Substrate-dependent accessibility changes in the Na+-dependent C4-dicarboxylate transporter, VcINDY, suggest differential substrate effects in a multistep mechanism

Members of the divalent anion sodium symporter (DASS) family (SLC13 in humans) play critical roles in metabolic homeostasis, influencing many processes including fatty acid synthesis, insulin resistance, adiposity, and lifespan determination. DASS transporters catalyse the Na+-driven concentrative uptake of Krebs cycle intermediates and sulfate into cells; disrupting their function can protect against age-related metabolic diseases and can extend lifespan. An inward-facing crystal structure and an outward-facing model of a bacterial DASS family member, VcINDY from Vibrio cholerae, predict an elevator-like transport mechanism involving a large rigid body movement of the substrate binding site. How substrate binding influences the conformational state of VcINDY is currently unknown. Here, we probe the interaction between substrate binding and VcINDY conformation using a site-specific alkylation strategy to probe the solvent accessibility of several broadly distributed positions in VcINDY in the presence and absence of substrates (Na+ and succinate). Our findings reveal that accessibility to all positions tested can be modulated by the presence of substrates, with the majority becoming less accessible in the presence of Na+. Mapping these solvent accessibility changes onto the known structures of VcINDY suggests that Na+ binding drives the transporter into an as-yet-unidentified intermediate state. We also observe substantial, separable effects of Na+ and succinate binding at several amino acid positions suggesting distinct effects of the two substrates. Furthermore, analysis of a solely succinate-sensitive residue indicates that VcINDY binds its substrate with a low affinity and proceeds via an ordered process in which one or more Na+ ions must bind prior to succinate. These findings provide insight into the mechanism of VcINDY, which is currently the only structural-characterised representative of the entire DASS family.


Introduction
The divalent anion sodium symporter (DASS) family of transporters are present in all domains of life and are responsible for the transport of several key compounds into cells, such as citrate, Krebs cycle intermediates and sulfate 1 . Cytoplasmic citrate plays a major role in the metabolism of eukaryotic cells, contributing to the regulation of fatty acid, cholesterol and low-density lipoprotein synthesis [2][3][4][5] . By maintaining and controlling the cytoplasmic citrate concentration, members of the DASS family (Transport Classification Database no. 2.A.47, SLC13 in humans) are key players in metabolic regulation in eukaryotes as demonstrated by the phenotypes associated with their functional disruption. Knockdown of a DASS family member in fruitflies and nematodes leads to phenotypes analogous to caloric restriction, most notably a substantial increase in the lifespan of the organism, hence the alternative name for this family, INDY, which stands for I'm not dead yet 6 .
In mice, knockout of a DASS family member (NaCT, SLC13A5 in humans) leads to protection against adiposity and insulin resistance 7 , and knockout of the equivalent transporter in human hepatocarcinoma cells substantially reduced hepatoma cell proliferation and colony formation 8 . In addition to metabolic diseases, mutations in a DASS family member (NaCT, SLC13A5) are associated with neurological disorders including developmental delay and early onset epilepsy 9 .
Thus, DASS family members are prime targets for therapeutics designed to tackle metabolic diseases including diabetes and obesity, liver cancer and neurological disorders.
DASS transporters are ion-coupled secondary active transporters. Secondary transporters can harness the energy stored in ion gradients (usually Na + or H + ) across the membrane to drive the energetically uphill movement of substrate against its concentration gradient. Secondary active transporters occupy at least two major conformational states, the inward-facing state (IFS) and the outward-facing state (OFS), which alternate in order to expose the substrate binding site from the cytoplasmic to the extracellular side of the membrane, and vice versa.
The DASS transporter family belongs to the Ion Transporter (IT) Superfamily 10,11 , and the majority of characterised DASS transporters are known to transport their anionic substrates coupled to the co-transport of multiple Na + ions 12 13-17 . All of our current structural and mechanistic understanding of the DASS family comes from studies on a bacterial family member, VcINDY, from Vibrio cholerae, which is the only DASS transporter for which we have high resolution structural information 18,19 .
