Tamoxifen metabolites acting via BK orchestrate the dynamics of K and Ca in breast cancer cells.

Results

BKCa is expressed in clinically relevant BC biopsies and functionally relevant in human BC cells
Bulk gene expression analysis of 551 human derived BC specimen revealed different K+ channel transcript levels in ER+ BC. Consistent with a previous report (7), KCNMA1 mRNA was significantly higher expressed in these BC specimens compared to other cancer-associated K+ channels (KCNA3, KCNH1, KCNH2, KCNK5, KCNK9, KCNN3, and KCNN4). Interestingly, KCNMA1-derived transcripts were also present in all individual samples analyzed (Fig. 1A). Transcript levels of KCNMA1 in the ER+ human BC cell lines MCF-7 (hBKCalow) and CAMA-1 (hBKCamedium) and the ER- human BC cell line MDA-MB-453 (hBKCahigh) confirmed highest expression level of BKCa channels in hBKCahigh BC cells (Fig. 1B). An investigation of BKCa channel subunits confirmed β-, γ-, and Lingo1 transcripts in the BC cell lines, which, however, showed significantly lower expression levels in hBKCahigh compared to the pore-forming α-subunit of BKCa (Fig. S1A).
To verify the functionality of BKCa channels in these human BC cell lines, whole-cell patch-clamp recordings were performed to obtain current density-voltage relationships in hBKCalow, hBKCamedium and hBKCahigh BC cells. At +120 mV and physiological intracellular Ca2+ concentrations ([Ca2+]i), K+ outward currents of hBKCahigh BC cells were significantly increased over hBKCamedium and hBKCalow BC cells (Fig. 1C). Representative current-over-time traces of hBKCalow, hBKCamedium and hBKCahigh (Fig. 1, D–I left) were normalized to cell size in current density calculations (Fig. 1, D–I right). K+ outward currents that were reliably detected in response to depolarization steps in hBKCalow cells were insensitive to the specific BKCa blocker paxilline (PAX) (Fig. 1D). In contrast, peak K+ outward currents were significantly inhibited by PAX treatment in hBKCamedium and hBKCahigh cells (Fig. 1, E and F), verifying previously reported patch-clamp recordings on hBKCa localized in the PM of MCF-7 and MDA-MB-453 cells (7, 9).
Additionally, immunofluorescence (IF) staining shows a higher abundance of BKCa in hBKCahigh compared to hBKCalow BC cells (Fig. S1B), confirming whole-cell patch-clamp experiments. Importantly, phalloidin staining indicates that the cell size of hBKCalow and hBKCahigh BC cells varies (Fig. S1B), explaining the differences in the conductance and the resulting current densities over time in the whole-cell patch-clamp recordings.
To further evaluate whether hBKCahigh activity is sensitive to the BKCa specific activator NS11021, the compound was added to the bath solution (Fig. 1G), resulting in higher K+ outward currents during patch-clamp recordings. Further, we observed that TAM+M (4-OH-TAM and endoxifen) elevated K+ outward currents in hBKCahigh cells massively, and even more than NS11021 (Fig. 1, H and I), suggesting that TAM+M impacts the ion homeostasis of BC cells by modulating the subcellular K+ homeostasis via BKCa channel activation. Importantly, we observed that the majority of successfully measured hBKCahigh cells indeed responded to 4-OH-TAM (72.73%) and endoxifen (61.54%) with increased K+ currents (Fig. S1C), which is in line with the variability of BKCa protein levels in individual hBKCahigh cells (Fig. S1B). Potential reasons for this “patchy” BKCa expression profile were not investigated, but might be explained by the previously demonstrated cell cycle-dependent variation of K+ channel expression in non-cancerous and cancer cells (6, 38, 39), as our IF and patch-clamp experiments were carried out in non-synchronized cells.
Together, these findings confirm the presence of BKCa channels in ER+ human BC tissue and hBKCalow/medium/high BC cells. In addition, functional BKCa channel expression in hBKCahigh correlated with TAM+M-mediated effects on K+ outward currents.

