Allosteric regulation of Na⁺/I⁻ symporter (NIS) activity by perchlorate

Abstract

The Na⁺/I⁻ symporter (NIS), the plasma membrane protein that actively transports I⁻ (stoichiometry 2Na⁺:1I⁻) in thyroid physiology and radioiodide-based thyroid cancer treatment, also transports the environmental pollutant perchlorate (stoichiometry 1Na⁺:1ClO₄⁻), which competes with I⁻ for transport. Until now, the mechanism by which NIS transports different anion substrates with different stoichiometries remained unelucidated. We carried out transport measurements analyzed using a statistical thermodynamics-based equation and electrophysiological experiments showing that the different stoichiometry of ClO₄⁻ transport is due to ClO₄⁻ binding to a high-affinity non-transport allosteric site that prevents Na⁺ from binding to one of its two sites. Furthermore, low concentrations of ClO₄⁻ inhibit I⁻ transport not only by competition but also, critically, by changing the stoichiometry of I⁻ transport to 1:1, which greatly reduces the driving force. The data reveal that ClO₄⁻ pollution in drinking water is more dangerous than previously thought.

NIS, encoded by the SLC5A5 gene, is the key plasma membrane protein that mediates active I⁻ uptake in the thyroid gland, the first step in the biosynthesis of the thyroid hormones (THs), of which iodine is an essential constituent¹. THs are critical during embryonic and postembryonic development and for cellular metabolism at all stages of life. The molecular characterization of NIS began in 1996 with our isolation of the cDNA that encodes NIS². NIS couples the “uphill” inward transport of I⁻ against its electrochemical gradient to the “downhill” inward translocation of Na⁺ down its electrochemical gradient, generated by the Na⁺/K⁺ ATPase^1,3. NIS is key in thyroid physiology and has been at the center of radioiodide treatment of thyroid cancer since the 1940s⁴. One of our first fundamental mechanistic discoveries was that NIS mediates I⁻ transport with a 2 Na⁺: 1 I⁻ stoichiometry⁵. One long-standing puzzle about transport by NIS is that the protein accumulates I⁻ from serum concentrations that are orders of magnitude below the K_M of NIS for I⁻. We have shown that there is a simple explanation for this required but unusual property: at physiological Na⁺ concentrations, 79% of all NIS molecules have two Na⁺ ions bound to them, and are therefore poised to bind and transport I⁻. As a result, even when very few NIS molecules bind I⁻, every bound I⁻ is transported⁶. Therefore, the mechanism of I⁻ transport by NIS appears to be an evolutionary adaptation to the scant amount of I⁻ in the environment⁷.

We have demonstrated that the binding of Na⁺ to one of the Na⁺ sites on NIS significantly increases the affinity of the other Na⁺ site for Na⁺ (ref ⁶). We have also shown, by scintillation proximity assays (SPA) using direct ²²Na⁺ binding experiments, that there is strong cooperativity between the two Na⁺ sites, and that the cooperativity between the two Na⁺ sites is lost when residues S353 and T354—Na⁺ ligands at the Na2 site—are replaced with alanine⁸, even though two Na⁺ sites remain present. Therefore, given that allosteric interactions require that the protein have access to multiple conformations, we inferred that other residues besides S353 and T354 participate in the coordination of Na⁺ at the Na2 site, a conclusion that was validated when we subsequently determined that residues S66, D191, Q194, and Q263 also participate in Na⁺ coordination at the Na2 site⁹.

Unlike Cl⁻ channels and transporters, which transport both Cl⁻ and I⁻ (with affinities in the mM range) ^10-12, NIS discriminates between these two anions in an astonishing way: it does not transport Cl⁻, an anion present at concentrations of ~100 mM in the extracellular fluid, but it does transport I⁻, which is present at submicromolar concentrations. NIS also translocates a variety of imaging substrates that can be detected by PET [positron emission tomography: ¹²⁴I⁻ and ¹⁸F-BF₄⁻] and SPECT/CT [single-photon emission computed tomography: ¹²³I⁻, ¹²⁵I⁻, ¹³¹I⁻, pertechnetate (^99mTcO₄⁻), and perrhenate (¹⁸⁶ReO₄⁻, ¹⁸⁸ReO₄⁻)]. Because NIS actively transports its substrates into the cell, it offers higher detection sensitivity than proteins that only bind their ligands stoichiometrically. As a result, NIS is becoming the counterpart of green fluorescent protein (GFP) or luciferase for imaging studies in humans^13,14.