Functional characterisation of VcINDY reveals it preferentially transports C4-dicarboxylates, for example, succinate, malate and fumarate 17 . The model VcINDY substrate, succinate, is transported in its dianionic form coupled to the transport of three Na + ion 17,20 . Our lab and others have revealed that VcINDY shares structural and functional characteristics with its mammalian homologues, suggesting that insight derived from the mechanism of VcINDY is directly applicable to the mammalian transporters 17,20,21 . The x-ray structures of VcINDY reveal that it forms a homodimer and each protomer consists of two domains; the scaffold domain that forms dimer interface contacts, and the transport domain that houses key substrate binding residue (Fig. 1A) 18,19 . All structures of VcINDY to date are captured in the same IFS conformation where the substrate is exposed to the cytoplasmic side of the membrane suggesting that the crystallisation conditions select for this lowest free-energy state conformation of VcINDY 18,19 . Using a combination of symmetry-based structural modelling and extensive biochemical and biophysical validation, we have described an OFS conformation of VcINDY in which the transport domain and the substrate binding site are vertically translocated through the membrane (Fig. 1A) 22 . Combined, the IFS structure and OFS model predict that VcINDY employs an elevator-like mechanism to achieve alternating access to the substrate binding site from both sides of the membrane.
The structures of VcINDY reveal that the substrate binding site within the transport domain is composed of backbone and side chain contacts from the tips of two re-entrant hairpin loops and an unwound helix (TM 7) 18 . This organisation is reminiscent of other elevator-like ion-driven transporters where local conformational changes of hairpin loops are required for ligand binding, controlling coupling ion and substrate access to the binding site, and as a means of preventing uncoupled transport 23,24 . The transport of substrate and coupling ion is tightly coupled, meaning that Na + transport cannot occur without succinate transport, and vice versa. In a simple kinetic scheme for VcINDY; succinate 2and 3 Na + ions bind to the OFS (Fig 1 B, step 1), facilitating a conformational change into the IFS via an intermediate occluded state that is yet to be structurally characterised (Fig 1 B, step 2, 3). The substrates are released from the IFS (Fig 1 B, step 4), and the empty transporter transitions back to the OFS via an occluded intermediate state (Fig 1 B, step 5, 6). The key to tightly coupled transport is that the IFS-to-OFS transition cannot occur unless the transporter either is ligand-free or carrying its full complement of substrate and coupling ions. What remains to be determined is how the presence of substrates influences the conformational state of VcINDY and whether local conformational changes of the re-entrant hairpins are required for transport.
Site-directed alkylation of single cysteine residues has been a valuable tool in elucidating the dynamic features of transporters, in particular the pioneering work on lactose permease by Kaback and co-workers 25,26 . Alkylation of single cysteine residues provides a readout of the accessibility of a particular position on the protein. The reactivity of single cysteine residues to hydrophilic thiolreactive reagents depends on the solvent accessibility of the amino acid residue in a given conformation. Thus, any change in the reactivity between the cysteine and the thiol-reactive reagent reflects a change in the local environment and/or solvent accessibility to that particular position in the protein. Here, we explore the substrate dependence of the elevator-like conformational changes of VcINDY.
To achieve this, we employ the hydrophilic, thiol-reactive reagent, methoxypolyethylene glycol maleimide 5K (mPEG5K) to probe the solvent accessibility of substituted single cysteines that are predicted to be accessible in the IFS or the OFS, but not both, according to the IFS crystal structure and OFS model. Our findings are consistent with VcINDY entering a structurally uncharacterised intermediate conformation upon binding substrates, with the majority of these accessibility changes being induced by binding Na + alone. Several positions towards the tips of the hairpin loops were identified that, unlike all other positions tested, displayed substantial sensitivity to the binding of succinate over and above the sensitivity to Na + binding. Taken together, these observations are consistent with a transport mechanism that involves multiple conformational changes that differ in their substrate dependence. In addition, our findings suggest that Na + must bind prior to succinate and that dicarboxylate substrates appear to bind with a very low affinity.

RESULTS
Experimental approach and generation of substituted cysteine panel. In this work, we sought to determine whether the presence of substrates drives VcINDY predominantly into its IFS, its OFS, or into an intermediate conformational state that has not yet been structurally characterised. To probe the conformational state of VcINDY, we devised a substituted cysteine solvent accessibility assay in which we could measure the rate of modification of substituted cysteine residues using a hydrophilic cysteine reactive mass tag, mPEG5K, in the presence and absence of substrates ( Fig.   2A). The reaction between an accessible single cysteine residue and mPEG5K would result in a 5 kDa increase in protein mass that is discernible on SDS-PAGE due to the PEGylated protein's retarded electrophoretic mobility (Fig. 2B). Digitisation and densitometric analysis of the distribution of bands in each sample allows us to quantify the extent of PEGylation and calculate a modification efficiency for each timepoint (Fig. 2C). Plotting the modification efficiency for each sample in a timecourse provides us with a means of monitoring the modification rate of a particular position on the protein. of a single substituted cysteine (pink spheres) in these different conformations. Cysteines that are more solvent exposed will react more readily with the mPEG5K (red spheres) resulting in increased rate of PEGylation.
