Functional characterization and discovery of modulators of SbMATE, the agronomically important aluminium tolerance transporter from Sorghum bicolor

About 50% of the world’s arable land is strongly acidic (pH ≤ 5). The low pH solubilizes root-toxic ionic aluminium (Al3+) species from clay minerals, driving the evolution of counteractive adaptations in cultivated crops. The food crop Sorghum bicolor upregulates the membrane-embedded transporter protein SbMATE in its roots. SbMATE mediates efflux of the anionic form of the organic acid, citrate, into the soil rhizosphere, chelating Al3+ ions and thereby imparting Al-resistance based on excluding Al+3 from the growing root tip. Here, we use electrophysiological, radiolabeled, and fluorescence-based transport assays in two heterologous expression systems to establish a broad substrate recognition profile of SbMATE, showing the proton and/or sodium-driven transport of 14C-citrate anion, as well as the organic monovalent cation, ethidium, but not its divalent analog, propidium. We further complement our transport assays by measuring substrate binding to detergent-purified SbMATE protein. Finally, we use the purified membrane protein as an antigen to discover native conformation-binding and transport function-altering nanobodies using an animal-free, mRNA/cDNA display technology. Our results demonstrate the utility of using Pichia pastoris as an efficient eukaryotic host to express large quantities of functional plant transporter proteins. The nanobody discovery approach is applicable to other non-immunogenic plant proteins.


Results
SbMATE exhibits broad substrate recognition -organic anion (citrate) and cation (ethidium) transport. We have previously reported the use of Xenopus oocytes to express SbMATE and measure its electrogenic transport activity 5,8 . We used this heterologous expression system to directly measure the efflux of 14 C-citrate from pre-loaded cells in the presence of inwardly-directed proton and sodium gradients, i.e. substrate-proton/cation antiport. Consistent with previous in planta experiments using transgenic A. thaliana lines, SbMATE-expressing oocytes mediated 14 C-citrate efflux, quantified as ~2.5-fold higher 14 C-citrate released from loaded cells relative to the basal levels from non-expressing control cells (Fig. 1B).
We further investigated substrate recognition and the proton antiport mechanism of SbMATE in Pichia pastoris. Intact yeast cells expressing SbMATE, empty vector-transformed control cells, and their respective uninduced control cells were washed and maintained on ice in phosphate-saline (PBS) buffer, pH 7.4. For the transport measurements, cells were incubated with 50 µM of ethidium bromide in buffer salt-concentration matched to PBS at pH 8.5 (i.e. 137 mM NaCl, 2.7 mM KCl). In this reversed, outwardly-directed proton gradient (assuming a near neutrality cytosolic pH) 14 , SbMATE exhibited uptake of ethidium into cells, as shown by a large increase in fluorescence (caused by the binding of ethidium to DNA) in the induced SbMATE-expressing cells, compared to control non-expressing, and uninduced cells ( Fig. 2A,B). We further measured the substrate-concentration dependence of SbMATE-mediated ethidium transport, which resulted in a half-maximal transport rate computed at 130 ± 17.8 µM ethidium bromide (mean ± SEM, n = 3; Fig. 2C). These quantitative analyses pertain to measurements of DNA-bound ethidium in cells, and not ethidium directly, a method that has previously been useful for studying transporter-substrate interactions 15,16 .
In order to further demonstrate that ethidium transport in our yeast system is SbMATE-specific, we engineered a mutation at the highly conserved aspartate residue on transmembrane helix 1, which was previously shown to be important for the protonation-deprotonation cycle crucial for substrate transport by other prokaryotic MATE family transporters 15,[17][18][19][20] . By sequence homology (Fig. 3A), the corresponding residue on SbMATE was found to be aspartate D127. We made a conservative asparagine substitution on this residue to create the mutant SbMATE D127N, and tested its ethidium transport capability. The levels of SbMATE D127N in the plasma membrane (determined by western blot analysis on equal amounts of membrane fraction preparations) were found to be comparable to a specific construct of WT SbMATE protein, which was cloned with a HRV 3C protease cleavable His 8 tag on the C-terminus (termed SbMATE-WT-3C-His) (Fig. 3B,C). The ethidium transport activity of the D127N expressing cells was found to be ~9% of those expressing SbMATE-WT-3C-His (Fig. 3D), verifying a role of D127 in organic cation transport.
