Introduction

Cancer represents a complex disease characterized by unconstrained cell proliferation and the spread of malignant cells in the body (Kalia, 2015; Seyfried and Shelton, 2010). It is one of the leading causes of death worldwide, with millions of new cases diagnosed each year(Sung et al., 2021). Globally, the most prevalent form of cancer represents breast cancer (BC) (Sung et al., 2021). Despite many available anti- cancer treatments which largely depend on the steroid and epidermal growth factor (HER2) receptor status (Dunnwald et al., 2007), cancer cells frequently escape from existing therapies due to adaptions (P. Wu et al., 2021). Therefore, the identification of novel targets and therapeutic strategies that confer benefits for at least a subset of patients, whose cancer displays specific molecular or cellular features, is of utmost relevance.

Important factors that emerged as new cancer targets are ion channels (Li and Xiong, 2011). Especially alterations in the expression levels and function of potassium ion (K⁺) channels are critically related to cancer malignancy and progression (Li et al., 2023, p. 0). One of these channels represents the calcium ion (Ca²⁺) and voltage-activated K⁺ channel of large conductance (BK_Ca) (Mohr et al., 2022). Canonical BK_Ca channels usually localize in the plasma membrane (PM) of cells, and contribute to the regulation of the cytosolic K⁺ content, the PM potential (ΔΨ_PM), cell cycle, -volume, and -motility (Burgstaller et al., 2022a). Opening of BK_Ca channels results in K⁺ efflux, increasing the electrochemical driving force for Ca²⁺ entry into the cancer cell and affecting pathological cell growth and death (Ouadid-Ahidouch and Ahidouch, 2013). Accordingly, in BC cells (BCCs), an upregulation of BK_Ca has been associated with increased malignancy (Huang and Jan, 2014; Mohr et al., 2022; Oeggerli et al., 2012). However, besides their localization in the PM, several K⁺ channels, including BK_Ca, have also been identified in the inner mitochondrial membrane (IMM) (mitoBK_Ca), a topic which has been extensively reviewed recently (Checchetto et al., 2021; Kulawiak and Szewczyk, 2022; Szabo and Szewczyk, 2023; Szewczyk et al., 2009; Wrzosek et al., 2020). mitoBK_Ca was first described by Siemen et al. in a human glioma cell line more than 20 years ago (Siemen et al., 1999). So far, mitoBK_Ca has further been found e.g. in bronchial epithelial cells (Dabrowska et al., 2022), neurons (Douglas et al., 2006), skeletal muscle cells (Skalska et al., 2008), and in cardiac myocytes (Xu et al., 2002). In the latter, a splice variant, the BK_Ca-DEC isoform containing a unique C-terminal exon of 50 amino acids forms the functional mitoBK_Ca channel at the IMM (Singh et al., 2013). Little, however, is known about the molecular identity of mitoBK_Ca in other cell types, and mitochondrial localization of BK_Ca in BCCs has not been demonstrated so far.

Cancer cells show increased energy demands due to their high proliferation rates. Thus, tumor cells compensate for their elevated energy demand by increasing metabolic activities and adapting to nutrient- poor metabolic niches in the tumor microenvironment, allowing them to overcome oxygen (O₂)-dependent mitochondrial metabolism (Eales et al., 2016; Gross et al., 2022; Jang et al., 2013; Nazemi and Rainero, 2020). This metabolic switch from oxidative phosphorylation to glycolysis, frequently referred to as the “Warburg effect”, describes the phenomenon that cancer cells rather secrete lactate to the extracellular matrix (ECM), instead of utilizing pyruvate to fuel the TCA cycle (Warburg, 1924). This lactate secretion towards the ECM was shown to promote multiple microenvironmental cues causing tumor progression (de la Cruz-López et al., 2019). Interestingly, extracellular K⁺ may also impair effector T-cell function, and, thus, the anti-tumor immune response (Eil et al., 2016), while, within the cancer cell, functions of several glycolytic enzymes rely on K⁺ (Bischof et al., 2021; Gohara and Di Cera, 2016). Further, the presence of mitoBK_Ca contributes to the K⁺ entry into the mitochondrial matrix to interfere with mitochondrial volume changes and the mitochondrial membrane potential (ΔΨ_mito) (Krabbendam et al., 2018). Consequently, (mito)/BK_Ca-dependent mechanisms may severely affect energy production pathways and the resulting energy supply of cancer cells.

To elucidate how BK_Ca contributes to increasing BCC malignancy (Oeggerli et al., 2012), we utilized BK_Ca pro- and deficient human BCC lines including MDA-MB-453 and MCF-7 cells, and primary murine BCCs derived from the mouse mammary tumor polyoma middle T-antigen (MMTV-PyMT)-induced wild type (WT) or BK_Ca knock-out (BK-KO) BC model (Mohr et al., 2022). In these cells, either isoforms of BK_Ca were expressed, or BK_Ca was pharmacologically blocked by paxilline or iberiotoxin, two frequently used inhibitors of BK_Ca that either penetrate the cell or act exclusively at PM localized channels, respectively (Candia et al., 1992; Zhou and Lingle, 2014). These approaches were complemented by patch-clamp experiments, and by measuring the cellular ion homeostasis and metabolism with fluorescence live-cell imaging, extracellular flux analysis and liquid chromatography-mass spectrometry (LC-MS)-based approaches.

Our results emphasize that the presence of BK_Ca in the IMM affects mitochondrial bioenergetics, thereby increasing BCC malignancy. This effect was specifically mediated by the DEC isoform of BK_Ca, i.e., mitoBK_Ca. Functional expression of BK_Ca-DEC was validated by single-channel patch-clamp recordings BCC-derived mitoplasts. Importantly, we also identified BK_Ca-DEC expression in a subset of BC patient biopsies using nanostring-based mRNA expression analysis. Finally, mitoBK_Ca crucially contributed to murine and human BCC proliferation and hypoxic resistance, and its activity increased lactate secretion resulting in higher extracellular acidification rates, even in the presence of O₂.

Combined, our analyses provide, for the first time, a mechanistic link between functionally relevant mitochondrial BKCa isoforms in cancer cells and the promotion of the Warburg effect.

Results

Functional characterization of BKCa expression in BCCs

First, we assessed the functional expression of BK_Ca in established human BCC lines by performing whole- cell patch-clamp experiments. Primary BCCs obtained from transgenic, BC-bearing MMTV-PyMT WT or BK-KO mice were used as positive or negative controls, respectively (Mohr et al., 2022). Patch-clamp experiments revealed that depolarizing stimuli delivered in 20 mV increments induced K⁺ outward currents that were larger in MMTV-PyMT WT compared to BK-KO cells (Figs. 1A and 1B, and Figs. S1A and S1B). To validate that these increased currents were elicited by BK_Ca, we pharmacologically inhibited the channel by using either paxilline or iberiotoxin (Candia et al., 1992; Zhou and Lingle, 2014). While the current remained unaffected by these treatments in MMTV-PyMT BK-KO cells (Fig. 1B and Fig. S1B), peak currents (I_max) were drastically reduced in MMTV-PyMT WT cells (Fig. 1A and Fig. S1A). Further, we analyzed the plasma membrane potential (ΔΨ_PM) in these cells using the voltage-sensitive fluorescent dye Dibac4(3) (Adams and Levin, 2012). The ΔΨ_PM was more polarized in MMTV-PyMT WT compared to BK-KO cells (Fig. 1C), as expected, due to the presence of hyperpolarizing BK_Ca channels in WT cells (N’Gouemo, 2014). Similar current-voltage relationships were recorded using the human BCC lines MDA-MB-453 and MCF-7, expressing either high or low levels of BK_Ca mRNA transcripts, respectively (Mohr et al., 2022). Analysis of the outward currents activated by depolarization demonstrated a paxilline- and iberiotoxin-sensitive current in MDA-MB-453 cells (Fig. 1D and Fig. S1C), indicative for functional BK_Ca channels in their PM, but not in MCF-7 cells (Fig. 1E and Fig. S1D), which is well in line with previous studies (Mohr et al., 2022). Based on the almost non-existent level of BK_Ca-mediated PM currents in MCF-7 cells, we utilized these cells to express different BK_Ca isoforms fused to a red fluorescent protein (RFP). Therefore, we either used the recently identified mitoBK_Ca isoform (Singh et al., 2013), namely BK_Ca-DEC^RFP, or the same channel lacking the C-terminal amino acids, referred to as BK ^RFP (Fig. 1F).

Figure 1:

We hypothesized, that, in analogy to cardiac myocytes (Singh et al., 2013), the expression of BK_Ca-DEC^RFP in MCF-7 cells may yield a functional channel present in the IMM. To test this, MCF-7 cells were first transiently transfected either with RFP fused to a glycosylphosphatidylinositol (GPI)-anchor (RFP-GPI) directing RFP to the PM, or with a cytochrome c oxidase subunit 8 (COX8) mitochondrial leading sequence fusion protein (mtRFP) (Fig. S1E). Analysis of the colocalization of mtRFP with MitoGREEN, a dye that specifically stains mitochondria, resulted in high colocalization scores, while a low overlap was observed for RFP-GPI transfected MCF-7 cells (Fig. S1E). Subsequently, the same experiments were performed with MCF-7 cells expressing BK^RFP or BK _Ca-DEC^RFP. Although the mitochondrial localization score was lower for BK_Ca-DEC^RFP compared to mtRFP, this BK_Ca variant showed a significantly higher overlap with the mitochondrial dye than BK_CaRFP (Fig. 1G). Nevertheless, as not all RFP signal in the BK_Ca-DEC^RFP transfected MCF-7 originated from mitochondria, we investigated the PM localization of the two channel isoforms by patch-clamp. Compared to native MCF-7 cells (Fig. 1E), both, BK ^RFP and BK -DEC^RFP, increased the PM outward current (Fig. 1H and Figs. S1F and 1G). Despite comparable expression levels of the RFP signals (Fig. S1H), presence of BK ^RFP caused significantly bigger currents across the PM compared to BK_Ca-DEC^RFP (Fig. S1H), indicative either for i) higher intracellular abundance or ii) major functional differences of BK_Ca-DEC^RFP. Importantly, the conductance of both channels was sensitive to the BK_Ca blockers paxilline and iberiotoxin, yielding comparable PM conductancés to native MCF-7 cells (Figs. 1E and 1H and Figs. S1F and S1G).

