Breast cancer is one of the leading causes of death among women worldwide due mainly to its ability to metastasize and develop chemoresistance. It has been estimated that one-third of breast cancer patients relapse at some point and that 25% of all cases are resistant to therapy (1). Chemoresistance is complex and involves several cellular and molecular events including alterations in the cell cycle, apoptosis or DNA damage repair pathways and a greater capacity to excrete chemotherapeutic drugs (2). The cell cycle regulator p21, which historically has been considered as a suppressor protein in normal cells, was recently linked to cancer progression and chemoresistance (3). For instance, the Nrf2-p21 axis has been associated with an increase in the resistant tumor cell population, activation of antioxidant mechanisms and chemoresistance in MCF-7, MDA-MB-231 and T47D cells (4). Moreover, ErbB2-dependent overexpression of p21 correlates with resistance to the chemotherapeutic drug Taxol in breast cancer (5), suggesting that in pathological conditions this cell arrest mechanism is triggered to protect the tumor cells from toxic treatments commonly used to target DNA division and/or induction of apoptosis. In recent years, drug expulsion has been considered another key mechanism of chemoresistance. The ATP-binding cassette (ABC) transporter family with its 49 members present in the human genome is one of the largest and oldest known protein families (6). One feature common to all members of this family is that they are membrane transporters that, by consuming ATP, are able to expel from cells a wide spectrum of substrates, including vitamins, lipids, hormones, metabolic waste products and xenobiotics such as toxins and drugs. Their expression and activity, in fact, are correlated with a decrease in the cytoplasmic concentration of drugs and consequent failure of therapy (7). In addition, in an analysis of cellular and population composition, the onset of chemoresistance has been linked to cancer stem cells (CSCs). CSCs are a cancer cell subpopulation that has been demonstrated to possess tumor-initiating properties and metastatic potential, and they are intrinsically chemoresistant (8). CSCs have been already described and characterized in several hematologic and solid tumors including breast cancer, where the CD44⁺/CD24⁻ surface marker profile has been considered a canonical CSC characteristic (9); although emerging evidence indicates that this profile is not exclusive to mammary cancer cells with CSC properties (10). Moreover, the origin of the CSC population is still controversial, and some other cellular events are associated with their stem-like profile as is the case for epithelial-mesenchymal transition (EMT). EMT in cancer is well documented and is characterized by a reversible conversion of cells with a polarized epithelial pattern into cells with a mesenchymal profile (11). At the molecular level, during EMT, epithelial cells lose adhesion molecules such as E-cadherin, lose their epithelial differentiation markers and acquire high motility by induction of vimentin and N-cadherin proteins. In fact, this transformation highly correlates with the CD44⁺/CD24⁻ profile and chemoresistance (12).

Several researchers have focused on improving chemotherapeutic treatment using natural molecules to limit chemoresistance and avoid significant increases in toxicity. Molecular iodine (I₂) is a chemical form of iodine that exerts significant antineoplastic effects on several types of cancer cells, and its actions could be mediated by multiple mechanisms. At moderately high concentrations, iodine induces a strong depolarization of mitochondrial membranes triggering mitochondrion-mediated apoptosis (13). Furthermore, I₂ is able to react with lipids and proteins producing several iodinated compounds. Among all the iodolipids 5-hydroxy-6-iodo-8,11,14,eicosatrienoic δ-lactone, also called 6-iodolactone (6-IL), has been confirmed to be an agonist of the peroxisome proliferator-activated receptor type γ (PPARγ). IL-6 promotes differentiation by decreasing the expression of specific markers associated with invasiveness and metastasis (14,15). Moreover, previous studies from our laboratory showed that when co-administered with doxorubicin (DOX), I₂ significantly improves conventional mammary cancer treatment in both women and rodents, and it diminishes the chemoresistance response (16,17). In the present study we developed a cell line resistant to low doses of DOX as a model to analyze in-depth how the I₂ supplement affects the chemoresistance response. DOX is an anthracycline antibiotic and is the most widely used chemotherapeutic drug in breast cancer treatment. Our results showed that after 30 days of exposure to 10 nM of DOX, MCF-7/D cells exhibited the same proliferation rate but higher expression of the p21, Bcl-2 and MDR-1 proteins associated with chemoresistance mechanisms in comparison with MCF-7/W. The molecular iodine supplement maintained its apoptotic effect in both types of cells, indicating that I₂ and DOX exert antineoplastic effects by different mechanisms. In addition, I₂ increased the intracellular retention of DOX and exerted a differential down-selection of the highly tumorigenic CD44⁺/CD24⁺ and E-cad⁺/vim⁺ subpopulations. The I₂ + DOX-selected cells showed a reserved tumorigenic competence in xenografts suggesting that the chemoresistance and invasive mechanisms were defective. All these I₂ actions were associated with a significant increase in PPARγ expression.