These results demonstrate that SbMATE is capable of transporting both organic acid anions, like citrate, and organic cationic compounds, such as ethidium.
Substrate transport by SbMATE can be cation and/or proton driven. We further substantiated our results suggesting a substrate-proton/cation antiport mechanism for SbMATE, by examining the cation and pH-dependence of the SbMATE electrogenic transport. In our previous findings obtained using two-electrode voltage clamp (TEVC) to measure whole-cell currents from Xenopus oocytes, SbMATE showed a large electrogenic transport in a bath media containing high Na + , with the transport being greater at pH 4.5 compared to pH 7.5, an effect not evident in control non-expressing cells 8 . In the present study, reducing [Na + ] in the bath media by lowering the NaCl concentration from 96 to 1 mM NaCl, thereby reducing the inwardly directed Na + gradient, resulted in a reduction of the SbMATE-mediated inward and outward currents and a shift in E rev to positive potentials (Fig. 4A). These results indicated the inward currents recorded in SbMATE cells were due to Na + influx. Substitution of NaCl by an equimolar concentration of KCl in the bath media did not significantly affect the SbMATE-mediated inward currents nor shifted the E rev (Fig. 4B), suggesting similar K + and Na + permeability. Taken together, these results suggest that SbMATE transport can also be coupled to other cations (e.g. Na + and K + ) in addition to H + . In bacteria, MATE mediated transport is typically coupled to the inwardly directed Na + gradient established by a primary Na + pump or a Na + -H + antiporter coupled with respiration, with a few exceptions that exhibit coupling to H + gradients. In eukaryotes, MATE transport is typically coupled to H + gradients, with the plasma membrane H + electrogenic pump being the primary source of the inwardly directed H + electrochemical driving force in plants. Therefore, we proceeded to further characterize the kinetics of SbMATE transport with respect to the proton-motive force. SbMATE-mediated currents were measured at increasing extracellular H + concentrations, with minimal Na + in the bath solution. As illustrated by the current-voltage relationships obtained at the different pH values (H + concentrations ranging from 30 nM to 31 µM), SbMATE-mediated inward currents increased as the bath was acidified, thereby increasing the inwardly directed H + gradient (Fig. 4C). The H + -dependent SbMATE currents at a given holding potential saturates within the experimental H + concentration range, at around 30 µM H + (Fig. 4D). The hyperbolic nature of these relationships suggests that only one H + binds to the transporter. The K m H calculated from the Michaelis-Menten relationship exhibited pronounced voltage dependence, with the apparent affinity constant exponentially increasing as the membrane potential was depolarized (Fig. 4E). These observations are consistent with the pH-dependence of the SbMATE-mediated electrogenic transport reported earlier for SbMATE expressed in X. oocytes 8 , as well as 14 C citrate permeability and pH dependence reported for other plant MATEs mediating Al-exclusion in maize 13 and rice bean 9 .
Similarly, the proton-dependence of ethidium transport into yeast cells by SbMATE was investigated by reducing the extracellular pH from 8.5 to 6.0 (i.e. thereby reducing the outwardly-directed proton gradient) during ethidium bromide incubation and subsequent washing steps. There was a notable decrease in the ethidium bromide transport rates to ~51% (at pH 7.4) and ~33% (at 6.0) of the rate at pH 8.5 (Fig. 5A). While the reduction of ethidium uptake could partly be due to increased competition between protons and ethidium for binding to a shared site on SbMATE, our citrate transport data also show a similar pH-dependence. We believe that citrate is unlikely to compete with protons for a shared binding site. Thus, we conclude that a major cause for the reduction in ethidium transport is likely a decrease in the outwardly-directed proton gradient, as the salt concentrations were kept identical in all pH treatments. The cytosolic pH of yeast cells is thought to remain stable around neutrality in response to shifts in external pH between 3.0 and 7.5 14 . Therefore, as described for in Xenopus oocytes (Fig. 4), the significant remnant transport activity by SbMATE in yeast cells under conditions with minimal proton gradients (Fig. 5A), is likely due to the use of an alternate ion gradient (e.g. K + or Na + ) as an energy coupling source, to power substrate transport in the absence of significant proton gradients.
