IK channel activation increases tumor growth and induces differential behavioral responses in two breast epithelial cell lines

Many potassium channel families are over-expressed in cancer, but their mechanistic role in disease progression is poorly understood. Potassium channels modulate membrane potential (Vmem) and thereby influence calcium ion dynamics and other voltage-sensitive signaling mechanisms, potentially acting as transcriptional regulators. This study investigated the differential response to over-expression and activation of a cancer-associated potassium channel, the intermediate conductance calcium-activated potassium channel (IK), on aggressive behaviors in mammary epithelial and breast cancer cell lines. IK was over-expressed in the highly metastatic breast cancer cell line MDA-MB-231 and the spontaneously immortalized breast epithelial cell line MCF-10A, and the effect on cancer-associated behaviors was assessed. IK over-expression increased primary tumor growth and metastasis of MDA-MB-231 in orthotopic xenografts, demonstrating for the first time in any cancer type that increased IK is sufficient to promote cancer aggression. The primary tumors had similar vascularization as determined by CD31 staining and similar histological characteristics. Interestingly, despite the increased in vivo growth and metastasis, neither IK over-expression nor activation with agonist had a significant effect on MDA-MB-231 proliferation, invasion, or migration in vitro. In contrast, IK decreased MCF-10A proliferation and invasion through Matrigel but had no effect on migration in a scratch-wound assay. We conclude that IK activity is sufficient to promote cell aggression in vivo. Our data provide novel evidence supporting IK and downstream signaling networks as potential targets for cancer therapies.

Breast cancer is one of the most common forms of cancer with roughly 230,000 new cases per year and is responsible for 40,000 deaths in the United States alone. About 1 in every 8 women will be diagnosed with breast cancer at some point in her lifetime. Women diagnosed with localized breast cancer have an excellent prognosis with over 98% 5 year survival. Unfortunately, patients with metastases have only a 25.9% 5 year survival highlighting the need for treatments aimed at metastasis [1]. A major hindrance to the development of better breast cancer treatments is a lack of understanding of what drives disease progression to metastasis. For decades it has been known that some cancers remain in a benign state for long time periods while others quickly progress to malignancy, and that generally only a subset of tumor cells are capable of metastasis. Although many metastasis-specific pathways have been described, our understanding of the upstream signals that activate these pathways, and the factors that cause certain cells to metastasize while nearby syngeneic cells do not, remain largely unknown [2].

Over the past decade ion channels have been receiving increased attention for their roles in cancer progression. Numerous cancer types have altered ion channel expression, with expression of certain channels correlating with cancer stage [3]. In various cases, inhibiting these channels leads to a decrease in cancer-associated behaviors including primary tumor growth and metastasis in vivo [4–6]. Most studies have focused on ion channels as downstream targets of signaling pathways that execute critical mechanical functions required for aggressive behaviors. For instance, inhibiting certain chloride and potassium channels responsible for generating changes in cell volume decreases cell migration and proliferation [7]. However, evidence suggests ion channels may have upstream regulatory roles as well, and little is known about the ability of ion channel activity to initiate signaling cascades to promote aggressive cancer behaviors [8, 9].

The intermediate conductance calcium-activated potassium channel (IK) is over-expressed in numerous cancer types including breast, prostate, uterus, stomach, colorectal, pancreas, pituitary gland, and brain cancers [10] and inhibiting IK decreases cancer cell proliferation, migration, and in vivo tumor growth and metastasis [11–16]. Based on these results, the widely held theory in the field is that IK is a downstream effector of signaling pathways and is required in the late steps of enacting aggressive cancer behaviors. However, IK may have additional upstream instructive roles and its activity may be sufficient to initiate aggressive behaviors through its effect on calcium dynamics. In prostate cancer cells, activation of IK with its agonist was sufficient to significantly increase intracellular calcium concentrations suggesting IK could regulate downstream calcium-dependent signaling pathways [17]. Furthermore, IK activation was sufficient to increase prostate cancer proliferation, providing additional evidence of the ability of IK to activate signaling pathways [12]. However, the possible sufficiency of IK to promote aggressiveness has not been previously studied in breast cancer cells.