TAM+M modulate K+ dynamics through BKCa independently of ER
Based on the observed effects on whole-cell currents by TAM+M (Fig. 1), we investigated K+ homeostasis utilizing lc-LysM GEPII1.0, a well-established cytosolic FRET-based K+ sensor (7, 40). Next to hBKCalow, hBKCamedium and hBKCahigh BC cell models, we additionally studied SKBr-3 (ER-) as hBKCalow cell line (Fig. 2). The latter was selected based on the RNAseq data in Cancer Cell Line Encyclopedia (CCLE) library and included in our analysis to address how the absence of ER (and low BKCa) in human BC cells affect the response to TAM+M.
Alterations in cytosolic K+ concentration ([K+]i) were first induced by the BKCa channel opener NS11021 in MCF-7 hBKCalow (ER+), CAMA-1 hBKCamedium (ER+) and MDA-MB-453 hBKCahigh (ER-) (Fig. 2, A and B). BKCa abundance of these cells correlated with the [K+]i loss, which was significantly different in hBKCahigh (ER-) compared to the [K+]i response seen in hBKCamedium (ER+) and hBKCalow (ER+), strongly suggesting that these [K+]i alterations occurred irrespective of the ER status of the cells (Fig. 2B). Importantly, in hBKCalow cells expressing RFP-tagged BKCa (+BKCa-RFPrescue) we consistently observed a massive drop in [K+]i compared to respective control hBKCalow (ER+) cells, regardless of whether RFP (+RFP) was present or not (Fig. 2A). In addition, we made use of a previously mapped panel of patient-derived BKCa loss- and gain-of-function mutations (41). Interestingly, one of the loss-of-function channel variants carries the G354S mutation (41), a site residing within the selectivity filter of BKCa, which is highly conserved among K+ selective ion channels (42). This loss-of-function mutant results in a “pore-death” variant of BKCa with dramatically reduced K+ conductance and a lower selectivity to K+, which causes symptoms like ataxia, dyskinesias and cognitive impairment in a young patient carrying the BKCa G354S mutation (43). We induced the G354S mutation in our established BKCa-RFP construct by site-directed mutagenesis. Importantly, the expression of BKCaG354S-RFP (+BKCaG354S) in hBKCalow (ER+) BC cells demonstrated levels comparable to the empty vector control (+RFP) and “wild type” BKCa-RFPrescue (Fig. S1D). FRET-based K+ recordings evoked by NS11021 in hBKCalow (ER+) BC cells expressing BKCaG354S were, however, significantly impaired compared to functional “wild type” BKCa-RFPrescue channels (Fig. 2A). Employing a related BKCa channel opener, NS1619, showed similar effects on [K+]i in hBKCalow (ER+) and hBKCahigh (ER-) BC cells (Fig. S1E).
To test if 4-OH-TAM and endoxifen impact the cell’s K+ dynamics in a BKCa- and/or concentration-dependent manner, we analyzed [K+]i in hBKCalow (ER+) and hBKCahigh (ER-) cell lines. Indeed, 4-OH-TAM induced a significantly higher drop of [K+]i in hBKCahigh (ER-) compared to hBKCalow (ER+) cells, an effect that was consistent for all concentrations tested (Figs. 2C and S1F). Furthermore, administration of endoxifen resulted in a concentration-dependent K+ loss in hBKCahigh (ER-) as indicated by the decreasing FRET ratio signal. At the maximal concentration tested, differences in [K+]i in hBKCalow (ER+) and hBKCahigh (ER-) reached the level of significance (Fig. 2D). In addition, we compared the FRET ratio signals recorded from SKBR-3 hBKCalow (ER-) and hBKCamedium (ER+) to hBKCalow (ER+) and hBKCahigh (ER-) BC cells treated with endoxifen. This analysis validated that the TAM+M-mediated K+ loss correlated with the BKCa expression level but occurred irrespective of the cell’s ER status (Figs. 2E and S1G). Subsequently, we excluded the potential interference caused by growth factors and steroid hormones. Therefore, cells were cultured in charcoal-stripped bovine serum (CCS), a reagent leading to significant hormone deprivation, before monitoring live-cell K+ changes by FRET imaging. In hBKCalow and hBKCahigh BC cells, we reproduced previous findings on TAM+M-induced [K+]i alterations, suggesting that the response to TAM+M is independent of media supplements, i.e. of endocrine pathways (Fig. S2, A and B).
To more directly evaluate whether steroid hormones influence the K+ homeostasis in a manner comparable to the anti-estrogenic TAM+M, we investigated changes in the K+ FRET ratio signal in response to E2. Pro-proliferative effects of E2 on human BC cells and stimulatory effects on BKCa channels have been reported (9, 18) but strikingly, at least in non-cancerous tissue, may exclusively depend on BKCa’s auxiliary subunit β1 (21, 22). In the observed timeframe, however, live-cell K+ imaging revealed only minor effects of E2 on the K+ balance in hBKCalow and hBKCahigh, and no differences between both BC cell models, suggesting E2 is linked to BKCa by a different mode of action (Fig. 2F).
To additionally validate and broaden our data on TAM+M-mediated [K+]i, endoxifen was administered to hBKCalow-expressing BKCa-RFPrescue or loss-of-function mutant BKCaG354S cells or cells solely expressing RFP as a control. Compared to hBKCalow, BKCa-RFPrescue expressing BC cells showed a massive decrease of [K+]i, whereas cells expressing only RFP did not exhibit a drop in [K+]i, i.e. FRET signals remained at the level of hBKCalow (Fig. 2G and S2C). “Pore-death” BKCaG354S expression in hBKCalow prevented the endoxifen-induced [K+]i loss compared to BKCa-RFPrescue channels (Fig. 2G), supporting the hypothesis of a TAM+M-promoted BKCa activity.
Finally, loss of [K+]i should lead to an increased extracellular K+ concentration ([K+]ex). To examine [K+]ex, we applied recombinant purified GEPII1.0 as previously described (40, 44). Gramicidin, a K+ ionophore, was used to induce maximal K+ release from cells (Fig. S2D). Also, we observed a clear trend for 4-OH-TAM and a significant increase of [K+]ex in hBKCahigh BC cells treated with endoxifen (Fig. S2E), confirming that the decrease in [K+]i ultimately results in an efflux of K+ to the extracellular space.
In sum, TAM+M treatment initiates a cascade causing a drop of [K+]i and consequently elevated [K+]ex levels. The K+ alterations observed are closely related to the cell’s BKCa channel abundance but occur independently of the ER status of all BC cell lines studied.

TAM+M-mediated effects on K+ dynamics are abolished by genetic or pharmacological inactivation of endogenous BKCa channels
In a BC mouse model based on the mouse mammary tumor virus (MMTV) polyoma middle T antigen (PyMT), genetic ablation of the BKCa α-subunit showed beneficial outcomes regarding tumor-free and overall survival in comparison to respective BKCa-proficient control littermates. Interestingly, and contrary to expectations from clinical use of the anti-estrogenic compounds, TAM+M treatment promoted the proliferative behavior of isolated murine BC cells obtained from the MMTV-PyMT model only in the presence of BKCa (9). Based on these previous studies, murine MMTV-PyMTtg/+ WT (mBKCaWT) and BKKO (mBKCaKO) derived BC cells were isolated from primary tumors and the cytosol-targeted K+ sensor lc-LysM GEPII1.0 was expressed to study basal [K+]i and changes induced by TAM+M. In agreement with the live-cell imaging experiments in human BC cells, endoxifen and 4-OH-TAM induced a BKCa-dependent drop in [K+]i (Fig. 3, A and B and Fig. S2F). Further, K+ efflux in mBKCaWT was sensitive to PAX, resulting in [K+]i loss equalizing with the one observed in mBKCaKO (Fig. 3A). Accordingly, TAM+M-induced BKCa-mediated [K+]i changes in hBKCahigh (ER-) (hereafter referred to as hBKCahigh) were sensitive to PAX, whereas hBKCalow (ER+) (hereafter referred to as hBKCalow) did not respond to pharmacological BKCa channel inhibition (Fig. 3, C and D and Fig. S2G).
These findings support the notion that murine and human BKCa channels respond similarly to TAM+M with respect to the effect of these compounds on the cell’s K+ balance. Further, genetic and pharmacological inhibition of endogenous BKCa prevented the observed alterations in [K+]i pointing towards a common TAM+M–BKCa axis.