VcINDY's colour scheme is the same as Figure 1 To allow us to differentiate between the IFS and OFS, we selected residues predicted to be accessible in the IFS or the OFS, but not both, using the IFS crystal structure and OFS repeatswapped model as guides 18,22 . While avoiding highly conserved residues or residues involved in substrate or cation binding, we selected 24 amino acids in VcINDY to be individually substituted for cysteine, using a functionally active cysteine-free version of VcINDY as a background (see Table  1 for a full list of mutants tested) 17 . Of the 24 single cysteine variants produced, we discarded those mutants that were not produced in sufficient quantities for analysis, were incapable of catalysing Na + -driven succinate transport, or were not reactive to mPEG5K under any conditions (Table 1).
Following this sieving step, we were left with 8 single cysteine mutants; VcINDYA120C OFS , VcINDYT215C OFS , VcINDYS381C OFS , VcINDYL384C OFS , VcINDYV388C OFS that are predicted to be more accessible in the OFS ( OFS denoting that they are predicted to be OFS-accessible cysteines); VcINDYT154C IFS , VcINDYM157C IFS , and VcINDYT177C IFS that are predicted to be more accessible in the IFS ( IFS denoting their predicted IFS accessibility); and a control cysteine mutant, VcINDYE42C, which is predicted to be equally accessible in both IFS and OFS. Importantly, all of the single cysteine mutants used in this analysis were capable of catalysing Na + -drive succinate transport, demonstrating that they can sample all transport-relevant conformations (Supplementary Figure 1). Using this panel of substituted cysteine mutants and the experimental approach described above, we sought to determine whether the presence or absence of substrates (Na + and succinate) drives VcINDY into the IFS, OFS, or a hitherto unidentified intermediate. In the apo state, we observe substantial PEGylation of the IFS-accessible VcINDYT154C IFS over the timecourse of the experiment resulting in PEGylation of ~65% of the protein (Fig. 3C, left, blue data). However, in the presence of saturating substrates, PEGylation of this position was almost completely prevented, suggesting that VcINDY favours a conformation in which T154 is not solvent accessible under these conditions (Fig. 3C, left, grey data).
Based on our current understanding of the structural mechanism of VcINDY, these data suggest that the presence of substrates drives VcINDY into a non-IFS conformation, which, based on our 2-state model, is the OFS. We reasoned that if this change in accessibility of T154C is indeed due to a rigid body elevator-like movement of VcINDY's transport domain, then we should observe the opposite effect of substrates on a single cysteine in a position predicted to be accessible in the OFS, but not in the IFS. To test this hypothesis, we analysed the PEGylation rate of the OFS-accessible cysteine variant VcINDYS381C OFS , in the presence and absence of substrates. In the absence of substrates, we observed a steady labelling rate of VcINDYS381C OFS over the course of our experiment ( Fig. 3B and C, middle panel). However, instead of seeing a further increase in the modification rate upon addition of substrate, which would be consistent with our hypothesis, we observe that labelling of VcINDYS381C OFS is almost completely blocked in the presence of Na + and succinate ( Fig. 3B and C, middle panel). To rule out the possibility that the presence of substrate could be directly diminishing the reactivity of mPEG5K through some unforeseen mechanism, we performed our PEGylation assay on VcINDYE42C, whose single cysteine is predicted to be equally accessible in both of the known conformations of VcINDY (Fig. 3A, right). For this control mutant, we observed equal PEGylation rates in the presence and absence of substrates, indicating that the effect on the modification efficiency of our IFS-and OFS-accessible mutants is indeed due to interaction of the substrate with VcINDY ( Fig. 3B and C, right panel).
Intriguingly, our data for VcINDYT154C IFS   Modification rates are influenced by the presence of Na + . To investigate the substrate-induced changes in modification efficiency in more detail, we sought to determine the individual effects of the coupling ion, Na + , and the substrate, succinate, on the modification efficiency of each substituted cysteine. To do this, we measured the PEGylation rate of our panel of single cysteine mutants in one of 4 different conditions; Na + alone; succinate alone; apo; or Na + and succinate. In each case, we ensured that the reactions were osmotically and ionically balanced using KCl, which is known not to catalyse transport nor interact specifically with VcINDY 17,20 .