BK_Ca modulates global and subcellular Ca²⁺ homeostasis in BCCs

As BK_Ca potentially affects cellular Ca²⁺ fluxes (Ouadid-Ahidouch and Ahidouch, 2013), we next investigated the cytosolic (Figs. 2A-C), endoplasmic reticulum (ER) (Figs. 2D-F), and mitochondrial (Figs. 2G-I) Ca²⁺ homeostasis in these cells. First, we assessed changes in the cytosolic Ca²⁺ concentration ([Ca²⁺]_cyto) over-time in response to cell stimulation with the purinergic receptor agonist adenosine-5’- triphosphate (ATP) (Müller et al., 2020) using the fluorescent Ca²⁺ indicator Fura-2 (Grynkiewicz et al., 1985), either under control conditions (Fig. 2A) or in the presence of paxilline (Fig. S2A) or iberiotoxin (Fig. S2B). Analysis of the Fura-2 fluorescence emission ratio showed significantly elevated basal (Fig. 2A) and ATP-elicited maximal [Ca²⁺]_cyto responses (Fig. 2C) in MMTV-PyMT WT compared to BK-KO cells under control conditions. Interestingly, the elevated basal [Ca²⁺]_cyto of WT cells was reduced to BK- KO cell levels in response to paxilline (Fig. 2A and Fig. S2A), but not by iberiotoxin (Fig. 2A and Fig. S2B), which is a peptide-based pore-blocking toxin that cannot pass the PM (Candia et al., 1992). Subsequently, basal and ATP-elicited maximal [Ca²⁺]_cyto transients were recorded in the human BCC line MDA-MB-453, where both BK_Ca blockers reduced the basal [Ca²⁺]_cyto (Fig. 2B), while maximal [Ca²⁺]_cyto was not altered by these treatments (Fig. S2D). Eventually, [Ca²⁺]_cyto was assessed in MCF-7 cells expressing or lacking the different BK_Ca isoforms (Fig. 2C). These experiments confirmed previous findings with the other BCC lines, as the expression of both splice variants increased the basal (Fig. 2C) and maximal [Ca²⁺]_cyto (Fig. S2E), although the effect was clearly more pronounced upon expression of BK_Ca-DEC^RFP.

Figure 2:

Next, we visualized changes in the ER [Ca²⁺] ([Ca²⁺]_ER). Therefore, BCCs were transfected with D1ER, an established genetically encoded, FRET-based Ca²⁺ sensor targeted to the lumen of the ER (Palmer et al., 2004). [Ca²⁺]_ER was depleted by extracellular Ca²⁺ removal and chelation by EGTA, followed by inhibition of the sarcoplasmic endoplasmic reticulum Ca²⁺ ATPase (SERCA) with BHQ and activation of inositol 1,4,5-trisphosphate (IP₃) receptors upon purinergic receptor stimulation with ATP (Lape et al., 2008; Müller et al., 2020; Salter and Hicks, 1995). Experiments were either performed under control conditions (Figs. 2D-F), or in the presence of paxilline or iberiotoxin for MMTV-PyMT (Fig. 2D and Figs. S2F and S2G), MDA-MB-453 (Fig. 2E), and MCF-7 cells expressing BK_CaRFP or BK_Ca-DEC^RFP (Fig. 2F). Throughout all BCCs investigated, expression of BK_Ca reduced [Ca²⁺]_ER, potentially indicating that the channel was i.) functional in this cell compartment and ii.) involved in regulating the Ca²⁺ homeostasis of the ER. In MMTV-PyMT WT cells, [Ca²⁺]_ER was restored by the cell-permeable BK_Ca inhibitor paxilline (Fig. 2D and Fig. S2F) but not by the cell-impermeable iberiotoxin (Fig. 2D and Fig. S2G), while both inhibitors restored the [Ca²⁺]_ER in MDA-MB-453 cells (Fig.2e). Accordingly, overexpression of both RFP-tagged BK_Ca isoforms in MCF-7 cells depleted the [Ca²⁺]_ER (Fig. 2F).

[Ca²⁺]_cyto and [Ca²⁺]_ER reportedly affect the mitochondrial [Ca²⁺] ([Ca²⁺]_mito) (Wacquier et al., 2019). Since an accumulation of K⁺ within the mitochondrial matrix may oppose mitochondrial Ca²⁺ uptake (Checchetto et al., 2021), we assessed the effects of functional BK_Ca expression on [Ca²⁺]_mito utilizing 4mtD3cpV, a genetically encoded Ca²⁺ sensor targeted to the mitochondrial matrix (Palmer et al., 2006). Again, BCCs were treated with ATP to investigate the [Ca²⁺]_mito upon cell stimulation. In MMTV-PyMT WT (Fig. 2G) as well as MDA-MB-453 cells (Fig. 2H), functional expression of BK_Ca reduced basal [Ca²⁺]_mito. Interestingly, in these two BCC types, basal and maximally elicited [Ca²⁺]_mito peaks increased in response to paxilline (Figs. 2G and 2H and Figs. S2H, S2J and S2K), but not iberiotoxin treatment (Figs. 2G and 2H and Figs. S2I-K). These findings confirm a role of intracellular located BK_Ca channels in modulating [Ca²⁺]_mito dynamics. Importantly, in MCF-7 cells, basal [Ca²⁺]_mito was only affected upon expression of the mitochondrially targeted BK_Ca-DEC^RFP, but not BK ^RFP (Fig. 2I), whereas neither of the two isoforms affected the maximal [Ca²⁺]_mito (Fig. S2L). Presumably, by facilitating K⁺ fluxes across the IMM, BK_Ca-DEC^RFP expression reduces the driving force for Ca²⁺ uptake and thus the resulting Ca²⁺ signals in the mitochondrial matrix.

In total, our [Ca²⁺] imaging approaches emphasize that different BK_Ca isoforms at different localizations, i.e. the PM or intracellular organelles, may either amplify or weaken [Ca²⁺]_cyto and [Ca²⁺]_ER signals, respectively. These opposing effects are expected as BK_Ca-mediated uptake of K⁺ into the ER should limit the Ca²⁺ uptake capacity of this subcellular compartment, while functional channel expression at the PM might increase the driving force for Ca²⁺ influx to fuel [Ca²⁺]_cyto. [Ca²⁺]_mito, however, seems to be exclusively and effectively suppressed by the activity of intracellularly located BK_Ca variants in MMTV- PyMT WT and MDA-MB-453 cells and solely by the expression of BK_Ca-DEC^RFP in MCF-7 cells.

The metabolic activity of murine and human BCCs is modulated by intracellular BK_Ca

Given the involvement of Ca²⁺ and K⁺ in regulating key features of cellular metabolism (Bischof et al., 2021; Dejos et al., 2020; Gohara and Di Cera, 2016), we proposed, that BK_Ca may alter metabolic activities of BCCs (Rossi et al., 2019). Hence, we analyzed the extracellular acidification rate (ECAR) as a measure of lactate secretion, i.e. glycolytic activity in MMTV-PyMT and MDA-MB-453 cells (Burgstaller et al., 2021a). ECAR measurements unveiled an increased basal ECAR of MMTV-PyMT WT compared to BK- KO cells (Figs. 3A and 3B). Subsequently, we performed a mitochondrial stress test by injection of Oligomycin-A, FCCP, and Antimycin-A, which increased ECARs in both cell types due to ATP-synthase inhibition, mitochondrial uncoupling and complex III blockade, respectively (Fig. 3A). Maximal ECAR, received upon FCCP injection, was elevated in WT compared to BK-KO cells (Fig. 3C). Besides this evidence derived from a gene-targeted BK_Ca channel-deficient model, we used paxilline or iberiotoxin as pharmacological BK_Ca modulators (Figs. S3A and S3B). Interestingly, only paxilline, but not iberiotoxin treatment equalized basal (Fig. 3B) and maximal ECARs (Fig. 3C) between WT and BK-KO, suggesting that intracellular BK_Ca channels stimulate the glycolytic activity of BCCs. Moreover, in MDA-MB-453 cells iberiotoxin treatment did neither affect basal (Fig. 3D and 3E) nor maximal ECARs (Fig. 3F). Paxilline, however, increased basal and maximal ECAR in these cells (Figs. 3E and 3F).

Figure 3:

Next, we investigated the oxygen consumption rates (OCRs) (Burgstaller et al., 2021a). (Fig. 3G). Interestingly, the increased basal and maximal ECAR (Figs. 3A-C) in MMTV-PyMT WT cells correlated with a higher basal (Fig. 3H) and maximal (Fig. 3I) OCR compared to BCCs lacking BK_Ca. Paxilline treatment reduced the basal and maximal OCR of BK_Ca proficient MMTV-PyMT cells to the BK-KO level (Figs. 3H and 3I and Fig. S3C), while iberiotoxin did not have any effect (Figs. 3H and 3I and Fig. S3D) again suggesting that that the latter toxin cannot reach the metabolism-relevant population of intracellular BK_Ca channels. Similar results were obtained in MDA-MB-453 cells, despite in these cells paxilline treatment increased the OCR (Figs. 3J-L), potentially due to the increased supply of oxidative phosphorylation with glycolytic substrates (Figs. 3D-F), while iberiotoxin had no effect.

To confirm that BK_Ca regulates the bioenergetic profile of BCCs, we subsequently applied LC-MS-based metabolomics according to the workflow presented in Figure S3E. We included typical analytes of glycolysis such as lactate (Fig. 3M) and pyruvate (Fig. 3N), as well as selected metabolites of mitochondrial metabolism including asparagine (Fig. 3O) and glutamine (Fig. 3P). Metabolomics data confirmed the metabolic differences between MMTV-PyMT WT and BK-KO cells and further demonstrated that BK_Ca inhibition by paxilline directly affected the concentrations of these metabolites (Figs. 3M-P). As observed before for the ECAR and OCR measurements, MMTV-PyMT and MDA-MB- 453 cells responded, however, differently to paxilline treatment (Figs. 3M-P). In summary, our data strengthen the notion that intracellular BK_Ca modulates the cellular energy balance of murine and human BCCs.