In the present study, our aims were (1) to investigate whether increased IK activity was sufficient to promote proliferation in breast epithelial cells and cancer cells and (2) to investigate whether an increase in IK was also sufficient to increase other aggressive cancer behaviors, including tumor growth and metastasis in vivo. We chose to use the metastatic breast cancer cell line MDA-MB-231 because IK inhibition studies demonstrated decreased proliferation, migration, and colony formation indicating IK has an important physiological role in aggressive behaviors in this cell line [18]. Surprisingly, we found that MDA-MB-231 in vitro proliferation, invasion, and migration were not affected by IK over-expression or activation. Interestingly, however, increased IK decreased proliferation and invasion of the spontaneously immortalized breast epithelial non-tumorigenic MCF-10A cell line but had no effect on migration. In contrast to the in vitro results, we found that over-expressing IK in MDA-MB-231 was sufficient to increase primary tumor growth and metastasis in mice. This study is the first to demonstrate the sufficiency of IK to increase cancer aggression in vivo and suggests the possibility of key differences in behavioral response to IK activation between tumorigenic and non-tumorigenic cells, although more cell lines must be tested to determine a potential trend. Our results indicate that IK plays an important instructive role in cancer progression and suggest the possibility of unique signaling mechanisms that could be used as specific targets.

In order to test the sufficiency of increased IK to induce increased aggression in the breast cancer cell line MDA-MB-231, we first generated cells with increased IK expression. Cells were infected by a retrovirus encoding either IK and red fluorescent protein (RFP) or RFP alone as vector control and selected for RFP using fluorescence activated cell sorting (FACS) (Supplementary Figure 1). MDA MB 231 have previously been reported to endogenously express IK [19](data accessible at NCBI Geo database GSE41678). We confirmed that IK was expressed in control cells (MDA-MB-231-RFP) by RT-PCR and that cells infected with IK virus (MDA-MB-231-IK) had significantly increased IK expression (p = 0.0027, 2-sample t-test, Figure 1A). Overexpression was further confirmed at the protein level by immunofluorescence (Figure 1B).

To validate functional activity of the over-expressed IK channel, the electric current density and Vmem of individual cells were measured in the presence or absence of 200 μM of the IK agonist 1- Ethylbenzimidazolinone (1-EBIO). It should be noted that 1-EBIO is also an agonist of the closely related small conductance calcium activated potassium channels (SK) [20]. However, SK transcripts were below the limit of detection using RT-PCR suggesting that IK is more prevalent. Furthermore, treatment of MDA-MB-231 with the IK specific antagonist TRAM-34 inhibits most calcium-sensitive potassium current indicating that SK has lower endogenous activity [18]. Electrophysiological recordings were acquired in the cell attached configuration using a perforated patch in order to best maintain the endogenous cytosolic contents and preserve Vmem. 1-EBIO treatment had no effect on current density of control MDA-MB-231-RFP cells (q = 0.15; 1-way ANOVA of 0 mV current density). Similarly, IK over-expression had no effect on current density in the absence of 1-EBIO (q = .31; 1-way ANOVA of 0 mV current density). However, in MDA-MB-231-IK cells, 1-EBIO treatment significantly increased the current density from 0.98 ± 2.32 pA/pF to 32.33 ± 13.60 pA/pF (p < .001, q = 12.30 1-way ANOVA of 0 mV current density; Figure 1C). Furthermore, the current density of MDA-MB-231-IK treated with 1-EBIO was significantly larger than either untreated or 1-EBIO treated control cells (q = 11.31 and q = 11.14 respectively, 1-way ANOVA of 0 mV current density). The reversal potential for MDA-MB-231-IK EBIO treated cells was -76.21 ± 2.58 mV, relatively near the potassium reversal potential predicted by the Nernst equation (−89.9 mV), indicating the current was comprised primarily of potassium flux.

These data suggest that MDA-MB-231-IK cells express significantly more functional IK channels than control MDA-MB-231-RFP cells.

