The complete nucleotide sequence of the cloned NIS cDNA and the deduced amino acid sequence are presented in reference (13). Beginning with Met at position 1, a long open reading frame codes for a protein of 618 amino acids (relative molecular mass 65,196). The hydropathic profile and initial secondary structure predictions of the protein led Dai et al. (13,17) to suggest an intrinsic membrane protein with 12 putative transmembrane segments (TMS). However, this model has subsequently been revised upon extensive experimental testing, as detailed below. The NH₂ terminus was originally placed on the cytoplasmic side, given the absence of a signal sequence. The COOH terminus, which was also predicted to be on the cytoplasmic side, was found to contain a large hydrophilic region of ∼70 amino acids. Levy et al. demonstrated experimentally the intracellular orientation of the COOH terminus by showing that an anti-COOH terminus antibody binds to its epitope only after cell permeabilization (18). Three potential Asn-glycosylation sites were identified in the deduced amino acid sequence at positions 225, 485, and 497. The first was located in a predicted intracellular hydrophilic loop, while the last two were located in the last hydrophilic loop, a segment predicted to be on the extracellular face of the membrane.

Levy et al. (19) obtained conclusive evidence showing that, contrary to previous suggestions (20), neither partial nor total lack of N-linked glycosylation impairs activity, stability, or targeting of NIS. Using site-directed mutagenesis, Levy et al. (19) substituted both separately and simultaneously the Asn residues (amino acids 225, 485, and 497) with Gln in all three putative N-linked glycosylation consensus sequences of NIS, and assessed the effects of the mutations. All mutants were active and displayed 50% to 100% of wild-type NIS activity, including the completely non-glycosylated triple mutant. They observed that the half-life of non-glycosylated NIS was similar to wild-type NIS and that the K_m value for I^– (∼20 to 30 μM) in non-glycosylated NIS was virtually identical to that of wild-type NIS. These findings demonstrate that, to a considerable extent, function, targeting, and stability of NIS are present even in the total absence of N-linked glycosylation. Therefore, a bacterial expression system, in which no N-linked glycosylation occurs, might be used to overproduce NIS for structural studies.

Levy et al. (19) also demonstrated that the putative N-linked glycosylation site at N225, which had originally been predicted to face intracellularly, is indeed glycosylated. Therefore, it is now clear that the hydrophilic loop that contains this sequence faces the extracellular milieu rather than the cytosol. They have proposed a 13-TMS model to be the most likely secondary structure for NIS (Fig. 4A.2). In contrast to the original model, in which the NH₂ terminus was predicted to face the cytosol on account of the lack of a signal sequence in NIS, in the current model, both the NH₂ terminus and the hydrophilic loop containing N225 are predicted to be on the extracellular side, and the COOH terminus faces the cytosol.

Levy et al. (19) subsequently demonstrated unequivocally that the NH₂ terminus faces the external milieu, as proposed in the current model. This conclusion was reached using two independent experimental approaches. First, these authors introduced a FLAG epitope at the NH₂ terminus. COS cells transfected with FLAG-containing NIS displayed undistinguishable I^– uptake accumulation from that of COS cells transfected with wild-type NIS. Immunofluorescence experiments revealed positive immunoreactivity with anti-Flag Ab in non-permeabilized COS cells transfected with FLAG-containing NIS. Positive immunoreactivity in non-permeabilized cells indicates that the NH₂ terminus faces externally. In contrast, immunoreactivity using anti-COOH Ab requires permeabilization because the COOH terminus faces the cytosol (18). The second technique took advantage of a previous observation that non-glycosylated NIS is active. The N-linked glycosylation amino acid sequence NNSS was introduced into the NH₂ terminus of non-glycosylated NIS (21). They observed glycosylation of NIS at the NH₂ terminus upon transfection of NNSS-containing NIS into COS cells, thus proving that the NH₂ terminus faces the lumen of the endoplasmic reticulum during biosynthesis and therefore faces the external milieu upon reaching the plasma membrane.

NIS, the product of the SLC5A5 gene, is a member of solute carrier family 5 (SLC5), which also includes – among many other proteins that mediate key physiologic transport processes – Na⁺/glucose co-transporter 1 (SGLT1) (22). The current 13-TMS model for NIS has been confirmed with the determination, by Faham et al. (23), of the structure of vSGLT, the Vibrio parahaemolyticus Na⁺/galactose transporter, with which NIS shares significant identity (27%) and homology (58%). vSGLT is the bacterial homologue of the human SGLT1.