BK_Ca alters the mitochondrial function of BCCs

To clarify how BK_Ca regulates BCC cell metabolic activities, we examined cellular bioenergetics in real time using single-cell fluorescence microscopy. First, we assessed the mitochondrial membrane potential (ΔΨ_mito) of MMTV-PyMT and MDA-MB-453 cells using TMRM under control and BK_Ca inhibitor-treated conditions, followed by depolarization of ΔΨ_mito using the proton ionophore FCCP (Joshi and Bakowska, 2011). These measurements unveiled, however, a less polarized ΔΨ_mito of MMTV-PyMT WT cells compared to BK-KO cells under control conditions (Figs. 4A and 4B). Paxilline, but not iberiotoxin, equalized ΔΨ_mito between MMTV-PyMT WT and BK-KO cells (Fig. 4B). Identical results were obtained in MDA-MB-453 cells (Figs. 4C and 4D).

Figure 4:

Interestingly, we found that ΔΨ_mito of MMTV-PyMT cells was tightly dependent on the extracellular glucose concentration ([GLU]_ex) (Fig. S4A). Under high [GLU]_ex conditions (25.0 mM), the difference in ΔΨ_mito between MMTV-PyMT WT and BK-KO cells disappeared, as ΔΨ_mito increased significantly in WT and decreased significantly in BK-KO cells compared to low [GLU]_ex (2.0 mM). Because F_OF₁-ATP- synthase (ATP Synthase) may hydrolyze ATP in an attempt to maintain ΔΨ_mito, these results strongly suggest that the forward (ATP producing) vs reverse (ATP consuming) mode of the ATP synthase is affected by the BK_Ca status of the cells (Naguib et al., 2018).

To unravel the substrate dependency of these BCCs for maintaining their energy homeostasis in dependency of BK_Ca, we measured the mitochondrial ATP concentration ([ATP]_mito) using a genetically encoded ATP sensor targeted to the mitochondrial matrix, mtAT1.03 (Imamura et al., 2009). Mitochondrial ATP reportedly responds most dynamically to energy metabolism perturbations (Depaoli et al., 2018). To assess the sources of ATP in the cells, we either deprived the cells of [GLU]_ex, or administered Oligomycin-A to inhibit the ATP-synthase (Depaoli et al., 2018) (Figs. 4E-J and Figs. S4B- G). In MMTV-PyMT cells, glucose removal as well as ATP-synthase inhibition reduced [ATP]_mito, albeit the effect was more pronounced upon [GLU]_ex depletion (Figs. 4E and 4F and Figs. S4B and S4C). Remarkably, paxilline, but not iberiotoxin treatment reduced the glucose, and increased ATP-synthase dependency of MMTV-PyMT WT cells for maintaining [ATP]_mito (Figs. 4E and 4F and Figs. S4B and S4C). These observations were confirmed in MDA-MB-453 cells, even though i.) these cells showed an overall higher dependency on the ATP-synthase, and ii.) the paxilline-sensitivity of [ATP]_mito maintenance was much less pronounced (Figs. 4G and 4H and Figs. S4D and S4E).

To validate these findings in a BK_Ca deficient model, these experiments were performed with MCF-7 cells either expressing exclusively RFP as control, BK ^RFP, or BK _Ca-DEC^RFP (Fig. 4I). Expression of BK _Ca- DEC^RFP, but not BK ^RFP, resulted in a high dependency on [GLU]_ex to maintain [ATP] while the [ATP]_mito rundown in the presence of Oligomycin A was identical for both BK_Ca splice variants. This suggests that BK_Ca-DEC^RFP specifically triggers a high [GLU]_ex sensitivity and simultaneously an independency on ATP derived from ATP-synthase for maintaining [ATP]_mito as demonstrated by the ratio of the respective changes under these experimental conditions (Figs. 4I and 4J and Figs. S4F and S4G). These findings suggest that intracellular (mitochondrial) BK_Ca contributes to the metabolic reprogramming of BCCs.

In an extension to extracellular flux analyses, LC-MS metabolomics (Fig. 3), and live-cell imaging experiments (Figs. 4A-J), pointing to a BK_Ca-dependent “oncometabolic” phenotype, we assessed the concentrations of mitochondrial hydrogen peroxide ([H₂O₂]_mito) using mitoHyPer3 (Bilan et al., 2013), a genetically encoded fluorescent indicator for monitoring H₂O₂ in the mitochondrial matrix (Figs. 4K-M). These experiments demonstrated increased levels of [H₂O₂]_mito in MMTV-PyMT WT (Fig. 4K) and MDA- MB-453 cells (Fig. 4L), or specifically upon BK_Ca-DEC^RFP expression in MCF-7 cells (Fig. 4M). Excessive reactive oxygen species (ROS) synthesis is caused by uncoupling of the respiratory chain and it is considered as an indicator of mitochondrial stress, which, among other reasons, promotes mutagenesis and BCC progression. In this regard, mitoBK_Ca may serve as an “uncoupling” protein (Gałecka et al., 2021), triggering ATP synthase to operate in reverse mode to consume instead of producing ATP, thereby counteracting the dissipation of the proton gradient and consequently the loss of ΔΨ_mito. Consequently, the lack of ATP must be compensated, e.g. by accelerating glycolysis. To investigate this assumption, we performed glucose uptake measurements using 2-NBDG, a fluorescent glucose analogue, which is taken up via glucose transporters (GLUTs), phosphorylated by hexokinase (HK) isoforms to generate 2-NBDG- 6-phosphate and subsequently remains within the cell (Bischof et al., 2021) (Figs. S4H and S4I). Unexpectedly, we found that MMTV-PyMT BK-KO cells showed a higher rate of glucose uptake and phosphorylation compared to WT cells under basal conditions (Fig. 4N). To unravel the role of mitochondria in contributing to glucose uptake by supplying ATP to HKs, we next performed these experiments in the presence of FCCP, which disrupts mitochondrial ATP production (Losano et al., 2017). If ATP-synthase works in forward mode, FCCP treatment should reduce or even prevent oxidative ATP production, and, subsequently, ATP-dependent 2-NBDG phosphorylation (Fig. S4H). Contrary, if ATP- synthase works in reverse mode, it may compete with HKs for ATP, and FCCP treatment should abolish this competition, leading to increased 2-NBDG phosphorylation in BCCs (Fig. S4I). Interestingly, our experiments unveiled increased 2-NBDG uptake in MMTV-PyMT WT and reduced uptake in BK-KO cells upon mitochondrial depolarization (Fig. 4N and Figs. S4J and S4K). To facilitate the interpretation, we calculated the FCCP-induced change in 2-NBDG uptake. These analyses revealed that mitochondria are less effective in assisting 2-NBDG uptake and phosphorylation in MMTV-PyMT WT cells, while they “support” these processes in BK-KO cells under control conditions (Fig. 4O). To test whether the changes in 2-NBDG uptake are sensitive to pharmacological modulation of BK_Ca, we performed a similar set of experiments in the presence of paxilline or iberiotoxin. While iberiotoxin treatment did not have any effect, paxilline treatment shifted the activity of the ATP-synthase to the ATP-supplying “forward” mode (Fig. 4O and Figs. S4J and S4K). Comparable effects were obtained in MDA-MB-453, despite paxilline treatment reduced basal accumulation of 2-NBDG in these cells (Figs. 4P and 4Q and Fig. S4L). Overall, the experiments performed so far suggest that mitochondria rather consume than generate ATP if BK_Ca was functionally expressed intracellularly in BCCs.

BK_Ca locates functionally in the IMM of murine and human BCCs

So far, our data suggest an important contribution of intracellularly located BK_Ca, possibly mitoBK_Ca, in reprogramming cancer cell metabolism. Thus, we applied an electrophysiological approach to provide functional evidence for endogenous mitoBK_Ca. Single-channel patch-clamp experiments were conducted using mitoplasts isolated from MMTV-PyMT WT and BK-KO, MDA-MB-453, and MCF-7 cells. In MDA-MB-453- and MMTV-PyMT WT-derived mitoplasts we indeed detected channels of large conductance (Fig. 5A and Fig. S5A). For these channels, the open probability did not significantly vary with the voltage (Fig. 5B). Only at very negative and positive voltages of approximately -150 mV and +150 mV differences in the open probabilities were observed (Fig. 5C). The bursts of single-channel openings showed an average conductance of 212 ± 2 pS (Fig. 5D). This large conductance and the sensitivity towards Ca²⁺- (Figs. 5E and 5F and Fig. S5B) and paxilline (Figs. 5E and 5F and Fig. S5C) pointed to mitoBK_Ca that exhibits the pharmacologic characteristics of canonical BK_Ca channels present at the PM. Overall, our electro-pharmacological experiments unveiled 3 different classes of channels with either small (≤100 pS), medium (≤150 pS) or large (∼210 pS) conductance, where the latter conductivity corresponds to mitoBK_Ca (Liu et al., 1999). MitoBK_Ca was detected at a frequency of 15% in 210 patches of MDA-MB-453 mitoplasts and at a lower frequency of 6% in 127 mitoplast patches of MMTV-PyMT WT cells. Importantly, this channel was absent in 76 mitoplast patches from MMTV-PyMT BK-KO and 58 mitoplast patches from MCF-7 cells (Fig. 5G). These findings provide evidence that the molecular entity for the K⁺ channel derives from the nuclear Kcnma1 gene, which is ablated in the MMTV-PyMT BK-KO.