Overall, the above results indicate that SbMATE mediates citrate and ethidium transport coupled to high-affinity H + binding and/or potentially other cations, such as, Na + or K + , validating a proton (H + )/cation -substrate antiport mechanism.
Substrate and charge selectivity of SbMATE. Citrate is a physiological transport substrate of SbMATE.
Since citrate buffers are used to maintain acidic pH ranges, we challenged SbMATE-expressing yeast cells with an external pH of 4.3 and measured the uptake of 50 µM ethidium bromide, in the presence of 20 mM citrate and the Uninduced and induced cells transformed with SbMATE WT or vector control were normalized to an OD 600 of 15, washed, and maintained in standard PBS buffer, pH 7.4. For the ethidium transport assay, 50 µM ethidium bromide and a pH 8.5 buffer (25 mM Tris-Cl, pH 8.5, 137 mM NaCl, 2.7 mM KCl) was used to generate an outward-proton gradient. Ethidium uptake was measured over a 20 min uptake period using flow cytometry, with 12,000 events collected per sample. Histograms were plotted for ethidium fluorescence vs normalized frequency (n = 5 replicates conducted with independently grown batches of cells, histograms shown are from a single experiment). (B) Median fluorescence intensities of uninduced or induced cells transformed with SbMATE WT were subtracted from the respective vector control cells (n = 5 independent replicates, data are mean ± SEM). (C) The ethidium transport assay described in Fig. 2B was performed with varying concentrations of ethidium bromide. The vector-subtracted median fluorescence intensities for SbMATE WTexpressing cells at the different ethidium bromide concentrations were used to calculate the transport rates, which were then normalized against the rate at 500 µM ethidium bromide set to 100% [n = 3 independent replicates, symbols are mean ± SEM, dotted line represents a hyperbolic fit, using y = P1x/(P2 + x), where P1 and P2 represent the maximum observed transport rate and concentration of EtBr (x) that produces halfmaximal transport rate, respectively].
SCIENtIFIC REPORTS | (2017) 7:17996 | DOI:10.1038/s41598-017-18146-8 same salt concentrations used in all of the above-described yeast experiments (i.e. 137 mM NaCl, 2.7 mM KCl). Amidst an almost complete lack of an outwardly-directed proton gradient, and despite an inwardly-directed sodium gradient, SbMATE still showed an appreciable degree of ethidium uptake (Fig. 5A). The transport rate was reduced to 25% of that observed at pH 8.5 (Fig. 5A), which is slightly lower than the 33% transport rate observed at pH 6.0 (Fig. 5A), and still above the background of non-expressing cells. These results suggest that the transport of organic cations by SbMATE can occur even in the background of highly competing levels of citrate, or other organic acid anions.
Similar to ethidium, its structural and functional analog, propidium, also binds to nucleic acids and exhibits an increase in fluorescence. However, in contrast to monovalent ethidium, propidium is a divalent organic cationic dye (Fig. 5B), which serves as a test compound to study the substrate charge preferences of SbMATE. Conducting the yeast transport assay with 50 µM and 100 µM propidium iodide, SbMATE-mediated transport was found to be negligibly low, compared to the transport of ethidium at the same concentrations (Fig. 5C,D). This observation suggests that SbMATE can distinguish between monovalent vs divalent organic cationic compounds, preferentially transporting the former. Propidium iodide is also routinely used to assess the viability of cells in flow cytometry experiments. Thus, the low fluorescence intensity of SbMATE-expressing cells with propidium ( Fig. 5C,D) also demonstrates that our cell sample preparations were intact and healthy. As such, the increase in ethidium fluorescence in our experiments (Figs 2A, 3C) was most likely due to transport, as opposed to alterations in overall membrane permeability.  Purification of functionally-folded SbMATE. Following the confirmation that SbMATE over-expressed in Pichia pastoris was transport-competent (Figs 2, 3 and 5), we proceeded with purifying the protein in detergent solution from 8 L fermenter cultures. Briefly, bDDM detergent-solubilised SbMATE was purified by affinity chromatography using the His 10 tag and size-exclusion chromatography. Each 8 L fermenter run produced an average yield of ~5 mg of purified and monodisperse SbMATE protein (Fig. 6A).