TAM+M stimulate a rise in [Ca2+]i levels through BKCa
Multiple hallmarks of cancer are influenced by alterations in K+ channels expression and Ca2+ ion dynamics including cell cycle control, migration and adhesion, as well as apoptosis (45, 46). Previous work indicated that cancer cell Ca2+ signaling is affected by aberrant K+ channel expression (47). Mechanistically, K+ channel activation and the resulting K+ efflux provide a driving force for Ca2+ influx (8, 48). Therefore, we examined the potential influence of TAM+M on Ca2+ homeostasis using Fura-2, a ratiometric (F340/F380) Ca2+ indicator. 4-OH-TAM administration elevated [Ca2+]i levels in a BKCa-dependent manner. Accordingly, the rise in Fura-2 ratio was significantly lower in hBKCalow and hBKCamedium cells compared to hBKCahigh cells (Fig. 4A). Application of the alternative TAM+M endoxifen confirmed these results as the largest increase in [Ca2+]i was induced in hBKCahigh (Fig. 4B). We also treated hBKCalow (ER-) cells with endoxifen and observed a similar increase in the Ca2+ signal compared to hBKCalow (ER+) cells (Fig. S3A), indicating TAM+M act on Ca2+, independent from ER-mediated genomic events.
Next, we monitored [Ca2+]i in hBKCahigh cells exposed to endoxifen and PAX for pharmacological BKCa inhibition. Under these co-treatment conditions, the endoxifen-induced increase in Fura-2 ratio was significantly diminished, suggesting that TAM+M require functional BKCa channels to affect [Ca2+]i (Fig. 4, C and D). To examine the Ca2+ accumulation in response to TAM+M in more detail, we depleted intra- and extracellular Ca2+ stores by administration of BAPTA AM (BAPTA) and EGTA, a cell-permeant and -impermeant chelator of Ca2+, respectively, in Ca2+-free buffer. Under both conditions Fura-2 ratio signals were significantly decreased during TAM+M treatment (Fig. 4, C and D), suggesting that Ca2+ mobilization across the PM from extracellular sources and from intracellular stores contributes to the observed increase in [Ca2+]i. In BC cells, Ca2+ release from the endoplasmic reticulum is recognized by the Ca2+ store sensor protein STIM and results in SOCE, which is an important mechanism to raise [Ca2+]i. Ultimately, SOCE is triggering a Ca2+ entry pathway involving PM Ca2+ channels (f. ex. ORAI) to restore [Ca2+]i (49). High log2 transcript expression counts of STIM1, ORAI1 and ORAI3 were identified by gene expression analysis of human BC biopsies (Fig. S3B). In subsequent Fura-2 experiments, we used SKF-96365 (SKF), an inhibitor of SOCE (50) and TRP/C channels (51). We recognized an initial increase in [Ca2+]i levels (Fig. 4C) which was also previously observed to be induced in nasopharyngeal carcinoma cells by SKF-mediated release of Ca2+ in the endoplasmic reticulum (52) or, by directly affecting Ca2+ entry in human leukemic HL-60 and MDCK cells (53, 54) and/or by enhancing Na+/Ca2+ exchanger function in glioblastoma cells (55). After Fura-2 fluorescence emission ratio signals stabilized in response to SKF, endoxifen treatment was still effective in increasing [Ca2+]i, albeit not to the same extent as in the absence of SKF (Fig. 4, C and D). We conclude from the Fura-2 measurements that STIM-ORAI complexes, as well as PM Ca2+ channels (TRP/C), contribute to the BKCa-dependent cellular [K+]i loss in response to endoxifen.
To investigate the influence of TAM+M on the mobilization of Ca2+ from intracellular stores, hBKCahigh were stimulated with adenosin-5′-triphosphate (ATP) that acts through purinergic receptors to subsequently trigger the release of Ca2+ from endoplasmic reticulum via inositol-triphosphate (IP3) (56, 57, 58). ATP was applied following pre-treatment of the cells with endoxifen under basal conditions or in the presence of PAX, EGTA or SKF (Fig. 4E from left to right). Statistical analyses revealed significantly lower levels of ATP-induced release of Ca2+ to accumulate as [Ca2+]i under PAX and EGTA conditions (Fig. 4F). These results indicate TAM+M-induced BKCa-dependent K+ dynamics increase the driving force for Ca2+ uptake from both extracellular and intracellular stores. Because PAX-mediated inhibition of BKCa decreased the ATP-induced rise in [Ca2+]i, we conclude that Ca2+ released from intracellular stores is functionally linked to Ca2+-activated BKCa channel activity and thus part of a feed-forward mechanism.
Using Fura-2 as Ca2+ sensitive probe highlights that TAM+M, through BKCa activation, causes a K+ loss across the PM, which enhances the driving force for Ca2+ influx via PM localized store- and non-store- operated Ca2+ channels. It additionally appears that Ca2+ release from intracellular stores, i.e. the endoplasmic reticulum, contributes to the observed rise in [Ca2+]i. hBKCalow BC cells or pharmacological channel blockade reduced all TAM+M-induced changes in [Ca2+]i, proving that BKCa activity is essential to functionally connect TAM+M to Ca2+ signaling.