Figure 5:

These experiments were further corroborated by immunoblotting experiments using whole-cell lysates and sub-cellular homogenates of different purity. We detected BK_Ca, the Na⁺/K⁺ ATPase as a marker of the PM, and cytochrome c oxidase subunit IV (COXIV) as a marker of mitochondria in these samples. A protein band corresponding to BK_Ca was identified not only in whole-cell lysates, but also in isolated mitochondria. Importantly, the Na⁺/K⁺ ATPase was absent in the latter protein fraction, confirming the purity of the mitochondrial preparation (Fig. S5D). These data confirm the presence of mitoBK_Ca, potentially BK_Ca-DEC, in the utilized BK_Ca proficient BCCs.

mitoBK_Ca promotes the Warburg effect, triggers cellular O₂ independency and stimulates BCC proliferation

Based on the observed influence of mitoBK_Ca on glycolysis and mitochondrial metabolism, we addressed, whether the channel contributes to the Warburg effect, commonly observed in cancer cells (Bischof et al., 2021; Warburg, 1924). Therefore, we assessed the Warburg index (WI) by investigating the cytosolic lactate concentration ([LAC]_cyto) over-time using Laconic, a FRET-based lactate sensor (San Martín et al., 2013). [LAC]_cyto was followed in response to either inhibiting mitochondrial metabolism by administration of NaN₃, a complex IV inhibitor, to stop pyruvate consumption (Leary et al., 1998), or upon subsequent inhibition of lactate secretion towards the ECM via monocarboxylate transporter 1 (MCT-1) using BAY-8002 (Quanz et al., 2018) (Figs. 6A and 6B). Indeed, MMTV-PyMT WT cells exhibited an increased WI compared to BK-KO cells under control conditions (Fig. 6C), indicating that the presence of BK_Ca favors lactate secretion rather than TCA-dependent utilization of pyruvate. Paxilline, but not iberiotoxin treatment reduced the WI in WT cells to the BK-KO level (Fig. 6C). The WI profiles of MDA-MB-453 (Fig. 6D) and MCF-7 cells (Fig. 6E) showed the same sensitivity towards paxilline, while in the latter the expression of BK_Ca-DEC^RFP, but not BK ^RFP, stimulated the WI. To validate the contribution of the different BK isoforms, endogenous BK_Ca transcripts in MMTV-PyMT and MDA-MB-453 cells were targeted using specific siRNAs targeting either the major BK_Ca isoforms or specifically the DEC exon of BK_Ca (Fig. S6A). Cell treatment with the respective siRNAs reduced the expression of BK_Ca or BK_Ca-DEC (Fig. S6B and S6C). Interestingly, the knockdown of BK_Ca interfered with the WI in these cells (Figs. 6F and 6G). This effect was reproduced by specific silencing of the BK_Ca-DEC isoform (Figs. 6F and 6G), indicating that BK_Ca-DEC-derived mitoBK_Ca channels stimulate the Warburg effect in BCCs.

Figure 6:

As glycolytic metabolites and ATP fuel cell proliferation, we investigated the proliferation rates of MMTV-PyMT WT, MMTV-PyMT BK-KO, and MDA-MB-453 cells over-time in the absence or presence of paxilline or iberiotoxin. Corroborating previous findings (Mohr et al., 2022), these analyses revealed faster proliferation of WT compared to BK-KO cells under control conditions (Fig. 6H). Interestingly, paxilline, but not iberiotoxin treatment, reduced the proliferation of MMTV-PyMT WT to the level of BK_Ca-deficient cells (Fig. 6H). These findings were confirmed in MDA-MB-453 cells (Fig. 6I), suggesting mitoBK_Ca as key-player in mediating the metabolic rewiring. Next, we studied whether the increased WI affects the proliferation rate of these BCCs at low O₂ tension (Nazemi and Rainero, 2020). Colony formation assays (CFAs) revealed an increased hypoxic resistance of MMTV-PyMT WT compared to BK-KO cells, as the number and size of colonies lacking BK_Ca was reduced upon O₂- deprivation (Fig. 6J and Fig. S6D). In sum, these data demonstrate a malignancy-promoting effect of mitoBK_Ca in BCC.

mitoBK_Ca is of potential clinical relevance

To determine the clinical relevance of our findings, we investigated whether BK_Ca-DEC transcripts are present in primary BC tissue. Therefore, the mRNA expression of BK_Ca and BK_Ca-DEC was analyzed by nanostring analysis in bulk tumor biopsies isolated from 551 BC patients. Remarkably, all 551 samples tested positive for BK_Ca mRNA expression, with 10 of the samples showing significant expression of BK_Ca-DEC above the log2 expression threshold of 5.5 (Figs. 6K and 6L). Importantly, if BK_Ca-DEC was expressed, the expression of BK_Ca well correlated with the expression of BK_Ca-DEC (R² = 0.6060, p = 0.008) (Fig. 6M).

In sum, our experiments emphasize the presence of BK_Ca in different intracellular organelles including the ER and vesicles of the secretory pathway, yielding its PM localization. At these sites, BK_Ca modulates the Ca²⁺ homeostasis and regulates ΔΨ_PM. K⁺ efflux across the PM may additionally affect glycolysis. Importantly, functionally relevant BK_Ca also locates in the IMM of BCCs, promoting, presumably by the K⁺ accumulation in the matrix following channel activation, ΔY_mito depolarization and consequently ATP- synthase activity in reverse mode as well as a depletion of [Ca²⁺]_mito. These profound ionic and bioenergetic changes ultimately trigger the proliferation of BCCs in a low oxygen environment, as found in solid tumors (Fig. 6N). Taken together, functional expression of mitoBK_Ca could possibly denote a prognostic or therapeutic marker for BC patients, and its pharmacologic modulation could represent a novel anti-cancer treatment strategy.

Discussion

Here, we demonstrate for the first time that BK_Ca-DEC (mitoBK_Ca) is functionally expressed in BCCs. BK_Ca-DEC modulates BCC metabolism, stimulates the Warburg effect, and accelerates cell proliferation rates in the presence and absence of O₂. These tumor- and malignancy-promoting effects were sensitive to BK_Ca inhibition using the cell-permeable BK_Ca inhibitor paxillin e(Zhou and Lingle, 2014), but not by the cell-impermeable blocker iberiotoxin (Candia et al., 1992), indicating that intracellular BK_Ca, presumably mitoBK_Ca, mediates malignant BCC behavior and tumor development.

In line with recent single-cell RNA sequencing data of 26 primary breast tumors (S. Z. Wu et al., 2021), we found high transcript levels for BK_Ca throughout the analyzed BC samples, in addition to BK_Ca-DEC, albeit in a much smaller subset of BC. Further, we confirmed functional BK_Ca expression in the PM of MMTV-PyMT WT and MDA-MB-453 cells, while MMTV-PyMT BK-KO and MCF-7 cells showed no or very low PM BK_Ca currents. Therefore, MCF-7 cells represented a suitable model to investigate the effects of BK_Ca over-/expression on the metabolic homeostasis of human BCC. Similarly to cardiac myocytes, expression of BK_Ca-DEC yielded mitochondrial localization of BK_Ca-DEC, although its abundance in the IMM appeared less pronounced compared to cardiomyocytes (Singh et al., 2013). While a direct comparison is difficult as plasmid transfections may have caused unexpected effects, our finding that BK_Ca-DEC^RFP caused significantly reduced currents across the PM compared to BK confirm its increased intracellular abundance. ^RFP putatively

Based on the potential impact of BK_Ca on the cellular ΔΨ_PM and ion balance (Burgstaller et al., 2022a), we conducted an in-depth investigation of the (sub)cellular Ca²⁺ homeostasis. Across the BCCs examined, we observed that functional BK_Ca expression modulated [Ca²⁺]_cyto dynamics. These alterations showed, however, differential sensitivities to BK_Ca inhibitors in MMTV-PyMT WT and MDA-MB-453 cells. While basal [Ca²⁺]_cyto in MDA-MB-453 cells was reduced by both, paxilline and iberiotoxin treatment, MMTV-PyMT WT cells were only sensitive to paxilline. Examination of [Ca²⁺]_ER confirmed the results from [Ca²⁺]_cyto, as [Ca²⁺]_ER levels increased with paxilline and iberiotoxin in MDA-MB-453, but only upon paxilline exposure in MMTV-PyMT WT cells, suggesting differential effects of intracellular and PM- localized BK_Ca on Ca²⁺ handling in these cells. Interestingly, in MCF-7 cells, both BK_Ca splice variants mediated the opposing effects on the basal [Ca²⁺]_cyto and [Ca²⁺]_ER levels, which either in- or decreased, respectively, upon transient BK ^RFP or BK -DEC^RFP expression. This is expected as the transitory hyperpolarization and the efflux of K⁺ due to the opening of PM BK_Ca provides the driving force for Ca²⁺ entry into cytoplasm, while a BK_Ca-mediated K⁺ increase within the ER, presumably through PM-directed channels crossing the ER membrane in the secretory pathway, would oppose the Ca²⁺ refiling (Burgstaller et al., 2022a). Moreover, these results are in line with the higher proliferative capability of BK_Ca proficient BCCs due to the manifold roles of Ca²⁺ as second messenger (Burgstaller et al., 2022a). Importantly, however, [Ca²⁺]_mito of BK_Ca proficient MMTV-PyMT WT and MDA-MB-453 cells was exclusively sensitive for paxilline, and it was specifically affected by the expression of BK_Ca-DEC^RFP, but not by BK ^RFP, in MCF-7 cells, suggesting that this Ca²⁺ pool is exclusively controlled by intracellular BK .

Subcellular Ca²⁺ alterations could reportedly alter cellular bioenergetics, as Ca²⁺ directly regulates metabolic enzymes and activities (Rossi et al., 2019). Indeed, extracellular flux analysis, LC-MS-based metabolomics and fluorescence-based live-cell imaging confirmed, that the observed alteration in sub- cellular Ca²⁺ homeostasis caused by BK_Ca, especially mitoBK_Ca, has severe effects on cell metabolism. Our data further emphasize, that the presence of mitoBK_Ca, as confirmed by mitoplast patch-clamp and Western blot analysis of isolated mitochondria, depolarizes BCC mitochondria, which is in line with a previous study showing an impact of mitoBK_Ca activation on ΔΨ_mito (Kicinska et al., 2016). mitoBKCa- dependent depolarization of ΔΨ_mito in turn triggers cellular glucose dependency and reverses the activity of the mitochondrial ATP-synthase to consume ATP for restoring ΔΨ_mito. Finally, BK_Ca-DEC-derived mitoBK_Ca channels promote the Warburg effect and ultimately stimulated proliferation rates of BCCs. Our data fit earlier findings from glioma cells, where an O₂-sensitivity of mitoBK_Ca was observed, which probably increased the hypoxic resistance of these cancer cells (Gu et al., 2014). Excessive production and release of lactate, a hallmark of the Warburg effect, leads to extracellular acidification, subsequently creating a microenvironment that promotes tumorigenesis and metastasis as well as the resistance to anti- tumor immune responses and therapy (de la Cruz-López et al., 2019; Nazemi and Rainero, 2020; P. Wu et al., 2021). Extracellular K⁺ [K⁺]_ex, in turn, accumulating within the necrotic core of solid tumors, was shown to interfere with effector T-cell function triggering immune escape of cancer cells (Eil et al., 2016). To elucidate whether mitoBK_Ca directly contributes to lactate-induced tumor aggressiveness or [K⁺]_ex, live-cell imaging of extracellular metabolites in 3D BCC models should be applied in future investigations (Burgstaller et al., 2022b, 2021b).

