Nitric Oxide Suppresses Tumor Cell Migration through N-Myc Downstream-regulated Gene-1 (NDRG1) Expression

Background: Expression of N-Myc downstream-regulated gene 1 inversely correlates with patient outcome. Results: Nitric oxide exposure leads to NDRG1 gene expression, which inhibits tumor cell migration. Conclusion: Nitric oxide-mediated sequestration of chelatable iron via dinitrosyliron complex formation is a major determinant of NDRG1 gene expression and phenotypic outcome. Significance: This mechanism of NDRG1 regulation is crucial for understanding the impact of •NO on metastasis. N-Myc downstream-regulated gene 1 (NDRG1) is a ubiquitous cellular protein that is up-regulated under a multitude of stress and growth-regulatory conditions. Although the exact cellular functions of this protein have not been elucidated, mutations in this gene or aberrant expression of this protein have been linked to both tumor suppressive and oncogenic phenotypes. Previous reports have demonstrated that NDRG1 is strongly up-regulated by chemical iron chelators and hypoxia, yet its regulation by the free radical nitric oxide (•NO) has never been demonstrated. Herein, we examine the chemical biology that confers NDRG1 responsiveness at the mRNA and protein levels to •NO. We demonstrate that the interaction of •NO with the chelatable iron pool (CIP) and the appearance of dinitrosyliron complexes (DNIC) are key determinants. Using HCC 1806 triple negative breast cancer cells, we find that NDRG1 is up-regulated by physiological •NO concentrations in a dose- and time-dependant manner. Tumor cell migration was suppressed by NDRG1 expression and we excluded the involvement of HIF-1α, sGC, N-Myc, and c-Myc as upstream regulatory targets of •NO. Augmenting the chelatable iron pool abolished •NO-mediated NDRG1 expression and the associated phenotypic effects. These data, in summary, reveal a link between •NO, chelatable iron, and regulation of NDRG1 expression and signaling in tumor cells.

its name implies, it is repressed by the proto-oncogenes N-Myc (neuroblastoma-derived myelocytomatosis) and Myc (3). The NDRG family of proteins is a member of the ␣/␤ hydrolase superfamily (4,5). These proteins contain mutations in all three catalytic residues of the active site, however, and possess no hydrolytic activity (4). Although the mechanism of action of NDRG1 has not been elucidated, its expression is associated with diverse physiologic processes ranging from developmental biology and endocrine signaling to immune responses and neuronal functioning (6). Pathologically, NDRG1 dysregulation has been linked to a host of disease states including neurological disorders. Most notably, however, is the strong association between NDRG1 and the metastatic progression of various cancers (2,7).
Originally discovered by a differential display study on homocysteine-treated human umbilical vein endothelial cells (8), NDRG1 mRNA and protein were subsequently found to be markedly decreased in breast, prostate, esophageal, glioma, and colon cancers when compared with normal tissue (9 -15). NDRG1 mRNA also was found to be more abundantly expressed in primary colon cancer tumors than their metastases (12). This protein was shown to be up-regulated by p53 (16 -18) and PTEN (the phosphatase and tensin homolog deleted on chromosome 10) (19); thus it was classified as a metastatic suppressor. Examination of other cancer types, however, found that NDRG1 is more abundantly expressed in cervical cancer, renal cancer, and hepatocellular carcinoma (20 -24). These findings highlight the context and tissue-specific functions of NDRG1. A recent study performing immunohistochemical analysis of tissue from prostate cancer patients found that when NDRG1 expression was examined in conjunction with KAI1 expression, the concomitant down-regulation of these genes was an independent prognostic marker of metastatic prostate cancer (25). Thus whereas NDRG1 expression may not be an effective prognostic indicator itself, it appears to be among a set of genes that have a characteristic expression indicative of metastasis in at least breast, prostate, and colon cancer.
Despite a knowledge gap in the mechanistic functioning of this protein and its associated phenotypes, numerous upstream regulatory effectors have been identified. These include generalized cell stress, small molecules (such as cAMP and Fe 2ϩ ), and numerous proteins including HIF-1␣, p53, PTEN, and the MYC family. A considerable body of work has been done that describes the effect of hypoxia and HIF-1␣ expression on NDRG1 up-regulation. There are also numerous reports implicating a role for metals in NDRG1 expression. Inducers of "chemical hypoxia" such as nickel (Ni 2ϩ ), cobalt (Co 2ϩ ), as well as chemical iron chelators, are known to strongly induce the expression of NDRG1 (NDRG1 regulation is well reviewed in Refs. 2, 6, and 7)). The link between metals, hypoxia, and NDRG1 up-regulation may be via HIF-1␣ accumulation. One mechanism by which Co 2ϩ and Ni 2ϩ mimic hypoxia is by substituting for the Fe 2ϩ atom in mononuclear non-heme iron oxygenases that contain the 2-His-1-carboxylate facial triad structural motif. For example, inhibition of HIF prolyl hydroxylase by this mechanism results in increased HIF-1␣ levels (26,27). Although compounds that elicit a hypoxic response have been shown to up-regulate NDRG1, both HIF-1␣-dependent and independent mechanisms of regulation have been described. NDRG1 expression was not observed in HIF-1␣ Ϫ/Ϫ mouse embryo fibroblasts (22), for example, although its expression was not hindered in HIF-1␣-deficient kidney cells (27).