We first analysed the effects of coupling ion or substrate on the modification rates of the IFS-and OFS-accessible mutants, VcINDYT154C IFS (Fig. 5A) and VcINDYS381C OFS (Fig. 5B). For these two mutants, we observed no changes in the modification rate in the presence of succinate alone when compared to apo conditions ( Fig. 5A and B). However, the presence of Na + alone substantially reduced the modification rate of these mutants, accounting for almost all of the reduction in the modification rate we observed in the presence of both Na + and succinate ( Fig. 5A and B). These data suggest that the presence of Na + alone is able to induce a shift in the protein's conformation that obscures these positions and reduces modification efficiency, whereas succinate binding by itself contributes minimally. In addition, the observation that succinate alone is unable to influence the modification rate of these positions indicates that one or more Na + ions must bind to VcINDY prior to succinate binding. We next tested the other members of our single cysteine panel to see if Na + binding also reduced their modification rate. In total, Na + binding alone accounted for the majority of substrate-induced modification rate changes in half of the single cysteine mutants, including; two IFS-accessible mutants, VcINDY154C IFS and VcINDY177C IFS , and two OFSaccessible mutants, VcINDYA120C OFS , VcINDYS381C OFS (Fig. 5C).
In addition to Na + -induced modification rate changes, 3 members of our mutant panel, VcINDY215C OFS , VcINDY384C OFS , and VcINDY388C OFS also exhibited further substantial modification rate reduction upon addition of succinate in the presence of Na + . These additional succinate-induced modification efficiency changes raises the possibility that the sequential binding of each substrate stabilises discrete conformations of VcINDY.
Interestingly, the amino acid positions that exhibit separable sensitivity to both Na + and succinate are located on the arms of the re-entrant hairpin loops that contribute to binding site formation and have been shown to perform a pivotal role in gating and coupling in other elevator-like transporters 23,24 . In contrast to all other positions tested, the modification efficiency of one of these re-entrant hairpin loop residues, VcINDYM157C IFS , was completely insensitive to the presence of Na + ions alone, but exhibited substantial sensitivity to the presence of succinate ( Fig. 5C and 6A).
Succinate did not induce changes in the modification rate of VcINDYM157C IFS in the absence of Na + , suggesting an ordered binding process in which one or more Na + must bind prior to succinate binding ( Fig. 5C and 6A). While our data are consistent with the different substrate conditions stabilising particular conformations of VcINDY, an alternative possibility is that, in the presence of substrates, the fully loaded transport undergoes rapid IFS-OFS interconversion; so rapid perhaps that the maleimide, which has a relatively slow rate of reaction, does not have sufficient time to react, which would result in apparent inaccessibility. To test this possibility, we performed our PEGylation assay on VcINDYM157C IFS in the presence and absence of succinate using MTS-PEG5K, whose methanethiosulfonate (MTS) moiety has a considerably faster reaction rate than maleimides 27 .
Under these conditions, we observed the same decrease in modification efficiency in the presence of succinate compared to the absence (Supplementary Figure 4), indicating that our observations with mPEG5K reflect conformational stabilisation rather than altered protein dynamics.
PEGylation rates of VcINDYM157C IFS suggest ordered low affinity binding of succinate. Due to the maverick nature of the VcINDYM157C IFS mutant, we investigated this variant in more detail.
To determine whether the apparent succinate-induced decrease in modification efficiency was due to a specific interaction between VcINDY and succinate, we performed our PEGylation assay on VcINDYM157C IFS in the presence of increasing concentrations of succinate ranging from no succinate up to 1 mM, while keeping a constant Na + concentration of 150 mM (Fig 6 B). We observed a dose-responsive decrease in modification efficiency of VcINDYM157C IFS with increasing succinate concentration, indicating that this effect is indeed due to succinate binding (Fig. 6B). However, substantial reduction in the modification efficiency is only apparent when the protein is incubated with ³100 µM succinate, suggesting an unexpectedly low affinity for succinate, considering the Km for transport is known to be 1 µM 17 .