Finally, our results demonstrate for the first time that BK_Ca-DEC transcripts are present in human BC biopsies. Although only a small proportion of patients was positive for BK_Ca-DEC expression, this finding could be of considerable clinical relevance considering the link between mitoBK_Ca function and BCC metabolism. Importantly, the design of our study likely underestimates the incidental number of BK_Ca- DEC positive BC, as i.) only hormone-receptor positive BC specimens were included, and ii.) bulk-tumor mRNA was analyzed, hampering the detection of low-abundant or tightly regulated transcripts against a strong background of non-cancer cells present in bulk tumor tissues. Finally, due to the small number of positive hits (N=10 positive versus N=541 BK_Ca-DEC negative specimens) and the lack of (sufficient) follow-up information in some of these cases, we are currently unable to correlate BK_Ca-DEC expression with, for example, treatment response or survival. Thus, future studies are warranted to show how the tumor’s BK_Ca-DEC status can help to predict or therapeutically improve standard chemo-endocrine treatments. The rather low abundance of BK_Ca-DEC in the clinical samples, however, is in agreement with our mitoplast patch clamp experiments with mitoBK_Ca-mediated K⁺ currents being detected at frequencies between 6 to 15%, suggesting that either a small proportion of mitochondria express functional mitoBK_Ca, or that the abundance of the channel per mitochondrion is low, requiring sensitive mechanistic approaches for its detection.

In summary, our data highlight a potentially druggable mitoBK_Ca isoform in BCCs, whose molecular entity is mainly formed by the Kcnma1 encoded BK_Ca-DEC splice variant. This channel promotes metabolic alterations in cancer cells, even under low-oxygen conditions, which may ultimately be of clinical interest for new anti-cancer therapies.

Material and Methods

Buffers and solutions

If not otherwise stated, all chemicals were purchased from Carl Roth GmbH, Karlsruhe, Germany. Buffers used in this study comprised the following:

Physiologic buffer for single cell live recordings contained (in mM): 138 NaCl, 5 KCl, 2 CaCl₂, 1 MgCl₂, 2 glucose, 10 HEPES, pH set to 7.4 with NaOH. No glucose was added during glucose removal experiments, while glucose was increased to 25 mM to investigate glucose dependency of the mitochondrial membrane potential. 0.1 mM Ethylene glycol bis(2-aminoethylether)-N, N, N’, N’-tetra acetic acid (EGTA) (Sigma Aldrich Chemie GmbH, Taufkirchen, Germany) instead of 2.0 mM CaCl₂ was added to obtain Ca²⁺ free buffer. The following compounds were added to yield the following final concentration: 3 µM Oligomycin-A, 200 nM Carbonyl cyanide-p-trifluoromethoxyphenylhydrazone (FCCP), 5 µM paxilline (all from (Santa Cruz Biotechnology, Dallas, USA), 30 nM iberiotoxin (Selleckchem, Planegg, Germany), 5 mM NaN₃, 3 µM BAY-8002, 15 µM 2,5-Di-(t-butyl)-1,4- hydroquinone (BHQ) (all from Sigma Aldrich Chemie GmbH), 100 µM adenosine-5’-triphosphate (ATP). For H₂O insoluble compounds, final DMSO concentration in the buffer did not exceed 0.1%. For experiments of glutamine and pyruvate removal, 2 mM GlutaMAX and 1 mM sodium pyruvate (both Thermo Fisher Scientific, Waltham, USA) were added to the starting buffer.

Intracellular buffer used for whole-cell patch-clamp experiments contained (in mM): 130 K-Gluconate, 5 KCl, 2 Mg-ATP 0.1, CaCl2, 0.2 Na2-GTP, 0.6 EGTA, 5 HEPES, pH = 7.2 with KOH.

Cell equilibration buffer fluorescence microscopy-based contained (in mM): 135 NaCl, 5 KCl, 2 CaCl₂, 1 MgCl₂, 2.6 NaHCO₃, 0.44 KH₂PO₄, 0.34 Na₂HPO₄, 10 glucose, 10 HEPES, 2 GlutaMAX, 1 sodium pyruvate, with 1x MEM amino acids and 1x MEM vitamins added (both Thermo Fisher Scientific). pH was adjusted to 7.4 using NaOH.

Buffers used for mitochondrial isolation, mitoplast preparation, and single channel patch-clamp comprised the following: The preparation solution contained (in mM): 250 sucrose, 5 HEPES, pH = 7.2. The mitochondrial storage buffer contained (in mM): 150 KCl, 0.1 CaCl₂, 20 HEPES, pH = 7.2. The hypotonic buffer contained (in mM) 0.1 CaCl₂, 5 HEPES, pH = 7.2. The hypertonic buffer contained (in mM): 750 KCl, 0.1 CaCl₂, 30 HEPES, pH 7.2. Low-Ca²⁺ solution (1 μM Ca²⁺) contained (in mM): 150 KCl, 1 EGTA, 0.752 CaCl₂, 10 HEPES, pH = 7.2.

Cell culture and transfection

Mouse mammary tumor virus polyoma middle T antigen (MMTV-PyMT) cells were isolated from tumors of MMTV-PyMT transgenic FVB/N WT or BK-KO mice. Tumor growth in vivo and biopsies were authorized by the local ethics Committee for Animal Research (Regierungspräsidium Tuebingen, Germany), and were performed in accordance with the German Animal Welfare Act. Animals were kept on a 12-hour light/ dark cycle under temperature- and humidity-controlled conditions with unlimited access to food (Altromin, Lage, Germany) and water. MMTV-PyMT cells used in this study were isolated from 3 – 7 different female breast-cancer bearing WT and 3 – 4 different female breast-cancer bearing BK-KO animals at an age of ∼ 12 – 14 weeks. Upon dissection, tumors were carefully minced into pieces using atraumatic forceps, lysed by 1 mg mL^-1 Collagenase-D (Roche, Basel, Switzerland) for 10 min, and cultured as follows: Cells were grown in modified improved minimal essential medium (IMEM) supplemented with 5% fetal bovine serum (FBS), 1 mM sodium pyruvate and 100 U mL⁻¹ penicillin and 100 µg mL⁻¹ streptomycin (all purchased from Thermo Fisher Scientific) at 37°C and 5% CO₂ in a humidified incubator. Fibroblasts were removed by exposure of the cultures to 0.25% trypsin-EDTA in PBS (Thermo Fisher Scientific) and short incubation at 37°C (∼ 1 minute). After gently tapping the plate, trypsin-EDTA with detached fibroblasts was removed and cells were further cultured in supplemented modified IMEM at 37°C and 5% CO₂ until subculturing.

MCF-7 and MDA-MB-453 cells were purchased from the Global Bioresource Center (ATCC). Cells were cultivated in Dulbecco’s modified eagle’s medium (DMEM) supplemented with 10% FBS, 1 mM sodium pyruvate and 100 U mL⁻¹ penicillin and 100 µg mL⁻¹ streptomycin (Thermo Fisher Scientific) at 37°C and 5% CO₂ in a humidified incubator.

Subculturing of cells was performed when cells reached a confluency of 80 – 90%. Therefore, cell culture medium was removed, cells were washed 1x with PBS, and trypsin-EDTA at a final concentration of 0.25% trypsin-EDTA in PBS was added. Subsequently, cells were incubated at 37°C and 5% CO₂ in a humidified incubator until cell detachment occurred (∼ 2-5 minutes). Trypsinization was stopped by adding supplemented cell culture media and cells were pelleted at 300xg for 5 minutes. The supernatant was removed, and cells were seeded to new cell culture dishes as required. For fluorescence microscopic live-cell imaging experiments, cells were either seeded in 6-well plates containing 1.5 H 30 mm circular glass coverslips (Paul Marienfeld GmbH, Lauda-Königshofen, Germany). All other vessels and serological pipettes used for cell culture were ordered from Corning (Kaiserslautern, Germany).

Transfection of cells was performed according to manufacturer’s instructions when cells showed a confluency of ∼70%, either using PolyJET DNA transfection reagent (SignaGen Laboratories, Maryland, USA) for plasmid DNA transfection or Lipofectamine 2000 (Thermo Fisher Scientific) for siRNA transfection or co-transfection of siRNA with plasmid DNA. Plasmid DNA amount was reduced to 1/3 for transfection of mitochondrial-targeted probes to ensure proper mitochondrial localization of the probes. Plasmid transfections were performed 16 hours, siRNA transfections were performed 48 hours before the experiments. Paxilline or iberiotoxin were added 12 hours prior to the experiments to the respective cell culture medium. DMSO served as a control.

Whole-cell patch-clamp

For whole-cell patch-clamp experiments, 30,000 cells were seeded on the day before the experiment in 35 mm glass bottom µDishes (ibidi GmbH, Graefelfing, Germany) and cultivated in the respective supplemented cell culture medium over-night at 37°C and 5% CO₂ in a humidified incubator. The next day, cell culture medium was removed, and cells were washed 2x and maintained in prewarmed physiologic buffer. Subsequently, recordings were performed using borosilicate glass capillaries (0.86x1.5x100mm) (Science Products GmbH, Hofheim am Taunus, Germany), with a resistance of 4-6 MW, which were pulled using a model P-1000 flaming/ brown micropipette puller (Sutter Instruments, California, USA) and filled with intracellular buffer. A MP-225 micromanipulator served for pipette control (Sutter Instruments). Recordings were performed in whole-cell mode. Currents were evoked by 15 voltage square pulses (300 ms each) from the holding potential of -70 mV to voltages between -100 mV and +180 mV delivered in 20 mV increments. For amplifier control (EPC 10) and data acquisition, Patchmaster software (HEKA Elektronik GmbH, Lambrecht, Germany) was used. Voltages were corrected offline for the capacity. Data analysis was performed using Fitmaster software (HEKA Elektronik GmbH), Nest-o-Patch software (http://sourceforge.net/projects/nestopatch, written by Dr. V Nesterov), and Microsoft Excel (Microsoft, Washington, USA).