Nitric oxide ( ⅐ NO, nitrogen monoxide) is a ubiquitous free radical signaling molecule that regulates many cellular processes including angiogenesis, smooth muscle tone, immune response, apoptosis, and synaptic communication (28). In addition to the many normal physiologic functions of ⅐ NO, it has been implicated in the etiology and progression of many diseases including cancer (29). Although ⅐ NO is produced in and around certain tumors, its unique physical and chemical properties dictate that under biological conditions it only reacts with a minority of chemical species; i.e. other radicals and transition metals (30). Of these biological targets, one of the potentially most significant and least studied is the chelatable iron pool (CIP). This small, but chemically significant, fraction of total cellular iron (0.2-3.0%, low M range) (31, 32) is methodologically defined because it is accessible to chemical iron chelators (33). More importantly, it has recently been demonstrated that when cells are exposed to ⅐ NO, the CIP is quantitatively converted into paramagnetic dinitrosyliron complexes with thiolcontaining ligands (DNIC) (34,35).
Nitric oxide is distinct from most signaling molecules in that it is not limited to classical receptor-ligand interactions, and it can directly target a wide variety of molecules within a cell (including heme and non-heme iron). Moreover, phenotypic consequences attributed to ⅐ NO are often the indirect result of higher nitrogen oxides formed from secondary reactions with other radical species. However, like the effects of iron chelators and divalent metals (Co 2ϩ and Ni 2ϩ ), ⅐ NO has been shown to disrupt iron homeostasis and inhibit mononuclear non-heme iron oxygenases such as prolyl hydroxylase (36). Although signaling responses of ⅐ NO can proceed through a multitude of possible mechanisms, it can be seen that there are distinctly recognizable biochemical similarities between the effects of ⅐ NO and those of chelators and metals (Co 2ϩ and Ni 2ϩ ) (37)(38)(39)(40). For these reasons we asked whether ⅐ NO might prove to be equally as efficacious as these other compounds at up-regulating NDRG1. This is the first report of NDRG1 expression in response to ⅐ NO. We note that triple negative breast cancer cells exposed to ⅐ NO demonstrate NDRG1 mRNA and protein up-regulation consistent with the sequestration of chelatable iron in the form of DNIC. DNIC have been detected in numerous human tissue types and tumors (41,42) yet explanations regarding their biological and potential therapeutic significance are limited. This may be significant because both ⅐ NO production and NDRG1 protein levels are known to be increased in breast cancers. Although the specific upstream target for ⅐ NO that can be attributed to NDGR1 expression has not been elucidated, the reaction of ⅐ NO with chelatable iron is a significant contributor to this response. Furthermore, we demonstrate that the inhibitory effect of ⅐ NO on triple negative breast cancer migration is a consequence of diminishing the chelatable iron pool and upregulation of NDRG1. The amount of iron in the CIP is inversely related to the amount of iron in DNIC, such that the increase in ⅐ NO-bound iron correlates with a proportional decrease in tumor cell migration and invasion.
Cell Culture-HCC 1806 triple negative breast cancer cells were obtained from the American Type Culture Collection (ATCC). The cells were grown to 80% confluence in 6-cm tissue culture plates in RPMI 1640 growth medium containing 10% fetal bovine serum and 1% penicillin/streptomycin. Prior to treatments, growth media was replaced with serum-free growth medium for 16 h. All experiments were conducted under these culture conditions.
Iron Supplementation-Cells were treated with ferric ammonium citrate (150 g/ml) for the indicated time points. Plates were washed with PBS to remove excess extracellular iron.
Western Blot Analysis-Briefly, cells were lysed using CelLytic TM M Cell lysis reagent (Sigma) with 1% protease inhibitor mixture (Calbiochem) and 1 mM PMSF (Sigma). Protein samples were separated on denaturing polyacrylamide gels (Bio-Rad) and transferred to PVDF membranes using the iBlot transfer system (Invitrogen). The membrane was blocked, and incubated overnight with primary antibodies for NDRG1 (Santa Cruz Biotechnology) and HIF-1␣ (Transduction Laboratories). The blots were finally analyzed in a Fluor Chem HD2 imager (Alpha Innotech) using SuperSignal West Femto Max-imal Sensitivity Substrate (Thermo Scientific). Figures are representative of n Ն 3 individual experiments.
Cell Migration, Invasion, and Viability-The xCELLigence DP system was used for measurement of migration, invasion, and cell viability. This system utilizes specialized culture plates that contain gold electrode arrays on the bottom of individual wells. Increased cellular contacts on the electrode surfaces increase the impedance across these gold arrays. This impedance value is measured by the DP system and is reported in the arbitrary unit of cell index.