The functional integrity of detergent-purified SbMATE protein was tested using two different approaches. First, we conducted an in vitro binding assay using label-free, surface plasmon resonance (SPR) to measure the binding kinetics of SbMATE to the transport substrate, ethidium. Purified SbMATE was observed to bind ethidium in a saturable manner (Fig. 6B), with a dissociation constant (K D ) of 47.1 ± 15.6 µM (mean ± SEM, n = 5) (Fig. 6B). This binding affinity is relatively high compared to the known SbMATE substrate, citrate, which has been tested at higher µM-mM concentrations. Consistent with our transport assays (Fig. 5C,D), we also found that the use of propidium iodide, at the same concentrations as ethidium in this assay, did not result in discernible, concentration-dependent binding traces (Fig. 6C).
Functional validation of detergent-purified SbMATE was further performed by reconstituting the protein into an artificial membrane environment and evaluating its electrogenic activity (i.e. the capacity to transport ions), using electrophysiological measurements. Incorporation of SbMATE resulted in single channel-type electrogenic activity, seen as discrete changes in current amplitude (with the unitary channel spending about half of the total recording time in a conducting open state of about 1 pA) as the transport protein transitioned between open and closed conformational states (Fig. 7). This type of single channel activity, is consistent with that recently reported for other plant MATE transporters overexpressed in Arabidopsis 21 .
High-affinity nanobodies against purified SbMATE. The availability of large amounts of functional purified protein is valuable for the discovery and screening of potential binders, such as antibodies, which can be useful as research reagent tools. Integral membrane proteins, however, are typically considered 'hard targets' for discovering antibodies via the traditional animal immunization approach. Additionally, plant-derived proteins, in general, are known to have substantially lower immunogenicity compared to other microbial/mammalian antigens that are used to immunize animals. To surpass these hurdles, we applied an in vitro approach 22 to enable the rapid discovery of single-domain antibodies, also called nanobodies (Nbs), against membrane protein targets.
Our Nb discovery process resulted in a panel of six Nbs ( Table 1) that were purified from E. coli and tested for binding to SbMATE. The Nb panel was diverse in binding affinities, ranging from low-affinity (µM) (Fig. 8A-D) to moderately high-affinity (nM) (Fig. 8E,F), as characterized by SPR. The higher affinity binders, Nbs 86 and 94, were further characterized to have dissociation constants, K D of 173.57 ± 45.53 nM and 99.53 ± 19.32 nM, respectively (mean ± SEM, n = 3) (Fig. 8E,F). These represent novel, first-generation Nbs obtained from a naïve camelid-derived Nb library, without conducting deliberate affinity maturation.
Nanobodies bind to the folded conformation of SbMATE. One advantage of in vitro Nb technologies is that discovery of Nbs occurs against the full-length, natively folded target proteins. Our purified SbMATE protein was functionally folded, as was demonstrated by its ability to bind the transport substrate ethidium, to distinguish the non-transported analog, propidium, and also its ability to mediate ion transport when reconstituted into an artificial lipid bilayer (Figs 6B,C and 7). In order to investigate whether some of our Nbs bind to a native fold of the antigen, we tested the effect that Nb binding might have on ethidium binding to SbMATE and on SbMATE-mediated electrogenic transport.
When we performed SPR using Nbs and ethidium at a molar ratio 1:10 Nb:ethidium, our two high-affinity Nbs, 86 and 94, reduced ethidium binding on SbMATE to 35.04 ± 13.39% and 71.40 ± 13.91%, respectively (mean ± SEM, n = 3) (Fig. 9A,B). In contrast, these Nbs did not yield a reduction of the non-specific ethidium signals obtained for the hydrophobic control protein, BSA (Fig. 9C). Furthermore, upon extracellular exposure of SbMATE-expressing Xenopus oocytes to a mixture of Nbs (Nbs 83, 86, 89, 94 and 124), a ~2-fold increase in the SbMATE mediated inward currents were observed. This agonistic effect of the Nbs mix was reversible upon re-establishing the control bath conditions (Fig. 10A). Likewise, intracellular Nb-SbMATE interactions were studied in SbMATE expressing cells preloaded with the Nb mix 2.5 hours prior to the electrophysiological recordings. This resulted in larger currents than control SbMATE-expressing cells preloaded with water (Fig. 10B). Importantly, neither extracellular nor intracellular Nbs had any effect of the current magnitudes recorded in non-expressing cells. These findings are interesting as they indicate an agonistic effect for our Nbs on SbMATE's transport functions.