Cloning and plasmid preparation

Cloning was performed using conventional PCR-, restriction- and ligation-based procedures. BK_Ca-DEC was a gift from Michael J. Shipston and was N-terminally attached to an RFP (BK_Ca-DEC^RFP) using KpnI and BamHI restriction sites after PCR amplification (NEB Q5 High-Fidelity DNA-Polymerase, New England Biolabs (NEB), Ipswich, USA). For generation of BK ^RFP, a PCR amplification of BK _Ca-DEC was performed, where the reverse primer omitted the last amino acids including the 50 amino acids encoding the DEC exon. Mitochondrial targeted TagRFP (mtRFP) was generated by fusing a double repeat of COX8 pre-sequence N-terminally to TagRFP. RFP-GPI was generated by fusing the membrane leading sequence (MLS) and the GPI-anchor signal of cadherin 13 N- and C-terminally to TagRFP, respectively. After PCR reactions, the DNA fragments were purified from gel electrophoresis using the Monarch DNA gel extraction kit (NEB), fragments and destination plasmid were digested using the respective restriction enzymes (NEB) and ligation (T4 DNA Ligase, NEB) and transformation (chemically competent NEB 5- alpha E. coli) were performed according to manufacturer’s instructions. Plasmids were verified by sequencing (Microsynth AG, Balgach, Switzerland). DNA maxipreps were performed using the Nucleobond Xtra Maxi kit (Macherey Nagel GmbH & Co. KG, Düren, Germany). Purified DNA was stored at 4°C.

Apotome imaging

For apotome imaging of mtRFP, RFP-GPI, BK_CaRFP or BK_Ca-DEC^RFP colocalization with mitochondria, MCF-7 cells were seeded on circular 18 mm glass coverslips (Marienfeld GmbH) in 12-well plates (Corning). Cells were transfected using PolyJet transfection reagent according to the manufacturer’s instructions. 16 hours after transfection medium was exchanged for fresh cell culture medium and cells were further cultivated for 24 hours. Subsequently, the medium was exchanged for fresh medium containing MitoGREEN (PromoCELL GmbH, Heidelberg, Germany) at a final concentration of 66.6 µM and cells were further incubated at 37°C and 5% CO₂ for 20 minutes. Subsequently, cells were washed 2x with cold PBS, paraformaldehyde (PFA) (Carl Roth GmbH) at a final concentration of 2% in PBS was added and cells were incubated for 30 minutes on ice. Next, cells were washed again 2x with PBS and mounted on glass slides (Th. Geyer GmbH & Co. KG, Renningen, Germany) with PermaFluor aqueous mounting medium (Thermo Fisher Scientific) and dried over-night in the dark at 4°C. Imaging was performed using a Zeiss axio imager Z1 equipped with an Apotome.2 (Carl Zeiss AG), a HBO 100 lamp and a 63x 1.4 plan apochromatic objective. Microscope control and image acquisition was performed using Zeiss zen 2.6 blue edition (Carl Zeiss AG). Image analysis was performed using the colocalization test in ImageJ with Fay randomization after cell selection by ROIs.

Fluorescence live-cell imaging

Cells were either analyzed using a Zeiss AXIO Observer Z1 or a Zeiss Axiovert 200m microscope (Carl Zeiss AG, Oberkochen, Germany). The Zeiss AXIO Observer Z1 was connected to a LEDHub high-power LED light engine (OMICRON Laserage, Rodgau-Dudenhofen, Germany) and equipped with a EC Plan- Neofluar 40x/1.3 Oil DIC M27 objective (Carl Zeiss AG), an Optosplit II emission image splitter (Cairn Research Ltd, Faversham, UK), and a pco.panda 4.2 bi sCMOS camera (Excelitas PCO GmbH, Kelheim, Germany). The microscope possessed a BioPrecision2 automatic XY-Table (Ludl Electronic Products, Ltd., New York, USA). Optical filters included a 459/526/596 dichroic mirror and a 475/543/702 emission filter for FRET- and TMRM-based measurements, and a 409/493/573/652 dichroic mirror combined with a 514/605/730 emission filter for Dibac4(3), Fura-2 and 2-NBDG based measurements (all purchased from AHF Analysentechnik, Tuebingen, Germany). The Optosplit II emission image splitter was equipped with a T505lpxr long-pass filter (AHF Analysentechnik). The LEDHub high-power LED light engine was equipped with a 340 nm, 385 nm, 455 nm, 470 nm and 505-600 nm LED, followed by the following emission filters, respectively: 340x, 380x, 427/10, 473/10 and 510/10 or 575/15 (AHF Analysentechnik). The Zeiss Axiovert 200m microscope was connected to a pe340^fura light source (CoolLED, Andover, UK), an Optosplit II emission image splitter (Cairn Research Ltd.) and a pco.panda 4.2 sCMOS camera (Excelitas PCO GmbH) and equipped with 340/26, 380/14 and switchable 427/10, 485/20 or 575/15 excitation filters (AHF Analysentechnik) in the light source, respectively, a 40x Fluar 1.30 oil immersion objective (Carl Zeiss AG), a 459/526/596 or 515LP dichroic mirror and a 475/543/702 or 525/15 emission filter (AHF Analysentechnik) in the microscope, and a T505lpxr (AHF Analysentechnik) in the Optosplit II. Image acquisition and control of both microscopes was performed using VisiView software (Visitron Systems GmbH, Puchheim, Germany). Perfusion of cells was performed using a PC30 perfusion chamber connected to a gravity-based perfusion system (NGFI GmbH, Graz, Austria) and a vacuum pump.

Fura-2 based Ca₂₊ measurements

For fura-2 based Ca²⁺ measurements, cells were taken from the humidified incubator at 37°C and 5% CO2, washed 1x with cell equilibration buffer and loaded with fura-2 AM (Biomol GmbH, Hamburg, Germany) at a final concentration of 3.3 µM in cell equilibration buffer for 45 minutes at room temperature.

Subsequently, cells were washed 2x with cell equilibration buffer and stored in equilibration buffer for additional 30 minutes prior to the measurements. Paxilline or iberiotoxin treatment of the cells was performed 12 hours prior to the measurements and both inhibitors remained present during the fura-2 loading procedure and the measurement at concentrations of 5 µM and 30 nM, respectively. Imaging experiments were either performed on the Zeiss AXIO Observer Z1 or the Zeiss Axiovert 200m microscope (Carl Zeiss AG) in physiologic buffer using alternate excitations at 340 nm and 380 nm. Emissions were captured at roughly 514 nm (Zeiss AXIO Observer Z1) or 525 nm (Zeiss Axiovert 200m). To evoke intracellular Ca²⁺ signals, cells were perfused with physiologic buffer containing ATP (Carl Roth GmbH) at a final concentration of 100 µM.

Genetically encoded sensor-based measurements

Cells were grown on 1.5 H 30 mm circular glass coverslips (Paul Marienfeld GmbH) for perfusion-based experiments. Cells were taken from the incubator, cell culture medium was removed, and cells were equilibrated for at least 30 minutes in cell equilibration buffer in ambient environment. Sensor plasmids used in this study comprised the following: D1ER (Palmer et al., 2004) (Addgene plasmid #36325) for measurement of [Ca²⁺]_ER and 4mtD3cpV (Palmer et al., 2006) (Addgene plasmid #36324) for measurement of [Ca²⁺]_mito were a gift from Amy Palmer & Roger Tsien, mtAT1.03 (Imamura et al., 2009) for measurement of [ATP]_mito was a gift from Hiromi Imamura, Laconic (San Martín et al., 2013) (Addgene plasmid #44238) for measurement of [LAC]_cyto and assessment of the Warburg index was a gift from Luis Felipe Barros, and mito-Hyper3 (Bilan et al., 2013) was a gift from Markus Waldeck-Weiermair. Experiments were either performed on the AXIO Observer Z1 or the Axiovert 200m microscope (Carl Zeiss AG). For experiments where paxilline (5 µM) or iberiotoxin (30 nM) were used, these compounds were added with the media change prior to the transfection with the PolyJet transfection reagent (SignaGen Laboratories) approximately 12 hours prior to the experiments. The compounds remained present throughout the experiment. FRET-based sensors were excited at 427/10 nm and emissions were collected simultaneously at roughly 475 and 543 nm. mito-Hyper3 was excited at 427/10 and 510/10 nm.

Extracellular flux analysis

Assessment of extracellular acidification rate (ECAR) and oxygen consumption rate (OCR) was performed using a Seahorse XFe24 analyzer (Agilent, Santa Clara, USA). The day before the assay, Seahorse XF24 cell culture microplates (Agilent) were coated with 0.5 mg mL^-1 poly-L-lysine (Sigma Aldrich Chemie GmbH) for 30 minutes at 37°C, washed 2x with PBS followed by cell seeding of 50,000 MMTV-PyMT WT, 50,000 MMTV-PyMT BK-KO or 100,000 MDA-MB-453 cells per well of the 24-well plate in a final volume of 250 µL, either in the presence or absence of 5 µM paxilline, 30 nM iberiotoxin or an equivalent amount of DMSO. 4 wells contained medium without cells and served as blank. Cells were cultivated over-night at 37°C and 5% CO₂ in a humidified incubator. The Seahorse XFe24 sensor cartridges were equilibrated over-night at 37°C in Seahorse XF Calibrant solution according to manufacturer’s instructions. The next day, cells were washed using Seahorse XF DMEM medium, pH 7.4 additionally supplemented with 5.5 mM glucose, 2 mM GlutaMAX and 1 mM sodium pyruvate (Thermo Fisher Scientific), with or without paxilline, iberiotoxin or DMSO according to manufacturer’s instructions. With the last washing step, a final volume of 500 µL was adjusted per well and cells were incubated at 37°C in the absence of CO₂ for 30 minutes. Meanwhile the Seahorse XFe24 sensor cartridge was loaded with the following compounds: 55 µL of 20 µM Oligomycin-A, 62 µL of 3 µM FCCP and 69 µL of 25 µM Antimycin-A (Santa Cruz Biotechnology), yielding final concentrations of 2 µM, 300 nM and 2.5 µM upon injection, respectively. For analysis, ECAR and OCR rates were normalized for the protein concentration (µg) per well, which was assessed using the Pierce BCA protein assay kit (Thermo Fisher Scientific) according to manufacturer’s instructions. Absorbance was measured at 540 nm using a TECAN multiplate reader (TECAN Group Ltd., Männedorf, Switzerland) and concentrations were assessed using a calibration curve.