IK over-expression without agonist treatment did not significantly alter the Vmem of MDA-MB-231 cells (control -26.3 ± 8.2 mV, MDA-MB-231-IK -33.5 ± 16.5 mV, q = 1.34, 1-way ANOVA). However, there was a greater range of Vmem values in the MDA-MB-231-IK population (range of -6.4 to -67.6 mV) compared to control cells (range of -13.9 to -40.6 mV; p = 0.045; F test to compare variance) with the IK population being skewed towards hyperpolarized (control -0.26, MDA-MB-231-IK -0.62, adjusted Fisher-Pearson standardized moment coefficient), suggesting IK over-expression leads to increased basal levels of conductance in some cells or is transiently active (Figure 1D). Application of 1-EBIO significantly hyperpolarized the Vmem of MDA-MB-231-IK cells to -73.9 ± -2.8 mV (p < .001, paired 2-sample t-test). 1-EBIO also hyperpolarized the Vmem in control cells, although this was not statistically significant (−40.3 ± 17.7 mV; p = 0.06 paired 2-sample t-test). Additionally, control cells exhibited a fluctuation in Vmem at the time of 1-EBIO application, suggesting the presence of active IK or SK channels (Figure 1E).

We next investigated the effect of IK expression and activity on in vitro behaviors of MDA-MB-231 cells. IK is endogenously activated by calcium signaling and its activation is highly temporally and spatially regulated which may be important to its downstream function [13]. We wanted to compare the effect of increased IK current responding to endogenous temporal and spatial activation as well as to low and high levels of continuous pan-IK activation. Cells over-expressing IK were expected to maintain endogenous activation patterning but with larger currents. We therefore compared the behavioral response of control and IK over-expressing cells treated or untreated with agonist.

It has been previously reported that IK over-activation by agonist treatment increases proliferation in prostate cancer cells [12]. Here, we monitored the proliferation of both control and IK over-expressing MDA-MB-231 cells over 4 days in the presence or absence of 1-EBIO. We found that there was no significant difference in proliferation across any of the conditions (Day 4: p = 0.33 1-way ANOVA; Figure 2A). We next evaluated in vitro behaviors associated with metastasis. The ability of cells to invade was assayed in vitro using a transwell assay in which the number of cells invading across a Matrigel-coated membrane in response to a serum gradient was quantified. There was no significant difference in the number of invading cells across any of the conditions (p = 0.32 1-way ANOVA; Figure 2B). Similarly, migration in a scratch wound assay was not altered, with no significant difference in the surface area of wound recovery after 10 h (p = 0.41 1-way ANOVA; Figure 2C). These results suggest that increased IK activity is not sufficient to increase in vitro measures of MDA-MB-231 cell aggression.

The ability to form colonies in soft agar is used as an in vitro measure of transformation and tumor forming ability. Inducing expression of oncogenes such as c-MYC or RAS in non-tumorigenic cell lines can confer colony forming ability which correlates with tumor formation in mouse xenografts demonstrating the capacity of this assay to assess conversion to a malignant phenotype [21]. MDA-MB-231 cells are highly malignant and therefore readily form colonies. Given that IK is over-expressed in many metastatic cancers, we anticipated increased IK activity would increase colony formation. Surprisingly, colony formation in IK-expressing cells was decreased to 64% as compared to untreated control and was further decreased to 30% by 1-EBIO treatment (Figure 3A–3B; p < 0.001 1-way ANOVA). 1-EBIO similarly reduced the colony formation of control MDA-MB-231 cells to 32% of untreated cells. These data suggest that treatment with 1-EBIO, which induces constant IK activation, more severely inhibits colony formation than elevated transiently activated IK induced by increased IK expression.

MDA-MB-231 cells are highly aggressive and so the lack of effect of IK expression/activity on cell proliferation, migration, and invasion may be due to an inability to further augment these already aggressive behaviors. We therefore tested the effect of increased IK expression and activation in the non-tumorigenic breast epithelial MCF-10A cell line. We created an IK over-expressing population of MCF-10A cells. As with MDA-MB-231, MCF-10A-IK cells had dramatically increased IK expression (p < 0.001, 2-sample t-test; Supplementary Figure 2A) and increased current density at 0 mV with 1-EBIO treatment from 0.483 +/− 0.344 pA/pF in control cells to 22.90 +/− 18.46 pA/pF in MCF-10A-IK (p < .001, q = 7.30 1-way ANOVA, Supplementary Figure 2B). IK over-expression without agonist treatment hyperpolarized the Vmem of MCF-10A cells from -21 ± 6.3 mV in control cells to -47 ± 14.7 mV in MCF-10A-IK cells (p < 0.001, 1-way ANOVA; Supplementary Figure 2C) indicating the overexpressed channels were activated by endogenous signaling. Application of 1-EBIO hyperpolarized MCF-10A control cell Vmem to -31 ± 7.4 mV (p < 0.01, paired t test) indicating that the cells have endogenous IK or SK channels. 1-EBIO treatment induced a larger effect on MCF-10A-IK cells, hyperpolarizing the Vmem to -69 ± 6.9 mV (p < 0.01, paired t-test). Although the effect of 1-EBIO treatment may have been partially mediated by SK channel activation, we were not able to detect SK transcripts by RT-PCR suggesting SK has either very low or no expression in MCF-10A cells. In summary, these data demonstrate that IK overexpression in MCF-10A cells induces a similar change in electrical properties to MDA-MB-231 cells.

