Expression of the Na⁺/l^-symporter (NIS) is markedly decreased or absent in gastric cancer and intestinal metaplastic mucosa of Barrett esophagus

Iodide (I^-) is an essential constituent of the thyroid hormones triiodothyronine (T₃) and tetraiodothyronine (T₄). These hormones are vital for the normal development and maturation of the central nervous system in the newborn and for multiple metabolic functions in the adult. I^- metabolism in humans appears to have adapted to provide sufficient I^- for normal thyroid function in the face of the environmental scarcity of I^-. A cornerstone of I^- metabolism is active I^- uptake in the thyroid, a process mediated by the sodium/iodide symporter (NIS)[1, 2]. NIS is an integral plasma membrane glycoprotein located in the basolateral membrane of the thyroid follicular cells[3]. Although NIS-mediated active I^- uptake has long been viewed as a distinctly thyroidal phenomenon, it is now clear that active I^- transport observed in extrathyroidal tissues such as salivary glands, lactating mammary gland, gastric mucosa, and placenta is also mediated by NIS [3–8]. The NIS cDNA cloned from these tissues is identical to thyroid NIS[5]. Indeed, deglycosylation with N-glycosidase F or methionine-specific CNBr cleavage of thyroid, stomach, and mammary gland NIS proteins has indicated that NIS is the same protein in each of these tissues[6]. NIS-mediated radioiodide uptake in the stomach and salivary glands is routinely observed in radioiodide/^99mTcO₄ ^-whole-body scintiscans (Fig 1A) [9].

The supply of I^- for thyroid hormone biosynthesis is governed by dietary I^- intake, I^- absorption, and thyroidal I^- uptake. I^- is presumed to be absorbed in the small intestine, but neither the anatomical location of its absorption nor its mechanism has been identified. Interestingly, gastric NIS mediates the active transport of I^- from the bloodstream to the gastric lumen, i.e., the active secretion of I^- into the gastric juice. Secreted I^- is then recirculated into the bloodstream when it is absorbed, along with newly ingested dietary I^-, in the small intestine. I^- is ultimately excreted mainly by the kidneys. The role of secreted I^- in the gastric juice is unknown, as is the function of NIS-mediated I^- secretion to the saliva in the salivary glands. By contrast, the functional role of NIS-mediated I^- translocation in the lactating mammary gland is crucial and very clear: the process results in I^- secretion to the milk, thus supplying the anion to the breast-fed newborn for his/her own thyroid hormone biosynthesis[6].

Data on NIS expression and function in regions of the gastrointestinal tract other than the stomach are still scant and somewhat controversial. The presence of the NIS transcript, as detected by RT-PCR, has been reported in both the colon[10] and the small intestine[11]. However, other investigators have been unable to amplify the NIS transcript in either of these two tissues[5, 12, 13]. By immunoshistochemistry, Spitzweg et al[5], Lacroix et al [13], and Wapnir et al [14] have observed some NIS protein expression in the colon; in contrast, Vayre et al [15] observed it only in the rectum but not in the rest of the colon. None of the investigations carried out to date have shown NIS expression in the esophagus[3]. Findings on NIS expression in gastrointestinal tumors have also been limited. Gastric carcinoma, unlike normal mucosa, has generally been reported to exhibit no I^- or pertechnetate (^99mTcO₄ ^-)accumulation [16–18] (pertechnetate is an anion with the same size and charge as I^- and is similarly transported by NIS), with the sole exception of Wu et al [19], who reported radioiodide uptake in gastric adenocarcinoma. In addition, in the 1960's and 70's, radioiodide or ^99mTcO₄ ^-gastric scintigraphy was studied as a possible method for the diagnosis of gastric neoplasia, based on the finding that malignant gastric tissue failed to transport I^-or ^99mTcO₄ ^-into the gastric juice [16–18]. Furthermore, during the same decade, Berquist et al used ^99mTcO₄ ^- scintigraphy to establish the diagnosis of Barrett esophagus, based on the characteristic replacement of the distal esophageal mucosa by the ^99mTcO₄ ^-concentrating gastric columnar epithelium[20, 21]. These observations suggest not only that NIS expression may be impaired as a result of malignant transformation, but also that the determination of NIS expression and function may be of diagnostic value in gastroesophageal disease.

Consistent with the above concept is our recent report of decreased or absent NIS expression in 27 gastric adenocarcinomas studied by high-density tissue microarrays[14]. Whereas tissue microarrays are optimal for high-throughput screening, the sampling with tissue cores is limited Thus, to expand our findings and more thoroughly examine the issue of NIS expression in gastroesophageal cancer and its possible diagnostic value, we have analyzed NIS expression in normal and malignant gastrointestinal tissue samples from 83 patients. Samples were obtained during surgical resection or endoscopic examination and were analyzed by immunohistochemistry on conventional tissue sections, given that this technique offers the advantage (over RT-PCR) of determining expression and cellular localization of the NIS protein instead of the NIS transcript[22]. We also assessed NIS expression by immunoblot analysis to ascertain the specificity of the observed immunoreactivity.

Snap frozen tissue samples were obtained for immunoblot analysis from 17 patients undergoing resections for gastric tumors (3 MALT and 14 adenocarcinomas from 5 female and 12 male patients; average age: 62). Corresponding normal peritumoral tissues were also collected.

We studied by immunohistochemistry tissue samples obtained from the gastrointestinal tracts of 66 patients (average age: 54; male/female ratio: 37/29). In the case of 20 cancer patients, samples were taken from tumors and their neighboring areas (tumor, tumor margin, and 1 cm and more than 3 cm from the tumor margin) immediately after resection in the operating theatre; the remaining 46 (66–20) were biopsies. Tissue samples from small and large intestines were obtained from all segments of the intestinal tract (jejunum, ileum, and right and left colon). Permission for the investigation was obtained from the local Ethical Committee of the St. George University Teaching Hospital Székesfehérvár, Hungary.

Immunohistochemical analysis of NIS in the thyroid, salivary glands, and stomach revealed very distinctly in which particular cells NIS is located in each tissue, namely the thyroid epithelial (Fig. 1C,D), salivary gland ductal epithelial (Fig. 1E,F), and gastric mucin-producing epithelial cells (Fig. 1G,H). In all three kinds of cells, NIS is clearly observed in the basolateral plasma membrane. By immunoblot analysis of membrane fractions, human NIS was detected mainly as a fully glycosylated mature ~90–120-kDa polypeptide in thyroid, salivary gland, and stomach (Fig. 1B and Fig. 2). We have previously shown that in rat tissues, NIS differences in electrophoretic mobility are due to different degrees of glycosylation[6]. The partially glycosylated ~50-kDa-precursor band and the ~180-kDa-dimer band observed in thyroid tissue were also detected in salivary glands and in the stomach upon longer exposure.

To gain further insight into NIS expression in the digestive tract, we analyzed NIS by immunohistochemistry in samples from different patients. A total of 155 tissue samples obtained from the gastrointestinal tracts of 66 patients (Table 2) was studied, including 42 esophageal, 53 gastric, 11 small-bowel, and 49 large-bowel tissue samples. NIS expression was detected only in the gastric mucosa. More specifically, we observed NIS expression in the basolateral region of normal gastric mucosal surface epithelial cells in the form of expressed, homogenous linear staining (Figs. 1G,H; and 3m). At the same time, we detected no NIS expression in the mucin-containing cells of the neck of the foveola, in the parietal and chief cells, or in the neuroendocrine cells. Similarly, the goblet, Paneth, and ciliated epithelial cells of gastric intestinal metaplasia, as well as the mucus-producing columnar cells of the colon and the squamous esophageal epithelium, proved NIS negative (Fig. 3b,q,t,w).

	Number of samples	NIS (+)	NIS (-)
Esophagus	42
- normal mucosa	4	0 (0 %)	4 (100 %)
-squamous cell carcinoma*	6	0 (0 %)	6 (100 %)
-adenocarcinoma	4	0 (0 %)	4 (100 %)
- Barrett– junctional/fundus type	5	5 (100 %)	0 (0 %)
- Barrett– intestinal metaplasia	5	0 (0 %)	5 (100 %)
Stomach	53
- normal	8	8 (100 %)	0 (0 %)
-adenocarcinoma*	8	0 (0 %)	8 (100 %)
-polyp (hyperplastic)	10	9 (90 %)	1 (10 %)
-gastritis	6	6 (100 %)	0 (0 %)
Small intestine	11
-normal	4	0 (0 %)	4 (100 %)
-inflammation	3	0 (0 %)	3 (100 %)
-polyp	2	0 (0 %)	2 (100 %)
-adenocarcinoma	2	0 (0 %)	2 (100 %)
Colon	49
-normal	8	0 (0 %)	8 (100 %)
-inflammation	3	0 (0 %)	3 (100 %)
-polyp	9	0 (0 %)	9 (100 %)
-adenocarcinoma*	8	0 (0 %)	8 (100 %)

* Samples obtained from the corresponding peritumoral tissues are shown in Table 3; however, they are included in the total number of samples in this table.

Nine out often hyperplastic gastric polyps were NIS-positive (Fig. 3n,o), the only negative being a colon-type metaplasia (Fig. 3p,q). The findings in samples from patients with Barrett esophagus were striking: five junctional and fundus-type columnar metaplasias were NIS positive (Fig. 3c,d), whereas five showing intestinal metaplasia were consistently negative (Fig. 3e,f).

The samples from patients with gastritis displayed NIS expression independently of the presence (4/6) or absence (2/6) of Helicobacter pylori (Table 2).

The field of extrathyroidal I^- transport has changed considerably since the extensive review published on the topic in 1961 by Brown-Grant. The main vertebrate nonthyroid tissues reported to accumulate I^- actively via NIS are the salivary glands, gastric mucosa, lactating mammary gland, placenta, choroid plexus, and ciliary body of the eye[3, 4, 8, 24]. With the exception of the lactating mammary gland, the physiological role of I^- or other anions transported by NIS in extrathyroidal tissues is unknown. These tissue-specific NIS-mediated transport systems exhibit functional similarities to their thyroid counterpart, such as a Km for I^- ranging from 10 to 30 μM and susceptibility to inhibition by perchlorate (CIO₄ ^-). In contrast to TSH-regulated thyroid NIS and lactationally regulated mammary gland NIS, both salivary and gastric NIS are constitutively expressed[6, 25, 26]. NIS has been detected in the basolateral membrane of all ductal epithelial cells in the salivary gland[6], and in the basolateral membrane of superficial mucin-secreting epithelial cells in the stomach[6, 13, 15, 27]. Although Spitzweg et al [27] reported NIS-specific immunostaining in parietal cells, we observed NIS expression exclusively in surface epithelial cells. Wapnir et al [14] have reported some NIS expression in a limited number of colon samples, but no NIS expression in colon was detected in the present study by either immunohistochemistry or immunoblot (not shown). There are at least three possible explanations for this discrepancy: NIS expression may be focal rather than widespread along the large bowel, variations in fixation methods may alter tissue antigenicity, or microarray overstaining may result from the "edge effect" of the small tissue cores.

NIS transports other anions with the following relative apparent affinities: I^- (1.00) ≥ SeCN^- (0.87) > SCN^- (0.34) > CIO₃ ^-(0.12) > NO₃ ^- (0.04)[28]. As indicated earlier, the roles of I^- or other anions secreted to the lumen of the gastrointestinal system by the constitutively expressed salivary and gastric NIS are unknown. I^- concentrated in the gastric juice is reabsorbed in the small intestine and used for thyroid hormone biosynthesis by the thyroid or excreted by glomerular filtration through the kidney. Albeit with a lower affinity than I^-, nitrate (NO₃ ^-) is one of the other anions transported by NIS. Nitrate can be reduced to nitrite (NO₂ ^-) by facultative anaerobic bacteria, and it can be acidified in the stomach, generating nitric oxide (NO), which has a strong bactericidal effect[29, 30]. Indeed, acidified NO₂ ^- is bactericidal against H. pylori[31]; this effect is amplified by thiocyanate (SCN^-), another NIS substrate[32]. H. pylori is frequently found in patients with gastric ulcers, suggesting a possible causative role for this bacteria[33]. Thus, one may speculate that if NIS-mediated NO₃ ^- secretion occurs in the stomach, NIS function might play a role in containing or preventing H. pylori-linked gastric ulcers or chronic gastritis. However, we observed no changes in NIS expression in samples from patients with gastritis as compared to normal subjects, irrespective of whether H. pylori was present.

NIS expression and function have been investigated in both thyroid and breast cancers. Whereas NIS expression is absent or decreased in 30% of thyroid cancers, NIS is actually overexpressed but not properly targeted to the plasma membrane in the remaining 70%[14, 22]. Significantly, over 70% of human breast cancers express NIS, raising the prospect of the possible use of radioiodide in the diagnosis and treatment of breast cancer[6], as is routinely and successfully done in the treatment of NIS-expressing thyroid cancer.

We ascertained NIS expression in gastrointestinal tumors by carrying out immunohistochemical analysis on biopsies obtained from 66 patients, partly during surgical interventions and partly during endoscopical examinations. In addition, 17 surgical samples were investigated by immunoblot analysis. We detected no NIS staining in patients with intestinalization or gastric cancer, indicating that malignant transformation is linked to decreased or suppressed NIS expression. Interestingly, the focal NIS staining observed in the direct vicinity of gastric tumors increased gradually and became linear as we proceeded away from the tumor. This suggests that loss of NIS expression may precede microscopically identifiable morphological changes. One could speculate that NIS expression in the apparently normal gastric mucosa could be used to define the true healthy margin, thus becoming a potentially useful indicator to decrease local/anastomotic recurrences.

In junctional and fundus-type columnar metaplasia Barrett mucosa, NIS expression was intact; in contrast, in Barret mucosa with intestinalization, NIS expression was absent. A similar phenomenon was observed in hyperplastic gastric polyps with intestinal metaplasia. These data provide the molecular explanation for the ^99mTcO₄ ^-scintigraphic results from 30 years ago[16–18, 20, 21].

Taken together, our findings underscore the prognostic and diagnostic significance of the absence of NIS expression in gastric alterations when intestinalization or cancer occurs. This is especially true in Barrett metaplasia, since the junctional and fundus-type metaplasias, in which NIS is normally expressed, pose a much lower risk of malignant transformation than intestinal metaplasia, which exhibits no NIS expression and gives rise to a higher number of dysplastic alterations and adenocarcinomas even in small (i.e., <3-cm) lesions[34]. In conclusion, this research suggests that the introduction of NIS immunohistochemical tests in gastric mucosa samples may be of considerable diagnostic value in Barrett esophageal and gastric polyps to evaluate intestinal metaplasia, and as an additional early molecular marker in the diagnosis of precancerous or/and cancerous gastroesophageal lesions.