E-plates-These single-use plates were used for measuring growth and viability. 80% of the 5.0-mm surface on the bottom of wells was covered with a gold electrode array. 2 ϫ 10 4 cells were added to individual wells. Cells were allowed to settle for 30 min before any treatments were added. Measurements of cell index were taken every 15 min.
Cellular Invasion and Migration (CIM) Plates-The CIM plate is a single-use two-chambered system similar to a standard Boyden chamber. The upper chambers are sealed at the bottom with a microporous polyethylene terephthalate membrane containing microfabricated gold electrode arrays on the bottom side of the membrane. The median pore size of this membrane is 8 m. Chamber diameter is 5.0 mm. 2 ϫ 10 4 cells were suspended in serum-free growth media and added to the top chambers. The lower chambers contained 10% serum media as a migratory stimulant. Cells were allowed to settle for 30 min before the addition of any treatments. Measurements of the cell index were taken every 15 min.
For invasion studies, the membrane on the bottom of the top chamber of a CIM plate assembly was coated with 30 l of a 1:40 dilution of Matrigel TM (BD Biosciences) in serum-free media and incubated at 37°C for 4 h. 4 ϫ 10 4 cells were added to the upper chamber in serum-free media. Subsequent steps were performed in the same manner as the cell migration assay.
Knockdown Cell Lines-NDRG1 and HIF-1␣ knockdown HCC 1806 cells were prepared by transfection with Mission plasmids (Sigma). Briefly, a plasmid vector containing either NDRG1-or HIF-1␣-specific shRNA was transfected into the HCC 1806 cell line using Lipofectamine 2000. Individual colonies were selected by growth in puromycin containing media. The cell lines were validated for knockdown by qRT-PCR and Western blot analysis.
Real-time ⅐ NO Measurements-Cells were grown in 15-cm plates. A ⅐ NO-selective electrode (amiNO-2000, innovative instruments) was connected to an Apollo 4000 free radical ana-lyzer (World Precision Instruments) and positioned ϳ1 mm above the monolayer and allowed to equilibrate for 2 h followed by addition of the ⅐ NO-donor DETA/NO (n ϭ 3).
Electron Paramagnetic Resonance-EPR was performed on a Bruker X-band EMX Plus EPR spectrometer. Samples were frozen and read in liquid N 2 . DNIC were detected at g ϭ 2.03, modulation amplitude 10 G, 200 G scan range, 90-s scan time, for 1 scan. For quantification, the double integral of the first derivative spectra was compared with that of a 15-point standard curve generated with synthetic diglutathion DNIC (supplemental Fig. S1). CIP was estimated by treating cells with 1 mM desferrioxamine for 4 h. The resulting Fe 3ϩ -desferrioxamine g ϭ 4.3 signal was read with a modulation amplitude of 10 G, 200 G scan range, 30-s scan time, for 4 scans. For quantification, the double integral of the first derivative spectra was compared with that of a 15-point standard curve generated with Fe 3ϩdesferrioxamine (supplemental Fig. S2).

RESULTS
Nitric Oxide Induces NDRG1 mRNA Transcription and Protein Translation-It is well established that NDRG1 is up-regulated in response to chemical iron chelators (37). We, and others, have recently demonstrated that ⅐ NO can react with the CIP to form cellular DNIC (34,35). Because ⅐ NO targets the same pool of iron as chelators (the CIP), we hypothesized that ⅐ NO might have a comparable effect on NDRG1 up-regulation via its ability to diminish cellular iron availability.
We first performed a time course analysis of NDRG1 protein ( To determine whether up-regulation of NDRG1 by ⅐ NO is dose-dependent, we examined changes in mRNA and protein levels over a range of DETA/NO concentrations (at 8 h). It can be seen in Fig. 1, C and D, that up-regulation of NDRG1 by ⅐ NO occurs in a concentration-dependent manner. Increases in NDRG1 mRNA and protein levels are evident with as little as 125 M DETA/NO ([ ⅐ NO Ϸ 5-50 nM] ss ). Maximal induction is achieved with 500 M DETA/NO ([ ⅐ NO Ϸ 225 nM] ss ). A further increase in the ⅐ NO concentration had no effect on the magnitude of NDRG1 up-regulation.
