Zinc transporter 10 (ZnT10)-dependent extrusion of cellular Mn²⁺ is driven by an active Ca²⁺-coupled exchange

Abstract

Manganese (Mn²⁺) is extruded from the cell by the zinc transporter 10 (ZnT10). Loss of ZnT10 expression caused by autosomal mutations in the ZnT10 gene leads to hypermanganesemia in multiple organs. Here, combining fluorescent monitoring of cation influx in HEK293-T cells expressing human ZnT10 with molecular modeling of ZnT10 cation selectivity, we show that ZnT10 is exploiting the transmembrane Ca²⁺ inward gradient for active cellular exchange of Mn²⁺. In analyzing ZnT10 activity we used the ability of Fura-2 to spectrally distinguish between Mn²⁺ and Ca²⁺ fluxes. We found that (a) application of Mn²⁺-containing Ca²⁺-free solution to ZnT10-expressing cells triggers an influx of Mn²⁺, (b) reintroduction of Ca²⁺ leads to cellular Mn²⁺ extrusion against an inward Mn²⁺ gradient, and (c) the cellular transport of Mn²⁺ by ZnT10 is coupled to a reciprocal movement of Ca²⁺. Remarkably, replacing a single asparagine residue in ZnT10 (Asp-43) with threonine (ZnT10 N43T) converted the Mn²⁺/Ca²⁺ exchange to an uncoupled channel mode, permeable to both Ca²⁺ and Mn²⁺. The findings in our study identify the first ion transporter that uses the Ca²⁺ gradient for active counter-ion exchange. They highlight a remarkable versatility in metal selectivity and mode of transport controlled by the tetrahedral metal transport site of ZnT proteins.

Introduction

The mammalian SLC30 (solute-like carrier 30)³ family of Zn²⁺ transporters (ZnT) includes 10 membrane-embedded proteins (1) and is part of the broader cation diffusion facilitator (CDF) family that spans across bacteria, fungi, and plants (2). ZnTs are associated with numerous Zn²⁺-related pathophysiologies, for instance Alzheimer's (5, 6), and type II diabetes (7, 8). Most mammalian ZnT members mediate cellular and vesicular Zn²⁺ transport by exchanging it with H⁺ (3, 4). Despite their modest general sequence homology, ZnT members share a conserved tetrahedral metal transport site with plant, fungal, and bacterial transporters (4, 9, 10).

In contrast to other ZnT members that transport Zn²⁺, ZnT10 emerges as a Mn²⁺ transporter (11–13). This novel function was first discovered by analysis of homozygous mutations in human ZnT10 gene. Patients carrying these mutations manifested severe hypermanganesemia associated with onset of parkinsonism, polycythemia, and chronic liver disease (14, 15). These studies were followed by functional analysis in neurons showing that ZnT10 is linked to Mn²⁺ efflux, which is critically determined by an asparagine (Asn) residue replacing a conserved histidine (His) at the tetrahedral metal transport site of ZnT10 (14, 15). How ZnT10 is able to mediate Mn²⁺ extrusion is unknown however. Manganese gradient across the cell membrane is very steep and can reach more than 3000-fold (16). Moreover, in contrast to Zn²⁺, Mn²⁺ is redox active and may induce oxidative damage (17, 18). Hence, removal of Mn²⁺ requires a counter ion distributed in an electrochemical gradient that can power the cellular extrusion of Mn²⁺. Although vesicular H⁺ gradient may reach 100-fold (19), the magnitude of the H⁺ gradient across the cell membrane is only ∼5- to 10-fold (20), which would be insufficient to support cellular Mn²⁺ extrusion. In contrast, Ca²⁺ is distributed at 4–5 orders of magnitude inward facing electrochemical gradient across the cell membrane and thus could potentially provide the driving force for Mn²⁺ efflux (21).

Here we show that ZnT10-dependent Mn²⁺ efflux is coupled to Ca²⁺ exchange. Hence, ZnT10 is the first known ion transporter that utilizes the steep transmembrane Ca²⁺ gradient. Finally, we demonstrate that a single mutation in the ZnT10 metal transport site renders ZnT10 exchanger to a cationic-like channel.