TAM+M-induced BKCa-activity mediates voltage-and Ca2+ alterations which sustain BKCa stimulation
As TAM+M affects BKCa-dependent K+ and Ca2+ dynamics, we next monitored [K+]i, again employing the FRET-based K+ probe lc-LysM GEPII1.0 while (i) chelating intracellular Ca2+ using BAPTA AM, (ii) under Ca2+-free conditions with extracellular EGTA, and after blockade of PM TRP/C and STIM/ORAI using SKF. Interestingly, none of these experimental conditions prevented induction of the TAM+M-mediated drop in [K+]i (Fig. 5A). Importantly, the initial drop in [K+]i induced by SKF did not lead to continuous BKCa activation in absence of endoxifen (Fig. S3C).
Hence, the systematic elimination of intra- and extracellular Ca2+ sources did not interfere with the TAM+M-induced BKCa-dependent K+ dynamics in hBKCahigh, indicating that the interaction of TAM+M with BKCa precedes any subsequent events on [Ca2+]i (Fig. 5B). To confirm these K+ live-cell imaging results, whole-cell patch-clamp experiments were performed under EGTA-buffered, Ca2+-free conditions to record current density-voltage relationships of hBKCahigh cells. In the absence of [Ca2+]ex, we again identified significantly elevated K+ outward currents in response to endoxifen (Fig. 5C), while the patch success rate decreased under EGTA-buffered conditions (Figs. S1C, 50%) compared to physiological [Ca2+]ex-buffered bath solution (Fig. 1I and S1C, 61.54%).
The observed changes in ion homeostasis and in particular the activation of BKCa should affect the membrane potential of BC cells. The fluorescent and slow-response voltage-sensitive probe DiBAC4(3) (59) was therefore used to detect changes in ΨPM (ΔΨPM) in hBKCahigh and hBKCalow BC cells in response to TAM+M. Endoxifen significantly increased DiBAC4(3) fluorescence intensity in hBKCahigh, but not in hBKCalow BC cells, suggesting ΨPM was more depolarized by TAM+M administration if BKCa was expressed (Fig. 5, D and F). Further, ΨPM was recorded during pharmacological inhibition of BKCa with PAX, intracellular depletion of Ca2+ by BAPTA AM and [Ca2+]ex-free conditions with EGTA (Fig. 5, E and F). Again, increased fluorescence intensity indicated a more depolarized ΨPM, but this TAM+M-induced effect in hBKCahigh cells was largely abrogated by pharmacological BKCa inhibition and considerably reduced in the presence of the Ca2+-chelators (Fig. 5F).
Combined, TAM+M influences [K+]i dynamics by BKCa activation, leading to depolarized ΔΨPM and mobilization of intracellular and extracellular Ca2+ sources to maintain BKCa stimulation.

TAM+M increases single-channel currents in a Ca2+-independent manner
As our data suggests that the TAM+M-BKCa axis includes mobilization of Ca2+ from multiple stores, we applied inside-out patch-clamp recordings to study TAM+M-mediated BKCa single channel behavior in the presence and absence of Ca2+. Accordingly, we used hBKCahigh BC cells to study the influence of endoxifen on channel open probability under Ca2+-free conditions or at physiological Ca2+ levels at either negative (-50 mV) or positive (+50 mV) voltages (Fig. 6, A and B). At both voltages and Ca2+ levels, endoxifen increased the number of channel openings, resulting in a significantly increased open probability (NPO). Single channel conductance yielded values of approximately 200 pS (data not shown), which is in the expected range of canonical BKCa channels. As the effect of endoxifen on NPO was independent of Ca2+, we conclude that TAM+M modulates BKCa function through a direct interaction and not indirectly by influencing Ca2+ handling mechanisms like transporters, channels or stores. Interestingly, the highest NPO in response to endoxifen was recorded at physiological Ca2+ levels and negative potentials (Fig. 6B), supporting our initial hypothesis that TAM+M-mediated BKCa activity results in a feed-forward mechanism, including Ca2+-dependent sustained BKCa channel activation.
In sum, these single-channel recordings confirmed TAM+M-promoted BKCa channel activity in a voltage-and Ca2+-independent manner. This suggests a direct interaction between TAM+M and BKCa channels, which results in an increased channel open probability.

Molecular modelling of TAM+M interaction with hBKCa
Docking of TAM+M against entire hBKCa protein structures (PDB Codes 6V38, 6V22 (60)), suggest two putative extracellular binding sites, one located close to the N-terminus (termed “WDF”, based on the amino acids forming major interactions with TAM+M) (Fig. 7, A and B and Fig. S3, D and E) and one posed between subunits close to the pore (termed “TSYF” based on the same principle) (Fig. 7, C and D and Fig. S3, D and E). Both pockets show high docking scores for 4-OH-TAM and endoxifen with virtually the same binding modes, indicating similar effects. The binding modes place the stilbene scaffolds in the hydrophobic part of the transmembrane region. They show π-π interactions with W22 and F252 for the “WDF” site, as well as with F131 and Y274 for the “TSYF” site. In the “WDF” site, the positively charged secondary or tertiary amines of endoxifen and 4-OH-TAM form a salt bridge to D261. Interestingly, in the “TSYF” site, the protonated tertiary amine of 4-OH-TAM appears to preferentially interact with the hydroxyl function of the side chain of S134, while the protonated secondary amine of endoxifen can, in principle, form two hydrogen bonds targeting the side chain hydroxyl group of S134 and the backbone carbonyl of F131. The phenolic hydroxyl functions of both metabolites are located in close proximity to the backbone of W23 (“WDF” site) or the backbone of T298 (“TSYF” site). In addition to the high docking scores in 6V38 (Fig. S3D), a homology model (Fig. S3F) showed the “WDF” binding site close to the recently described binding site of E2 in the structure of BKCa with β1-subunit in smooth muscle cells (21).
In sum our results reveal that TAM+M activates BKCa by potential binding sites in the α-subunit of BKCa to trigger changes in the [K+]i
vs [K+]ex balance that in turn allows elevated [Ca2+]i by facilitating Ca2+ influx across the PM and by Ca2+ release from the endoplasmic reticulum. Ca2+ accumulation and depolarized ΨPM subsequently maintain BKCa activity in terms of a feed-forward mechanism (Fig. 7E).