⅐ NO-donor compounds, like DETA/NO, enable continuous controlled treatment of cells with steady-state ⅐ NO concentrations. The concentration of any ⅐ NO-donor compound, however, is not indicative of the actual steady-state ⅐ NO concentration that the cells are exposed to (43). For this reason we used electrochemical detection to measure, in real-time, steadystate ⅐ NO concentrations in the media of DETA/NO-treated cells. The steady-state ⅐ NO concentrations from 500 M DETA/NO were well within the physiologic range (Ϸ80 -750 nM over the 24-h period, Fig. 2). Stability of NDRG1 mRNA and Protein after ⅐ NO Exposure-Nitric oxide is often constitutively or intermittently increased in solid tumors (29,45). We hypothesized that NDRG1 protein DECEMBER 2, 2011 • VOLUME 286 • NUMBER 48 stability in response to its up-regulation by ⅐ NO could be an important downstream phenotypic determinant of its metastatic suppressive function. Therefore, we set out to determine whether the elevated NDRG1 protein and mRNA levels persisted after ⅐ NO removal. In Fig. 1 it can be seen that 4 -24 h of ⅐ NO exposure leads to an increase in NDRG1 mRNA and protein. It is not clear, however, if these increases were in response to continuous ⅐ NO exposure or whether NDRG1 transcription might continue after shorter initial treatments. We evaluated this by treating cells with ⅐ NO (DETA/NO 500 M) for various lengths of time (1-8 h), removing the ⅐ NO source, and measuring changes in NDRG1 protein levels at 8 h for all samples. This data indicated that a minimum of 4 h of ⅐ NO exposure was required for NDRG1 up-regulation, but protein levels continued to rise with increasing duration of exposure (Fig. 3A).

Nitric Oxide Suppresses Tumor Cell Migration
Next we set out to determine the stability of NDRG1 mRNA and protein after its up-regulation by ⅐ NO. Cells were exposed to ⅐ NO for 8 h (time ϭ 0, Fig. 3, B and C), the ⅐ NO source was then removed, and samples were collected at various time points to determine changes in NDRG1 mRNA and protein levels. Although it can be seen that NDRG1 protein levels are continually elevated for Ͼ48 h post ⅐ NO removal, mRNA levels return to baseline values by 16 h.
Chelatable Iron Is Involved in the Regulation of NDRG1 by ⅐ NO-NDRG1 is strongly up-regulated by treatment with iron chelators (39). Upon cellular ⅐ NO exposure, the CIP is quantitatively converted into paramagnetic DNIC. The amount of DNIC formed is related to the dose and duration of ⅐ NO exposure (35). To date, almost all known ⅐ NO signaling pathways are through a process that involve heme binding or S-nitrosothiol formation. Although we cannot directly refute these mechanisms, our data suggest that the effect of ⅐ NO on NDRG1 expression involves an alternate mechanism. Therefore, we asked if the ability of ⅐ NO to alter cellular iron availability could be sufficient to regulate NDRG1 mRNA and protein.
This hypothesis was tested by augmenting the CIP in the presence of ⅐ NO. We treated cells with iron (ferric ammonium citrate (FAC)) for various lengths of time and measured changes in the g ϭ 4.3 signal by EPR (Fig. 4, A and B). This data  confirmed that the CIP could be increased by exogenous iron administration and that measured increases in the CIP were proportional to the length of time the cells were treated. We then measured DNIC levels in these iron-loaded cells after ⅐ NO exposure (500 M DETA/NO) and determined that increases in DNIC were proportional to increases in the CIP (Fig. 4, A and  B). Finally, we compared the amount of DNIC formed in ironloaded cells after treatment with two different concentrations of ⅐ NO (500 versus 1,000 M DETA/NO, Fig. 4C). At the higher dose of ⅐ NO, the amount of DNIC approached the level of the augmented CIP. This indicates that the concentration of ⅐ NO may be a limiting factor for DINC formation at high CIP levels (i.e. at low ⅐ NO concentrations less of it will react with the CIP versus competing reactions with other cellular targets).
Although the biological activity of DNIC remains largely unknown, one clear consequence of their formation is a diminished CIP. It seems probable that sequestration of chelatable iron by ⅐ NO may profoundly affect enzymatic signaling pathways requiring iron ion cofactors. To verify the role of chelatable iron in NDRG1 regulation, we measured changes in NDRG1 protein levels following treatment of HCC 1806 cells with two different metal chelators and with ⅐ NO (Fig. 5A). The magnitude of NDRG1 gene expression induced by ⅐ NO or iron chelators was equivalent.
If ⅐ NO regulates NDRG1 via iron sequestration in the form of DNIC, it follows that greater amounts of ⅐ NO should be neces-sary to elicit the same response when the CIP is increased. As a proof of principle, we treated HCC 1806 cells with iron (FAC) for varying lengths of time to make stepwise increases in the CIP (as in Fig. 4). These cells were then exposed to ⅐ NO for 8 h and changes in NDRG1 mRNA (Fig. 5C) and protein levels ( Fig.  5B) were measured. For a given amount of ⅐ NO, the degree of NDRG1 gene expression in response to ⅐ NO was less when the CIP was augmented. In a similar set of experiments, we further verified these findings by treating HCC 1806 cells with or without iron for 16 h to maximize the CIP. Iron loading was followed by measuring changes in NDRG1 mRNA levels subsequent to treatment with increasing concentrations of ⅐ NO (DETA/NO 250 -1,000 M) for 8 h (Fig. 5D). This figure clearly demonstrates that when the concentration of chelatable iron in a cell is artificially increased, greater amounts of ⅐ NO are necessary to obtain equivalent levels of NDRG1 mRNA up-regulation compared with basal iron levels.