Eskandari et al. (24) have examined the mechanism, stoichiometry, and specificity of NIS by means of electrophysiologic, tracer uptake, and electron microscopic methods in X. laevis oocytes expressing NIS. These authors obtained electrophysiologic recordings using the two-microelectrode voltage clamp technique. They showed that an inward steady-state current (i.e., a net influx of positive charge) is generated in NIS-expressing oocytes upon addition of I^– to the bathing medium, leading to depolarization of the membrane. Since the recorded current is attributable to NIS activity, this observation confirms that NIS activity is electrogenic. Simultaneous measurements of tracer fluxes and currents revealed that two Na⁺ ions are transported with one anion, demonstrating unequivocally a 2:1 Na⁺/I^– stoichiometry. Therefore, the observed inward steady-state current is due to a net influx of Na⁺ ions.

Eskandari et al. (24) determined that the turnover rate of NIS is ∼36 s^-1, and reported that expression of NIS in oocytes led to a ∼2.5-fold increase in the density of plasma membrane protoplasmic face intramembrane particles, as ascertained by freeze-fracture electron microscopy. This is the first direct electron microscopy visualization of ostensible NIS molecules present in the oocyte plasma membrane.

Similar steady-state inward currents were generated by a wide variety of anions in addition to I^– (including ClO₃^–, SCN^–, SeCN^–, NO₃^–, Br^–, BF₄^–, IO₄^–, and BrO₃^–), indicating that these anions are also transported by NIS. However, ClO₄^–, the most widely characterized inhibitor of thyroidal I^– uptake, was surprisingly found not to generate a current, strongly suggesting that it was not transported (24). Yoshida et al. (25) have reported, similarly, that ClO₄^– did not induce an inward current in Chinese hamster ovary (CHO) cells stably expressing NIS, as measured using the whole-cell patch clamp technique. The most likely interpretation of these observations, at the time, seemed to be that ClO₄^– was not transported by NIS, although the unlikely possibility that ClO₄^– was translocated by NIS with an electroneutral 1:1 Na⁺/ClO₄^– stoichiometry could not be ruled out. To unequivocally elucidate whether ClO₄^– is translocated or not by NIS, flux experiments needed to be performed with ClO₄^– labeled with ³⁶Cl; however, the unavailability of ³⁶ClO₄^– made such experiments impossible. Yoshida et al. (26) thereafter showed that ClO₄^– elicited no change in the membrane current in the highly functional rat thyroid cell line FRTL-5, as revealed by the whole-cell patch-clamp technique, consistent with the notion that ClO₄^– was not transported into FRLT-5 cells and supporting both their previous observations in CHO cells (25) and supporting the results of Eskandari et al. (24) in oocytes. All these results, taken together, led researchers to conclude that ClO₄^– was a potent inhibitor of NIS that acted as a blocker, not a substrate.

However, a study carried out by Dohán et al. (27) using a bioassay has conclusively – and surprisingly – demonstrated that ClO₄^– in fact is actively translocated by NIS after all. These authors employed an indirect approach in a polarized in vitro bicameral model system. The model recapitulated a polarized epithelial monolayer with a basolateral and an apical surface. In the model, a monolayer of polarized epithelial Mardin-Darby canine kidney (MDCK) cells stably expressing exogenous NIS (engineered to be targeted apically in these cells) separated the lower (basolateral) from the upper (apical) chamber. The authors added I^– to the apical chamber and followed the change in I^– concentration in both chambers over time. I^– was rapidly transported from the apical to the basolateral chamber, and the gradient generated was maintained for >6 hours. When I^– and ClO₄^– were simultaneously added to the apical chamber, I^– was also translocated to the basolateral chamber, except that transport occurred after a considerable delay (>1 hour), suggesting that the ClO₄^– concentration had to decrease beyond a certain point before NIS could start translocating I^–.