Initial characterization of the low-dose DOX-resistant model showed that at 10 nM DOX, the MCF-7/W cell culture maintained its proliferation rate at 60% of the untreated control, whereas 20 nM DOX induced a total block of proliferation at 96 h (Fig. 1A). The established DOX-resistant model required two and four times longer (8 and 14 days) to reach 80% confluence after the first and second subcultures (passages), but within 30 days, the duplication rate had returned to the control value (Fig. 1B). The acute treatment (96 h) with 10 nM DOX decreased the proliferation rate (% change) only in MCF-7/W cells (Fig. 1C). DOX adaptation was accompanied by significant increases in the expression of the chemoresistance markers p21, Bcl-2 and MDR-1 (Fig. 1D, MCF-7/D DOX). Removal of chronic DOX treatment from MCF-7/D cells decreased p21 and MDR-1 expression (MCF-7/D n.t.).

Fig. 2 shows the effect of 200 µM I₂ alone or co-administered with 10 nM DOX. Iodine alone inhibited proliferation similarly in both types of cells, and the magnitude of this inhibition was also similar to that observed in the MCF-7/W cells treated with 10 nM DOX. Co-administration of I₂ + DOX exerted an additive effect on both cellular populations, as indicated by the coefficients of drug interaction (CDI). Gene analysis (Fig. 3) showed that the antineoplastic effect of I₂ per se was associated with a decrease in Bcl-2 and an increase in PPARγ expression in both the MCF-7/W and MCF-7/D cells. These effects were also observed with I₂ + DOX, but in this case I₂ also impaired cell cycle arrest (canceled the increase caused by DOX) and intensified the decrease in Bcl-2 expression, thereby enhancing apoptosis induction (BAX/Bcl-2 index). Survivin expression did not show any change.

Fig. 4 shows the effect of I₂ on the expression of two of the most important drug expulsion membrane transporters. Iodine did not modify the expression of MDR-1 in the MCF-7/W cells but blocked its induction by DOX in the MCF-7/D cells. In contrast, the I₂ supplement showed significant induction of ABCg2 transporter expression in all conditions (Fig. 4A). The DOX functional retention assay showed an increase in the intracellular concentration of DOX (fluorescence) when I₂ was administered for 72 h, with a tendency observed at low concentrations, but a clear and significant increase at 500 nM DOX (Fig. 4B).

Phenotypes of mammary CSCs (CD44/CD24) and the EMT process (E-cad/vim) were analyzed in MCF-7/D cells. Fig. 5 shows a significant decrease in the CD44⁺/CD24⁺ population in favor of the double-negative cell population in I₂-treated cells with and without DOX. The CD44⁺/CD24⁻ phenotype, which is the scarcest subtype (<4%) observed in these resistant cells, showed a modest but significant increase (6%) after I₂ treatment. Fig. 6 shows that in terms of EMT classification, the most abundant population in the MCF-7/D cells corresponded to E-cad⁺. Iodine treatment was accompanied by a significant decrease in E-cad⁺/vim⁺ in favor of E-cad⁺/vim⁻, and again, this was independent of DOX presence. RT-PCR analysis show that the I₂ supplement diminished vimentin expression (Fig. 6B).

To analyze the in vivo tumorigenic capacity of MCF-7/D subpopulations, athymic mice were inoculated with DOX-resistant cells pre-incubated for 96 h with 10 nM DOX (MCF-7/D) or 200 µM I₂ + 10 nM DOX (MCF-7/I₂ + D). Each animal was inoculated with both subpopulations on the left or right side, respectively. Fig. 7 shows that MCF-7/D cells induced xenograft beginning on day 4 and maintained a rapid growth until day 12, whereas with MCF-7/I₂ + D, its growth rate and tumor size were significantly less.