We next studied the effect of IK expression and activity on in vitro measures of aggression in MCF-10A cells. IK over-expression alone had no effect on MCF-10A cell proliferation, but activation of IK by 1-EBIO treatment significantly decreased proliferation of control MCF-10A cells by 33 ± 4 % and IK expressing cells by 43 ± 16 % (p < 0.01, 1-way ANOVA; Figure 4A). MCF-10A invasion was significantly decreased by both IK over-expression and activation (p < 0.001, 1-way ANOVA; Figure 4B). It should be noted that in the control condition, less than 1% of the cells loaded in the upper chamber invaded to the lower chamber reflecting similar basal invasion levels as has been reported elsewhere and significantly reduced as compared to MDA-MB-231 [22–24]. However, no significant change in cell migration using a wound scratch assay was observed across any condition (p = 0.28, 1-way ANOVA; Figure 4C). Finally, we tested the effect of IK expression and activity on colony-forming ability. MCF-10A is an immortal but not transformed cell line and is unable to form colonies in soft agarose but can be induced to form colonies by expression of strong oncogenes [25]. Neither MCF-10A control nor MCF-10A-IK cells were able to form colonies with or without 1-EBIO treatment (Supplementary Figure 3).

We initially hypothesized that the differential proliferative response between MCF-10A and MDA-MB-231 cells was due to a difference in apoptosis sensitivity. High potassium currents have been reported to induce apoptosis and one possibility was that MCF-10A cells are sensitive to large potassium currents while MDA-MB-231 cells could be protected by anti-apoptotic signaling that is often active in highly metastatic cells [26]. To test this, we measured the percentage of apoptotic and dead cells to determine if the decrease in MCF-10A cell number following IK activation was due to an increase in cell death. Control and IK over-expressing cells were cultured for 24 hours in 1-EBIO or vehicle control and stained with Annexin-V-FITC, which binds external phosphatidylserine an early marker of apoptosis, and violet dead cell stain. FACS was used to quantify the percentage of labeled cells. There was no significant difference in the percentage of apoptotic or dead cells across all cell conditions for both MCF-10A and MDA-MB-231 cells (Figure 5A, 5C). We next investigated changes in the cell cycle distribution in the same samples using propidium iodide staining. 1-EBIO treatment increased the percentage of control and IK-overexpressing MCF-10A cells in the G2 phase (from 14.80 ± 3.71 % to 18.48 ± 3.27 % in control MCF-10A cells and from 16.08 ± 2.18 % to 21.4 ± 5.1 % in MCF-10A-IK cells; p < 0.01, 1-way ANOVA; Figure 5C). However, no significant differences were found in any of the conditions for MDA-MB-231 cells (p = 0.29, 1-way ANOVA; Figure 5D). The increased G2 accumulation was therefore observed in the same populations that had decreased proliferation.

These data suggest that in MCF-10A cells the G2/M checkpoint includes a requirement for decreased IK activity and/or depolarization and that MDA-MB-231 cells are able to evade this requirement, possibly due to strong pro-growth signaling active in this cell line.