LC-MS-based metabolomics

Formic acid, acetic acid, acetonitrile and methanol of Ultra LC-MS grade were supplied by Carl Roth (Karlsruhe, Germany). Ammonium hydroxide solution (Suprapur® quality 28.0 - 30.0% NH₃ basis) was purchased from Sigma-Aldrich (Merck, Taufkirchen, Germany). Deionized water was purified by a Purelab ultrapurification system (ELGA LabWater, Celle, Germany). Uniformly (U-) ¹³C-labeled yeast extract of more than 2 × 109 Pichia pastoris cells (∼15 mg; strain CBS 7435) was obtained from ISOtopic Solutions (Vienna, Austria). All standards used were purchased from Sigma-Aldrich Chemie GmbH. Stock solutions of the individual calibrants were prepared at concentrations of 1 mg mL^-1 and used for further dilution. The individual stocks were stored at -80°C until use.

Targeted LC-MS analysis was performed using an Agilent 1290 Infinity II series UHPLC system from Agilent Technologies (Waldbronn, Germany) equipped with a binary pump, autosampler, thermostated column compartment and a QTrap 4500 mass spectrometer with a TurboIonSpray Source from SCIEX (Ontario, Canada). The samples were filled into homogenization tubes. Internal standards were added prior to sample preparation. For the extraction of the analytes, 1 mL of the ice-cold extraction solvent (50% methanol and 50% water) and 0.15 g of zirconia/glass beads were added to the cell pellets (which were slowly thawed on ice). The samples were homogenized (6,800 rpm, 1 min at 4 °C, 10 × 10 s, pause 30 s) with Cryolys Evolution using dry ice cooling (Bertin Technologies, France). The samples were then spun down for 5 min (16,100xg at 4 °C). The supernatant was carefully removed, transferred into fresh tubes and evaporated to dryness overnight under nitrogen using a high-performance evaporator (Genevac EZ-2) (Genevac, Ipswich, UK). The dry residue of the extract was reconstituted in 10 µL water and 90 µL acetonitrile, followed by 3 cycles of vortexing and sonication (30 seconds each). The samples were centrifuged at 18928xg for 5 min and the supernatant was used for further analysis.

Chromatographic separation was performed on a Waters (Eschborn, Germany) Premier BEH Amide column (150 x 2.1 mm, 1.7 µm). For metabolite analysis in ESI⁺ mode, mobile phases A and B were adjusted to a pH of 3.5 with formic acid and consisted of 20 mM ammonium formate in water and acetonitrile, respectively. In ESI-, the chromatographic conditions differed. Mobile phase A and B were adjusted to a pH of 7.5 with acetic acid and consisted of 20 mM ammonium acetate in water and acetonitrile, respectively. The gradient elution profile was the same for both positive and negative ionization modes (0.0 min, 100% B; 13 min 70% B; 15 min 70% B; 15.1 min 100% B; 20 min 100% B) and was carried out at a flow rate of 0.25 mL min^-1 and a constant column temperature of 35 °C. The injection volume was 5 µL. The autosampler was kept at 4°C. Ion source parameters were as follows: nebulizer gas (GS1, zero grade air) 50 psi, heater gas (GS2, zero grade air) 30 psi, curtain gas (CUR, nitrogen) 30 psi, source temperature (TEM) 450 °C, ion source voltage +5,500 V (positive mode) and - 4,500 V (negative mode). Due to the large number of transitions monitored simultaneously, the Scheduled-MRM function was enabled. A window of 30 seconds was set around the designated metabolite-specific retention time and the total cycle time was 1 s. Blank solvents (mobile phase A and B) followed by QCs were injected in the beginning of the chromatographic batch to ensure proper column and system equilibration.

Plasma- and mitochondrial membrane potential measurements

DY_PM was determined using Bis-(1,3-dibutylbarbituric acid)trimethine oxonol (Dibac4(3)) (Thermo Fisher Scientific). Cells were cultivated on 30 mm glass coverslips. On the day of analysis, cells were equilibrated in cell equilibration buffer containing Dibac4(3) at a concentration of 0.25 µg mL^-1 for 30 minutes and subsequently analyzed by fluorescence microscopy using 485/20 excitation at the Zeiss Axiovert 200m microscope. Emission was captured at roughly 525 nm.

DY_mito was assessed using Tetramethylrhodamin-Methylester (TMRM) (Thermo Fisher Scientific). After cell cultivation on 30 mm glass coverslips in the presence or absence of paxilline, iberiotoxin or DMSO for 12 hours, the cell culture medium was replaced with cell equilibration buffer containing 200 nM TMRM and cells were incubated in ambient environment for 30 minutes. Subsequently, cells were washed with physiologic buffer containing 2 mM or 25 mM glucose and 200 nM TMRM, and cells were equilibrated for further 30 minutes. The glass coverslips were transferred to the PC30 perfusion chamber and experiments were started using the gravity-based perfusion system (NGFI GmbH). Paxilline, iberiotoxin or DMSO (control) and 200 nM TMRM remained present throughout the experiment. Mitochondrial depolarization was induced by the perfusion of a buffer containing 200 nM FCCP. For analysis, the fluorescence emission at ≥600 nm upon excitation at 575/15 nm of a region of interest (ROI) above mitochondria was divided by a ROI in the nucleus (mitochondria free area) and the ratio was plotted over-time, or basal values were given.

2-NBDG based glucose uptake measurements

Glucose uptake was assessed using 2-(N-(7-Nitrobenz-2-oxa-1,3-diazol-4-yl)Amino)-2-Deoxyglucose (2- NBDG) (Biomol GmbH). Therefore, cells were seeded on 30 mm glass coverslips (Marienfeld GmbH) in 6-well plates (Corning) and cultivated over-night at 37°C and 5% CO₂ in a humidified incubator. The next day, cells were taken from the incubator, cell culture medium was replaced for cell equilibration buffer and cells were equilibrated for 30 minutes at ambient environment. Subsequently, cells were washed 3x with glucose free physiologic buffer, and glucose free physiologic buffer containing 100 µM 2-NBDG, with or without paxilline and 200 nM FCCP or an equivalent amount of DMSO, was added to the cells, followed by incubation at 37°C for 30 minutes. Next, cells were washed 3x with glucose free physiologic buffer to remove any residual 2-NBDG and were analyzed by fluorescence microscopy using 485/20 excitation light. Emission was captured at roughly 525 nm (Zeiss Axiovert 200m).

Mitochondrial single-channel patch-clamp measurements

For mitoplast electrophysiology, mitochondria were isolated from BCC grown to confluency (>90%). Adherent cells were washed 2x with PBS, scraped from the dish, collected in a tube and centrifuged at 400xg for 5 minutes. Subsequently, the cell pellet was resuspended in preparation solution, followed by homogenization using a glass-glass homogenizer. Next, the homogenate was centrifuged at 9,200xg for 10 minutes. The resulting pellet was resuspended in preparation solution and centrifuged at 780xg for 10 minutes. The supernatant was collected, followed by centrifugation at 9,200xg for 10 minutes and resuspension of the mitochondrial fraction in the mitochondrial storage buffer. All procedures were performed at 4°C.

An osmotic swelling procedure was used for the preparation of mitoplasts from the isolated mitochondria. Therefore, mitochondria were added to the hypotonic buffer for ∼1 minute to induce swelling and breakage of the outer mitochondrial membrane. Subsequently, isotonicity of the solution was restored by the addition of hypertonic buffer at a dilution of 1:5.

Patch-clamp experiments on mitoplasts were performed as previously described (Bednarczyk et al., 2013). The experiments were carried out in mitoplast-attached single-channel mode using borosilicate glass pipettes with a mean resistance of 10-15 MΩ. The patch-clamp glass pipette was filled with mitochondrial storage buffer. This isotonic solution was used as a control solution for all experiments. The size of the pipettes and the formation of the gigaseal were monitored by measuring electrical resistance. Connections were made with Ag/AgCl electrodes and an agar salt bridge (3 M KCl) for the ground electrode. The current was recorded using an Axopatch 200B patch-clamp amplifier (Molecular Devices, California, USA). To apply substances, a self- made perfusion system containing a holder with a glass pipe, a peristaltic pump, and Teflon tubing was used. All channel modulators were added as dilutions in the isotonic solution containing 100 μM CaCl₂.

The presented single-channel current-time recordings are representative for the most frequently observed conductances under the given conditions. The conductance was calculated from the current-voltage relationship. The probability of channel openings and the current amplitude were determined using the Single-Channel Search mode of the Clampfit 10.7 software (Molecular Devices).

Western blotting

Mitochondria for western blotting were prepared as described earlier(Pallotti and Lenaz, 2007) with some modifications. Cells were washed and collected in PBS and centrifuged at 500xg for 10 min. The pellet was frozen in liquid nitrogen and stored at -80°C. The next day, pellet was resuspended in ice-cold isolation buffer containing (in mM): 210 mannitol, 70 sucrose, 1 PMSF, 5 HEPES and bovine serum albumin (2.5 mg mL^-1), pH 7.2. Digitonin was added to a final concentration of 0.02–0.04% for additional membrane permeabilization. After 1 min of incubation, digitonin was diluted with isolation buffer and the cells were centrifuged at 3,000xg for 5 min at 4°C. The pellet was resuspended in ice-cold isolation buffer. The cells were homogenized using a glass/glass homogenizer and centrifuged at 1,000xg for 5 min at 4°C followed by another homogenization. The homogenates were centrifuged at 1,000xg for 5 min at 4°C to remove cell remnants. The supernatants were collected and centrifuged for 60 min at 10,000xg, at 4°C. Next, the pellet was resuspended in the isolation buffer without BSA and centrifuged at 10,000xg for 30 min at 4°C, followed by resuspension in isolation buffer without BSA and centrifugation at 10,000xg for 15 min at 4°C. This step was repeated once more. The pellet containing crude mitochondria was resuspended in 0.8 mL of 15% Percoll (in isolation buffer without BSA) and layered on top of a Percoll step gradient (23% and 40% Percoll layers). The suspension was centrifuged at 30,000xg for 30 min at 4°C. Mitochondrial fraction located between the 23% and 40% Percoll layer was collected, diluted with isolation buffer without BSA and centrifuged at 10,000xg for 15 min at 4°C. The final mitochondrial pellet was resuspended in storage buffer containing 500 mM sucrose and 5 mM HEPES, pH 7.2. Each mitochondrial isolation was performed using between 15-30 x 10⁶ cells.