These findings suggest that our in vitro-discovered Nbs bind to the transport-competent, folded form of SbMATE, thereby highlighting the suitability and advantages of such in vitro Nb discovery technologies when studying integral membrane proteins, including those from plants.

Discussion
Worldwide food demand is projected to experience a 40% increase by 2030 23 . Studies on crop yields predict that this is unlikely to be met by applying traditional agricultural methods based on conventional plant breeding and additional use of fertilizer and chemicals. Novel technologies will instead need to focus directly on improving voltage on the apparent affinity constants for protons (K m H ) derived from (D). The curve shown represents a single exponential function ([H] = [H 0 ] exp (V/τ 0 ), with H and V being substrate and voltage, respectively, and extrapolated. Fitting parameters were H 0 = 1 ± 0.1 µM and τ 0 = 30 ± 2 mV. All electrophysiological recordings were performed in the absence of exogenous intracellular citrate or ethidium bromide (i.e. cell which were not microinjected with any substrate prior to the recordings).
SCIENtIFIC REPORTS | (2017) 7:17996 | DOI:10.1038/s41598-017-18146-8 various agronomic properties of crops. For example, a molecular breeding approach based on the use of markers linked to SbMATE was used to improve sorghum Al tolerance and yields on acid soils 24 and biotechnological approaches were applied to improve Al-tolerance in barley, via transformation with the SbMATE gene 6 . A detailed understanding of substrate specificities and transport mechanisms of agronomically important transporters, such  Fig. 2B. The transport rate was calculated as described for Fig. 2C, with the rate at pH 8.5 set to 100% (n = 3 independent replicates, data are mean ± SEM). The pH 4.3 assay conditions were obtained using 20 mM citrate buffer. (B) Chemical structures of (i) ethidium (+1) and (ii) propidium (+2) reveal a high degree of similarity, except for the additional charged moiety on propidium. (C) Transport assay was performed as described in Fig as SbMATE, will provide new avenues for engineering more effective (Al-tolerant) SbMATE-type transporter proteins and aid the success of these approaches.
To date, the characterization of the subgroup of plant MATE transporters associated with Al resistance and Fe transport/homeostasis has been solely performed in relation to their organic anion transport activity (specifically citrate) which underlies their major in planta roles. MATE family proteins such as NorM, however, were initially identified as proteins that transported a wide range of organic cationic substrates 25 . In fact, in plants, a separate clade of plant MATEs have been associated with the transport of a broad variety of secondary metabolites Figure 6. Purified SbMATE binds ethidium bromide, but not propidium iodide. (A) Detergent purified SbMATE elutes from size exclusion chromatography as a symmetrical monodispersed peak (FPLC gel filtration trace on left) that is deemed to be ~95% homogeneous (coomassie stained SDS-PAGE of pooled FPLC peak fractions shown on right). (B) Purified SbMATE was covalently immobilized on a CMD surface plasmon resonance (SPR) sensorchip. 100 µL of varying concentrations of ethidium bromide were injected, with each concentration having an association and dissociation time of 120 s. The chip was regenerated between each injection using a short 10 µL injection of 50 mM glycine-HCl, pH 2.1. Traces on the sensorgram represent double-referenced data, obtained by subtracting response signals from a blank surface and buffer injections (n = 5 replicates performed using independently purified batches of protein, traces represent data from a single experiment). (C) SPR was conducted as described in Fig. 6B, except with varying concentrations of propidium iodide instead of ethidium bromide (n = 3 independent replicates, sensorgrams are from a single experiment).