In addition to succinate, VcINDY can transport a number of other C4-dicarboxylates, including malate, fumarate, and oxaloacetate, but not shorter dicarboxylates, such as oxalate, nor tricarboxylates, such as citrate 17 . If this effect on PEGylation is indeed due to VcINDY's specific interaction with its substrates, we would expect other known substrates to be similarly influential, whereas non-transported dicarboxylates should have no effect. Compared to Na + alone conditions, substituting succinate with malate in the PEGylation reaction resulted in a significant increase in the modification rate, whereas substituting succinate with the non-substrate oxalate resulted in modification rates akin to those observed in the absence of any dicarboxylate substrate (Fig. 7A).
These data indicate that the reduction in modification rate of VcINDYM157C IFS are due to specific substrate interactions and not due to unforeseen indirect effects of dicarboxylates. However, we noted with interest that the presence of malate led to substantially and significantly reduced protection of the cysteine from PEGylation compared to succinate (Fig. 7A). Intrigued by this result, we extended the compound range in our VcINDYM157C IFS PEGylation assay to include the other known VcINDY substrates, fumarate and oxaloacetate, and the nonsubstrate citrate (Fig. 7B). To test the effects of our extended compound range on the VcINDYM157C IFS PEGylation rate, we determined the modification efficiency after 60 min incubation with mPEG5K and 1 mM of each test compound (Fig. 7C, open bars). As expected, the presence of succinate reduces the modification to the greatest extent, whereas oxalate, citrate, and surprisingly, oxaloacetate induce negligible changes to the modification efficiency compared to the absence of substrates (Fig. 7C). The presence of 1 mM fumarate and malate reduce modification efficiency substantially, but only half as much as succinate (Fig. 7C). We reasoned that these variations in the extent of labelling in the presence of different substrates could merely reflect differences in the affinity for each of the different substrates. To test this possibility, we performed the same PEGylation assay, but with a final concentration of 10 mM of each compound (Fig. 7C, closed bars). In the presence of increased concentrations of each compound, we observe no change in the modification efficiency in the presence of either oxalate nor citrate (Fig. 7C).
Increasing the concentration of oxaloacetate to 10 mM results in significantly decreased modification efficiency compared to 1 mM, and while increasing the concentration of fumarate and malate to 10 mM leads to further reduction in the modification efficiency, they were still significantly less effective than 10 mM succinate (Fig. 7C). These data suggest that the differences in the modification efficiency of VcINDYM157C IFS in the presence of different substrates is due primarily to differences in VcINDY's binding affinity for each substrate. In addition, these results reveal that 13 VcINDY's substrate binding site is not saturated in the presence of 1 mM substrate, suggesting that under these experimental conditions, VcINDY has a remarkably low affinity for its substrates.

DISCUSSION
In this work, we have described the first foray into determining the substrate-dependent conformational changes of VcINDY, which is a structural and mechanistic representative of the entire DASS transporter family. Using a site-specific alkylation strategy on detergent-solubilised protein, we have demonstrated that the modification rate, which serves as a proxy for solvent accessibility, of several broadly distributed positions in VcINDY can be modulated by the presence of substrates. We demonstrate that the majority of these changes in modification efficiency can be induced by the presence of Na + alone. However, we also observe substantial, separable effects of succinate binding on the modification efficiency of multiple positions, suggesting discrete effects of the two substrates. Furthermore, we have identified a position whose modification rate is insensitive to the presence of Na + , but highly sensitive to succinate. Further analysis of this variant indicates that VcINDY binds its substrate with a low affinity and proceeds via an ordered process in which one or more Na + ions must bind prior to the substrate, succinate. Mapping these solvent accessibility changes onto the IFS X-ray structure and OFS repeat-swapped model suggests that the binding of substrates drives the protein into an as-yet-unidentified intermediate state that obscures all of the positions tested.

VcINDY likely undergoes multiple large-and small-scale conformational changes during
transport. There are now many examples of structurally characterised transporters that are predicted to employ an elevator-like transport mechanism 24,28-37 , which has revealed common features amongst many of them, including; distinct scaffold and transport domains, a broken transmembrane helix containing an intramembrane loop that contributes to the substrate binding site, and two re-entrant hairpin loops that enter the membrane but do not fully span it. While structural differences exist between the predicted elevator-type transporters, they are all predicted (or have been shown) to undergo a substantial vertical translocation of the transport domain, usually accompanied by a pronounced rotation of this domain, to expose the substrate binding site to both sides of the membrane.
Our data suggest that multiple positions predicted to be solvent accessible in either the IFS or the OFS, but not both, of VcINDY become less accessible in the presence of Na + (Fig. 4) 40 . This blockage of the cation only-bound state is crucial to tight coupling in secondary active transporters and prevents cation leak, which could prove catastrophic to the cell.