Discussion
We combined live-cell imaging with electrophysiological and pharmacological approaches in human and murine BC cell systems and obtained novel insights into the interaction of clinically relevant TAM+M with the “oncochannel” BKCa, a K+ channel that is suspected to confer a negative influence on endocrine treatment in BC patients (9).
Supported by molecular docking calculations, our analyses of TAM+M-mediated effects in human and murine MMTV-PyMT-based BC cell models expressing either no, low, medium or high levels of BKCa show a close correlation between the cells’ BKCa status and their response to endocrine agents.
We suggest that the TAM+M-BKCa axis in BC cells directly controls the intracellular K+ and Ca2+ balance, as the observed effects of endocrine agents on ion homeostasis occurred within seconds or minutes and were limited to BKCa-proficient cells, and were independent from genomic actions of the ER. In fact, BKCa but not ER status of the BC cells had an influence on whether TAM+M exposure caused a K+ drop in the cells and consequently a Ca2+ accumulation.
Our observed TAM+M-mediated activation of BKCa in BC is consistent with our previous findings that low concentrations of TAM+M promote rather than inhibit BC cell growth in the presence of functional BKCa channels (9). Accordingly, high KCNMA1 expression levels resulted in lower recurrence-free survival rates in ER+ BC patients (9). By quantifying K+ channel transcripts from human ER+ BC biopsies and human BC cells, we confirmed KCNMA1 expression (7) is aberrantly increased in bulk tumor samples and selected BC cell lines.
When we validated the functional expression of BKCa channels in our BC cell models, we observed a significant PAX-sensitive K+ current component in hBKCamedium and hBKCahigh cells that was not present in hBKCalow BC cells, thereby confirming previous reports of BKCa expression in hBKCalow/high cell models (7, 9). In contrast, pharmacological modulation of BKCa by the channel opener NS11021 led to an increase in current densities in whole-cell patch-clamp analysis and a drop in [K+]i, from which we conclude that human BC cells with varying BKCa channel levels represent a suitable model to study the putative effects of TAM+M on BKCa channel activity and ion homeostasis.
In this current work, we further established a correlation between the BKCa expression level, the reduction of [K+]i and the elevation in [Ca2+]i in response to TAM+M. Our BKCarescue experiments in hBKCalow BC cells confirmed TAM+M-mediated BKCa activity by restoring [K+]i dynamics. Additionally, we demonstrated that the “pore death” mutant BKCaG354S expressed in hBKCalow BC cells did not mediate TAM+M-provoked [K+]i dynamics. These defined amendments in BKCa expression and activity that we implemented either as BKCa-RFPrescue or BKCaG354S confirmed a direct interaction of TAM+M with the pore-forming α-subunit. Importantly, ER-independent actions of TAM are frequently reported, especially in the context of cancer cell proliferation (61, 62, 63), however, these have not been examined at the level of K+ and Ca2+ homeostasis.
In contrast to our previous observations where prolonged E2 exposure increased BC cell proliferation in a BKCa-dependent manner (9), our current work revealed that E2-mediated effects on K+ dynamics were independent of BC BKCa channels, at least during the few-minute time frame considered here. Thus, follow-up studies are required to address and explain these differences in the mode of action of E2 and TAM+M on BKCa. For the time being, we suggest that the BKCa-dependent pro-proliferative effects of E2 may be mediated by an alternative signaling pathway that is independent of any direct interaction between E2 and BKCa‘s α-subunit or ER. This view is in line with findings in non-cancerous cells, where the expression of auxiliary β1-subunits was essential for the activation of BKCa by E2 (18, 21, 22). It is also in line with findings in triple negative breast cancer cells (absence of ER, PR and HER2), where E2-dependent activation of G protein-coupled estrogen receptor 1 (GPER) stimulated a non-genomic pathway causing cell proliferation (64, 65). The role of BKCa in this context, however, was not specifically investigated.
Interestingly, TAM+M-induced [Ca2+]i dynamics were decreased by [Ca2+]i/ex-depleted conditions, TRP/C and SOCE inhibition and by BKCa blockade with PAX, giving rise to the view that BKCa-mediated K+ loss triggers Ca2+ elevation. To this end, we showed that the presence of [Ca2+]i/ex or intact SOCE is dispensable for the TAM+M-induced response of BKCa, as FRET recordings demonstrated an unaltered TAM+M-provoked K+ efflux, whole-cell patch-clamp experiments showed increased current densities even under Ca2+-free EGTA-buffered conditions, and the NPO of BKCa in inside-out patches was raised independent of Ca2+ stores or transporting mechanisms. It should be emphasized that physiological Ca2+ conditions contributed to the highest increase in NPO, confirming an amplified TAM+M effect on BKCa activity by Ca2+.
To address whether the TAM+M-BKCa axis plays a role for intracellular Ca2+ mobilization, we induced Ca2+ mobilization from the endoplasmic reticulum by ATP. The K+ loss induced by TAM+M caused an increase in [Ca2+]i under [Ca2+]ex-depleted conditions, probably due to an increase in Ca2+ driving force.
Hence, BKCa gating in BC cells is also linked to the homeostasis of this intracellular Ca2+ pool, and thus influences cellular signaling processes, as previously demonstrated for a wide range of different cell types (66). This is further corroborated by the PAX-induced BKCa inhibition, which impaired a significant part of the ATP-elicited [Ca2+]i transients after TAM+M pre-treatment (Fig. 4, E and F). These findings are in line with those previously reported by us, where PAX reduced the basal [Ca2+]i levels in MDA-MB-453 BC cells, while the mean peak amplitude of the [Ca2+]i signals induced by ATP remained unaltered (7).
We also considered that the TAM+M-BKCa axis would affect the ΨPM and thus voltage-dependent BKCa activation in the presence and absence of [Ca2+]ex/i. Indeed, we triggered a strong ΨPM depolarization with TAM+M that depended on the BKCa status and was sensitive to genetic or pharmacological inhibition of the channel. The measured effects on ΨPM were less prominent when [Ca2+]i or [Ca2+]ex were selectively depleted, indicating that ΨPM depolarization is a consequence of the (initial) TAM+M-BKCa interaction. Although this may be counterintuitive at first, as the BKCa-mediated K+ efflux should cause a greater membrane polarization rather than depolarization, we need to be aware of, that DiBAC4(3) is a slow-acting voltage-sensitive dye (59). We conclude that we have not been able to demonstrate the initial and rapid hyperpolarizing effect of TAM+M, as only the subsequent Ca2+ influx leading to membrane depolarization is detected by DiBAC4(3).
Finally, for the TAM+M binding sites suggested by molecular docking, additional homology modelling of the β1 subunit to the β4 subunit (in 6V22) revealed, that the extracellular “WDF” binding site would lie close to the amino acids W163 and F166 of the β1-subunit, which were found to be critical for the effect of E2 on hBKCa channel complexes (21). Both binding sites are close to binding sites of the BKCa channel activator CTIBD (4-(4-(4-chlorophenyl)-3-(trifluoromethyl)isoxazol-5-yl)benzene-1,3-diol) (67, 68), with CTIBD occupying the same space as TAM+M in the “WDF” binding site. According to our simulations, both binding sites are located in the extracellular region, fitting well to the recently reported activation of BKCa by a membrane-impermeable form of TAM in smooth muscle cells (69).
In conclusion, we show that TAM+M induces considerable changes in the K+ and Ca2+ homeostasis in human and murine BC. This is mediated by direct activation of BKCa oncochannels. TAM+M binding to one or two postulated extracellular binding sites results in BKCa activation, followed by the elevation of [Ca2+]i and a more depolarized ΨPM. Both effects consequently lead to a sustained feed-forward mechanism of BKCa activation. Considering that BKCa is involved in cell cycle control, proliferation, invasion, migration and malignancy of BC cells, it may be inferred from these changes that in patients with BC they may affect the response to endocrine treatment and promote the development of drug treatment resistance. A future perspective for translation of our results into clinical practice should be the analysis of BKCa channel expression in tumor biopsies, upfront onset of TAM therapy to predict TAM-modulation based on ER-independent interaction with oncochannels. In addition, we envision that the pharmacological inhibition of BKCa could be promising as a novel therapeutic in targeted anti-cancer concept, either in patients with endocrine resistance, or in patients with BC having aberrant BKCa channel expression.