Results

Functional analysis of Mn²⁺ and Zn²⁺ transport by ZnT10

The properties of ZnT10 metal selectivity are not fully resolved. Specifically it is not clear whether ZnT10 transports both Zn²⁺ and Mn²⁺ (22, 23) or only Mn²⁺ (24). To determine whether ZnT10 mediates Zn²⁺ transport we monitored cellular Zn²⁺ influx in HEK293-T cells expressing human ZnT10 or transfected with vector alone (pcDNA). Cells were preloaded with the high-affinity fluorescent Zn²⁺ dye, FluoZin-3AM and were then superfused with a 50 μm Zn²⁺-containing, Ca²⁺-free, Ringer's solution. Despite the presence of high Zn²⁺ concentration in the solution, and in agreement with previous studies (11, 13, 25), no increase in cellular Zn²⁺ influx was observed in cells expressing ZnT10 (Fig. 1, a and b). Fura-2AM is a widely used Ca²⁺ reporter that was also previously used as a Mn²⁺-sensitive dye when excited at 360 nm wavelength (26). When cells were perfused with Mn²⁺-containing, Ca²⁺-free, Ringer's solution, the Mn²⁺ influx rate in ZnT10-expressing cells was increased by more than 10-fold compared with control cells (Fig. 1, c and d). Using the same paradigm we then performed Mn²⁺ dose-response analysis (Fig. 1, e and f) and found that ZnT10 is capable of conducting high-affinity Mn²⁺ transport with a K_½ = 1.235 ± 0.064 μm.

Figure 1.

ZnT10 mediates Mn²⁺ but not Zn²⁺ transport. a, representative traces of Zn²⁺ influx in ZnT10 (red) versus pcDNA (black) transfected cells loaded with FluoZin-3 and monitored for cytoplasmic Zn²⁺ transport. Cells were initially superfused with Ringer's solution and then with Ringer's solution containing Zn²⁺ (50 μm) at the time intervals indicated by the horizontal bar. b, mean rates of cytoplasmic Zn²⁺ uptake for ZnT10 (n = 7) and pcDNA (n = 9) taken from a (unpaired t-test). c, representative traces of Mn²⁺ influx in ZnT10 (red) versus pcDNA (black) transfected cells loaded with Fura-2AM and monitored for cytoplasmic Mn²⁺ transport. Cells were initially superfused with Ringer's Ca²⁺-free solution and then with Ringer's Ca²⁺-free solution containing Mn²⁺ (5 μm) applied as indicated by the horizontal bar. d, mean rates of cytoplasmic Mn²⁺ uptake for ZnT10 (n = 52) and pcDNA (n = 29) taken from c (unpaired t-test; ****, p < 0.0001). e, representative traces of Mn²⁺ dose-response analysis of ZnT10 transfected cells. Mn²⁺ was added at the indicated concentrations and Mn²⁺ influx was monitored as described in c. f, mean rates of cellular Mn²⁺ uptake taken from e as a function of Mn²⁺ concentration (allosteric sigmoidal fit; n = 4; K_½ = 1.235 ± 0.064, μm; V_max = 4.931 ± 0.477, a.u.).

ZnT10 does not utilize the H⁺ or Na⁺ transmembrane gradient for Mn²⁺ transport

All known ZnT family members use a H⁺-driven gradient for metal exchange (4). We therefore asked if ZnT10 also utilizes transmembrane H⁺ gradients for transport of Mn²⁺. We used the same protocol for Mn²⁺ uptake as described in Fig. 1c while superfusing the cells with Ringer's solutions titrated at an extracellular pH range of 6–8 (Fig. 2a). At pH values of 7.4 and 8, Mn²⁺ influx was maximal, diminished at pH 7 and was totally blocked at the acidic pH 6 (Fig. 2, a and b). Such modulation of Mn²⁺ influx rate by pH can be either linked to regulation by pH or to direct H⁺ coupled transport. To distinguish between these two mechanisms, we first asked if ZnT10 can utilize the H⁺ gradient to support Mn²⁺ efflux. We loaded the cells with Mn²⁺ exploiting the Mn²⁺ influx mode of ZnT10 at pH 7.4, by superfusing ZnT10-expressing cells with Mn²⁺-containing Ringer's solution. This was followed by superfusion with Mn²⁺-free Ringer's, and comparison of rates of Mn²⁺ efflux at pH 7.4 versus pH 6 (Fig. 2, c and d). No change in Mn²⁺ efflux rate at any of these pH values was observed (Fig. 2, c and d). Another criterion for H⁺-coupled exchange, previously shown for ZnT members and YiiP (4), is a metal-driven counter H⁺ transport manifested by pH_i change. We therefore compared cytosolic pH changes in ZnT10- or vector- expressing cells loaded with the intracellular fluorescent pH indicator BCECF. No Mn²⁺-dependent pH_i changes in either ZnT10- or vector-expressing cells were observed (Fig. 2, e and f). Altogether, this set of experiments indicates that although ZnT10 can be regulated by extracellular pH it does not utilize the H⁺ gradient for mediating a counter H⁺/Mn²⁺ transport.

Figure 2.

ZnT10 is regulated by pH but does not conduct H⁺/Mn²⁺ or Na⁺/Mn²⁺ exchange. a, representative traces of Mn²⁺ (5 μm) uptake in ZnT10 transfected cells. Cells were superfused with Ringer's solution at the indicated pH values. Mn²⁺ influx was monitored as described in Fig. 1c. b, mean rates of cellular Mn²⁺ uptake derived from a (one-way ANOVA test; n = 3; ***, p < 0.001; **, p < 0.01). c, Mn²⁺ efflux in cells expressing ZnT10 is unaffected by extracellular pH changes. Representative traces of Mn²⁺ (5 μm) transport in ZnT10 (blue and red) and pcDNA (black) transfected cells that were loaded with Fura-2AM and monitored at the indicated pH values. Cells were first superfused as in Fig. 1c with pH 7.4 Ringer's solution containing Mn²⁺ (5 μm) as indicated by the left horizontal bar, then as indicated by the right horizontal bar with Mn²⁺-free Ringer at either pH 7.4 Ringer's solution (blue) or pH 6 (red and black) Ringer's solution. d, mean rates of cytoplasmic Mn²⁺ fluxes of cells superfused with a Ringer's solution at pH 7.4 (n = 11) or pH 6 (one-way ANOVA test, ZnT10 n = 11; pcDNA n = 4) taken from c. e, representative traces of cellular pH_i values in ZnT10 (red and cream) or pcDNA (black) transfected cells preloaded with BCECF-AM. Ringer's solution with (red and black) or without (cream) Mn²⁺ (5 μm) that was added as indicated by the horizontal bar. f, mean rates of cytoplasmic pH changes taken from e of Mn²⁺-treated ZnT10 (n = 21), pcDNA (n = 13), and Mn²⁺ untreated ZnT10 (n = 5) transfected cells (one-way ANOVA test). g, effect of Na⁺ on Mn²⁺ transport by ZnT10. Representative traces of Mn²⁺ (5 μm) influx in ZnT10 (red and brown) versus pcDNA (black and gray) transfected cells loaded with Fura-2AM and monitored for cytoplasmic Mn²⁺ transport as described in c. Cells were superfused with Na⁺ full Ringer's solution containing Mn²⁺ (red and black) or Na⁺-free NMDG buffer containing Mn²⁺ (brown and gray) as indicated in the graph. h, mean rates of cellular Mn²⁺ uptake of ZnT10 transfected cells in Na⁺ full (red, n = 4) or Na⁺-free (brown, n = 3) Ringer's solution. Data are derived from g and normalized to the background (unpaired t-test).

Previous studies reported a Na⁺-dependent Zn²⁺ efflux (27). Further, transmembrane Na⁺ gradient is used by numerous Na⁺ coupled exchangers. We therefore asked if ZnT10 conducts Na⁺/Mn²⁺ exchange. We used the same paradigm for Mn²⁺ uptake as described in Fig. 1c, perfusing cells in the presence or absence of extracellular Na⁺ that was isoosmotically replaced by NMDG. No difference in Mn²⁺ influx rate was monitored in cells superfused with Na⁺-containing or Na⁺-free Ringer's, indicating that ZnT10 does not conduct Na⁺-dependent Mn²⁺ transport (Fig. 2, g and h).

ZnT10 is the first known transporter that utilizes the Ca²⁺ gradient

Because H⁺ or Na⁺ failed to support ZnT10-dependent Mn²⁺ transport, we asked if the transport site can bind Ca²⁺. Applying a structural/modeling-based comprehensive analysis, we scanned the RCSB Protein Data Bank and extracted all the protein structures containing bound Ca²⁺, in addition to Mn²⁺ or Zn²⁺ (see “Experimental procedures”), and then analyzed for each metal ion to which residues it tends to bind and in what coordination number (Fig. 3, a and b; full data in Tables S1–S3). This analysis shows clear differences between the binding patterns of each metal: Zn²⁺ tends to be bound in a coordination number of 4, whereas both Mn²⁺ and Ca²⁺ are usually bound in a higher coordination number (5–6 and 6–7, respectively). In terms of preferred residues, Ca²⁺ shows very low preference to bind histidine residues, whereas Zn²⁺ shows high preference to histidine and cysteine, and Mn²⁺ is mostly bound to aspartate, followed by glutamate and histidine.

Figure 3.

Data analysis poses Ca²⁺ as a possible candidate to support Mn²⁺ exchange by ZnT10. a, coordination number statistics of (left to right) Zn²⁺ (dark purple), Mn²⁺ (purple), and Ca²⁺ (pink) show strong preferences for 4-coordination of Zn²⁺ and higher coordination number preferences for both Mn²⁺ (5, 6) and Ca²⁺ (6, 7). b, selected amino acid tendencies of (left to right) Zn²⁺ (dark purple), Mn²⁺ (purple), and Ca²⁺ (pink). Filled parts refer to the percentage of the side-chain's polar atoms that are bound to the metal. c, structural model of ZnT10 was calculated based on the X-ray structure of the CDF protein YiiP (PDB code: 3H90) (48). Each monomer is represented in a different color, A-site residues are presented in sticks. Unstructured domains that are not presented in the model are residues 1–7, the His-rich loop (residues145–226), and the C-terminal (residues 386–485). d, magnification of the tetrahedral binding site: ZnT10 WT site occupied with Mn²⁺ ion in yellow (radius of 0.9 Å), ZnT10 Asn-43 substitution to His (ZnT10 N43H) creates tighter site and different coordination ability of the metal, ZnT10 Asn-43 substitution to Asp (ZnT10 N43D) maintains the cavity size but adds an extra charge, ZnT10 Asn-43 substitution to Thr (ZnT10 N43T) creates similar-sized polar cavity and ZnT10 Asn-43 substitution to Ala (ZnT10 N43A) creates a bigger cavity but neutralizes the site.

The canonical tetrahedral metal-binding site of the Zn²⁺-transporting ZnT proteins is composed of a 2His–2Asp formation (4, 9). In contrast, the ZnT10 site is composed of Asn-Asp-His-Asp (Asn-43–Asp-47–His-244–Asp-248) (see Fig. 3, c and d and Fig. S1). As shown in Fig. 3b, Ca²⁺ has the highest preference to the Asn residue, followed by Mn²⁺, whereas Zn²⁺ shows poor interaction with this residue. Thus, the ZnT10-binding site suggests more efficient transport of Mn²⁺ and Ca²⁺, than of Zn²⁺.

To determine whether Ca²⁺ drives Mn²⁺ transport by ZnT10 we applied the same functional criteria that we used to interrogate the H⁺ or Na⁺ transport. To monitor and distinguish between Mn²⁺ and Ca²⁺ fluxes we used Fura-2AM that can very effectively distinguish between Mn²⁺ and Ca²⁺ when excited at 360 nm or 340 nm/380 nm, respectively (28–30). To ascertain that we can distinguish between fluxes of these cations using a 360 nm versus 340 nm/380 nm spectral analysis, we monitored Mn²⁺ and Ca²⁺ influx (Fig. S3) induced by Ca²⁺ store depletion. Consistent with previous studies (28–30), Ca²⁺, but not Mn²⁺, influx was fluorescently monitored at a ratiometric 340 nm/380 nm excitation. Conversely Mn²⁺, but not Ca²⁺, influx was monitored when cells were excited at 360 nm (Fig. S3). Using this experimental paradigm, we then asked if extracellular Ca²⁺ can modulate the rate of Mn²⁺ transport by ZnT10 (Fig. 4, a–c). We found that Mn²⁺ influx was totally blocked when the cells were superfused with Mn²⁺ (5 μm) and Ca²⁺ (1.8 mm, initial phase of purple trace, Fig. 4, a and c) containing solution. Note that both Ca²⁺ and Mn²⁺ concentrations used are within the physiological range. In contrast, Mn²⁺ influx was observed in Ca²⁺-free Ringer's solution (initial phase of red trace, Fig. 4, b and c). Accordingly, in cells loaded with Mn²⁺ (in the absence of Ca²⁺), Mn²⁺ efflux depended on the presence of physiological Ca²⁺ (late phase of red trace, Fig. 4, b and c). Furthermore, Ca²⁺-dependent Mn²⁺ efflux persisted even in the presence of extracellular Mn²⁺ (late phase of brown trace, Fig. 4, b and c). Thus, extracellular Ca²⁺ can effectively drive active cellular Mn²⁺ efflux, even against a steep inward Mn²⁺ gradient (Fig. 4b). If ZnT10 is mediating Ca²⁺/Mn²⁺ exchange, it should manifest Ca²⁺ fluxes reciprocal to Mn²⁺ transport. We therefore monitored cytosolic Ca²⁺ by ratiometrically monitoring Fura-2AM at the Ca²⁺-sensitive 340 nm/380 nm ratiometric excitation wavelength. We found that addition of Mn²⁺ to Ca²⁺-free solution, led to a drop in cytosolic Ca²⁺ concentrations in ZnT10-expressing cells but not in vector transfected cells (Fig. 4, d and e). In the ZnT10-expressing cells, subsequent addition of Ca²⁺ led to robust Ca²⁺ influx (Fig. 4, d and e), thus manifesting a clear pattern of Mn²⁺/Ca²⁺ exchange. Finally, we conducted a dose-response analysis of the extracellular Ca²⁺ concentrations required for driving Mn²⁺ efflux in cells loaded with Fura-2AM, excited at 360 nm, and found that the K_½ for Ca²⁺ is 0.97 ± 0.03 mm (Fig. 4, f and g). Thus our results indicate that the functional properties of ZnT10 are tailored for a physiological Ca²⁺-driven Mn²⁺ exchange, promoting active cellular Mn²⁺ extrusion.

Figure 4.

ZnT10 Mn²⁺ efflux is driven by Ca²⁺ influx. a and b, representative traces of Ca²⁺-dependent Mn²⁺ exchange. Mn²⁺ transport was monitored in ZnT10 (purple, red, and brown) or pcDNA (black) transfected cells preloaded with Fura-2AM and excited at 360 nm as described in Fig. 1c. Cells were first superfused with Ringer's solution containing Mn²⁺ (5 μm) in the presence or absence of Ca²⁺ (1.8 mm) as indicated by the left horizontal bar, then superfused with Ringer's solution containing Ca²⁺ (1.8 mm) in the presence or absence of Mn²⁺ (5 μm) as indicated by the right horizontal bar. c, mean rates of cellular Mn²⁺ influx (unpaired t-test; red, n = 8; purple, n = 10) and efflux (one-way ANOVA test; red, n = 8; brown, n = 8; purple, n = 3; black, n = 13) values taken from a and b (****, p < 0.0001; ***, p < 0.001; *, p < 0.05). d, representative traces of Mn²⁺-dependent Ca²⁺ transport. Changes in intracellular Ca²⁺ levels were monitored in ZnT10 (red) versus pcDNA (black) transfected cells. Cells were preloaded with Fura-2AM and excited at 340 nm and 380 nm (see “Experimental procedures”). Ringer's solution containing Mn²⁺ (5 μm) and Ca²⁺ (1.8 mm) were added as indicated by the horizontal bars. e, mean rates of cellular Ca²⁺ efflux and influx rates in ZnT10 (red, n = 19) versus pcDNA (black, n = 25) transfected cells taken from d. (unpaired t-test, ****, p < 0.0001). f, representative traces of Ca²⁺-dependent Mn²⁺ fluxes dose-response analysis in ZnT10 transfected cells. Cells were loaded with Fura-2AM and excited at 360 nm as described in Fig. 1c. Cells were first superfused with Ca²⁺-free Ringer's solution containing Mn²⁺ (5 μm) as shown by the left horizontal bar and then superfused with Ringer's solution containing the indicated Ca²⁺ concentrations. g, mean rates of cellular Ca²⁺-dependent Mn²⁺ efflux taken from f at the indicated concentrations (allosteric sigmoidal fit; n = 4; K_½ = 973.6 ± 29.98, μm; V_max = 1.234 ± 0.08, a.u.).

ZnT10 tetrahedral Asn-43 position determines a channel versus exchanger mode of transport

The canonical ZnT tetrahedral site contains a His residue, which is bigger than the Asn residue found in ZnT10 tetrahedral site (Asn-43, Fig. 3d), hence making the canonical ZnT site tighter for metal binding. This suggests that Asn-43 may be a key residue to control the mode of ion transport by ZnT10. We therefore compared the rates of transport by ZnT10 constructs mutated at the Asn-43 position. Note that all constructs were expressed, as determined by Western blot analysis (Fig. S2). We initially monitored Mn²⁺ influx (as described in Fig. 1c) in ZnT10 N43H, expressing cells, simulating the canonical site, and found that it does not mediate Mn²⁺ transport (N43H blue trace, Fig. 5, a and b). Similarly the ZnT10 N43D, mutant simulating the bacterial site (10), was nonfunctional (N43D orange trace, Fig. 5, a and b). In contrast, mutating the Asn-43 to a noncharged alanine (N43A brown trace, Fig. 5, a and b) or to a polar threonine (N43T green trace, Fig. 5, b and c) enhanced Mn²⁺ influx compared with WT ZnT10. We then asked if these mutations affect the coupling of Mn²⁺ to Ca²⁺ transport by monitoring Mn²⁺-dependent Ca²⁺ efflux. We found that in cells expressing the ZnT10 N43A mutation, Mn²⁺-dependent Ca²⁺ efflux was ∼3-fold higher compared with WT ZnT10 (brown trace, Fig. 5, d and f), whereas the N43T mutation resulted in a significant decrease in Ca²⁺ efflux (green trace, Fig. 5, d and f). Our results therefore indicate that the Asn-43 position of ZnT10 controls the coupling of Mn²⁺ to Ca²⁺. The N43A mutant enhances coupling between Mn²⁺ and Ca²⁺ exchange, whereas the N43T mutant induces uncoupled channel-like Mn²⁺ transport activity (Fig. 5, c and d). Consistent with this channel-like mode of the N43T mutant, comparison to WT ZnT10 kinetics showed that the apparent affinity of ZnT10 N43T to Mn²⁺ is lower and its maximal rate of Mn²⁺ transport is higher with respect to WT ZnT10 (Fig. 5, e and f and Table S5). Finally, we studied the role of the two additional residues on the tetrahedral site, ZnT10 His-244 and ZnT10 Asp-47. We found that even the conservative substitutions of ZnT10 H244D (mimicking other known bacterial Mn²⁺ CDF transporters; see Table S4) or ZnT10 D47E, were sufficient to eliminate Mn²⁺ transport (Fig. S3), indicating that similarly to other ZnTs these positions are essential and indispensable for metal cation selectivity and transport (4, 9, 31). Other mutations at these positions (ZnT10 D47A, ZnT10 H244A, ZnT10 D248A, ZnT10 D248A, and ZnT10 D248E) impaired ZnT10 expression and therefore are not included in the analysis (Fig. S3). Altogether, the results of this set of experiments identified ZnT10 Asn-43 position as a critical residue not only for promoting Mn²⁺ transport by ZnT10 but also for controlling an exchanger versus channel-like mode of ZnT10 transport (Fig. 5g).

Figure 5.

ZnT10 43 position controls exchange versus channel-like mode of operation by ZnT10. a, representative traces of Mn²⁺ influx in cell transfected with either ZnT10 WT (red), ZnT10 N43H (blue), ZnT10 N43D (orange), ZnT10 N43T (green), ZnT10 N43A (brown), or pcDNA (black) preloaded with Fura-2AM and excited at 360 nm. Cells were superfused with Ca²⁺-free Ringer's solution and then Ringer's solution containing Mn²⁺ was added when indicated by the horizontal bar. b, mean rates of cellular Mn²⁺ influx for ZnT10 WT (n = 31), pcDNA (n = 21), ZnT10 N43H (n = 7), ZnT10 N43D (n = 4), ZnT10 N43T (n = 11), ZnT10 N43A (n = 7) taken from a (one-way ANOVA test; *, p < 0.05; **, p < 0.01; ****, p < 0.0001). c, representative traces of cytosolic Ca²⁺ efflux as described (Fig. 4d) in ZnT10 WT (red), pcDNA (black), ZnT10 N43T (green), and ZnT10 N43A (brown) transfected cells. d, mean rates of cellular Ca²⁺ efflux of ZnT10 WT (red, n = 19), pcDNA (black, n = 25), ZnT10 N43T (green, n = 10), or ZnT10 N43A (brown, n = 11) expressing cells taken from c (one-way ANOVA test; *p < 0.05; ****, p < 0.0001). e, dose-response analysis of Mn²⁺ transport by ZnT10 N43T mutant. Representative traces of Mn²⁺ influx in ZnT10 N43T transfected cells preloaded with Fura-2AM and exited at 360 nm. Cells were superfused with Mn²⁺ at the indicated concentration and Mn²⁺ influx was monitored as described in Fig. 1e. f, mean rates of cytoplasmic Mn²⁺ influx taken from e (Michaelis-Menten's fit; n = 4; K_m = 4.61 ± 1.815, μm; V_max = 14.25 ± 2.294, a.u.). g, schematic representation of ZnT10 and ZnT10 N43T acting as an exchanger and channel-like, respectively.

Discussion

The results presented in this study show that ZnT10 is the first known transporter that harnesses the steep transmembrane Ca²⁺ gradient for active removal of Mn²⁺, which is playing a double-edged sword role in cells. On the one hand, Mn²⁺ is an essential cofactor of many enzymes. On the other hand, the presence of Mn²⁺ can be deleterious, triggering generation of harmful oxygen radicals, because of its physiological redox activity (32, 33). Therefore, cellular Mn²⁺ concentration must be tightly controlled. Mn²⁺ can enter cells via several cation channels and zip transporters (34, 35). Removal of Mn²⁺ is more challenging because it requires transport against its steep gradient. Previous studies revealed that ZnT10 plays an important role in maintaining Mn²⁺ homeostasis. Mutations that diminish expression of ZnT10 have been linked to toxic accumulation of Mn²⁺ (14, 15). How ZnT10 can pump out Mn²⁺ against a major electrochemical gradient remained a major unanswered question. The hypothesis that Mn²⁺ extrusion is powered by Ca²⁺-dependent exchange is supported by the following findings: 1) In the absence of extracellular Ca²⁺, Mn²⁺ strongly permeates into ZnT10 expressing cells. 2) In the presence of both Ca²⁺ and Mn²⁺ in the extracellular solution, Mn²⁺ influx is blocked in ZnT10 expressing cells. 3) Addition of physiological Ca²⁺ concertation subsequent to Mn²⁺ influx reverses this process and triggers a Ca²⁺-dependent Mn²⁺ efflux mediated by ZnT10. Thus ZnT10 manifests an exchange mechanism with a “classical” reverse and direct mode of operation. 4) Remarkably, Ca²⁺ can support Mn²⁺ influx by ZnT10 against an inward physiological gradient, thus supporting its role as a secondary active transporter, harnessing the Ca²⁺ gradient to power an active against-gradient extrusion of Mn²⁺. 5) Transport of Mn²⁺ by ZnT10 is coupled to reciprocal Ca²⁺ transport. Thus, we find that influx of Mn²⁺ is linked to Ca²⁺ efflux, whereas ZnT10-dependent Mn²⁺ extrusion is linked to cellular Ca²⁺ influx. By all these criteria ZnT10 is acting as the first metal exchanger powered by extracellular Ca²⁺. What is the physiological rational of switching ZnT10 mode of transport from H⁺ coupled exchange (used by other ZnT members) to Ca²⁺-dependent Mn²⁺ exchange? Although cytoplasmic level of Zn²⁺ is estimated in pm range, Mn²⁺ level may reach several orders of magnitude higher than Zn²⁺ (which is estimated in μm range). Moreover, although Zn²⁺ is relatively redox inert, Mn²⁺ has a physiologically relevant redox potential that may be highly toxic. Thus, the steep Ca²⁺ gradient compared with the modest H⁺ gradient can support a more efficient exchange required to handle the higher Mn²⁺ concentrations. Our analysis (Fig. 3, a–d) further shows that the absence of His pairing at the metal transport site of ZnT10, replaced by Asn, supports a tighter Mn²⁺ than Zn²⁺ binding and may therefore facilitate the rate of Mn²⁺ transport compared with Zn²⁺ transport by other ZnT members (Table S3).

The versatility of the tetrahedral metal-binding site provides an excellent model for a comparative interrogation of structural-functional requirements of Mn²⁺ versus Zn²⁺ transport. We found that ZnT10 fully discriminates between Mn²⁺ and Zn²⁺ and does not conduct any apparent Zn²⁺ transport.

Our results show that pH has a strong regulatory effect on Mn²⁺ transport by ZnT10. The presence of a histidine residue in the transport site, which allows efficient binding only when unprotonated, might be one of the reasons for this. Such regulatory effect is shared by many other transporters and channels, for example NHE and Orai, and suggests a physiological role and pathophysiological implications. For example, cellular acidosis could be followed by impaired ZnT10 transport activity leading to toxic accumulation of Mn²⁺. Intriguingly, previous studies linked brain acidosis to severity of neurodegenerative syndromes such as Alzheimer's disease (36–38) in which high Mn²⁺ concentrations are also found (39–41). Further studies are required to determine whether impaired Mn²⁺ homeostasis during neurodegeneration is linked to modulation of ZnT10 activity.

Finally, our results indicate that apart from conferring Mn²⁺ selectivity the ZnT10 Asn-43 position has an additional function. Substituting it with Thr (ZnT10 N43T) triggers rapid Mn²⁺ or Ca²⁺ fluxes that are independent of the presence of a counter trans ion. The specific change to Thr, compared with Ala and Asp, creates a similar-sized cavity and maintains the polarity of the site without adding extra charge, yet lowers the possible coordination number. This modification may preserve the capability to attract the cations but also allows a better, channel-like release of the ions. Thus, this position is required to couple Ca²⁺ and Mn²⁺ in the exchange mode mediated by ZnT10. Intriguingly, ZnT10 shares similar cation selectivity with Orai1. Both are highly selective for Mn²⁺ and Ca²⁺, rejecting other metals, and the transport rates of both transporters are very modest compared with classical channels (35). A phylogenic analysis suggested that Orai1 shares a similar origin with CDF proteins (42); thus future physiological studies will determine whether they also share common transport sites or mode of operation.

Experimental procedures

Cell culture

HEK293-T cells were cultured in DMEM, supplemented with 10% FCS, 1% streptomycin and 1% penicillin. Cells were grown in either 25 cm² or 75 cm² flasks, in a humidified CO₂ incubator, at 37 °C. For live-cell imaging and immunocytochemistry experiments, cells were transferred onto glass coverslips, in 60-mm cell culture dishes. For immunoblotting, cells were transferred to 100-mm cell culture dishes.

Plasmid transfection

Plasmid transfection was performed as described previously (43). Amounts of plasmid (0.67 μg) used for transfection were calibrated by Western blotting.

Generation of mutants

The plasmids used in this study are variations of hZnT10 double-stranded plasmid. Site-directed mutagenesis was performed using the QuikChange Site-Directed Mutagenesis Kit (Stratagene) according to the manufacturer's protocol. All primers were designed using the primer design tool, on the University of Washington server, and manufactured by Sigma. The primers were as follows: ZnT10 N43H, CGACTCCTTCCACATGCTCTCCGACC; ZnT10 N43A, CCGACTCCTTCGCCATGCTCTCCGACC; ZnT10 N43T, CCGACTCCTTCACCATGCTCTCCGACC; ZnT10 N43D, CCGACTCTTCGACATGCTCTCCGACC; ZnT10 D47A, CAACATGCTCTCCGCCCTGATCTCGCTGTG; ZnT10 D47E, CATGCTCTCCGAGCTGATCTCGCTG; ZnT10 H244A, CAGAGGTGTACTTTTGGCTGTGATGGGAGATGC; ZnT10 H244D, CAGAGGTGTACTTTTGGATGTGATGGGAGATGCC; ZnT10 D248A, CTTTTGCATGTGATGGGAGCAGCCCTGGGGTCCGTGGTTG; ZnT10 D248E, CTTTTGCATGTGATGGGAGAAGCCCTG.

Immunoblot analysis

Cells were extracted using 200 μl of hot lysis buffer (1% SDS, 10 mm Tris-HCl, pH 8) per 100-mm plate and transferred to ice. A protease inhibitor mixture (Boehringer complete protease inhibitor mixture; Roche Applied Science) was added to the lysates, and protein concentrations were determined using the modified Lowry procedure (44). SDS-PAGE and immunoblot analyses were performed, using the anti-ZnT10 (Abbexa) and anti-FLAG (GenScript) antibodies at dilutions of 1:2000. Secondary anti-mouse and anti-rabbit antibodies (Jackson ImmunoResearch Laboratories) were used at dilutions of 1:20000.

Fluorescence imaging

The imaging system consisted of an Axio Vert 100 inverted microscope (Zeiss), Polychrome 4 monochromator (TILL Photonics, Planegg, Germany), and a SensiCam cooled charge-coupled device (PCO, Kelheim, Germany).

Fluorescent imaging measurements were acquired with the Imaging Workbench 6 software (Axon Instruments, Foster City, CA) and analyzed using Microsoft Excel, KaleidaGraph, and GraphPad Prism 6. Cytoplasmic ion transport was determined in cells loaded with either 0.5 μm FluoZin-3 (Zn²⁺ assay), 2 μm Fura-2AM (Mn²⁺ and Ca²⁺ assays), or 1 μm BCECF-AM (H⁺ assay). For Ca²⁺ assays, Mn²⁺ rates were monitored at excitation wavelengths of 360 nm as described previously (28). Ca²⁺ levels were monitored at 340 nm/380 nm (Fura-2AM). Note that while occupied by Ca²⁺, Fura-2AM florescence will rise. In contrast, Mn²⁺ binding will reduce Fura-2AM florescence. Excited at the Ca²⁺ isosbestic point (360 nm), Fura-2AM is no longer sensitive to Ca²⁺ and reflects only Mn²⁺-related changes. The 340 nm/380 nm fluorescence measurements would signal Ca²⁺ while canceling out most of the nonrelated Ca²⁺ signals (45).

Imaging experiments were conducted in the following way: Cells were superfused using Ringer's solution containing 130 mm NaCl, 20 mm Hepes, 15 mm glucose, 5 mm KCl, and 0.8 mm MgCl₂, with the pH adjusted to 7.4 (unless stated otherwise), supplemented with 1.8 mm CaCl₂ in Ca²⁺ Ringer's solution, 5 μm Mn²⁺ in Mn²⁺ Ringer's solution, or 50 μm Zn²⁺ in Zn²⁺ Ringer's solution.

For all single-cell imaging experiments, traces of averaged responses recorded from 5 to 25 cells in each experiment are shown. The rate of ion transport was calculated from each graph (summarizing an individual experiment) by a linear fit of the change in the fluorescence (ΔF/Δt).

Sequence alignment

Sequences for data comparison were obtained using BLAST (46) and sequence alignments were constructed using the multiple sequence alignment program ClustalW (47).

Metals' amino acid preferences and coordination number analysis

Similarly to Barber-Zucker and others (31) a search for all known protein structures was conducted at the RCSB Protein Data Bank on May 2016. For each metal (Mn²⁺, Zn²⁺, and Ca²⁺), the next parameters were used for PDB files filtration: Deposit structure had to contain protein, X-ray resolution was between 0 and 2.5 Å (so only X-ray structures with reliable resolution were used), and the PDB file had to contain a LINK entrance with the Atom label (Mn², Zn²⁺ or Ca²⁺) in a metal coordination connection. For structures with more than 90% sequence identity, only a representative structure was retrieved.

In each structure, if there were symmetric binding sites, only one representative binding site for each metal was taken under consideration. If any ligand other than water (such as other metal ions, DNA molecules, or other small organic ligands) was bound to a metal cation, this metal was excluded from the statistic. Only metals that were bound by at least two atoms from a protein were considered.

Structural model of ZnT10

ZnT10 model was built using the SWISS-MODEL automatic modeling mode (13, 24, 25, 48), based on the structure of Escherichia coli YiiP (4) (PDB code: 3H90). Mutations have been implemented by Swiss-PdbViewer 4.1.0 (22) and the structural model figures were prepared using PyMOL (PyMOL Molecular Graphics System, Version 1.7.4, Schrödinger, LLC).

ZnT10 illustration

Illustration of ZnT10 was performed using the Servier Medical Art illustration resources.

Statistical analysis

All data were statistically analyzed by using GraphPad Prism 6 program. Statistical significance was determined by using the unpaired t-test with Welch's correction (confidence level = 95%) or one-way analysis of variance (ANOVA) test (Tukey's multiple comparisons test, confidence interval = 95%). Data dispersion were calculated as mean ± S.E.

Author contributions

M. L., N. E., and E. H. data curation; M. L., N. E., S. B.-Z., E. H., R. Z., and M. H. formal analysis; M. L. and N. E. investigation; M. L., R. Z., M. H., and I. S. writing-original draft; E. H. and I. S. conceptualization; E. H. methodology; R. Z., M. H., and I. S. writing-review and editing; M. H. and I. S. funding acquisition; I. S. supervision.