Other anions transported by NIS include thiocyanate (SCN⁻), nitrate (NO₃⁻), tetrafluorborate (BF₄⁻), hexafluorophosphate (PF₆⁻)^5,15-17 and, crucially, the environmental pollutant perchlorate (ClO₄⁻). That ClO₄⁻ is actively transported by NIS, rather than merely acting as a blocker¹⁸, strongly suggests that the adverse health effects of ClO₄⁻ contamination of drinking water are more serious than previously thought, especially for the most vulnerable populations, which include pregnant and nursing women, fetuses, and nursing infants^18-20.

We have previously carried out numerous measurements of NIS-mediated transport of I⁻ and oxyanions, and analyzed the data using a statistical thermodynamics–based formalism that, through global fitting, provided a platform for estimating all necessary constants and for identifying the molecular species most relevant for transport⁶. Here, we present extensive experimental data, analyzed with the same statistical thermodynamics formalism, except including an allosteric site. Remarkably, the mere addition of this one site in one of its two states (occupied or unoccupied) to the existing model reveals the mechanism by which oxyanions change the transport stoichiometry not only for themselves, but also, surprisingly, for I⁻. This has significant physiological consequences: when the 2 Na⁺ : 1 I⁻ electrogenic stoichiometry becomes electroneutral (1 Na⁺ :1I⁻), it decreases the driving force for I⁻ transport, lowers I⁻ accumulation, and imperils biosynthesis of the thyroid hormones. Our results strongly suggest that ClO₄⁻ pollution in drinking water is more dangerous than previously thought¹⁸.

RESULTS

Theoretical background.

NIS has an unusual property for a transporter: it transports different anions with different stoichiometries. Although it translocates its physiological substrate, I⁻, using two Na⁺ ions per I⁻ transported (Fig. 1ai), NIS actively transports other substrates (such as ReO₄⁻ and ClO₄⁻) with a 1 Na⁺ : 1 anion stoichiometry^5,6,18,21. The mechanism by which NIS transports different anions with different stoichiometries has not yet been elucidated. Thus, on the basis of the statistical thermodynamics formalism and our previous knowledge about the NIS transport mechanism, we proposed several hypotheses that could account for the different stoichiometries: 1) the larger oxyanions binding to their transport (T) site may inhibit Na⁺ from binding to the NaA site (Fig. 1aii); in addition to their T site, the oxyanions may bind 2) to a non-transport (NT) site that partially overlaps with the NaA site (Fig. 1aiii); or 3) to an NT site that allosterically inhibits the binding of Na⁺ to the NaA site (Fig. 1aiv). In the physiological 2:1 transport, only the species with all three sites occupied can transition from outwardly to inwardly open. Whichever of these three hypotheses is correct, a NIS species in the outwardly facing conformation with an anion at the T site and only one Na⁺ bound must be able to transition to the inwardly facing conformation, as long as the NT site is occupied. We show here that hypothesis 3 is correct. Strikingly, occupation of the NT site changes not only the stoichiometry of oxyanion transport but also that of I⁻ transport.

Fig. 1. (a) Possible hypotheses as to how NIS transports different substrates with different stoichiometries.

