are 210 and 290 µM, respectively.

Assuming mono-cationic histamine is the transported substrate, the calculated reversal potentials were listed in Supplementary Fig. 9. If, however, di-cationic histamine is the substrate, then the expected reversal potentials would be −84.1, −75.4, −71.1, and −68.4 mV for the H⁺/substrate stoichiometries of 4:3, 3:3, 2:3, and 1:3, respectively. Notably, the reversal potential for di-cationic histamine is calculated according to Eq. (2) as follows:

$$E_{rev}=\frac{-E_0}{\frac{n}{m}-2}\left\{\frac{n}{m}log\frac{{\left[H^{+}\right]}_{in}}{{\left[H^{+}\right]}_{out}}-log\frac{{\left[S^{2+}\right]}_{in}}{{\left[S^{2+}\right]}_{out}}\right\}$$ 2

Based on our data shown in Supplementary Figs. 9 and 10, the observed reversal potential for histamine is supportive of an H⁺/substrate stoichiometry of 4:3, with mono-cationic histamine being the substrate. This result is concordant with a cytosolic pH of 7.4, whereat histamine is ~95% (at 20 °C) or ~97% (at 35 °C) mono-cationic³⁰.

Cryo-EM grid preparation and data collection

The gel-filtration peak fractions containing the tag-free hVPAT^EM were concentrated (MWCO 100 kd) to ~3 mg/ml. Four microliters of the protein sample was then added to freshly glow-discharged holey carbon, Quantifoil R1.2/1.3 300-mesh copper grids (EM Sciences)¹⁵. The cryo-EM grids were blotted for 4 s at 4 °C and under 100% chamber humidity, and plunge-frozen into liquid ethane by using a Vitrobot Mark IV (FEI). Afterwards, these grids were loaded into a Titan Krios (FEI) electron microscope operating at 300 kV, which is equipped with a K3 Summit detector (Gatan) in CDS mode with a BioContinuum GIF energy filter (20 eV slit width) at the Hormel Institute, University of Minnesota. Data were collected by using the EPU software (Fisher Scientific) with a pixel size of 0.664 Å (magnification of ×130,000) and a nominal defocus value ranging from −1.0 to −2.0 µm. Each movie consists of 40 dose-framed fractions and was recorded with a dose rate of ~32 e⁻/Ų/s and a total dose of ~51.2 e⁻/Ų. Cryo-EM data collection parameters and statistics are all listed in Supplementary Table 1.

Cryo-EM image processing

A total of 8118 and 6534 dose-fractionated movies were collected for samples incubated with histamine and serotonin, respectively. The images were processed by using the patch-based beam-induced motion correction software MotionCor2³⁸ and contrast transfer function (CTF) estimation program CTFFIND-4.1.13³⁹, as implemented in cryoSPARC v4.5.1²⁰. Images with defocus values outside the range −0.4 to −3.2 µm or CTF fit resolutions worse than 7 Å were excluded from further data processing. The cryo-EM data processing workflow is exemplified in Supplementary Fig. 3.

For the histamine-treated protein sample, a total of 8,266,630 particles were picked by using both the Blob picker and the Template picker and subjected to the Remove Duplicate Particles Tool. Junk particles were removed through two rounds of 2D classifications. A total of 1,413,303 particles from the good 2D classes were used for ab initio reconstruction of four or three maps and further heterogeneous refinement (3D classification). Three rounds of 3D classification generated 162,629 particles for the third round of 2D classification to continuously remove bad particles. A total of 127,728 particles from the good class were subjected to further non-uniform and CTF refinements to generate a final map at a nominal resolution of 3.93 Å (Supplementary Fig. 3).

Notably, map resolution was estimated by employing the gold-standard Fourier shell correlation (FSC) threshold at 0.143, i.e., FSC_0.143, between the two independently calculated half-maps from CryoSPARC. Local resolution variation was estimated from the two half-maps by using CryoSPARC or Phenix⁴⁰. The two independent half maps produced from CryoSPARC were improved by using the software spIsoNet²¹ to correct the data anisotropy. Of note, the two half maps were independently processed by spIsoNet to allow for valid FSC calculation and the estimation of resolutions. We also attempted to use spIsoNet and RELION4⁴¹ to correct the particle misalignment owing to data anisotropy²¹, but to no avail, i.e., no improvement in the resulting density maps. The cryo-EM images obtained from the serotonin-incubated sample were processed similarly to those for the histamine-treated one, with data collection and processing details provided in Supplementary Table 1.