A given amount of sample solubilized in Laemmli buffer (Bio-Rad) was separated by 10% Tris-tricine- SDS-PAGE and transferred onto polyvinylidene difluoride (PVDF) membranes (Bio-Rad). After protein transfer, the membranes were blocked with 10% nonfat dry milk solution in Tris-buffered saline with 0.2% Tween 20 and exposed to one of the following antibodies: anti-BKα antibody (NeuroMabs, USA, clone L6/60, diluted 1:200), anti-COXIV (Cell Signaling Technology, Leiden, Netherlands, no. 4844, 1:1,000), or anti-alpha 1 sodium potassium ATPase (Abcam, Berlin, Germany, no. ab7671, 1:1,000). This was followed by incubation with a secondary anti-rabbit (Thermo Fisher Scientific, no. 31460) or anti-mouse antibody (Thermo Fisher Scientific, no. SA1-100) coupled to horseradish peroxidase. The blots were developed using enhanced chemiluminescence solution (GE Healthcare). To estimate the molecular weight of the analyzed proteins PageRuler Prestained Protein Ladder (Thermo Fisher Scientific) was used.

qPCR analysis

mRNA was isolated from 6-well plates showing a cell confluency of ∼90% using the Monarch Total RNA Miniprep kit (NEB). siRNA transfection was performed 48 hours prior to RNA isolation using the Lipofectamine 2000 transfection reagent (Thermo Fisher Scientific) according to manufacturer’s instructions. qPCR reactions were performed using the GoTaq 1-Step RT-qPCR System (Promega GmbH, Walldorf, Germany). 100 ng of isolated mRNA were used per reaction. Primers were designed to span exon – exon junctions and to recognize both, human and murine BK_Ca sequences. Different primer pairs were used for human and murine b-tubulin. Primer and siRNA sequences are listed in Supplementary Table 1 and Supplementary Table 2. Primers were purchased from Thermo Fisher Scientific, siRNAs were purchased from Microsynth AG. qPCR reactions were run in a CFX connect real-time PCR instrument (Bio-Rad Laboratories GmbH, Feldkirchen, Germany). Sample (Ct) values were normalized to Ct’s of housekeeper gene (b-Tubulin) and calculated as 2-Ct (normalized).

Proliferation assays

Proliferation assays were performed using 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) (Thermo Fisher Scientific) assay. For the assay, 2.500 MMTV-PyMT WT and BK-KO cells or 7.500 MDA-MB-453 cells were seeded per well in flat bottom 96-well cell culture plates (Corning) in their respective cell culture medium. Per run, condition, and time-point, 7 technical replicates were performed. One well served as a blank per time-point and condition and did not contain cells. The next day, medium was exchanged for fresh cell culture medium either containing 5 µM paxilline, 30 nM iberiotoxin or an equivalent amount of DMSO as a control and the MTT assay was started for time-point 24 hours directly thereafter. For each time-point (24, 48, 72 and 96 hours), MTT reagent was added with a medium change (10 µL + 100 µL / well) at a final concentration of 455 µg mL^-1 and cells were incubated for 4 hours. Subsequently, 85 µL of medium were removed from each well, 85 µL of DMSO (Carl Roth GmbH) was added and wells were mixed by pipetting thoroughly. 85 µL of this mixture were transferred to a new 96-well plate (Corning) and absorbance was measured at 540 nm using a TECAN microplate reader (TECAN Group Ltd.). Absorbances at the different time-points (48, 72 and 96 hours) were normalized to the absorbances after 24 hours.

Colony formation assays

For colony formation assays, cell suspensions containing 400 cells mL^-1 of MMTV-PyMT WT or BK-KO cells were prepared. 1 mL of this suspension was spread drop-per-drop per well of a 6-well cell culture plate (Corning) already containing 1 mL of cell culture medium using 1 mL syringes and 23G needles. Culture plates were gently moved left/right and back/forth to equally distribute the cells across the well, plates were kept at room temperature for 20 minutes to ensure uniform cell settling, and cells were subsequently cultured over-night at 37°C and 5% CO₂ in a humidified incubator. The next day, medium was exchanged for 2 mL of fresh cell culture medium, and cells were kept in the presence or absence of O₂ at 37°C and 5% CO₂ in humidified incubators for 7 days. After 7 days, cells were washed carefully 2x with PBS and fixed for 30 minutes on ice using 2% PFA in PBS. Meanwhile, a solution of crystal violet (0.01% w/v) was prepared in ddH₂O, cells were washed 2x with PBS and incubated in the crystal violet solution (Sigma Aldrich Chemie GmbH) for 60 minutes after fixation, followed by excessive washing with ddH₂O until a clear background was obtained. Plates were dried at room temperature for at least 12 hours before images were captured using an Amersham Imager 600 (GE Healthcare UK Limited, Buckinghamshire, UK). Analysis was performed using ImageJ.

Breast cancer patient study

A subset of 551 primary tumor specimens with available archival tumor blocks were obtained from a prospective, observational multicenter adjuvant endocrine treatment study of 1286 post-menopausal HR- positive breast cancer patients recruited between 2005 and 2011 (German Clinical Trial Register DRKS 00000605, “IKP211” study). Study inclusion and exclusion criteria have been previously described (Schroth et al., 2020). Patients had received standard endocrine treatment, i.e. tamoxifen, aromatase inhibitors, or switch regimens between both. Ethics approval was obtained from the Medical Faculty of the University of Tuebingen and the local ethics committees of all participating centers in Germany. Informed patient consent was obtained from all participants as required by institutional review boards and research ethics committees. All patient data were de-identified prior inclusion in this study.

Nanostring nCounter gene expression analysis

Gene expression analysis was performed according to manufacturer’s protocol. In brief, total RNA from human primary tumors of the IKP211 study (Quick-DNA/RNA FFPE, ZymoResearch, Freiburg, Germany) was extracted from 10 µm sections of FFPE tissues with at least 20% tumor cell content. Samples were enriched for tumor tissue by microdissection. A total of 250 ng RNA was subsequently hybridized with target-specific capture and color-coded reporter probe sets in a pre-warmed thermal cycler at 65°C for 20 h. In the post-hybridization process, the total volume was increased to 32 µl with RNase- free water. Fluorescence count measurements were immediately conducted in the Nanostring nCounter System. Data were analyzed using nSolver 4.0 and normalized to housekeeping genes. ABCF1, NRDE2, POLR2A, PUM1 and SF3A1 served as housekeepers. Probes used for Nanostring nCounter gene expression analysis are listed in Supplementary Table 3.

Statistical analysis

Statistical analysis was performed using Prism8 software (GraphPad Software, Boston, USA). All data were tested for normal distribution using D’Agostino and Pearson omnibus normality test. A two-tailed Unpaired t-test or Welch’s t-test were used for statistical comparison of normally distributed data, depending on whether the variances of the populations were comparable or significantly different as tested by an F-test. A two-tailed Mann-Whitney test was used for pairwise comparison of non-normally distributed data. Comparison of >2 data sets was done either using One-way ANOVA followed by Tukey’s multiple comparison (MC) test or Brown-Forsythe and Welch ANOVA test followed by Games-Howell’s MC test for normally distributed data, depending on whether the variances were comparable or significantly different. A Kruskal-Wallis test followed by Dunn’s MC test was performed if data were not normally distributes. The statistical tests used are indicated in the figure legends. p-values of ≤0.05 were considered as significant, where * or #p≤0.05, ** or †p≤0.01 and *** or ‡p≤0.001. No priory sample size estimation was performed.

Acknowledgements

This work was funded by the Deutsche Forschungsgemeinschaft (DFG) with individual grants to RL and LM. RL is a member of the GRK2381: "cGMP: From Bedside to Bench", DFG grant number 335549539. SM, WS, MS and RL acknowledge financial support from the ICEPHA Graduate Program “Membrane- associated Drug Targets in Personalized Cancer Medicine”. HB is a fellow of the Austrian Science Fund (FWF) funded Erwin-Schrödinger-Program, project number J-4457. SB acknowledges financial support from the Fritz Thyssen Stiftung. The authors thank Clement Kabagema-Bilan, Michael Glaser, Monika Zochowska and Antoni Wrzosek for excellent technical support and Peter Ruth for valuable discussions. This work was partially supported by the Nencki Institute of Experimental Biology, the Polish National Science Centre grant no. 2019/34/A/NZ1/00352 (AS), and by the FWF, project number I-3716 (RM).

Author contributions

HB and RL initiated and designed this study ; HB, SM, PK, BK, SB, JJ, KS, WS and PB performed experiments and were involved in data curation; HB, SM, PK, BK, SB, JJ, KS and WS assisted in formal analysis; HB, SM, SB and PB visualized data; HB, SB, WS, LM, MS, PB, AS and RL acquired financial support; HB, SM, PK, BK, SB, KS, WS, ML and PB designed the methodology; HB and RL performed project administration; PK, WS, LM, MS, ALB, RM, ML, PB, AS and RL provided resources; RL supervised the project and the personnel; HB and RL wrote the original draft. All authors critically reviewed the manuscript and stated comments.

Data availability

Raw data for the graphs can be found in source data files.

Declaration of interests

The authors declare no competing interests.

Supplementary Information

Supplementary Figure 1:

Supplementary Figure 2:

Supplementary Figure 3:

Supplementary Figure 4:

Supplementary Figure 5:

Supplementary Figure 6:

Supplementary Table 1:

Supplementary Table 2:

Supplementary Table 3: