Breast cancer is one of the most common cancers in women in North America, where incidence is highest in the world [1]. In the United States, more than 200,000 newly diagnosed cases were estimated in the year 2005. While treatment with surgery, chemotherapy, hormonal and radiation therapy, and antibodies targeting the Her-2/neu growth promoting protein have slowed disease progression or resulted in remission in many patients, metastatic disease still causes death in the majority of affected patients within 5 years of diagnosis. Thus, continued discovery of novel methods of detecting and treating residual and metastatic breast cancer are of extremely high importance.

The sodium/iodide symporter (NIS) is a plasma membrane glycoprotein serving as the molecular basis for iodide accumulation in the thyroid gland, where it mediates active uptake of iodide from the bloodstream for incorporation into thyroid hormones [2]. Radioiodine is routinely administered to patients for the diagnosis and treatment of thyroid carcinomas, resulting in a 10-year survival rate of greater than 90% [3, 4].

NIS is also expressed in the lactating mammary gland, where it facilitates uptake of iodide from the maternal bloodstream into milk for utilization by the developing neonatal thyroid gland [5, 6]. While NIS has been detected in approximately 80% of human breast cancers using immunohistochemical staining [7], few tumors have clinically evident iodide uptake ability. In one study, although 24 out of 25 breast cancer patients had detectable NIS mRNA expression in their primary tumors, positive radionuclide uptake was only noted in four of the tumors (17%) and was correlated with NIS mRNA expression level [8]. A second study demonstrated that only 25% (2/8) of breast cancer metastases expressing NIS protein had clinically evident radionuclide uptake [9]. NIS-mediated radionuclide imaging and therapy has exciting potential as a safe and effective method of breast cancer diagnosis and treatment, and delineating factors underlying the discordance between NIS expression and function is critical in moving this modality into the practical realm. To achieve this aim, mechanisms of NIS regulation in breast cancer must be uncovered.

We previously found that phosphatidylinositol-3 kinase (PI3K) and cAMP signaling pathways were associated with NIS expression in transgenic mouse models of breast cancer [10]. However, cAMP alone was not sufficient to induce NIS protein expression in cultured MCF-7 human mammary carcinoma cells. Our current study was to evaluate the role of PI3K signaling in NIS expression, trafficking and iodide uptake ability in MCF-7 cells and in human breast tumors, as activating mutations of the PI3K p110α subunit are frequent in human breast cancers [11] and cell lines [12]. We found that stable or transient activation of PI3K p110α in MCF-7 cells increases underglycosylated NIS protein levels with impairment of NIS cell surface trafficking, which is required for iodide uptake activity. PI3K activation also interferes with cell surface trafficking and function of all-trans retinoic acid and hydrocortisone (tRAH) induced endogenous NIS as well as transiently expressed exogenous Flag-tagged NIS. Finally, an association was found between PI3K activation and NIS expression in human breast tumor tissues.

The importance of PI3K signaling in NIS regulation was studied using MCF-7 human mammary carcinoma cells, which are the only immortalized breast cancer cells with inducible NIS expression and function [20]. While NIS protein was not detectable in untreated MCF-7 cells, it was markedly induced by all-trans retinoic acid combined with hydrocortisone (tRAH) [16] as the glycosylated 86 kDa form (Figure 1A). In contrast, only the underglycosylated 50 kDa NIS form was present in MCF-7 cells stably expressing activated PI3K p110α^CAAX. Cell surface NIS protein was detectable in MCF-7 cells treated with tRAH, but not in the MCF-7 cells stably expressing PI3K p110α^CAAX (Figure 1B). NIS immunofluorescence was performed to further assess the effect of PI3K activation on NIS subcellular localization in MCF-7 cells. While intense cell border NIS protein was evident in the tRAH-treated MCF-7 cells (arrows, left panel, Figure 1C) and sparsely located in the perinuclear area, it was not apparent in the MCF-7/PI3K p110α^CAAX cells, where NIS appeared dispersed throughout the intracellular compartment (right panel, Figure 1C).

PI3K activation also appears to interfere with the ability of tRAH to induce glycosylated NIS protein expression. As shown in Figure 2A, while abundant glycosylated NIS protein was induced by tRAH treatment in parental MCF-7 cells (lane 1 vs. lane 2), underglycosylated 50 kDa NIS was the dominant form in MCF-7/PI3K p110α^CAAX cells, which negligibly responded to tRAH treatment with glycosylated NIS protein (lane 3 vs. lane 4). Transient expression of PI3K p110α^CAAX also induced the underglycosylated 50 kDa NIS form (lane 1 vs. lane 5), and responsiveness to tRAH treatment with glycosylated NIS protein was decreased by approximately 45% (lane 2 vs. lane 6) even though the transfection efficiency was only 20–25%. Interestingly, tRAH treatment slightly decreased the overall level of NIS protein in MCF-7/PI3K p110α^CAAX cells.

MCF-7/PI3K p110α^CAAX cells initially had modest basal iodide uptake ability, as published in our previous study [10], but most of this function was lost over cell passage even though NIS protein remained detectable. Radioiodide uptake ability in our current study correlated well with fully glycosylated (surface) NIS protein levels. The stably transfected MCF-7/PI3K p110α^CAAX cells had minimally increased NIS-mediated radioiodide uptake over parental MCF-7 cells and did not significantly respond to tRAH treatment (Figure 2B). Transient transfection of activated PI3K p110α^CAAX in parental MCF-7 cells modestly increased NIS-mediated radioiodide uptake over empty vector-transfected controls but decreased tRAH-induced radioiodide uptake activity (Figure 2C).

A similar inhibitory effect of PI3K activation on NIS surface trafficking and function was observed upon forced expression of exogenous NIS. Immunofluorescence for exogenous Flag-tagged NIS protein demonstrated that Flag-NIS cell surface trafficking was largely inhibited in MCF-7/PI3K p110α^CAAX cells as compared to parental MCF-7 cells (Figure 3A). Additionally, radioiodide uptake activity conferred by transient transfection of exogenous Flag-NIS was decreased by 4-fold in MCF-7/PI3K p110α^CAAX cells versus parental MCF-7 cells (Figure 3B). Thus, cell surface trafficking of both endogenous and exogenous NIS were decreased by PI3K activation.

To determine whether PI3K upregulation plays a role in the discordance between the increase in NIS expression and the absence of radioiodide uptake activity in human breast tumors, we evaluated a human breast cancer tissue microarray for NIS expression and localization using immunohistochemistry. This array consisted of randomly selected breast tissues from patients at The Ohio State University Medical Center with a diagnosis of breast carcinoma during the years 1992–1994. Tumors were 51% ER-positive, 57% PR-positive, 53% Her-2/neu-positive, and 27% with loss of PTEN tumor suppressor, each of which had no statistical correlation with NIS expression. All tumor specimens were primary, although 29% of the patients also had local lymph node metastases. None had known distant metastases at the time of biopsy. Patients had an average age of 42.26 years and were 89% Caucasian and 11% African-American.

We correlated NIS expression/localization with the presence of PI3K activation as indicated by phosphorylation and nuclear translocation of Akt, which is commonly used to estimate PI3K pathway activation. Similar to previously published data, only 20% of tumors were NIS-negative (Figure 4A). Overall, 58% of tumors had primarily intracellular NIS expression (Figure 4B), while 22% had some degree of plasma membrane NIS expression (Figure 4C). NIS-positive tumors were three times more likely than NIS-negative tumors to have PI3K activation as indicated by Akt phosphorylation (Table 1 and Figures 4A–C versus 4D–F). Interestingly, all tumors with loss of the PTEN (PI3K antagonist) and concomitant pAkt expression had marked intracellular NIS expression without notable cell surface NIS expression.

This study is the first to uncover a potential explanation for the discordance between NIS expression and iodide uptake ability in human breast cancers, a roadblock to widespread application of NIS-mediated radioiodide imaging and ablation of residual and metastatic tumors. We showed that both transient and chronic PI3K activation increase underglycosylated NIS protein levels but impair NIS cell surface trafficking, thus accounting for the absence of NIS-mediated radioiodide uptake activity. NIS expression and intracellular localization were also associated with pAkt expression in human breast cancer tissues. Our study suggests that the increased intracellular NIS protein in human breast cancer specimens [7] is likely a consequence of PI3K activation.

Many human breast cancers have PI3K activation as a consequence of Her-2/neu or Src oncogene expression [21], or by activating mutations of the PI3K p110α subunit [11], any of which could lead to NIS expression. In our breast cancer tissue microarray, PI3K activation might be explained by Her-2/neu expression in 48% of the pAkt-positive tumors or by loss of PTEN in 19% of pAkt-positive tumors. Activating point mutation E545K in the helical domain of PI3K p110α has been described in many cultured breast cancer cells, including MCF-7 [12]. In our study, both chronic and transient PI3K p110α overexpression resulted in 2–3 fold greater pAkt protein levels than found in parental MCF-7 cells.

The lack of NIS protein glycosylation in PI3K p110α^CAAX-expressing MCF-7 cells is consistent with impairment of processing for NIS cell surface trafficking. Membrane glycoproteins are processed by glycosylation beginning in the endoplasmic reticulum and Golgi apparatus with transport using machinery such as the AP-1B adaptor complex and clathrin-coated vesicles and final transport to the cell surface by carrier vesicles [22]. While it is not currently known how PI3K activation interferes with NIS cell surface trafficking, it is known that one subtype of PI3K interacts with clathrin [23].

It has been shown that abolishing three glycosylation sites in human NIS resulted in a 50 kDa NIS protein with decreased NIS activity after exogenous delivery into COS cells [24]. The fact that this mutant still had 50% of wild-type iodide uptake ability indicates that some NIS protein was able to reach to the cell surface. However, it was unclear whether the decrease in radioiodide uptake was due to less efficient cell surface trafficking or to a decrease in NIS activity. We interpret the lack of NIS glycosylation in our MCF-7 model to be a consequence of protein retention at a pre-glycosylation stage rather than a cause of trafficking failure. However, we cannot exclude that a lack of glycosylation itself might impair trafficking and function.

In our current study, tRAH-responsive NIS induction was nearly abolished in MCF-7/PI3K p110α^CAAX cells and reduced in parental MCF-7 cells transiently transfected with PI3K p110α^CAAX. The greater inhibition of NIS cell surface trafficking and function in PI3K-overexpressing stable clones vs. PI3K-transfected MCF-7 cells could be related to chronicity of PI3K activation, which is a likely scenario in human breast cancer. Additionally, transient transfection efficiency is only 25–30% in MCF-7 cells.

In summary, this study advances our knowledge of NIS regulation in breast cancer by demonstrating that PI3K signaling increases underglycosylated NIS protein levels without an increase in NIS-mediated radioiodide uptake activity due to impairment in NIS cell surface trafficking. Because many oncogenes in human breast cancer upregulate PI3K signaling, and PI3K somatic mutation was frequently detected in human breast cancers, this pathway could realistically account for the high intracellular NIS expression and lack of radioiodide uptake in most human breast cancers. The fact that PI3K signaling appears to have differential effects on NIS expression in the breast versus thyroid suggests that a strategy may be developed to selectively increase NIS expression in breast tissue. Further investigation into the mechanism(s) of impairment of NIS cell surface trafficking by PI3K will be critical in realizing the application of NIS-mediated radioiodide imaging and ablation in breast cancer.