Moreover, to investigate whether ClO₄^– reached the opposite chamber, ClO₄^– alone (without I^–) was added to the apical chamber, the basolateral chamber, or both. Knowing that NIS was exclusively expressed apically in these cells, the authors took aliquots from the apical or basolateral chamber and added them to separate NIS-expressing MDCK cells to assess whether the aliquots would inhibit I^– transport, indicating that the aliquots contained ClO₄^–. The authors found that only aliquots from the basolateral chamber were markedly inhibitory, even when ClO₄^– was only added to the apical chamber, demonstrating that ClO₄^– was actively translocated to the basolateral chamber by NIS (27).

Dohán et al. extended their in vitro observations by injecting ¹³¹I^– or ¹³¹I^–/ClO₄^– into lactating dams but not into their pups and by showing that thyroidal I^– uptake in both dams and pups was markedly lower in the ClO₄^–-treated animals than in the controls. Furthermore, to investigate whether ClO₄^– not only inhibited NIS function but also was actually translocated by NIS into the milk, the authors injected ClO₄^– into lactating dams and tested whether milk samples inhibited NIS-mediated I^– transport in MDCK cells expressing NIS. They observed that only milk samples from animals that received ClO₄^– inhibited I^– transport. In conclusion, it is clear that ClO₄^– is not a non-transported blocker of NIS. Instead, it reaches the milk by NIS-mediated active transport, not by passive diffusion.

It is now evident that ClO₄^– elicited no currents in electrophysiology experiments because Na⁺-dependent ClO₄^– transport is electroneutral and, therefore, its kinetic analysis could only be carried out by direct substrate flux measurements (which, as indicated above, are not feasible). Thus, Dohán et al. (27) instead analyzed the kinetic parameters of NIS-mediated transport of a structurally similar anion, perrhenate (ReO₄^–), taking advantage of the higher specific activity of the radioisotope ¹⁸⁶ReO₄^–. The initial rates of NIS-mediated, Na⁺-dependent ReO₄^– transport yielded a hyperbolic curve indicative of an electroneutral stoichiometry, strongly suggesting that the NIS-mediated Na⁺/ClO₄^– transport stoichiometry is also electroneutral. This finding stands in stark contrast to the electrogenic 2:1 Na⁺/anion stoichiometry observed with most other NIS substrate anions, demonstrating that NIS translocates different substrates with different stoichiometries, a remarkable property that had not previously been discovered in any transporter. A simple mathematical model accurately predicts the degree of competition between ClO₄^– and I^–, and that ClO₄^– is transported with a V_max 3 times lower and a backflow rate 10 times lower than that of I^– (27). The results of Dohán et al. have been confirmed by Tran et al. (28), who showed, using a sensitive chromatography–electrospray ionization-tandem mass spectrometry method, that ClO₄^– is actively transported in FRTL-5 cells.

The above-cited findings on NIS-mediated transport of ClO₄^–, in addition to their significance in the area of structure/function of plasma membrane transporters regarding the existence of different stoichiometries for different substrates, are also having a considerable impact on an ongoing debate on the public health effects of ClO₄^– pollution. Numerous sources of drinking water in the United States are contaminated with ClO₄^– as a result of industrial production of the anion, which is used as a propellant and as an explosive. As a result, low-level exposure to ClO₄^– is common in the U.S. population and has been associated with increased serum TSH and decreased T₄ levels in women with a low iodine status (29). The demonstration that NIS actively transports ClO₄^– and mediates its translocation to the milk indicates that ClO₄^– may pose a health risk, particularly to nursing newborns, although recent studies have not shown an effect of low levels of ClO₄^– on thyroid function in first trimester pregnant women (including those with a urinary iodine less than 100 μg/L) (30,31) and in neonates of nursing mothers (32). Exposure of a nursing mother to high levels of ClO₄^– would lead to both lower I^– and higher ClO₄^– in the milk, ultimately compromising the newborn’s own thyroidal I^– uptake and thyroid hormone biosynthesis. These effects would seriously impair neurologic and overall physical development in utero.

Three different transcription factors have been implicated in thyroid-specific gene transcription (33,34): (1) Thyroid transcription factor 1 (TTF-1), a homeodomain (HD)-containing protein present in the developing thyroid, lung, forebrain, and pituitary and in the adult thyroid and lung; (2) TTF-2, a forkhead protein detected in developing thyroid and anterior pituitary and in the adult thyroid; and (3) Pax8, a nuclear protein member of the murine family of paired-domain (PD) containing genes, present in the developing thyroid, kidney, and mid-brain boundary and in the adult thyroid and kidney. Specific combinations of these factors have been proposed to regulate transcription of the thyroid-specific proteins thyroglobulin (Tg) and thyroperoxidase (TPO) and also of the TSH receptor (TSHr).

The hNIS cDNA was identified on the expectation that hNIS would be highly homologous to rNIS. Using primers to the cDNA rNIS sequence (specifically rNIS nucleotide sequences 362 to 480 and 671 to 900), Smanik et al. (35) amplified a cDNA fragment of hNIS from human papillary thyroid carcinoma tissue by PCR, utilized this cDNA fragment to screen a human thyroid cDNA library, and isolated a single cDNA clone encoding hNIS. The nucleotide sequence of hNIS revealed an open reading frame of 1,929 nucleotides, which encodes a protein of 643 amino acids. hNIS exhibits 84% identity and 93% similarity to rNIS. hNIS differs from rNIS only on account of a five amino acid insertion between the last two hydrophobic domains (amino acids 485 to 488 and 499) and a 20 amino acid insertion in the COOH terminus (amino acids 618 to 637). Subsequently, Smanik et al. (36) examined the expression, exon–intron organization, and chromosome mapping of hNIS. Fifteen exons encoding hNIS were found to be interrupted by 14 introns (Fig. 4A.3), and the hNIS gene was mapped to chromosome 19p. The human NIS promoter has been sequenced by three different groups (37,38,39).

It has long been established that TSH stimulates NIS activity via cAMP (40,20,41) and more recently that it upregulates NIS
mRNA levels (42). However, to fully understand these mechanisms it is necessary to study the transcriptional regulation of NIS. As for rat NIS, Endo et al. (43) localized a TTF-1 binding site between -245 and -230 bp, in the proximal rNIS promoter (2 kb), that confers thyroid-specific transcription but only exerts a modest effect. The same group (44) subsequently identified, in the 5′-flaking region between -1,968 and +1 bp, a novel TSH-responsive element (TRE) between -420 and -385 bp upstream of the TTF-1 site in the rNIS promoter, that upregulated 2- to 3-fold NIS expression. The TSH effect is cAMP-mediated and thyroid-specific. They showed that the protein that binds this site is different from TTF-1, TTF-2, Pax8, or other known transcription factors, and the authors named the putative binding protein in the TRE site NTF-1 (NIS TSH-responsive factor 1). However, as the TSH upregulation of NIS via this TRE site is lower than the regulation that the same group reported previously (∼6-fold) (42), they suggested that other transcription binding sequences may be present upstream from the 1968 bp region they studied.

Ohno et al. (45) showed that the rNIS regulatory region contains an enhancer located between -2,264 and -2,495 bp that recapitulates the most relevant aspects of NIS regulation (Fig. 4A.4). This enhancer mediates thyroid-specific gene expression by the interaction of Pax8 with a novel cAMP-dependent pathway. The NIS upstream enhancer (NUE) stimulates transcription in a thyroid-specific and cAMP-dependent manner. NUE contains: Two Pax8 binding sites, two TTF-1 binding sites that have no effect on rNIS transcription, and a degenerate CRE (cAMP-responsive element) sequence (5′-TGACGCA-3′), which is important for NUE transcriptional activity (Fig. 4A.4). In NUE, both Pax8 and the CRE-like binding factor act synergistically to obtain full TSH/cAMP-dependent transcription. However, this enhancer is also able to mediate cAMP-dependent transcription by a novel PKA-independent mechanism (45). Transcriptional regulation of the Tg, TPO, and TSHr genes by TSH/forskolin is mediated by cAMP. However, no CRE sequences have been identified for any of these, except the TSHr gene. Thus, the novel PKA-independent mechanism reported by Ohno et al. (45) is highly significant, as it established a direct relationship between thyroid-specific transcription factors and the TSH/cAMP
regulation in rNIS. This picture clearly separated rNIS gene regulation from that of true thyroid-restricted genes (Tg and TPO) and also from TSHr, a notion consistent with the fact that NIS is not exclusively expressed in the thyroid gland. Subsequently, Chun et al. (46) demonstrated that the CRE-like element within NUE interacts with different b-Zip molecules, such as Fos/Jun and CREB/ATF. Although the CRE-like element plays a central role in mediating cAMP-induced stimulation of enhancer activity, it requires the cooperation of at least one of the two Pax8 binding sites. Although the synergistic effect of Pax8 and the b-Zip molecules requires that they bind to adjacent sites in NUE, they do not appear to physically interact with each other directly.