HIF-1␣ Is Not Required for Up-regulation of NDRG1 by ⅐ NO-Previous reports have demonstrated both HIF-1␣-dependent and -independent mechanisms of NDRG1 up-regulation (6,7,46). Although HIF-1␣ is strongly up-regulated under conditions of hypoxia, there are also numerous reports documenting the rapid and robust accumulation of HIF-1␣ in response to ⅐ NO and iron chelators under normoxic conditions (47).
Although ours is the first report of NDRG1 being up-regulated by ⅐ NO, it is also known to be a downstream target of HIF-1␣ under certain conditions. Therefore, it was important to specifically evaluate the contribution of HIF-1␣ signaling to NDRG1 up-regulation by ⅐ NO. To this end we generated two stably transfected HCC 1806 cell lines. In one cell line we knockeddown NDRG1 mRNA (HCC 1806 N) and in the other we knocked-down HIF-1␣ mRNA (HCC 1806 H). Then we treated the cells with ⅐ NO and measured the magnitude of NDRG1 and HIF-1␣ up-regulation at the mRNA and protein levels (Fig. 6, A  and B). In the NDRG1 knockdown cells, the HIF-1␣ protein was still strongly up-regulated in response to ⅐ NO, whereas the NDRG1 protein, as expected, was undetectable (Fig. 6A). Conversely, in the HIF-1␣ knocked down cells, NDRG1 mRNA and protein were significantly up-regulated, therefore, still responsive to ⅐ NO. Under basal conditions HIF-1␣ is constitutively expressed and it is not transcriptionally regulated by ⅐ NO. Not surprisingly, we did not see any increases in HIF-1␣ mRNA in response to ⅐ NO exposure in either cell type (Fig. 6B). As predicted, however, NDRG1 mRNA was not increased in the HCC 1806 N cell line but was strongly increased in the HCC 1806 H cells. These data strongly suggest that NDRG1 gene expression as a result of ⅐ NO exposure is not HIF-1␣-mediated in this cell type.
NDRG1 Suppresses Tumor Cell Migration-In cancer biology NDRG1 is generally regarded as a metastasis suppressor protein. Because cell migration is a critical element in the metastatic process, we set out to determine whether changes in NDRG1 expression in response to ⅐ NO would modulate this process. To evaluate this we used the HCC 1806 cell line because it is highly migratory in response to a serum stimulus. This enabled us to contrast differences in the migratory potential of HCC 1806 cells with that of the NDRG1 and HIF-1␣ knockdown cell lines we generated (HCC 1806 N and H). In response to serum, the NDRG1 knockdown cells migrated almost twice as much as either the wild-type or HIF-1␣ knockdown cells, suggesting that basal NDRG1 levels are an important inhibitor of cell migration (Fig. 7A). The HIF-1␣ knockdown cells migrated slightly more than, but similar to, the wild-type cells. This is expected because basal HIF-1␣ protein Having determined the migratory profile of these cell lines, we set out to examine the influence of ⅐ NO on their behavior. Fig. 7B indicates that ⅐ NO had a strong inhibitory effect on HCC   DECEMBER 2, 2011 • VOLUME 286 • NUMBER 48 cells was also suppressed by ⅐ NO, but to a lesser extent (Fig. 7C). Interestingly, ⅐ NO had almost no suppressive effect on cell migration in the HIF-1␣ knockdown cells (Fig. 7D). Fig. 7E illustrates the inhibitory effect of ⅐ NO on each cell type when compared with their untreated control. Fig. 7F demonstrates the inhibitory effect of ⅐ NO on each cell type when compared with the wild-type (HCC 1806).

Nitric Oxide Suppresses Tumor Cell Migration
These data indicate that basal levels of NDRG1 alone are sufficient to restrain cell migration, and HIF-1␣ alone has a suppressive effect on migration. However, marked inhibition of migration is only seen when NDRG1 is strongly up-regulated in conjunction with HIF-1␣. Cumulatively then, ⅐ NO-mediated suppression of migrations requires HIF-1␣. When HIF-1␣ is present, the majority of these suppressive effects are due to NDRG1 and basal NDRG1 expression can suppress migration to an extent.
In addition to migration, invasion is a critical component of the metastatic process. In support of our hypothesis, therefore, we examined the ability of ⅐ NO to modulate cell invasion. Fig. 7G illustrates that ⅐ NO similarly suppresses tumor cell invasion. Cell viability was assessed for all conditions in Fig. 7 via standard growth plates (E plates). Knockdown cell lines and ⅐ NO treatments had no effect on viability to 80 h (supplemental Fig. S3).
Chelatable Iron Regulates ⅐ NO-mediated Migration-As we observed, increasing the chelatable iron pool dampened the effect of ⅐ NO on NDRG1 induction. Therefore, we asked if excess iron would similarly effect the responsiveness of migration to ⅐ NO. We treated HCC 1806 cells with iron (FAC) and then measured the suppressive effect of ⅐ NO on migration (Fig.  7H). As shown, iron supplementation did not increase the migration of cells in the absence of ⅐ NO. In support of our hypothesis, however, iron supplementation did reduce the overall suppressive effect of ⅐ NO on migration when compared with non-supplemented control cells. This effect could not be attributed to changes in cell viability, which remained at 100% for the duration of the experiment (Fig. 7I).