Cryo-EM post-processing and structure refinement

Starting from the two independent half maps obtained from spIsoNet, we used various map sharpening or enhancement software from the CCP-EM suite, e.g., LocScale^42,43 and Phenix, e.g., Autosharpen, Resolve, Local-aniso-sharpen^44–46, as well as the deep-learning-based software CryoFEM²², EMReady2²³, and DeepEMhancer⁴⁷, to improve the density features. We evaluated the improvement of maps, or the lack thereof, based on the quality of the protein density features, the performance of structure refinement (protein stereochemistry, model-to-map CC and FSC, etc.), and the Q scores²⁴. According to this evaluation, CryoFEM and EMReady2 were chosen to calculate the final cryo-EM maps for model building and data deposition.

The published model of spermine-bound hVPAT^EM (PDB 9D7X) was placed into the cryo-EM maps by using Phenix, which was followed by iterative manual model rebuilding using the software Coot⁴⁸. Ligands were placed into the cryo-EM density maps by using the software ChemEM⁴⁹. Ligand restraints were generated by using eLBOW⁵⁰, as implemented in Phenix. Both structures were refined in real space with secondary structure and Ramachandran restraints by using Phenix⁴⁰. We used both the unmodified experimental cryo-EM maps and the enhanced maps for model assessment. In comparison with the unmodified maps, we found that the enhanced density maps generally yielded structures with better protein stereochemistry, higher model-to-map CC values or Q scores (calculated for the same unmodified map). Therefore, we selected the enhanced maps for figure preparation and data deposition.

The validation of both structural models was performed by using MolProbity⁵¹ implemented in Phenix. All the structural figures were prepared using ChimeraX⁵². During structure refinement, we also randomly altered the coordinates of each rebuilt model by 0.5 Å by using the pdb tools in Phenix, and refined this “shaken” model against one of the unmodified experimental half maps produced by CryoSPARC¹⁵. We compared the refined structure with the other half map to ensure that our model adjustment is consistent with the un-enhanced, experimental maps¹⁵. The cryo-EM density for mVenus or nanobody in this study is generally weaker than that for the antiporter; thus, only the transporter model was refined and deposited to ensure circumspect handling of the data. To avoid over-interpretation of the density maps, the following residues in hVPAT were not included in the deposited models due to weak density, which was likely caused by the conformational flexibility within the luminal loop between TM1 and TM2: residues 56-71. Moreover, the cryo-EM density for residues D140 and D255 in the histamine-bound structure; residues E83, E171, D255, and serotonin 5HT1 in the serotonin-bound form is suboptimal; therefore, their rotamer assignments remain uncertain. Notably, the relatively weak cryo-EM or electrostatic potential density observed for the Asp and Glu side-chain carboxylate groups likely arises from their deprotonation under our experimental conditions at pH 8.0 and/or susceptibility to radiation damage^53–55.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

M.L. conceived the study. M.L. and Y.G. prepared the protein samples and carried out the biochemical studies. B.L. and G.Y. prepared the cryo-EM grids and collected and analyzed the cryo-EM data. M.L. and J.C. performed cryo-EM map post-processing and structure refinement. J.S. contributed to data analysis. M.L. wrote the manuscript with input from B.L.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Funding

This work was in part supported by the NIH (R01GM145642 to M.L.), the Hormel Institute, University of Minnesota (to B.L.), and the DOE (KC0304000 to J.S.).

Data availability

The cryo-EM density maps (experimental half maps as well as the enhanced maps) have been deposited in the Electron Microscopy Data Bank (EMDB) under accession codes: EMDB-73561 [https://www.ebi.ac.uk/pdbe/entry/emdb/EMD-73561] (the histamine-bound structure); and EMDB-73562 [https://www.ebi.ac.uk/pdbe/entry/emdb/EMD-73562] (the serotonin-bound structure). The atomic coordinates have been deposited in the Protein Data Bank (PDB) under accession codes: PDB9YWU (the histamine-bound structure); and PDB9YWV (the serotonin-bound structure). The atomic coordinates that we used for structure analysis also included PDB9D7X (the spermine-bound structure of human VPAT); PDB9D7U (the low pH structure of human VPAT). The source data underlying Figs. 3 and 5, Supplementary Figs. 2, 6, 9, 10, 11, and 12 are provided alongside this manuscript as a Source Data file. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-75564-x.