(i) Electrogenic transport takes place when 2 Na⁺ ions and I⁻ are translocated. Hypotheses for electroneutral transport: (ii) Hypothesis 1: Oxyanion transport is electroneutral (1 Na⁺: 1 oxyanion) because binding of the oxyanion to its transport (T) site (yellow circle) causes a partial overlap with the NaA site (light green square), thus inhibiting binding of Na⁺ to this site; according to this hypothesis, I⁻ could not be transported electroneutrally, because I⁻ binding to its T site (blue circle in panel i) would not cause the latter to partially overlap with the NaA site (light green square), (iii) Hypothesis 2: Oxyanion transport is electroneutral because binding of the oxyanion to a non-transport (NT) site (red hexagon) causes the site to partially overlap with the NaA site (light green square), thus inhibiting binding of Na⁺ to this site; according to this hypothesis, I⁻ could be transported electroneutrally in the presence of an oxyanion, but high concentrations of Na⁺ would outcompete the oxyanion at the NT site (red hexagon) and would restore electrogenic transport of I⁻; (iv) Hypothesis 3: Oxyanion transport is electroneutral because binding of the oxyanion to an NT site (red hexagon) allosterically inhibits binding of one Na⁺ to the NaA site (green) (no overlap); according to this hypothesis, I⁻ could be transported electroneutrally in the presence of an oxyanion, and high concentrations of neither Na⁺ nor of I⁻ would restore electrogenic transport of I⁻, (b) ClO₄⁻ changes the stoichiometry of NIS-mediated I⁻ transport. Initial rates (4-min time points) of I⁻ transport at different concentrations of Na⁺ (0-280 mM) and ClO₄⁻ (0-5 μM). Carrier-free ¹²⁵I⁻ was used as a tracer to determine the effect of ClO₄⁻ on the mechanism of I⁻ transport. (i) Points represent the average of duplicate or triplicate ¹²⁵I uptake data. The surface, calculated with equation 1a, represents the rate of transport (cpm/μg DNA), expressed as ${\mathrm{V}}_{\max }$ times the sum of the fraction of NIS species that can transport I⁻, i.e., the fraction of NIS molecules that are occupied by 2 Na⁺ ions and the fraction occupied by 1 Na⁺ and ClO₄⁻ at the NT site. (ii) Data from panel a, showing only the experimental points at 0 μM and 5 μM ClO₄⁻ (i.e., the sections of the surface in panel b (i) at the concentrations indicated). (c) The change in the stoichiometry of I⁻ transport from electrogenic to electroneutral brought about by ClO₄⁻ persists even at high concentrations of I⁻. Initial rates (4-min time points) of I⁻ transport at different concentrations of I⁻ (0.75-160 μM), at a constant concentration of ClO₄⁻ (5 μM), and as a function of the Na⁺ concentration (0-280 mM). Data are expressed as pmol of I⁻/μg DNA). The surface was calculated using equation 1b.

If the difference in stoichiometry between I⁻ and oxyanion transport by NIS reflects the existence of a high-affinity NT site for the oxyanions (referred to as oX2 when occupied) that either directly or allosterically inhibits the binding of Na⁺ to the NaA site (Fig. 1aiii,aiv) ⁶, the total number of binding sites in NIS is (at least) four: the anion T site (referred to as I⁻ when occupied by I⁻ and as oX1 when occupied by an oxyanion), the two Na⁺ sites (NaA and NaB), and the oxyanion NT (oX2) site. Crucially, if so, only the oxyanions have a high affinity for the NT site; the affinity of I⁻ for this site is extremely low. Furthermore, when the NT site is occupied by an oxyanion, Na⁺ cannot bind to the NaA site. Under these assumptions, the species that are able to bind ions from the extracellular fluid, which are named according to the sites within them that are occupied, are (see also Table 1):

Table 1.

Species present in the experiments described and their statistical weights ($\xi$)

NIS species	Statistical weight ($\xi$)
Empty	1
NaA	KNA · [Na+]
NaB	KNB · [Na+]
oX1	KoX1 · [XO4−]
aI−	KI− · [I−]
oX2	KoX2 · [XO4−]
NaA·NaB	KNA · KNB · ϕNA,NB · [Na+]2
oX1·oX2	KoX1 · KoX2 · ϕoX1,oX2 · [XO4−]2
aI−·oX2	KI− · KoX2 · ϕI−,oX2 · [XO4−] · [I−]
bNaA·oX2	0
NaA·oX1	KNA · KoX1 · ϕNA,oX1 · [Na+] · [XO4−]
aNaA·I−	KNA · KI− · ϕNA,I− · [Na+] · [I−]
NaB·oX1	KNB · KoX1 · ϕNB,oX1 · [Na+] · [XO4−]
NaB·oX2	KNB · KoX2 · ϕNB,oX2 · [Na+] · [XO4−]
aNaB·I−	KNB · KI− · ϕNB,I− · [Na+] · [I−]
NaA·NaB·oX1	KNA · KNB · KoX1 · ϕNA,NB · ϕNB,oX1 · ϕNB,oX1 · [Na+]2 · [XO4−]
aNaA·NaB·I−	KNA · KNB · KI− · ϕNA,NB · ϕNA,I− · ϕNB,I− · [Na+]2 · [I−]
bNaA·oX1·oX2	0
a,bNaA· I−·oX2	0
bNaA·NaB·oX2	0
NaB·oX1·oX2	KNB · KoX1 · KoX2 · ϕNB,oX1 · ϕNB,oX2 · ϕoX1,oX2 · [Na+] · [XO4−]2
aNaB·oX2·I−	KNB· KoX2 · KI− · ϕNB,oX2 · ϕNB,I− · ϕI−,oX2 [Na+] · [Xo4−] · [I−]
bNaA·NaB·oX1·oX2	0
a,bNaA·NaB·I−·oX2	0