Regarding the human NIS gene, a thyroid-specific, TSH-responsive, far-upstream (-9,847 to -8,968) enhancer – highly homologous to the rat NUE – has been reported by two groups (47,48). It contains putative Pax8 and TTF-1 binding sites and a CRE-like sequence. The TTF-1 binding site is not required for full activity (Fig. 4A.4) (47,48).

TSH is the primary hormonal regulator of thyroid function overall, and has long been known to stimulate thyroidal I^– accumulation. Most actions of TSH take place through activation of adenylate cyclase via the GTP binding protein Gα. This cascade of events is initiated by the interaction of TSH with its receptor on the basolateral membrane of the follicular cells. Early observations made prior to the isolation of the NIS cDNA suggested that TSH stimulation of I^– accumulation results, at least in part, from the cAMP-mediated increased biosynthesis of NIS (40,41). Once the rat NIS cDNA was isolated (13) and anti-NIS Abs were generated, Levy et al. demonstrated in rats that NIS protein expression is upregulated by TSH in vivo (18). Consistent with these findings is a later observation by Uyttersprot et al. (49) that the expression of NIS mRNA in dog thyroid (∼3.9 kb) is dramatically upregulated by goitrogenic treatment (i.e., propylthiouracil (PTU) treatment, which leads to elevated TSH circulating levels in vivo).

Upregulation of thyroid NIS expression and I^– uptake activity by TSH has been demonstrated not only in rats in vivo (18), but also in the rat thyroid–derived FRTL-5 cell line (41) and in human thyroid primary cultures (50,51). Marcocci et al. (52), Kogai et al. (42) and Ohno et al. (45) have all shown that TSH upregulates I^– uptake activity by a cAMP-mediated increase in NIS transcription. After TSH withdrawal, a reduction of both intracellular cAMP levels and I^– uptake activity is observed in FRTL-5 cells (41). This is a reversible process, as I^– uptake activity can be restored either by TSH or agents that increase cAMP (41,45). To investigate NIS biogenesis, Levy et al. (18) carried out metabolic labeling and immunoprecipitation experiments in the presence of TSH, and observed that NIS is synthesized as a precursor (∼56 kDa) (19). After a 60-minute chase period, a broad, fully processed polypeptide band (∼90 kDa) also became apparent.

Kaminsky et al. (40) made the surprising observation that I^– uptake activity is present in membrane vesicles (MVs) prepared from FRTL-5 cells that, when intact, had completely lost I^– uptake activity due to prolonged TSH deprivation (Fig. 4A.5). This suggests that mechanisms other than transcriptional may also operate to regulate NIS activity in response to TSH.

Kogai et al. (50) have shown that TSH markedly stimulates NIS mRNA and protein levels in both monolayer and follicle-forming human primary culture thyrocytes, whereas significant stimulation of I^– uptake is observed only in follicles. These interesting observations indicate that, in addition to TSH stimulation, cell polarization and spatial organization are also crucial for proper NIS activity, and suggest that NIS may be regulated by such posttranscriptional events as subcellular distribution. Riedel et al. (53) have demonstrated conclusively by immunoblot analysis that NIS is present in FRTL-5 cells as late as 10 days after TSH withdrawal and that de novo NIS biosynthesis requires TSH (53). Therefore, it is clear that any NIS molecules detected in TSH-deprived FRTL-5 cells had to be synthesized prior to TSH withdrawal. This is consistent with NIS being a protein with an exceptionally long half-life, as suggested by Kogai et al. (42) and Paire et al. (20). Riedel et al. (53) determined by pulse-chase analysis that the NIS half-life is ∼5 days in the presence of and ∼3 days in the absence of TSH. Even though the NIS half-life in the absence of TSH is 40% shorter than in the presence of the hormone, it is still sufficiently long to account for the persistence of significant I^– uptake activity in MV from cells deprived of TSH (53).