Experimental procedures

Cell culture
Human BC cell lines MDA-MB-453 (hBKCahigh (ER-)), MCF-7 (hBKCalow (ER+)), CAMA-1 (hBKCamedium (ER+)), SKBr-3 (hBKCalow (ER-)) were purchased from ATCC: The Global Bioresource Center. Cells were cultivated in Dulbecco’s modified eagle’s medium (DMEM), 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin (P/S) (Thermo Fisher Scientific). MDA-MB-453 and MCF-7 were additionally supplemented with 1 mM sodium pyruvate (Thermo Fisher Scientific). Murine BC cells were isolated from breast tumor-bearing MMTV-PyMT transgenic FVB/N WT (mBKCaWT) and BKKO (mBKCaKO) mice (9). Animal experimental procedures were approved by the local ethics Committee for Animal Research (Regierungspräsidium Tübingen) and in accordance with the German Animal Welfare Act. Animals had access to food and water ad libitum and were kept on a 12 h light/dark-cycle under controlled humidity and temperature. Spontaneous tumor induction and development was monitored at an age of 12 to 14 weeks, and murine BC cells were isolated from n = 5 mBKCaWT and n = 6 mBKCaKO mice. Breast cancer cells were isolated from primary tumors as previously described (7, 70) and cultured in modified improved minimal essential medium (IMEM) supplemented with 5% FBS and 1% P/S. Human and murine BC cells were cultivated at 37 °C, 5% CO2 in a humidified incubator.
Subculturing of cells was performed at 80 to 90% confluency with 0.05% trypsin-EDTA diluted in PBS (Thermo Fisher Scientific) for 5 min in the incubator. After centrifugation at 1000 rpm for 5 min, cells were seeded into new cell culture flasks or dishes for subsequent experiments. Cell culture materials were purchased from Sarstedt or Corning (Kaiserslautern, Germany). In experiments that aimed to exclude the influence of growth factors and hormones, cells were cultured 72 h in serum-free medium followed by charcoal-stripped serum (CCS) enriched medium.