Given the various effects of IK expression and activity on cellular aggressiveness in vitro, we next assessed the effect of IK over-expression on in vivo tumor growth and metastasis. MDA-MB-231 control and MDA-MB-231-IK cells were injected into the mammary fat pad of 6 wk old female Rag2−/− Il2rg−/− mice. Caliper measurements were used to estimate the tumor volume over 4 wks of tumor growth. Primary tumor growth was significantly increased in MDA-MB-231-IK tumors beginning 17 days post cell injection (p < 0.05, 2-sample t test; Figure 6A). To demonstrate that the tumors were derived from the injected cells, tumor sections were stained for human nuclear antigen (HNA) and cells stained positively throughout the bulk of the tumor (Figure 6B). MDA-MB-231-IK tumors showed higher IK immunoreactivity compared to control tumors, suggesting that IK over-expression was retained in vivo (Figure 6B). Despite the increase in tumor size, there was no significant difference in the number of Ki67 positive proliferating cells in 28 day tumor sections (p = 0.27, 2-sample t-test; Figure 6C). There was also no significant difference in the number of cleaved caspase-3 positive apoptotic cells in tumor sections at day 28 (p = 0.15, 2-sample t-test; Figure 6E). However, there was a trend towards increased Ki67+ cells and reduced caspase-3+ cells in the IK-overexpressing tumors, suggesting that increased proliferation and/or reduced apoptosis may contribute to the increased size of these tumors. In addition, as these measurements were taken at the end of the experiment, they may underestimate a larger difference in proliferation or apoptosis that occurred earlier on in the growth of the tumor.

We next investigated whether the primary tumors also displayed differences in aggressive phenotypes. Staining of primary tumor sections with H&E revealed similar local invasion into surrounding muscle and fibroadipose tissue from both MDA-MB-231 control and MDA-MB-231-IK primary tumors (Figure 7A). To assess vascular density, primary tumor sections were stained with the endothelial marker CD-31 and the number of vessels per square millimeter was quantified. There was no significant difference in the vascular density between control and MDA-MB-231-IK tumors and surprisingly if anything there was a trend towards denser vascularization in control cells (p = 0.14, 2-sample t-test, Figure 7B).

We also assessed metastatic dissemination by comparing the number of cells metastasizing to the lung. Lungs were sectioned and stained with anti-HNA antibody and the number of metastasizing cells was quantified. There was a significant increase in the number of metastasizing cells/mm2 in MDA-MB-231-IK injected mice as compared to control (p < 0.01, 2-sample t-test; Figure 8A–8B). In summary, IK over-expression significantly increased primary tumor growth as well as dissemination of cells to lung tissue.

Despite decades of research, metastatic cancer continues to have high rates of recurrence and patient mortality. A better understanding of the molecular pathways that promote aggressive behaviors is needed in order to develop treatments aimed at preventing metastasis. IK channels play important functions in proliferation, migration, and invasion as shown by a decrease in these behaviors in vitro and in vivo with IK inhibition or knock-down [11, 14, 27, 28]. In this study we investigated whether IK was also sufficient to promote these behaviors, as sufficiency would suggest IK plays important roles in upstream signaling pathways driving disease progression. We found that over-expression of IK was sufficient to increase primary tumor growth and metastasis but that it had cell line dependent differential effects on aggressive behaviors in vitro. This study adds to data from inhibition studies suggesting differences in behavioral response to potassium channel activation between cell lines of different tumor aggressiveness [15, 29]. Furthermore, there was no significant change in MDA-MB-231 proliferation in response to IK activation, which is in contrast to the previously reported increased proliferation of prostate cancer cell lines [12, 17]. These results suggest important differences exist in the signaling networks downstream of IK activation between different cell lines and support the investigation of primary tissue and tumors to determine if these differences correlate with disease state.

In support of the initial hypothesis that increased IK activity promotes cancer progression, in vivo primary tumor growth and metastasis of MDA-MB-231 to the lung was increased by IK over-expression. MDA-MB-231 control and MDA-MB-231-IK cells developed similarly aggressive primary tumors with no observable difference in local tissue invasion or vascularity. This may be because MDA-MB-231 cells are endogenously highly invasive and it may be difficult to observe potentiation in primary tumors. The in vivo data are in contrast to the effect of IK over-expression on in vitro proliferation and aggressive behaviors of MDA-MB-231 cells. While there was no significant effect of IK on proliferation, migration, or invasion, it is interesting to note that there was a trend of MDA-MB-231-IK having the highest value in each assay. It is possible that IK over-expression results in a small increase of each behavior that is challenging to detect within the inherent variability of the assays. But in vivo where the effect of all of these behaviors is combined during tumor growth, the overall effect is amplified and becomes statistically significant. Alternatively, we speculate the opposing in vitro and in vivo results could reflect a requirement for signaling from the surrounding tumor microenvironment. It is well known that many in vivo factors with important effects on cancer cell behavior are not recapitulated in vitro and often treatments that reduce aggressive in vitro behaviors have no effect in vivo [30–32]. Microenvironmental factors that influence IK activation and signaling have not been studied extensively and may have important roles. For example, growth-promoting effects of IK signaling may require highly specialized spatio-temporal activation that is not recapitulated in vitro.

IK signaling may be initiated by interactions with stromal cell types, activation of specific adhesion molecules, or microenvironmental factors such as hypoxia or inflammatory cytokines, as potassium conductance has been shown to be sensitive to all of these factors [33, 34]. As a specific example, signaling from the human eag-related gene 1 potassium channel (hERG1) is dependent on interactions with integrin receptors, demonstrating one mechanism by which potassium channel activity is linked to a feature of the surrounding microenvironment [35, 36]. Further experiments are needed to determine how IK activity is responsive to microenvironmental factors.

Migration of both MCF-10A and MDA-MB-231 cells was unaffected by IK over-expression and activation suggesting IK activity is insufficient to drive migration. Prior studies have reported that IK inhibition decreases migration in MDA-MB-231 and have supported a role for activation of IK specifically on the lagging edge [13, 18]. Activation lacking spatial regulation, such as occurs with addition of agonist, may have counteracting effects. The effect of increased IK expression on spatially restricted activation is not known. Activation of IK has been reported to increase intracellular calcium [17], allowing for the possibility of a positive feed-back loop whereby activation of some channels increases intracellular calcium which then activates additional channels. High expression of IK could increase positive feedback mechanisms and cause spatially restricted signals to be propagated to larger portions or the entire cell.

IK expression and activation decreased in vitro invasion of MCF-10A but not MDA-MB-231 cells. This effect was most pronounced in IK over-expressing cells treated with agonist and thus the degree to which MCF-10A invasion was inhibited correlated with the level of IK activity. Given that there was no change in MCF-10A migration, the decreased invasion is likely related to either an inability to degrade the ECM or dysregulation of volume control preventing volume reduction required to pass through the pores in the membrane. Further studies are needed to distinguish these two possibilities as well as to understand why MCF-10A but not MDA-MB-231 invasion is sensitive to increased IK currents. Surprisingly, IK expression and activation decreased MDA-MB-231 colony formation. Further studies are needed to delineate a mechanism for this result.

In contrast to increased proliferation seen in prostate cancer cells, neither IK over-expression nor activation with 1-EBIO caused a change in MDA-MB-231 proliferation, suggesting the mechanism downstream of IK activation may not be conserved across cancer types [12]. Additionally, proliferation of the non-tumorigenic cell line MCF-10A was decreased by IK activation in both control and IK over-expressing cells but not by IK over-expression alone. There was no change in apoptosis or cell death suggesting the diminished cell number was due to a decrease in the rate of cell division. MCF-10A control and IK over-expressing cells treated with 1-EBIO had a significant increase in the percentage of cells in G2 while IK over-expression alone had no effect on cell cycle distribution. MDA-MB-231, which had no change in proliferation across any condition, also had no significant difference in the percentage of cells in G2. Thus, the accumulation of MCF-10A cells in G2 mirrored the decrease in proliferation, suggesting constant IK activation decreases proliferation of MCF-10A by inhibiting progression from G2 to mitosis. Changes in potassium channel activity and Vmem are associated with progression through check points of the cell cycle with depolarization and a decrease in potassium channel activity typically occuring during G2 [5]. Activation of IK may prevent depolarization during late G2 and progression to mitosis. Further investigation is needed to determine if G2 accumulation is due to hyperpolarization specifically, as this would for the first time indicate a requirement for depolarization as part of the G2 checkpoint. The different response seen between MDA-MB-231 and MCF-10A cells suggests the cell lines may have distinct bioelectric profiles with altered sensitivity to bioelectric events.

This may be a result of differences in expression of bioelectric sensors including proteins related to calcium signaling, or a difference in the ability to compensate for the increased potassium current. Understanding the bioelectric requirements of the G2 checkpoint may help elucidate novel therapies designed to utilize differences in sensitivity to target specific cell populations.

One possible mechanism to explain the contrasting behavioral response between MCF-10A and MDA-MB-231 is a difference in expression of IK binding partners and other closely associated proteins that differ between the two cell types. IK signaling is dependent on interactions with beta-1-integrin and TRP member calcium channels in other cell types [17, 37]. Little is known about variability of associated proteins and how it affects behavioral response. There are many members within the TRP calcium channel family each with unique activation properties and thus IK associated with different members may have a different effect on calcium dynamics and downstream signaling. While some studies have investigated the requirement of associated proteins to IK signaling, none have investigated the effect of substituting related family members to gauge the response to a more subtle perturbation of the system. Additionally, a recent study found anomalous downstream responses to different IK inhibitors likely caused by maintenance of high intracellular calcium in response to one inhibitor but not the other [38]. Together with our work, these findings suggest the downstream response to IK modulation is complex and point to mechanisms that may make the response cell-type specific. Analysis of additional cell lines and primary cells is needed to determine if the divergent behavioral responses to IK stimulation are characteristic of specific cell types or disease states and to elucidate the mechanism driving the different behaviors.

Prior studies have placed the cancer-associated ion channel IK within a growing number of channels that are required to support aggressive tumor behaviors. This study is the first to demonstrate that IK is also sufficient to promote in vivo cancer growth and metastasis. Our findings suggest that IK has a greater impact on signaling pathways driving cancer progression, likely calcium-sensitive pathways, than previously thought. Ion channels are an intriguing potential mechanism of metastasis initiation as they are highly sensitive to microenvironmental factors and therefore could explain why only a small percentage of syngeneic cells go on to form metastases. The results of this study support the ability of IK to increase aggression of already transformed cancer cells. The ability of IK to promote tumor growth and metastasis in vivo points to the potential of IK-inhibiting drugs to decrease cancer progression. Furthermore, the opposing behavioral response between cancerous MDA-MB-231 and non-tumorigenic MCF-10A cell lines warrants further study to determine if there are unique mechanisms conferring different sensitivities that could be utilized to formulate targeted therapies.

The retroviral expression plasmid pMIG and a plasmid containing tagRFP (RFP- red fluorescent protein) were purchased from Addgene (#12282 and #37537). A plasmid containing the IK complete coding sequence was purchased from Genecopoeia (GC-0G00902). The IK coding sequence in tandem with P2A and tagRFP was inserted into pMIG to replace the IRES and green fluorescent protein (GFP) sequence using the Gibson cloning method following manufacturer instructions (NEB). Briefly, primers were designed to amplify pMIG excluding the IRES-GFP sequence (forward primer 5′ GCCAAGCTTATCGATAAAATAAAAGA, reverse primer 5′ AATTCCGGCGCCTAGAGA). Primers for the channels and tagRFP were designed with 15-40 bp overhanging sequences to match the adjacent DNA sequence of the final plasmid and to insert the P2A sequence (IK forward primer 5′ CTAGGCGCCGGAATTACCATGGGCGGGGATCTGG, reverse primer 5′ TCTCCTGCTTGCTTTAACAGAGAGAAGTTCGTGGCTCCGGATCCCTTGGACTGCTGGCTGGG; tagRFP forward primer 5′ TCTCTCTGTTAAAGCAAGCAGGAGACGTGGAAGAAAACCCCGGTCCTGTGTCTAAGGGCGAA GAGCTGA, reverse primer 5′ CCTACAGGTGGGGTCTCACTTGTACAGCTCGTCCATGCC). All fragments were amplified by PCR and gel purified. A Gibson reaction was performed using three DNA fragments, the pMIG backbone, channel, and tagRFP, to create the final plasmid (pMIG-IK). A plasmid with only tagRFP inserted was generated as a control (pMIG-RFP).

Two days prior to transfection, 2 × 106 293 Plat GP cells (Cell Biolabs) were plated on a 10 cm dish. Transfection was performed with 60 μL of lipofectamine 2000, 6 μg of VSV packaging plasmid, and 12 μg of pMIG-Channel-P2A-RFP plasmid per manufacturer instructions. Cells were incubated in the DNA:liposome complex overnight in a final volume of 6 mL. At 72 hours post transfection, virus media was collected and filtered through a 45 μm filter. Human breast epithelial cell lines were transduced by incubating 50% confluent cells with virus in media supplemented with 6 μg/mL polybrene for 24 hours. At 72 hours post-transduction, retrovirus treated cells were placed in media containing 50 μg/mL G418 for selection. Retrovirus infected cells were expanded for later FACS sorting.