In both VcINDY and GltPh, the tips of these re-entrant loops form part of the substrate binding site, coordinating both the coupling ion and substrate 18,19,42  In the IFS structure and OFS model of VcINDY, the substrate is solvent exposed and straddles the interface of the scaffold and transport domains 18,19,22 , potentially precluding the need for hairpincoordinated gating in DASS transporters. In this study, we have probed the substrate-dependent accessibility of several residues in the re-entrant hairpin loops of VcINDY; M157 and T154 in HP1, and S381, L384, and V388 in HP2 (Fig. 8A). Interestingly, we observe different effects of coupling ion and substrate binding within the same hairpin. As with other tested positions, T154 and S381 are mostly sensitive to the presence of Na + , whereas L384, V388, and in particular M157 exhibit considerable sensitivity to the presence of succinate (Fig. 5C). While further experimental work is required to provide a full picture of the dynamics of VcINDY's re-entrant hairpins, these results suggest discrete effects of the coupling ion and substrate on the conformation of the hairpins.
Analogous to the situation with GltPh and its homologues, binding of Na + is likely required for high affinity binding of succinate, and binding of the succinate molecule is required to stabilise the "closed" conformation of the re-entrant loop, which allows the elevator-like conformational change to take place 23,[43][44][45][46] . The suggestion that the re-entrant hairpins undergo conformational changes in response to succinate binding is strengthened by the observations made during a series of solvent accessibility studies on the eukaryotic DASS transporter, NaDC1 47,48 . In concordance with our work presented here, the binding of Na + and succinate had substantial and separable effects on the solvent accessibility of residues in the arm of re-entrant hairpin 2 in NaDC1 (in the same region and S381, L384 and V388 in VcINDY), suggestive of discrete conformational changes upon each Na + and succinate binding event 47,48 . In contrast to our observations with VcINDY, the HP2 re-entrant loop residues become more accessible in the presence of Na + alone, and then less accessible upon addition of succinate 47 . The differences between these two systems may be explained by the experimental set-up; NaDC1 was probed in the lipid bilayer, in the presence of gradients and a membrane potential, and in a background with 10 native cysteines present, whereas, in our study, each VcINDY variant was detergent-solubilised and contained only one cysteine residue. Cysteine accessibility studies on another bacterial DASS family member, SdcS from Staphylococcus aureus, revealed that D329 (F291 in VcINDY), which is predicted to be on the outward façade of the protein, is accessible in the presence of Na + , but not in the absence 49 . However, the equivalent position of VcINDY was not tested in this study. Curiously, the same SdcS study, N108 (N94 in VcINDY) was shown to be accessible from both the external and cytoplasmic sides of the membrane. However, in both the IFS crystal structure and the repeat-swapped model of VcINDY, there is dense protein blocking access to this site from the cytoplasmic side of the membrane 49 . This perhaps suggest that the mechanism of SdcS is considerably different to the current structural model based on VcINDY, or that the conformational changes during transport are substantially more extreme than currently thought. Nevertheless, these studies combined suggest that there are discrete conformational changes upon binding Na + and succinate to DASS transporters. Indeed, solvent accessibility studies on NaDC1 prior to the elucidation of the structure of VcINDY revealed that the accessibility changes in HP2 were temperature insensitive, indicative that the modulation in accessibility was not due a large conformational change, but perhaps due to blockade of these possible binding site residues by succinate 47 . However, our current structural understanding of the DASS family reveals that these residues are in the arm of HP2 and do not form part of the binding site. Therefore, these observed accessibility changes of HP2 residues could be explained by subtle, local conformational changes of the hairpin in response to succinate binding. More structural information is required to fully realise the DASS transporter mechanism, in particular, a structure of the apo state of VcINDY will be especially illuminating.
Potential implications to the mechanism of VcINDY. While further work is needed to fully realise the dynamics of VcINDY, our data are consistent with VcINDY having a complex mechanism that consists of multiple large-scale and local conformational changes as described by the following hypothetical mechanistic scheme (Fig. 8B). For simplicity, we will start this transport cycle with VcINDY in its open OFS. However, as VcINDY is a secondary-active transporter, it has the ability to work in the reverse direction depending on the direction and magnitude of the electrochemical gradients. In the open OFS conformation (Fig. 8B, state i), VcINDY is initially ligand free and the overall architecture resembles the structure described by the repeat swapped model 22 . One or more Na + ions bind, inducing a shift in the transport domain conformation into an intermediate state in which the majority of sites tested in this study are obscured to some extent (Fig. 8B, state ii).
However, HP2 remains "open" allowing succinate to still access the binding site. Binding of succinate stabilises the "closed" state of HP2 (Fig. 8B, state iii), allowing the transport domain to move into the closed IFS (Fig. 8B, state iv). HP1 opens (Fig. 8B, state v), releasing the substrates into the cytoplasm (Fig. 8B, state v and vi). HP1 closes and the ligand-free transport domain can translocate to form the OFS and restart the cycle. Substrate binding appears to be ordered and low affinity. In this study, we have identified a position in the arm of HP1, M157, whose accessibility is insensitive to the addition of Na + , but undergoes substantial accessibility changes in the presence of succinate (Fig. 5C and 6A). Probing this mutant in more detail, we discovered that the accessibility effects were only induced by the addition of compounds known to be transportable by VcINDY, for example, succinate, malate and fumarate, demonstrating specific binding was required (Fig. 7). In addition, we did not observe any changes in accessibility upon addition of succinate in the absence of Na + , demonstrating that Na + must bind first before succinate can bind, as has been suggested in previous studies on DASS transporters 50 . This ordered binding is consistent with other Na + -driven elevator transporters. The archetypal elevator transporter, GltPh, transports aspartate coupled to the co-transport of 3 Na + ions 51,52 , with 2 Na + ions binding first to the apo transporter to 'prime' the binding site before the aspartate and final Na + bind 46,53 , allowing the hairpin to close and the OFS-to-IFS transition to occur.
Our data suggest a similar ordered binding mechanism for VcINDY. However, more detailed analysis of the coupling ion and substrate binding is required to illuminate this process further.
When titrating succinate and monitoring the dose-responsive change in modification of VcINDYM157C IFS , we only observed substantial shifts in the modification efficiency once the succinate concentration reached ~100 µM (Fig. 6B). This concentration is surprisingly high considering VcINDY's Km for succinate transport in proteoliposomes in the presence of a similar Na + concentration is 1 µM 17 . To put this in the context of other elevator-like transporters; GltPh also has Km for transport of ~1 µM 54,55 , but has a Kd of 100 nM 46 . Since VcINDY and GltPh have a similar mechanism and identical Kms, it is not unreasonable to predict they would have similar Kds.
However, if this were the case, we would expect VcINDY's binding site to be saturated at 100 µM; the fact that the we continue to observe decreases in the modification rate with increasing succinate concentration demonstrates that it is not. These results indicate that in these experimental conditions VcINDY has a low affinity for substrate. Transporters with Kms lower than their Kd are rarely observed, one notable example being CaiT, which has a Km of ~80 µM and Kd of ~3 mM 56 . As is the case with CaiT, VcINDY's apparent low binding affinity may be due to the protein containing more than one functionally significant substrate binding site, as has been suggested for other DASS transporters 57 . Alternatively, this apparent low affinity may be due to VcINDY being in a detergent-solubilised state. Further functional analysis in the presence of a lipid bilayer are needed to resolve this issue.
We performed this study on detergent-solubilised protein in order to make the results directly comparable to the crystal structure of VcINDY, which was crystallised in its detergent solubilised form. In addition, we selected our panel of single cysteine mutants based on the crystal structure and repeat-swapped model of VcINDY in order to report on the these particular conformations.
Using lipid bilayer mimetics, such as detergents, may produce different results to what is seen in a membrane environment, where there is an absence of lateral pressure, other membrane proteins, specific protein:lipid interactions and electrochemical gradients. Indeed, single molecule FRET (smFRET) experiments have demonstrated that GltPh has appreciably different dynamic behaviour in detergent micelles versus lipid bilayer 40,58 . However, site-directed alkylation studies on other transport proteins have directly shown that alkylation rates are similar in detergent compared to those measured in lipid bilayer 59 , giving us confidence that the qualitative differences we observe in the presence and absence of substrates are mechanistically relevant. The elevator-like mechanism requires large-scale conformational changes, which is almost certainly affected by the lipid composition (both lipid head group and hydrocarbon chain length and saturation). The dynamics and overall transport rate of the only other well characterised elevator-like transport, GltPh, is influenced by the presence and composition of the lipid environment and molecular dynamic simulations have recently shown large scale bilayer deformation can be induced by elevator-like transporter conformational changes 40,58,[60][61][62][63] . So, aside from any specific protein:lipid interactions that may be influencing VcINDY's function, the physical properties (e.g. flexibility) of the bilayer, which is dictated by the lipid composition, will clearly have a strong effect on its mechanism. How the lipid environment and electrochemical gradients across bilayers influences protein dynamics and substrate-dependent conformational changes of VcINDY is of great interest and is the subject of ongoing studies in our lab.

MATERIALS AND METHODS
Molecular biology and cysteine mutants generation. All single cysteine variants were generated in a previously characterised cysteine-free background in which all three native cysteines had been substituted for serine 17 . Substitutions were made with a Quikchange II site-directed mutagenesis kit (Agilent Technologies). Expression plasmids were fully sequenced to ensure that the desired codon substitution occurred and no unwanted secondary mutations were introduced.

Expression of VcINDY.
Over the course of this work, we modified the expression protocol of VcINDY three times in an effort to maximise its yield, which led to an increase in yield from 0.2 mg/L culture to >5 mg/L culture for wildtype VcINDY. Changes in expression levels of VcINDY did not affect the quality of the protein produced, the PEGylation efficiency for each variant, nor the transport activity of each variant (data not shown). In all three protocols, VcINDY and its variants were expressed in-frame with an N-terminal decahistidine tag form a modified pET vector 64 .
VcINDY was initially expressed as described previously 17,18 . Briefly, BL21-AI (Invitrogen) cells harbouring the expression vector were grown in LB supplemented with 50 µg/ml kanamycin and incubated at 37 o C until it reached an A600 of 0.8. Cells were rapidly cooled in an ice bath for 20 min, at which point expression was induced by addition of 10 µM IPTG and 6.6 mM L-arabinose. The cells were incubated at 19 o C and grown for approximately 16 hours before being harvested and resuspended in Lysis Buffer (50 mM Tris, pH 7.5, 200 mM NaCl, and 10% (v/v) glycerol). Following disappointing yields via the above method, we adopted the MemStar method described by Drew and co-workers 65  The resultant proteoliposomes were collected by ultracentrifugation, resuspended in Inside Buffer to a concentration of 8 mg/ml lipid, freeze thawed thrice, and stored at -80 o C.
In vitro transport assays. For transport assays, the required amount of proteoliposomes were thawed, extruded 11 times through a 400 nm filter, collected by ultracentrifugation, and resuspended to a final concentration of 80 mg/ml lipid. Transport assays were performed by rapidly mixing the prepared proteoliposomes with Reaction Buffer (20 mM Tris, pH 7.5, 100 mM NaCl, 100 PEGylation timecourse. To perform the PEGylation timecourse, detergent solubilised protein was thawed and buffer exchanged using Zeba Spin Desalting Columns (Thermo Fisher Scientific) to remove the 0.5 mM TCEP and exchange the protein into PEGylation Buffer (50 mM Tris, pH 7, 5.4 mM DM, 5% (v/v) glycerol). The substrate-free apo sample was formulated by mixing protein solution with 1 M KCl and PEGylation Buffer to generate a final protein concentration of 10 µM and a final KCl concentration of 150 mM. The "Na + alone" sample was formulated in the same way except KCl was substituted for NaCl. The same approach was used for the "Na + + succinate" samples except that succinate was added to a final concentration of 1 mM. For the "succinate alone" sample, protein was mixed with 150 mM KCl and 1 mM succinate. The protein/substrate mixtures were incubated for at 10 minutes at room temperature, at which point the PEGylation reaction was started by addition of methoxypolyethylene glycol maleimide 5K (mPEG5K) (Sigma Aldrich). Samples were collected at various timepoints and the reaction was terminated by addition of SDS-PAGE sample buffer containing 100 mM methyl methanethiosulfonate (MMTS) (Sigma Aldrich). The PEGylation reaction samples were analysed using non-reducing polyacrylamide gels, which were stained with Coomassie Brilliant Blue dye to visualise the protein.
Densitometric analysis. The intensities of the bands corresponding to the unmodified and PEGylated VcINDY protein bands were quantified using ImageJ software 67,68 . For each timepoint, the modification efficiency was calculated using the following equation; (%) = 1 + : × 100 For the final data sets, replicates (n≥3) of the modification efficiency for each timepoints were averaged and the standard error of the mean (SEM) was calculated. Where indicated, statistical significance was examined using unpaired t-tests.