Electrophysiology
For whole-cell and excised inside-out patch-clamp experiments, 300,000 cells of hBKCamedium/high and 150,000 cells of hBKCalow were seeded in 35 mm dishes the day before measurement. For whole-cell patch-clamp experiments, BC cells were washed twice with pre-warmed PBS and extracellular buffer (Bfex) containing (in mM): 143 NaCl, 2 CaCl2, 1 MgCl2, 10 HEPES, 10 D-Glucose, pH 7.4 adjusted with NaOH (all from Carl Roth) was applied. A P-1000 Flaming/Brown micropipette puller (Sutter Instruments) was used to pull micropipettes with 3.5 to 4.5 MΩ from borosilicate glass capillaries (BM150–10P, Science Products GmbH, Hofheim, Germany). Glass micropipettes were polished using an MF-830 Micro Forge (Narishige International Ltd) and filled with intracellular solution (in mM: 130 K-gluconate, 10 KCl, 5 MgCl2, 0.6 EGTA, 5 HEPES, 0.1 CaCl2, 0.2 Na2-GTP, 2.0 Mg-ATP, pH 7.4 adjusted with KOH). Osmolality of intra-and extracellular buffer was adjusted to 300 mOsm/kg. Single-channel recordings in excised inside-out patches were performed in symmetrical isotonic KCl buffer containing (in mM): 10 NaCl, 110 KCl, 2.69 CaCl2, 5 EGTA, 2 MgCl2, 10 HEPES, pH 7.4 adjusted with KOH to obtain physiological [Ca2+]i levels (0.1 μM free Ca2+) (71) and Ca2+-free conditions were achieved by depleting CaCl2. Micropipettes with 5.0 to 7.0 MΩ were pulled and polished as mentioned above. For both configurations, a Nikon Tc2R microscope equipped with a Nikon S Plan Fluor 40x/0.6 objective (Nikon Instruments Inc.) and a DFK 33U x 174 camera (The Imaging Source Europe GmbH) was used for visual guidance. The recording setup was further comprised of a MP-225 micromanipulator (Sutter Instruments) and an EPC10 amplifier (HEKA Elektronik GmbH). PatchMaster software (HEKA Elektronik GmbH) and nest-o-patch (http://sourceforge.net/projects/nestopatch, Dr V. Nesterov) were used for data recording and analysis. In whole-cell mode, currents were recorded in response to 300 ms voltage steps ranging from −100 to +180 mV (hBKCahigh) and −100 to +120 mV (hBKCalow) in 20 mV increments from a holding potential of −40 mV. Recordings were repeated after addition of PAX (1 min), NS11021 (3 min) and TAM+M (8 min). Whole-cell capacity and access resistance were compensated. Cells for which access resistance changed by more than 20% during the recording were excluded from the analysis (<10%). The mean current amplitude during the last 16.2 ms of the depolarizing pulse was measured and normalized to cell capacity. Single-channel currents were recorded using voltage steps ranging from −80 to +80 mV in increments of 10 mV for 15 s. Recordings were repeated after addition of TAM+M (5 min). For current analysis using Nest-o-Patch, the recordings were filtered at 500 Hz. The open probability from a group of channels (NPO) in the patches represents channel activity. Representative single-channel recordings and analysis of NPO are described for −50 and +50 mV at physiological or Ca2+-free conditions.

FRET-based K+ imaging
BC cells were seeded to a confluency of 70 to 80% on glass coverslips (1.5 H 30 mm, Paul Marienfeld GmbH) in 35 mm dishes (Corning). For FRET-based live-cell K+ imaging, cells were transfected 16 h prior to the experiments with PolyJet (SignaGen Laboratories) according to the manufacturer’s instructions. Plasmids used in this study are as follows: FRET-based K+ sensor cytosolic lc-LysM GepII1.0 (40) as a gift from Roland Malli; BKCa channel overexpression was performed with BKCa-RFP (7) as well as respective control by the empty RFP plasmid (72). Loss-of-function mutant BKCaG354S previously reported by (43), was generated by site-directed mutagenesis using: BKCa-RFP plasmid as template, the primers: BKG354S forward 5′-ACA ATG TCT ACA GTG AGT TAT GG-3′ and BKG354S
reverse 5′-CAT AAA CGT CCC CAT AAC TCA-3′, Q5 High-Fidelity DNA Polymerase (NEB) followed by digestion with DpnI (NEB) according to manufacturer’s instructions. Plasmid preparation was performed by NucleoBond Xtra Maxi Kit according to the manufacturer’s protocol (Macherey-Nagel). The successful mutagenesis was validated by sequencing of the insert (Microsynth SeqLab).
Glass coverslips were transferred to a perfusion chamber (NGFI GmbH, Graz, Austria) and washed twice with Bfex. To obtain intracellular Ca2+-depleted conditions, BC cells were incubated for 30 min prior to the recording with cell-permeant BAPTA-acetoxymethyl ester (BAPTA AM). Extracellular Ca2+-free conditions were achieved by the addition of EGTA (Sigma-Aldrich) and omission of CaCl2. Measurements were performed under the presence of different compounds with stock solutions diluted in DMSO at the following final concentrations: 1 μM endoxifen, 1 μM 4-OH-TAM (Sigma-Aldrich), 10 μM NS11021, 5 μM paxilline, 10 μM SKF-96365 (Tocris). Dilution series were performed with 1 μM endoxifen and 1 μM 4-OH-TAM to perform concentration-dependent experiments. Controls were treated with DMSO.
A Zeiss Axio Observer Z1 (Carl Zeiss AG) microscope was used for FRET-based measurements as previously described (7, 70, 73). The imaging platform contained a BioPrecision2 automatic XY-Table (Ludl Electronic Products, Ltd) and a Plan-Neofluar 40x/1,30 Oil immersion objective (Carl Zeiss AG). An external light source (LED-Hub, Omicron Laserage) equipped with LED at 340 nm, 380 nm, 455 nm, 470 nm and 505 to 600 nm and emission filters: 340x, 380x, 427/10, 473/10 and 575/15 (AHF, Analysentechnik) allowed live-cell imaging by FRET. For FRET measurements, an Optosplit II (Cairn Research) together with dichroic mirrors (459/526/596 nm) and emission filters (475/543/702 nm) was utilized for simultaneous recording of FRET and CFP emissions (all from AHF Analysentechnik). Images were captured by a pco.panda 4.2 bi sCMOS camera (Excelitas PCO) and VisiView software (Visitron Systems). Addition of buffers and treatments was enabled by a gravity-based perfusion system (NGFI GmbH).