SCIENtIFIC REPORTS | (2017) 7:17996 | DOI:10.1038/s41598-017-18146-8 ( Fig. 1A; also see 2 ). Our findings, indicating that SbMATE can also transport organic cationic compounds, such as ethidium, brings these clades closer together under the larger umbrella of MATE family transporters. It is interesting, however, that propidium, which is a bivalent structural and functional analog of ethidium, was not transported as efficiently (Fig. 5). Although propidium's mass is slightly larger than that of ethidium, substrates of MATE family transporters generally have a wide range of molecular weights 25 . It is thus likely that the lack of propidium transport by SbMATE suggests an inherent limitation on the number of positive charges that can be accommodated within SbMATE's substrate-binding site(s), and is not simply an effect of differential mass, given the diverse sizes of MATE-family substrates.
The discovery of the MATE family of transporters was initially accompanied by mechanistic studies on NorM, showing Na + /substrate-antiport energetics 26,27 . On the other hand, studies on mammalian MATE family transporters revealed H + /substrate antiport mechanics 28,29 . Although K + is the primary cationic osmoticum that is essential in maintaining a positive turgor pressure in plant cells, in some instances Na + can partially fulfil this role. Our mechanistic studies on SbMATE strongly suggest that its transport cycle could be driven by both, proton and/or cation (K + or Na + ) gradients, which indicates a degree of flexibility in the ion-binding site(s). However, given the high affinity for H + , and the fact that H + constitute a main electrochemical driving force in plants, it is likely that H + constitutes the main coupling ion for SbMATE. Similar ionic flexibility was recently shown for the prototypical bacterial MATE transporter, NorM 15 , suggesting that such energetics could be more widespread among other MATE family transporters. The ethidium transport activity of SbMATE in yeast was greatly reduced (to 33%) but not abolished, when the extracellular pH was lowered from pH 8.5 to 6.0 (Fig. 5A). The residual transport activity at pH 6.0 indicates a markedly low pKa for protonation sites, such as D127, on SbMATE compared to other known MATE transporters. This is not surprising, given that SbMATE is activated in sorghum crops growing in acidic soils at pH < 5.0.
While SbMATE-mediated H + influx in Xenopus oocytes was found to be electrogenic ( Fig. 4) 8 , H + /ethidium antiport by SbMATE is expected to be electroneutral, similar to related MATE transporters 25 , and likely involves cytoplasmic H + exchange for ethidium uptake by cells. Perhaps the most interesting among our findings in yeast is that ethidium uptake continued to occur amidst (i) an almost non-existent outwardly-directed proton gradient (pH out of 4.5), (ii) a substantial inwardly-directed sodium gradient at 157 mM Na + out , and (iii) in the presence of high levels of the physiological substrate, citrate. At this point, it is unclear as to what exactly drives this residual ethidium transport observed under these conditions. Yeast have dedicated carboxylate transporters and citrate present in the culture medium is generally bioavailable. Citrate should have been readily available for transport via the membrane-embedded SbMATE protein [30][31][32] . Although it is possible that citrate may have been transported by endogenous pumps into yeast cells and its efflux by SbMATE was exchanged for ethidium (i.e. Et + /citrate antiport), the data also bears similarity to the transporter-mediated facilitated downhill uniport fluxes observed in the absence of proton-motive forces for secondary transporters of the Major Facilitator Superfamily (MFS) 33 . While additional experimentation is required to gain further mechanistic insights, our data clearly suggest multiple energetic modes for SbMATE, depending on the substrate being transported, in addition to implicating a strong physiological relevance for organic cationic substrates of SbMATE that might interfere/compete with citrate in acidic soils.
The SbMATE residue D127 is highly conserved as a negatively charged residue throughout bacteria and all plant MATE transporters associated with organic acid transport, while being highly conserved as a neutral (serine or threonine) residue in plant MATE transporters associated with the transport of other plant metabolites (Fig. 3A). In the bacterial MATE transporters, the corresponding carboxylate residue is D32 in NorM from Vibrio parahaemolyticus, D36 in NorM from Vibrio cholerae, D41 in NorM from Neisseria gonorrheae, D41 in the PfMATE transporter from Pyrococcus furiosus, and D40 in DinF-BH from Bacillus halodurans 15,17,18,20,28 . These residues have been shown to be crucial for transport activity, and are likely involved in proton/sodium and/or ion-substrate competition. Consistent with these findings, our protonation-mimetic mutation, D127N, severely impaired SbMATE's ability to transport ethidium in yeast (Fig. 3). Thus, the eukaryotic SbMATE transporter appears to show remarkable structure-function homology with bacterial MATE transporters, even with a considerably extended intracellular N-terminal region that is absent in the bacterial MATE transporters studied to date. Future studies aimed at evaluating the effect of D127N in the electrogenic H + /Na + /K + transport recorded in Xenopus oocytes expressing SbMATE should prove interesting.