^aterms with I⁻; I⁻ does not bind to the allosteric NT site (oX2).

^bterms with oX2 and NaA occupied (statistical weight = zero).

No ions: NIS-empty

One ion: NIS-NaA, NIS-NaB, NIS-I⁻, NIS-oX1, and NIS-oX2

Two ions: NIS-NaA-NaB, NIS-NaA-I⁻, NIS-NaA-oX1, NIS-NaB-I⁻, NIS-NaB-oX1, NIS-NaB-oX2, NIS-I⁻-oX2, and NIS-oX1-oX2.

Three ions: NIS-NaA-NaB-I⁻, NIS-NaA-NaB-oX1, NIS-NaB-oX2-I⁻, and NIS-NaB-oX1-oX2 (competent transport species)

States contributing to transport.

Note that there are no species with I⁻ at the NT site or with Na⁺ at the NaA site when an oxyanion is bound to the NT (oX2) site. The species listed are necessary and sufficient to account for all the conditions investigated experimentally. The statistical weights $\xi$ of the different species are shown in Table 1, which includes, for completeness, the species with a zero statistical weight. Note also that, as expected, the statistical weights are a function of the ion concentrations. Given that the conformational change required for the transport process is slow (36·sec⁻¹)⁵ or much less frequent than the binding and unbinding of ions, NIS can be considered in equilibrium with the extracellular ions, and the fractions of the different species as a function of the concentrations of Na⁺ and of the anion(s) can be obtained from the equations derived using the statistical thermodynamics formalism (see Methods) by means of the constants shown in Table 2. In brief, the partition function $\mathit{Z}$ is obtained as the sum of all statistical weights ($\mathbf{Z}=\Sigma {\xi }_{\text{all}\text{species}}$), and the rate of transport $\mathbf{v}$ is proportional to the sum of the statistical weights of the species that can carry out transport under the specified experimental conditions divided by the partition function (Equation 1):

$$\mathsf{V}={\mathsf{V}}_{\max }\cdot (\Sigma {\xi }_{\text{species}\text{that}\text{can}\text{transport}{\mathrm{I}}^{{}^{{}^{\text{-}}}}}∕\mathsf{Z})$$ Equation 1

detailed equations are presented in the Methods section, and can be obtained by replacing the in ${\xi }_{\mathrm{s}}$ Equation 1 with the corresponding explicit expressions in Table 1.

Table 2.

Global adjustment of the data.

	Kd				ϕ
	NaA (mM)	NaB (mM)	oX1 (μM)	oX2 (μM)	NaA,NaB	oX1,oX2	oX2,NaB	oX1,NaA	oX1,NaB
ClO4−	138.89 ± 5.6	526.32 ± 13.4	8.93 ±0.4	1.67 ± 0.5	12.23 ± 0.3	3.70 ± 0.5	3.35 ± 0.3	3.00 ± 0.2	2.50 ± 0.1
ReO4−	140.06 ± 4.7	526.32 ± 12.5	2.01 ± 0.1	0.79 ± 0.1	7.38 ± 0.2	2.29 ± 0.3	2.90 ± 0.3	2.93 ± 0.1	2.90 ± 0.1

When a competent anion is bound to the NT site, Na⁺ cannot bind to the NaA site, but NIS can transport an anion (I⁻ or an oxyanion) at the T site electroneutrally (1:1), driven by transport of one Na⁺ at the NaB site. Therefore, two of the species with ions bound to them can transport their bound anion: the species with 2 Na⁺ ions and one anion bound to them, and the species with one Na⁺ at the NaB site, an anion at the NT site, and the same or a different anion at the T site. In the case of carrier-free ¹²⁵I⁻ transport, the population of NIS molecules with I⁻ bound to them is negligible. Nevertheless, it can be assumed that the rate of transport is proportional to the sum of the fractions of species that are primed to bind I⁻ and transport it, i.e., the species with 2 Na⁺ ions bound to them and the species with an oxyanion at the NT site and a Na⁺ at the NaB site.