Nitric Oxide Regulates NDRG1 Expression in Other Cell Types-In addition to breast cancer, ⅐ NO is known to be upregulated in a variety of diverse tumor types. We looked at five other tumor cell lines to determine the universality of the nitric oxides ability to regulate NDRG1 expression. We treated breast (MDA-MB-231 and MCF-7), colon (HT-29), ovarian (SKOV-3), and brain (SH-SY5Y) cell lines with ⅐ NO and examined changes in NDRG1 mRNA and protein (Fig. 8, A and B). In all cell types except the SH-SY5Y cells, NDRG1 mRNA and protein are both significantly up-regulated. It is of interest that the N-Myc is well known to be up-regulated in neuroblastomas (48); therefore, it is likely that NDRG1 expression is suppressed to such a degree that ⅐ NO has no effect on its regulation.

The Role of MYC Genes in NDRG1 Regulation-It is known that NDRG1 is down-regulated by both N-Myc and c-Myc (3).
For this reason it was important to determine whether NDRG1 up-regulation by ⅐ NO was in response to changes in N-Myc or c-Myc expression. First, we analyzed changes in N-Myc and c-Myc protein and mRNA in response to ⅐ NO in HCC 1806 cells (Fig. 9, A and B). Although it is known that N-Myc is absent in HCC 1806 cells (49), we verified this by Western blot and qRT-PCR ( Fig. 9A and data not shown). We were, however, able to detect a concentration-dependent increase in c-Myc mRNA and protein levels upon ⅐ NO exposure. We also measured changes in c-myc mRNA in 5 additional cell lines (supplemental Fig. S4). Because NDRG1 is purported to be down-regulated by c-Myc, we suspect that NDRG1 up-regulation by ⅐ NO is via alternate, dominating mechanisms.
Excluding Other ⅐ NO-mediated Mechanisms for NDRG1 Up-regulation-The major biological mechanisms by which ⅐ NO signals is through activation of soluble guanylyl cyclase (sGC) to form cGMP. We therefore wanted to rule out sGCmediated signaling as a mechanism on ⅐ NO-induced NDRG1 up-regulation. We treated HCC 1806 cells with the cGC inhibitor ODQ (1H- [1,2,4]oxadiazolo[4,3-a]quinoxalin-1-one) and measured the ability of ⅐ NO to up-regulate NDRG1. We also treated these cells with YC-1 ((3-(5Ј-hydroxymethyl-2Ј-furyl)-1-benzylindazole), a ⅐ NO-independent activator of sGC. Inhibition of sGC did not diminish the magnitude of NDRG1 upregulation by ⅐ NO and activation of sGC in the absence of ⅐ NO did not up-regulate NDRG1 (supplemental Fig. S5).

DISCUSSION
Mechanism for NDRG1 Regulation via ⅐ NO Is Novel-Many studies of NDRG1 gene expression have focused on hypoxia (HIF-1␣), hypoxia mimetics (Co 2ϩ and Ni 2ϩ ), and metal chelators as mechanisms of regulation (2, 6, 7). The commonality between these positive regulators of NDRG1 suggests that a metal-coordinated or O 2 -binding protein may be involved. Chemical metal chelators primarily target the chelatable iron pool within a cell and inhibit the catalytic function of nonheme, non-iron-sulfur iron-requiring proteins. Divalent metals such as Co 2ϩ and Ni 2ϩ are known to substitute for iron in mononuclear non-heme, iron oxygenases and will similarly inhibit enzyme function. Nitric oxide has many biological targets, one of which is the chelatable iron pool (28,34,35). The reaction of ⅐ NO with the CIP results in iron sequestration in the form of DNIC. A logical result of DNIC formation, therefore, would be the observation of distinct phenotypic outcomes similar to treatment with metal chelators that are separate from other well defined ⅐ NO signaling pathways. For these reasons, we tested the ability of ⅐ NO to up-regulate NDRG1 and suppress metastasis in a CIP-dependant manner.
It is well known that ⅐ NO signaling is disseminated through several dominant pathways. These include ⅐ NO-heme interac-

Nitric Oxide Suppresses Tumor Cell Migration
DECEMBER 2, 2011 • VOLUME 286 • NUMBER 48 tions (50), binding and destruction of iron-sulfur proteins (51), and S-nitrosothiol formation (52). The activation of heme proteins such as guanylyl cyclase by ⅐ NO and the subsequent biochemical effects of these interactions have been studied at length (28). Iron-sulfur proteins are well known targets of ⅐ NO mainly under stress conditions at high ⅐ NO concentrations. These interactions can lead to disassembly of the cluster and possibly release of iron. The formation of S-nitrosothiols on key protein residues can have a direct or allosteric effect on protein function. We have recently demonstrated that under specific conditions DNIC can be quantitatively the largest ⅐ NO-derived cellular adduct (35). It is reasonable to conclude that perturbations in iron bioavailability secondary to DNIC formation will have phenotypic consequences. Although DNIC may not directly regulate NDRG1, their formation results in chelatable iron sequestration by ⅐ NO, and this sequestration correlates strongly with NDRG1 up-regulation. This suggests that an upstream requirement for iron may exist for regulators whose target genes are suppressors of NDRG1 itself or genes that encode regulators for NDRG1. We cannot completely exclude other ⅐ NO-mediated pathways as contributors to NDRG1 regulation. It is clear from our data, however, that lack of iron availability subsequent to ⅐ NO exposure is a major upstream driving force.
Diffusion of ⅐ NO and its reaction with chelatable iron is fast. One consequence of this is the rapid accumulation of the constitutively expressed HIF-1␣ protein. Similarly, we have shown that up-regulation of NDRG1 by ⅐ NO requires the interaction of ⅐ NO with chelatable iron. Unlike HIF-1␣, however, there is a 4 -8-h delay before the NDRG1 gene products accumulate. These kinetics are consistent with previous studies measuring NDRG1 up-regulation by iron chelators (2). Our findings suggest that the iron-dependent target for ⅐ NO resulting in NDRG1 gene expression is well upstream of NDRG1 itself.
When HCC cells are treated with ⅐ NO, DNIC form at concentrations that are equivalent to the concentration of the CIP (Fig. 4B). This is in agreement with work done by other groups, which demonstrated that the iron necessary for DNIC assembly is derived from the CIP (34). Under ⅐ NO treatment conditions that resulted in DNIC formation, we saw NDRG1 strongly upregulated at the mRNA and protein levels to the same extent as achieved with chemical iron chelators. We were able to make incremental increases in the CIP by incubating cells with FAC. The magnitude of the increase in the CIP was proportional to the length of time the cells were incubated with iron. We noted that as the CIP was increased ⅐ NO had a comparatively lesser effect on NDRG1 mRNA and protein induction and at the same time the ability of DNIC to reach equivalent concentrations as the CIP was diminished. Therefore, if the CIP concentration is high, a greater proportion of iron remains in the CIP as opposed to forming DNIC in the presence of ⅐ NO. In effect, this maintains the bioavailability of "free iron" leaving it accessible to support NDRG1 suppressive pathways.
In a similar set of experiments, we compared the amount of ⅐ NO necessary to achieve maximal NDRG1 mRNA expression in cells with a high versus normal CIP (Fig. 5D). We noted that a greater amount of ⅐ NO was required to achieve the same magnitude of NDRG1 up-regulation when the CIP was elevated compared with cells with basal iron levels. Maximal NDRG1 up-regulation by ⅐ NO was not observed in iron-supplemented cells until the DNIC concentration increased to approximate the concentration of the CIP. These results further emphasize that the magnitude of NDRG1 expression in response to ⅐ NO is not simply a function of the ⅐ NO concentration but of the DNIC to CIP ratio.
Finally, we ruled out many known NDRG1 regulatory pathways. Although N-Myc is absent in HCC 1806 cells, c-Myc is up-regulated by ⅐ NO. Previous reports have demonstrated, however, that c-Myc suppresses NDRG1 (3), and is thus not likely to be the means through which our observed changes are occurring. We also were able to rule out activation of sGC as being involved in NDRG1 gene expression, which is perhaps the best studied mechanism for ⅐ NO signaling. RNA knockdown of HIF-1␣ revealed that in this cell type HIF-1␣ is not required for NDRG1 expression. This is a crucial observation because under normoxic conditions (21% O 2 ) ⅐ NO is known to strongly induce HIF-1␣ accumulation at concentrations consistent with NDRG1 up-regulation.
There are some reports of p53 being an upstream regulator of NDRG1 (7,18) and it is also well known that p53 is post-translationally regulated by ⅐ NO (47). This pathway could be excluded in our model because HCC 1806 are p53 null (53).
The mode of NDRG1 regulation that we propose depends on iron availability, and thus may have occurred through mechanisms involving the iron-regulatory proteins. Iron-regulatory proteins (IRP1 and Ϫ2) are important regulators of intracellular iron. In response to changes in cellular iron status, these proteins bind to conserved iron-responsive elements (IREs) in the 3Ј-and 5Ј-untranslated regions (UTRs) of specific mRNAs (54). These mRNAs generally code for proteins involved in iron homeostasis such as transferrin, ferritin, ferroportin, and DMT1 (55,56). Under conditions of low cellular iron, IRE-IRP interactions act as both translational repressors and mRNA stabilizers resulting in increased transferrin (TfR1) and decreased ferritin translation. It is known that ⅐ NO favors the binding of IRPs to IREs by directly targeting the [4Fe-4S] cluster of IRP1 resulting in its disassembly (57). An attractive explanation, therefore, for NDRG1 up-regulation by NO would be through IRE-IRP interactions. This was ruled out, however, because the stem-loop structure of the 3Ј-UTR of NDRG1 has been previously examined and determined not to show structural similarities to any other known highly conserved IRE structures (58). It was therefore concluded that it would not be capable of binding IRPs (58,59).
Although additional experiments will be needed to elucidate specific iron-dependent pathways of NDRG1 regulation, these data represent the first demonstration of a phenotypic singling effect associated with DNIC formation. This may represent a new category of ⅐ NO signaling independent of cGMP, S-nitrosothiol, or heme interaction mechanisms.
NO, NDRG1, and Cancer-Triple negative breast cancers, along with many other aggressive cancers, are highly resistant to conventional treatments such as cyclophosphamide, methotrexate, and 5-fluorouracil (CMF) (60). Thus there is a pressing need to understand the underpinnings of the metastatic progression in these cancers to develop new therapies. Although this reality is certainly appreciated, obtaining molecular indices of metastatic progression will never be as simple as finding a single oncogene and will always remain highly tissue specific. These studies examine the regulation of a metastasis repressor gene by ⅐ NO. This is of considerable interest because both ⅐ NO and NDRG1 have both been shown to be positively and negatively correlated with metastatic progression (29,44). Examined together, however, potential explanations for their seemingly contradictory behavior emerge.
Although NDRG1 expression has been found to be increased in certain cancers, it appears to function as originally defined in breast cancers, as a metastasis suppressor. In this setting it seems likely that loss of NDRG1 expression is a cause not a result of metastasis. Thus finding novel means to up-regulate its expression in these tumors is of considerable interest. When the highly metastatic breast cancer cell line HCC 1806 was exposed to ⅐ NO, there was marked suppression in cell migration. Conversely, in the absence of ⅐ NO, knocking down NDRG1 mRNA resulted in a doubling in the rate of cell migration. This indicates that basal levels of NDRG1 limit the migratory potential of cells and up-regulation of this protein by ⅐ NO is suppressive. The rate of cell migration in NDRG1 knockdown cells exposed to ⅐ NO was equivalent to the rate of wild-type cell migration in the absence of ⅐ NO. This indicates an NDRG1independent means of migration suppression by ⅐ NO. Interestingly, the migration of HIF-1␣ knockdown cells was not suppressed by ⅐ NO even when NDRG1 was up-regulated. Taken together, this implies that there are HIF-1␣-mediated, NDRG1independent, mechanisms of migration suppression. Although HIF-1␣ accumulation is not directly involved in NDRG1 protein expression in these experiments, we observed an enhanced suppressive effect on migration when both proteins were present.
Under conditions of iron supplementation, where DNIC formation was less than the CIP, the suppressive effect of ⅐ NO on migration was completely abolished. This is consistent with our hypothesis that the chelation of iron by ⅐ NO leads to the up-regulation of NDRG1, and implies that the multitude of other ironindependent, ⅐ NO-induced pathways are not involved in suppression of migration. Furthermore, we were able to rule out cytotoxic effects of ⅐ NO as an explanation for its ability to diminish cell migration indicating that this is a true metastatic suppressive effect. Our findings are consistent with NDRG1 as a metastasis suppressor protein, but also hint at conditions where it might not be effective. When the HIF-1␣ knockdown cells were treated with ⅐ NO, there was a clear increase in NDRG1 mRNA and protein. Despite this increase, ⅐ NO had no effect on cell migration. It may be that in cases where NDRG1 expression is lost, metastasis becomes highly likely, but its expression might not be capable of suppressing migration in certain genetic backgrounds. Thus its high expression is more likely to be a failed compensatory mechanism in cases where it is associated with metastasis.
The redox status and microenvironmental conditions of any ⅐ NO-producing tissue can greatly impact both the concentration of ⅐ NO and the duration of its production. Our results demonstrate that Ն4 h of ⅐ NO exposure is required for NDRG1 up-regulation, suggesting that persistent rather than transient iron sequestration is essential. Although NDRG1 mRNA levels degrade rapidly after ⅐ NO removal, the amount of protein remains elevated for Ն48 h. Together, these results indicate that short exposure to ⅐ NO can have long lasting phenotypic effects. This may have important implications toward establishing ⅐ NO-attributable effects on tumor behavior even under conditions where there is discontinuous ⅐ NO production.
Conclusions-These results demonstrate that ⅐ NO, via its interaction and sequestration of chelatable iron, results in the up-regulation of NDRG1 in HCC 1806 (and other) cells. This represents a novel, previously unknown, mechanism of NDRG1 up-regulation and the first demonstration for a functional role of DNIC formation. Nitric oxide is up-regulated in a variety of tumor types and it is also generated by several classes of experimental chemotherapeutics. NDRG1 will likely be an effective therapeutic target in certain cancers. In fact, a recent study found that up-regulation of NDRG1 by thiosemicarbazones completely inhibited pancreatic tumor xenograft growth (46). In addition to the relevance of this data to cancer progression, these results may be extended to explain normal physiologic conditions where ⅐ NO is synthesized and NDRG1 is expressed. Continued research will be needed to confirm the existence of the actual chelatable iron-dependent target and to better understand the relevance of ⅐ NO signaling to mammalian pathophysiological and physiological states.