The multistep protocol has been commonly used to establish in vitro models to study chemoresistance. Based on this method, resistant cells are selected by treating with a sequence of increasing DOX concentrations starting from low to over 1 mM depending on the author. However, in some cases this multi-step selection is accompanied by a loss of identity of cellular origin (20,21) or has only a weak correlation with clinical reality, where tumors undergo neither such step-by-step exposure nor such high DOX concentrations (22,23). For that reason, we used a single-step treatment extended for one month, a period that resembles the interval that separates one treatment cycle from the next. In our experience, DOX at a low concentration (10 nM) favors drug adaptation, as suggested by cell proliferation to a normal rate and decreasing sensitivity to DOX. At the molecular level, the antineoplastic effect of DOX results from a variety of actions; the best known are its ability to intercalate into DNA and to form a complex with topoisomerase II and DNA which triggers apoptosis, apparently via the p53-caspase pathway (24). In agreement, numerous studies indicate that DOX-resistant cells respond by decreasing topoisomerase II expression and increasing the expression of membrane drug transporters and the anti-apoptotic signal (22,25,26). Some of these mechanisms were observed in our MCF-7/D cells, such as cell cycle arrest (p21 upregulation), efficient drug expulsion (upregulated MDR-1 expression), and apoptosis evasion (BAX/Bcl-2 ratio decrease), indicating that these DOX-adapted cells can be considered as a chemoresistant cell model. The apparently paradoxical increase in p21 expression in response to DOX in both cell types agrees with recent studies showing that p21 can exert both anti- and pro-apoptotic effects in response to antitumor drugs, depending on cell type and cellular context (3). Cytotoxic drugs commonly act in mitotically active cells where they trigger apoptosis by inducing DNA damage (27). From this, it is reasonable to assume that early cellular alterations in reaction to such drugs may include apoptosis evasion and quiescence. Although the general observation that MCF-7/D cells return to the same proliferative rate as the wild-type, careful analysis reveals that these DOX-resistant cells include several subpopulations that could have different proliferation rates. Studies in our laboratory designed to confirm this hypothesis are now in progress.

The primary objective of the present study was to analyze for the first time the effect of iodine on the chemoresistance acquisition to DOX. Previous studies from our laboratory and others have shown that I₂ exerts antiproliferative and apoptotic effects in different models of cancer (16,17,28,29). Specifically, in mammary cell lines it has been demonstrated that cancerous (MCF-7, MDA-MB134, MDA-MB157 and MDA-MB436) and normal (MCF-10, MCF-12F) lines exhibit different sensitivity to I₂, but they all have a lower rate of proliferation when iodine is present (28,29). The most sensitive cell line is MCF-7, which is the focus of the present study as it represents the most frequent breast cancer in women (luminal, estrogen-positive) (30). Molecular iodine exerts a direct apoptotic effect by mitochondrial membrane depolarization and/or an indirect action via 6-iodolactone (6-IL). This iodolipid is generated by iodination of arachidonic acid; by activating PPARγ, 6-IL induces apoptosis and differentiation effects in MCF-7 cells (16,17). In the present study, I₂ maintained its apoptotic effect independent of the DOX-resistance mechanisms acquired by the cells. Several studies have shown that DOX-adaptation confers resistance to other drugs, due mainly to the overexpression of ABC transporters (22). It is possible that these membrane transporters cannot expel iodine. Indeed, our results showed that I₂ alone inhibited MDR-1 expression only weakly, but it significantly impaired MDR-1 upregulation by DOX treatment in MFC-7/D cells, suggesting that the changes associated with I₂ treatment were capable of interfering with the installation of DOX-resistance. One interesting observation is the significant increase in ABCg2 expression in both types of cells treated with I₂. It is well documented that PPARγ activation inversely modulates MDR-1 and ABCg2. Although the MDR-1 gene does not contain response elements to PPARs, these receptors can inhibit the Wnt/β-catenin pathway, which is directly involved in MDR-1 regulation (31). In contrast, ABCg2 is directly stimulated by PPARγ agonists (32) and although these transporters are overexpressed in some tumor types, they have also been detected in several normal tissues such as intestine, liver, brain, placenta and mammary glands (7). Moreover, this breast cancer resistance protein (ABCg2) is strongly induced in the mammary gland during pregnancy and lactation and is responsible for pumping vitamin B2 into milk, suggesting a physiological role in differentiated mammary cells (33). These facts, along with the observation that I₂ treatment is accompanied by significantly higher intracellular retention of DOX, suggest that the antineoplastic effect of iodine could be related to PPARγ activation resulting in maintaining drug sensitivity (downregulation of MDR-1 and, therefore, lower drug expulsion) and the induction of cell differentiation. It is well established that MCF-7 cells can respond to synthetic agonists of PPARγ by increasing lipid accumulation, terminating cell growth and undergoing changes characteristic of a less malignant state (14,34,35). These re-differentiation responses were also described by our group in mammary (MCF-12 and MCF-7), prostate (RWPE-1, LNCaP and DU-145) and neuroblastoma (SKN-AS and SKN-SH5Y) cell lines after I₂ or 6-IL administration (28,36,37). In this context, it is possible that the significant increase in the ABCg2 transporter corresponds more to an induction of differentiation than of chemoresistance. The analysis of CSC and EMT markers showed that the canonic CSC profile CD44⁺/CD24⁻ expected to be enriched in drug-resistant cells was poorly represented (<4%) in MCF-7/D cells, whereas the most abundant populations were the CD44⁺/CD24⁺ and CD44⁻/CD24⁺ subtypes (~40% each). The supplementation with I₂ showed a discrete increase in CD44⁺/CD24⁻ (~6%), no change in CD44⁻/CD24⁺ and a clear differential selection against CD44⁺/CD24⁺ with a significant increase in the double-negative population (CD44⁻/CD24⁻). Previous studies have described that the canonic CD44⁺/CD24⁻ profile is not the only profile that corresponds to an invasive phenotype. Indeed, in a recent study, using sphere-promoting (Mammocult; Stem Cell Technologies, Vancouver, BC, Canada) conditions, this double-positive subpopulation was found to be the most representative group in the MCF-7 CSC culture (30). Increases in the double-positive population were found to be associated with a worse outcome in salivary gland (38), pancreatic carcinomas (39), and in colorectal cancer this double-positive population represents the specific marker for CSCs (40). Controversial results have been reported in relation to the CD44⁻/CD24⁺ profile. In various studies, increases in CD24⁺ cells were found to be correlated with the most aggressive phenotype (41–43), whereas in others there was no correlation with prognosis (44,45). In contrast, the double-negative phenotype had no prognostic significance in breast cancer patients (45), and in preclinical studies these cells showed reduced capacity to induce tumor growth in soft agar and xenografts in mouse models (46), suggesting that these cells are less invasive. This less-aggressive profile found in I₂ + DOX cells was confirmed by the enrichment of E-cad⁺/vim⁻ cells. Indeed, the expected EMT profile (E-cad⁻/vim⁺) was absent in MCF-7/D cells, and the double-positive population was significantly diminished in favor of the E-cad⁺/vim⁻ subpopulation when these cells were treated with I₂. E-cadherin is a transmembrane glycoprotein involved in epithelial adherens junctions, and its loss could be sufficient to promote the invasion-metastasis cascade, activating specific downstream signal transduction pathways that bestow high motility on the cells by inducing vimentin and N-cadherin proteins (47). In contrast, vimentin which is the most commonly expressed and highly conserved member of the type III intermediate filament protein family is considered the main EMT marker. High vimentin expression is observed in several aggressive breast cancer cell lines. In MCF-7 cells, vimentin overexpression is accompanied by increases in motility and invasiveness. These characteristics were reduced by vimentin antisense oligos in MDA-MB-231 cells, which constitutively express this protein (48). Congruently, our results showed that I₂-treated cells exhibited the lowest vimentin expression and that the I₂ + DOX-treated subpopulation was powerless to initiate tumor xenografts, corroborating its weak invasive potential. The EMT process, which is triggered by factors such as transforming growth factor-β (TGF-β), SNAIL and TWIST, is reverted by PPARγ activation (49,50). Studies have shown the antineoplastic effects of PPARγ ligands in various preclinical models (51); however, agonists of these receptors used as monotherapy failed to exert therapeutic benefits in advanced stage breast patients (52). Notably, PPARγ agonists in combination with the conventional antineoplastic drugs, such as carboplatin or tumor necrosis factor-related apoptosis-inducing ligand (TRAIL), showed synergistic effects, indicating that differentiation induced by PPARγ activation restored sensitivity to the cytotoxic drug (53,54). These synergistic effects were replicated in cells treated with DOX + I₂ in both preclinical (16) and clinical studies (17), supporting the notion that some I₂ effects are mediated by PPARγ activation.

In conclusion, the use of molecular iodine at a moderately high concentration restored the sensitivity of mammary cancer cells MCF-7/D to DOX. Impaired DOX expulsion and decreased expression of the chemoresistance markers p21, Bcl-2 and MDR-1 resulted in the selection of a less aggressive population, suggesting the potential of I₂ as a clinically useful anti-chemoresistance agent.