Estimation of the binding and interaction constants.

The model based on the statistical thermodynamics formalism requires the estimation of a number of constants. They include those for the binding of each ion to its binding sites in empty NIS (K_a,X = 1/K_d,X), and the constants that reflect the increase in affinity at one site (γ) when another site (x) is occupied (ϕ_X,Y)⁶. The constants can be obtained by fitting the experimental transport data under a large number of conditions (Fig. 1b and Supplementary Fig. 1). The number of constants, however, is too large, and pairs of constants may be correlated for the data to be fitted by (e.g.) least-squares refinement adjusting all the parameters simultaneously. Fortunately, some of the constants, such as the affinity of empty NIS for the Na⁺ ions and I⁻ and the constants of interaction among these bound ions, as well as the number of binding sites for these ions, have already been obtained by fitting transport data from previous experiments using our statistical thermodynamics formalism and by direct Na⁺ binding experiments^6,8. Using these values, the new constants were obtained by a combination of least-squares refinement of small groups of parameters and manual fitting. The improvement in the value of χ² was used to monitor the progress of the refinement. The values of the constants are shown in Table 2. In comparing the previously reported values for the K_ds of ClO₄⁻ and ReO₄⁻ for the T site with the values in Table 2, it must be noted that the estimation of the previous values did not take into account the binding of the oxyanions to the NT site. Since the K_d values for the NT site are lower than those for the T site, the previous values represent binding to the T site when the NT site is mostly occupied. Given that binding to the NT site increases the affinity of the oxyanion for the T site by ϕ_oX1,oX2, the affinities equivalent to those obtained in previous experiments are K_d (previously reported) ≈ K_d (Table 2)/ϕ_oX1,oX2. Using the values in Table 2, K_d,ClO4− = 8.93/3.70 = 2.41 μM and K_d,ReO4− = 2.01/2.29 = 0.88 μM—values highly similar to those reported previously (1.5 and 2.3 μM, respectively)⁵.

Low concentrations of an oxyanion (ClO₄⁻ or ReO₄⁻) change the stoichiometry of NIS-mediated I⁻ transport from electrogenic to electroneutral.

To decide among our hypotheses (Fig. 1a), transport kinetics of carrier-free ¹²⁵I⁻ were measured as a function of the Na⁺ concentration for different concentrations of ClO₄⁻ (Fig. 1b) or ReO₄⁻ (Supplementary Fig. 1). In the absence of any oxyanion, the dependence of I⁻ transport on the Na⁺ concentration was clearly sigmoidal, as would be expected given the 2 Na⁺:1 I⁻ stoichiometry of NIS-mediated I⁻ transport. As the concentration of ClO₄⁻ or ReO₄⁻ was increased from 0 to 5 μM, three effects were observed: 1) the dependence of I⁻ transport on the Na⁺ concentration became progressively less sigmoidal, eventually switching to hyperbolic, i.e., indicating a shift toward an electroneutral 1 Na⁺:1 I⁻ stoichiometry, likely due to binding of the oxyanion to the proposed NT site; 2) the apparent affinity for Na⁺ increased (K_M 134.6 ± 12.5 mM in the absence of ClO₄⁻ and 43.5 ± 1.9 mM at 5 μM ClO₄⁻). This is an expected effect, because the binding of one of the substrates increases the affinity of NIS for the other substrate^6,8; 3) at the higher oxyanion concentrations, I⁻ transport was also inhibited by competition between the oxyanion and I⁻ for binding to the T site; this was reflected in a decrease in the ${\mathrm{V}}_{\max }$. These observations are perfectly in line with the existence of an NT site that, when occupied by an oxyanion, inhibits Na⁺ from binding to the NaA site (Fig. 1aiii,aiv). The affinities of the NT site for ClO₄⁻ and ReO₄⁻, estimated using the general statistical thermodynamics–based Equation 1, are 1.6 μM and 0.79 μM, respectively (Table 2).