In vitro, functional biochemistry and structural biology of plant-derived integral membrane proteins have been hampered due to the lack of reliable, large-scale recombinant protein production systems. Our P. pastoris yeast system is capable of yielding milligram levels of purified SbMATE protein, which was found to be functional in terms of binding to the transport substrate, ethidium, but not to propidium (Fig. 6). The channel-like current activity observed in our electrophysiological experiments extends recent reports of other active transporters exhibiting similar currents 21,34 . These protein preparations can be used to screen additional transport substrates, and perhaps even transport activators that could improve Al-tolerance phenotypes.
Our panel of SbMATE-binding Nbs (Fig. 8), including some that compete with ethidium, display agonistic effects on the SbMATE-mediated electrogenic transport in oocytes (Fig. 10), when administered extra-or intracellularly. These results suggest that one or more Nb-binding epitope(s) on SbMATE are accessible from either side of the membrane, such as the substrate-binding sites in the central cavity of the transporter. As such, the agonism observed could possibly be a result of Nb-binding induced proton/sodium transport, similar to substrate-induced proton transport seen in NorM, PmpM, DinF, and PfMATE 15,17,35,36 . Such panels of Nbs are useful, not only as tools to study protein trafficking under various environmental conditions, but to also potentially alter transporter efficiencies, and therefore, their functional roles in planta.

Conclusion
In this study, we have used a combination of Xenopus oocytes and a yeast-based heterologous recombinant expression system to functionally characterize the transporter protein, SbMATE from Sorghum bicolor, which is responsible for imparting Al-tolerance in sorghum. In doing so, we have extended the substrate recognition profile of SbMATE, demonstrating a H + /cation coupling transport system with the ability to not only transport citrate, but also organic monovalent cations, such as ethidium. We have also shown that yeast-expressed SbMATE can be purified in large quantities, and that the purified protein is functionally-folded and amenable as an antigen for in vitro antibody discovery. The overall Nb discovery methodology is also applicable to other plant-derived membrane proteins, allowing for the rapid development of reagent tools to probe protein localization and function.

Methods and Materials
All methods were carried out in accordance with relevant guidelines and regulations. All experimental protocols were approved by a UC San Diego Office of Research committee.
Expression and Functional characterization of SbMATE in Xenopus oocytes. cRNAs were prepared using the mMessage mMachine T7 in vitro transcription kit (Ambion, http://www.ambion.com/) with 1 µg of ScaI-linearized T7TS plasmid template containing the SbMATE coding region, flanked by the 3′-and 5′-untranslated regions of a Xenopus β-globin gene. Xenopus laevis defolliculated oocytes were purchased from Xenopus1 (Dexter. MI, USA). Oocyte microinjection and whole-cell recordings under two electrode voltage clamp methods were performed as described previously 37 . Briefly, V-VI stage oocytes were injected with 48 nL of water containing 20 ng of cRNA encoding SbMATE. To measure 14C Citrate efflux, control and SbMATE-expressing oocytes (2 days after cRNA injection) were microinjected with 23 nL of 460 µM 14C citrate (1 nCi/oocyte). The cells (6 cells per replicate; n = 4) were allowed to recover for 3 min in ice-cold Ringer solution (pH 7.5) and then transferred into 1.5 mL Ringer solution (pH 7.5 or 4.5) at room temperature. At the indicated time points, 0.75 mL of the bathing solution was sampled and replaced with fresh buffer. At the end of each experiment the cells were disrupted in scintillation fluid. Radioactivity from sampled efflux buffer at the various time points and remaining radioactivity in the oocytes was counted using full-spectrum DPM counting in a Beckman Coulter LS6500 Liquid Scintillation counter. For two-electrode voltage clamp, whole-cell currents were     . Higher affinity Nbs inhibit ethidium binding to SbMATE. (A) SPR was performed as described in Fig. 6B, with the following modifications. 