Fura-2-based Ca2+ measurement
For the Fura-2 recordings, the BC cells were seeded 24 h before the measurement with a confluence of 50% on glass cover slips in 35 mm dishes. Stocks of 15% Pluronic F-127 (Sigma-Aldrich) and 1 mM Fura-2 AM (Biomol GmbH, Hamburg, Germany) were mixed in a 1:3 dilution (following Fura-2). BC cells were washed with Bfex and loaded for 20 min at RT with a final concentration of 3.3 μM Fura-2 AM. Subsequently, cells were washed with Bfex for 10 min at RT. Imaging was performed on a Zeiss Axiovert 200M microscope equipped with a 40x Fluar 1.30 oil immersion objective (Carl Zeiss AG). Further, illumination was provided by a pE-340fura LED (CoolLED), together with excitation filters 340/26 and 380/14 (AHF Analysentechnik). Dichroic mirror 515LP and emission filter 525/15 (AHF Analysentechnik) enabled specific detection of Fura-2 based Ca2+ signals. A pco.panda 4.2 sCMOS camera (Excelitas PCO) and VisiView were used for image acquisition. Perfusion of treatments was achieved by a gravity-based system. Next to already mentioned treatments, 100 μM ATP (Carl Roth) was applied.

Plasma membrane potential measurements
Bis-(1,3-dibutylbarbituric acid)Trimethine Oxonol (DiBAC4(3)) (Thermo Fisher Scientific) was used in a concentration of 0.25 μg/ml to determine ΔΨPM. BC cells were seeded on 30 mm glass cover slips, washed with Bfex and incubated with DiBAC4(3) for 35 min at RT. During the measurement at the microscope Zeiss Axio Observer Z1, DiBAC4(3) was constantly present in the buffer. Excitation was performed by 470 nm LED and a dichroic mirror (409/493/573/652 nm), and an emission filter (514/605/730 nm) enabled DiBAC4(3) specific recordings.

qRT-PCR
Gene expression levels of BKCa subunits in human BC cells were analyzed using total mRNA isolated from hBKCalow, hBKCamedium and hBKCahigh BC cells. BC cells were seeded to a confluency of 80% in 100 mm dishes. Isolation, cDNA synthesis and qRT-PCR were performed as previously described (74). RNA was isolated using the NucleoSpin RNA kit (Macherey-Nagel) followed by cDNA synthesis of 0.5 μg RNA using iScript cDNA synthesis kit (Bio-Rad) according to manufacturer’s instructions, respectively. SsoAdvanced Universal SYBR Green Supermix (Bio-Rad) and target-specific primer pairs were used to perform qRT-PCR with CFX Connect Real-Time PCR Detection System (Bio-Rad).

Human breast cancer biopsies and nanostring nCounter gene expression analysis
An endocrine treatment study of 1286 BC patients between 2005 and 2011 (German Clinical Trial Register DRKS 00000605, “IKP211” study) was previously described for the analysis of 551 primary tumor biopsies from hormone receptor positive, post-menopausal patients with Nanostring nCounter gene expression analysis (7). In brief, patients were treated by TAM, aromatase inhibitor or switched endocrine therapy in between. Total RNA was isolated from FFPE tissues (Quick-DNA/RNA FFPE, ZymoResearch, Freiburg, Germany) of their breast cancer biopsies. Target-specific capture probes and color-coded reporter sets for ion channel genes and accessory subunits were designed and hybridized with 250 ng RNA in a thermal cycler at 65 °C for 20 h. The Nanostring nCounter system was used to measure fluorescence counts. Data analysis was performed with nSolver 4.0. Normalization to five housekeeping genes (ABCF1, NRDE2, POLR2A, PUM1, and SF3A1) was performed and cut off of log2 expression count was set to 5.5. We have previously published some information on the BC patient-related BKCa channel expression profile in (7).

Molecular docking simulation
TAM+M were docked against two protein cryo-EM structures of BKCa under Ca2+-bound and Ca2+-bound in complex with the β4 subunit conditions (PDB Codes 6V38, 6V22, respectively, (60)) using Schrödingers software suite (Schrödinger Release 2024–4). Using the Protein Preparation Wizard (75), bond orders were reassigned, hydrogens replaced and sensible protonation states at pH 7.4 were determined using PROPKA. Each protein was subjected to a restrained minimization using the OPLS4 force field (76) and a 0.3 Å RMSD convergence criterion. Ligands were prepared using LigPrep at pH 7.0. We used SiteMap (77, 78) with shallow settings to identify a multitude of binding sites and split large binding sites using the DBSCAN clustering algorithm. This resulted in 277 and 368 potential binding sites for BKCa under Ca2+-bound and Ca2+-bound in complex with β4 conditions, respectively. TAM+M were docked into each binding site using GLIDE (79) with standard precision. After the identification of putative binding sites, the extended induced fit docking protocol (80, 81) with extra precision in the structure of BKCa-Ca2+-bound (6V38) was performed. For the refinement step, an implicit membrane was defined based on the experimental data.

Statistical analysis
Statistical analysis was performed with GraphPad Prism 9 software (GraphPad, Software). All data were tested for outlier. Normal distribution was verified using Shapiro-Wilk or Kolmogorov-Smirnov test. If the variances of the populations were comparable, a two-tailed Unpaired t test was used; if F-test was significantly different, Welch’s test was applied for the statistical comparison of normally distributed data. Normally distributed data of paired measurements were evaluated with Paired t test, while the Wilcoxon-test was used for non-normally distributed data. For the comparison of >2 groups either One-way ANOVA followed by Dunnett’s or Tukey’s multiple comparisons for normally distributed data was used, always considering F-test variances, or Two-way ANOVA followed by Dunnett’s, Šídák’s or Tukey’s multiple comparisons test was performed. If data were not normally distributed, a Kruskal-Wallis test followed by Dunn’s multiple comparisons test was performed. Statistical tests are indicated in the Figure legends and described in detail in the Supplementary Tables (Tables S1–S9). p-values of ≤ 0.05 were considered as statistically significant with ∗p ≤ 0.05, ∗∗p ≤ 0.01 and ∗∗∗p ≤ 0.001.

Data availability
All data are present in the article or the Supporting Information.

Supporting information
This article contains supporting information (7, 9, 21, 44, 82).

Conflict of interest
The authors declare that they do not have any conflicts of interest with the content of this article.