Moreover, if hypothesis 1 were correct (Fig. 1aii), not only would high I⁻ concentrations outcompete the oxyanion at the anion T site, but high concentrations of Na⁺ would also outcompete the oxyanion because of the proposed overlap between the T and NaA sites. Thus, high concentrations of either I⁻ or Na⁺ would result in electrogenic I⁻ transport even in the presence of an oxyanion; in fact, neither of these outcomes was observed in our results. Instead, kinetic experiments carried out at different Na⁺ and I⁻ concentrations in the presence of 5 μM ClO₄⁻ (Fig. 1c) or 5 μM ReO₄⁻ (Supplementary Fig. 2) showed that the dependence of I⁻ transport on the Na⁺ concentration remains hyperbolic (i.e., transport is electroneutral with a 1 Na⁺: 1 I⁻ stoichiometry) for I⁻ concentrations as high as 160 μM or Na⁺ concentrations as high as 280 mM, clearly ruling out hypothesis 1. That high concentrations of Na⁺ did not have an effect on the ${\mathrm{V}}_{\max }$ of ReO₄⁻ transport⁶ is also consistent with this conclusion.

Similarly, if hypothesis 2 (Fig. 1aiii) were correct (i.e., if the NaA site and an NT site partially overlapped), high concentrations of Na⁺ would displace the oxyanion from the NT site and render I⁻ transport electrogenic—which, as mentioned, did not occur even at 280 mM (Fig. 1c), ruling out hypothesis 2 as well. On the other hand, that I⁻ transport remained electroneutral even at high concentrations of Na⁺ is consistent with the notion that the oxyanion binds to the NT site and causes it to allosterically inhibit the binding of Na⁺ to the NaA site, as proposed in hypothesis 3 (Fig. 1aiv). In summary, low concentrations of an oxyanion (<5 μM) change the stoichiometry of NIS-mediated I⁻ transport from electrogenic to electroneutral in WT NIS; in addition, I⁻ does not bind to the NT oX2 site.

Oxyanions markedly decrease the K_m of WT NIS for I⁻, but have no effect on the K_m of G93T NIS for I⁻.

We have reported previously that, in contrast to WT NIS, the G93T NIS mutant transports ClO₄⁻ and ReO₄⁻ electrogenically, even at an oxyanion concentration of 1 mM^5,21 (a concentration 200 times that required to saturate the NT site), indicating that when Gly 93 is replaced with Thr, the mechanism of NIS undergoes a dramatic change, such that the oxyanion does not inhibit Na⁺ binding to the NaA site⁶. We have also reported that the K_M of G93T NIS for I⁻ is higher (K_M = 303.8 ± 24.1 μM, i.e., a lower apparent affinity for I⁻) than that of WT NIS (K_M = 23.12 ± 1.30 μM) (Fig. 2a,b, blue), ²¹. We now demonstrate here that ClO₄⁻ (5 μM) decreased the ${\mathrm{V}}_{\max }$ of I⁻ transport mediated by WT and G93T NIS (Fig. 2a,b, orange), because ClO₄⁻ binds to the T site in both. In contrast, whereas ClO₄⁻ (5 μM) had virtually no effect on the K_M of G93T NIS for I⁻ (K_M = 307.7 ± 34.5 μM) (Fig. 2b, orange), it did increase the K_M of WT NIS for I⁻ as much as 6-fold (K_M = 137.9 ± 18.9 μM) (Fig. 2a, orange). This is because ClO₄⁻ switches the Na⁺: I⁻ stoichiometry from 2:1 to 1:1 in WT NIS (Fig. 2c) but not in G93T NIS (Fig. 2d). Taken together, these data strongly suggest that, in G93T NIS, either ClO₄⁻ cannot bind to the allosteric NT site or, if it does bind to it, the resulting conformational change that would prevent Na⁺ from binding to the NaA site is impaired.

Fig. 2. Differential effects of ClO₄⁻ on the transport stoichiometry of WT and G93T NIS and on their K_M values for I⁻.

Initial rates of I⁻ transport at different concentrations of I⁻ (1.25-640 μM) at 140 mM Na⁺ in the absence (blue line) or presence (orange line) of ClO₄⁻ (5 μM): (a) WT NIS and (b) G93T NIS. Initial rates of ¹²⁵I⁻ transport at different concentrations of Na⁺ (0-280 mM): (c) WT NIS using carrier-free ¹²⁵I⁻ in the absence (blue line) or presence (orange line) of ClO₄⁻ (5μM). Data are expressed as cpm/μg DNA) (d) G93T NIS using 750 μM ¹²⁵I (specific activity 35 μCi/μmol) in the absence (blue line) of ClO₄⁻ or in the presence (orange line) of ClO₄⁻ (5 μM). Data are expressed as pmol I⁻/μg DNA.

ClO₄⁻ elicits NIS-mediated currents when its concentration is lower than that necessary to saturate the NT site.

With respect to ClO₄⁻ transport by WT NIS, our hypothesis (Fig. 1aiv) predicts that at ClO₄⁻ concentrations lower than the K_d of ClO₄⁻ for the NT site (1.6 μM, Table 2), ClO₄⁻ should be transported, at least partially, with an electrogenic 2 Na⁺ : 1 ClO₄⁻ stoichiometry. This is because, under these conditions, the NT site will only be partially occupied by ClO₄⁻ and some NIS molecules will bind Na⁺ to the NaA and NaB sites and ClO₄⁻ to the T site (Fig. 1aiv). To test this key prediction, we carried out electrophysiological experiments in X. laevis oocytes heterologously expressing NIS. First, we showed that the addition of I⁻ (160 μM) to the medium elicited negative currents (indicating positive charges moving into the interior of the oocyte) due to the 2 Na⁺ : 1 I⁻ stoichiometry of transport^5,21 (Fig. 3a, red). Second, in the presence of only 0.3 μM ClO₄⁻, the currents elicited by I⁻ were substantially reduced (Fig. 3a, blue), suggesting that ClO₄⁻ had bound to the allosteric NT site without saturating it, and that, as a result, the stoichiometry of I⁻ transport had become nearly electroneutral. Third, in perhaps the most dramatic validation of hypothesis 3, we showed that the prediction laid out above was correct: ClO₄⁻ at low concentrations (0.3 μM, i.e., lower than the K_d of ClO₄⁻ for the NT site) actually elicited currents (Fig. 3b, blue), the hallmark of electrogenic transport of ClO₄⁻. Our previous reports^5,21 that ClO₄⁻ did not elicit currents in NIS-expressing X. laevis oocytes [albeit at high ClO₄⁻ concentrations (500 μM)] would make our current finding of electrogenic ClO₄⁻ transport at low ClO₄⁻ concentrations all the more unexpected, were it not for our newly proposed hypothesis 3. That these currents were mediated by NIS was demonstrated by showing that no currents were elicited by ClO₄⁻ when the Na⁺-dependent myo-inositol transporter (SMIT)²², another member of the SLC5 family (encoded by SLC5A3), was expressed in the oocytes instead of NIS (Fig. 3b, black). It is not surprising that ClO₄⁻ currents are 40 times smaller than those generated by I⁻, because the I⁻ concentration (160 μM) was >500 times the concentration of ClO₄⁻ (0.3 μM). Moreover, saturating the allosteric NT site by increasing the concentration of ClO₄⁻ abolished the ClO₄⁻-induced currents (Fig. 3c), as predicted by hypothesis 3 (Fig. 1aiv).

Fig. 3. ClO₄⁻ at low concentrations is transported electrogenically by NIS.

(a) Current/voltage relationships recorded by TEVC in oocytes expressing NIS upon addition of I⁻ (160 μM), in the presence or absence of ClO₄⁻ (0.3 μM) in the bath solution. (b) Effect of ClO₄⁻ (0.3 μM) on current/voltage relationships recorded by TEVC in oocytes expressing NIS or SMIT. (c) Current/voltage relationships recorded by TEVC in oocytes expressing NIS with 0.1-500 μM ClO₄⁻ in the bath solution. Error bars indicate SEM (n = 8-10).

DISCUSSION

In summary, by showing that oxyanions (such as ClO₄⁻) at concentrations as low as 5 μM or less change the stoichiometry of WT NIS-mediated I⁻ transport from electrogenic to electroneutral, we have revealed the existence of an NT site that, upon binding an oxyanion, allosterically inhibits the binding of Na⁺ to the NaA site. Our experiments were carefully designed to shed light on the regulation of the transport stoichiometry and mechanism of NIS; furthermore, we developed a general statistical thermodynamics-based equation²³ that contains terms representing all species contributing to transport under all conditions, and used it to demonstrate the existence of the NT site. In our previous research, we focused on mechanistic aspects of the simplest case of NIS activity: transport of a single anion⁶. Here, we have investigated more complex cases of NIS-mediated transport, including oxyanion binding, transport, and changes in Na⁺/anion stoichiometry. The omission of a number of species ($\xi =0$; Table 1)—NIS species with an oxyanion at the NT site and Na⁺ at the NaA site, and NIS species with I⁻ at the NT site (as posited by hypothesis 3)—is key to the success of the model. As described, the model allows the same equation (in its two forms—Equations 1a and 1b) to be used to fit the data from all the experiments performed. Being able to use a single equation is valuable indeed, as this obviates the need to make experiment-specific assumptions (i.e., different and independent ${\mathrm{V}}_{\max }$ and K_M values for each curve) in interpreting the results.

Even low concentrations of ClO₄⁻ radically change the mechanism of NIS-mediated I⁻ transport, by changing the driving force from 2 Na⁺ to 1 Na⁺ and increasing the K_M of NIS for I⁻ 6-fold; less transport of I⁻ will result in a decrease in the biosynthesis of THs. This reveals that contamination of drinking water with ClO₄⁻ is even more dangerous than previously thought—especially for the most vulnerable populations, including pregnant and nursing women and their fetuses and nursing newborns. NIS expressed in the lactating breast mediates active I⁻ transport into the milk²⁴; therefore, if a nursing mother drinks water containing ClO₄⁻, not only will there be less I⁻ in her milk, but also ClO₄⁻ will actually be actively translocated into the milk. As a result, I⁻ uptake for TH biosynthesis by the newborn will itself be inhibited by ClO₄⁻, both because it will compete for the T site and because it will change the Na⁺/I⁻ stoichiometry from 2:1 to 1:1—at precisely those early stages of life when THs are essential for the maturation of the central nervous system, lungs, and skeletal system¹⁷. NIS is also expressed in the placenta^25,26, where it transports I⁻ from the mother into the fetus’ bloodstream—which is crucial, as the fetus synthesizes 70% of its own THs. Therefore, if a pregnant woman drinks water contaminated with high concentrations of ClO₄⁻, it can gravely harm the developing fetus.

One intriguing question that deserves future investigation is whether or not the allosteric NT site evolved to maintain homeostasis by binding a physiological anion when TH levels are high. It will likewise be of great interest to determine whether other transporters have an allosteric NT substrate binding site that regulates their activity similarly to how the NT site modulates NIS function (i.e., a site completely different from the second substrate binding site proposed for vSGLT, PutP, LeuT or MhsT)^27-29. That the second substrate binding sites in LeuT, vSGLT, PutP, and MhsT are different from the NT allosteric oxyanion site in NIS is clear. Those sites have been proposed to bind leucine, galactose, proline, and hydrophobic amino acids respectively, and binding of these substrates does not change the stoichiometry of transport—for example, leucine binding by LeuT and leucine transport by MhsT as a function of the Na⁺ concentration at saturating concentrations of leucine are still sigmoidal (2 Na⁺)^29,30. In stark contrast, the allosteric site in NIS does not bind I⁻, it only binds the oxyanions, and this oxyanion binding does change the mechanism of transport from electrogenic to electroneutral. If the allosteric site in NIS were similar to the proposed second substrate binding sites in LeuT, vSGLT, PutP, and MhsT, it would bind I⁻, contrary to fact.

Finding allosteric regulation of the transport stoichiometry of other membrane transporters by NT binding sites would not only deepen our understanding of these transporters but also might eventually make it possible to develop novel drug targets.