50 µM ethidium bromide was injected either directly on SbMATE alone, or following injections of 5 µM Nb 86 or Nb 94 without a subsequent regeneration step. The ethidium trace (top/black) represents double-referenced data as described in Fig. 6B, while the ethidium +Nb traces (Nb 86 -bottom/red; Nb 94 -middle/blue) were generated using a similar double-referencing method, except the respective Nb injections alone were used instead of a buffer injection for the blank subtraction (n = 3 independent replicates, sensorgram is from a single experiment). (B) The response value (in RUs) at the end of the association phase (i.e. at ~118 s) of the injections described in Fig. 9A were normalized, with the RU value for the ethidium-SbMATE trace set to 100%. Response RUs were reduced to 35.04 ± 13.39% for ethidium-Nb86-SbMATE and to 71.40 ± 13.91% for ethidium-Nb94-SbMATE (n = 3 independent replicates, mean ± SEM). (C) SPR was performed as described in Fig. 9A, except the non-specific hydrophobic control protein, BSA was immobilized on the sensor chip, instead of SbMATE. To obtain comparable ethidium signals, 2x higher levels of BSA than SbMATE were immobilized.
P. pastoris cells (expressing SbMATE, mutants, or the non-expressing vector control) were washed with standard phosphate buffered saline (PBS), pH 7.4, and diluted to an OD 600 of 15 per 1 mL of experimental sample. The assay buffer for experiments conducted at pH 8.5 was 25 mM Tris-Cl, pH 8.5, 137 mM NaCl, and 2.7 mM KCl. Buffers for the different pH conditions were: pH 7.4 -PBS; pH 6.0-50 mM MES; pH 6.0, 137 mM NaCl, 2.7 mM KCl; pH 4.5-20 mM sodium citrate; pH 4.3, 137 mM NaCl, 2.7 mM KCl. Drugs, ethidium bromide or propidium iodide were incubated with cell suspensions for 20 min at RT, followed by washes in respective ice-cold buffer(s) and measurements by flow cytometry (Novocyte, Acea Bioscience). Approximately 12,000 events were collected per sample and the appropriately gated data were displayed qualitatively as histograms, and quantified through median fluorescent intensities. Additional experimental details are available in the figure legends.
Purification of SbMATE from Pichia pastoris. Cells were lysed at 40 Kpsi by a single pass through a constant cell disrupter. Cellular debris was removed by centrifugation (12,500 x g, 20 minutes, 277 K) and crude membranes prepared by centrifugation at 38,000 x g for 2-3 hours at 277 K. Expressed SbMATE was bDDM detergent-solubilised from membranes and purified by affinity chromatography using the histidine tag and subsequent size-exclusion chromatography.
Surface Plasmon Resonance Binding Studies. SPR measurements were performed on the BiOptix 404Pi instrument using CMD sensorchips. Detergent-purified SbMATE was covalently immobilized using EDC-NHS chemistry to 5, 000 resonance units (RUs) per sensor channel in a buffer containing 10 mM MES pH 6.0 and 0.05% bDDM. The running buffer was switched to 50 mM HEPES, pH 8.0, 150 mM NaCl, and 0.03% bDDM for the binding experiments with ethidium bromide, propidium iodide, or Nbs. The drug solutions were prepared and Nbs were desalted into the running buffer to minimize buffer-related effects on the SPR traces. Regenerations were performed using short injection pulses of 50 mM glycine, pH 2.1. All data were analyzed using the Scrubber (BioLogic) software, and subjected to the software-associated double-subtraction processing, curve-fitting, and fit statistics. Additional experimental details are available in the figure legends. Recordings were in SbMATE-expressing oocytes that were preloaded with a Nb mixture or water at 2.5 h prior to the recording. Preloading was achieved by injecting each cell with 48 nL of water or a 100 fold Nbs mix used in A. All electrophysiological recordings were performed in the absence of exogenous intracellular citrate or ethidium bromide (i.e. cell which were not microinjected with any substrate prior to the recordings).