Mapping the functional expression of auxiliary subunits of KCa1.1 in glioblastoma

Glioblastoma (GBM) is the most aggressive glial tumor, where ion channels, including KCa1.1, are candidates for new therapeutic options. Since the auxiliary subunits linked to KCa1.1 in GBM are largely unknown we used electrophysiology combined with pharmacology and gene silencing to address the functional expression of KCa1.1/β subunits complexes in both primary tumor cells and in the glioblastoma cell line U-87 MG. The pattern of the sensitivity (activation/inhibition) of the whole-cell currents to paxilline, lithocholic acid, arachidonic acid, and iberiotoxin; the presence of inactivation of the whole-cell current along with the loss of the outward rectification upon exposure to the reducing agent DTT collectively argue that KCa1.1/β3 complex is expressed in U-87 MG. Similar results were found using human primary glioblastoma cells isolated from patient samples. Silencing the β3 subunit expression inhibited carbachol-induced Ca2+ transients in U-87 MG thereby indicating the role of the KCa1.1/β3 in the Ca2+ signaling of glioblastoma cells. Functional expression of the KCa1.1/β3 complex, on the other hand, lacks cell cycle dependence. We suggest that the KCa1.1/β3 complex may have diagnostic and therapeutic potential in glioblastoma in the future.

Even though radiosensitivity is altered by K Ca 1.1, there is only scarce evidence whether channel modulation could potentiate GBM chemosensitivity 7 .
The pore-forming α subunit of K Ca 1.1 is associated with auxiliary subunits. These subunits modify the biophysical characteristics of the channel and responsiveness to pharmacological modulators 8 . Moreover, the expression of a particular β subunit can be characteristic for certain pathological conditions. For example, we have showed earlier that CD44 + fibroblast-like synoviocytes isolated from rheumatoid arthritis patients display increased β3 subunit expression and augmented whole-cell K Ca 1.1 currents 9 . The REpository of Molecular BRAin Neoplasia DaTa (REMBRANDT) glioma database indicates that the KCNMB3 gene, encoding the K Ca 1.1 β3 subunit, is expressed in a higher copy number in high-grade gliomas leading to a poorer prognosis compared to tumors expressing KCNMB2, the gene encoding the β2 subunit 10 . For comparison, the expression of the α subunit (KCNMA1) is upregulated only in ≈10% of GBM patients, and its overexpression does not correlate with overall patient survival 11 . However, there is no functional data to support the expression of K Ca 1.1 α subunits in complex with β subunits in human glioblastoma cells or glioblastoma cell lines. To address this, we characterized the β subunits of K Ca 1.1 in primary patient-derived GBM cells as well as in U-87 MG glioblastoma cell line using the combination of molecular biology, biophysics (electrophysiology) and pharmacology. Moreover, we investigated whether these auxiliary subunits regulate functional aspects and downstream effects of K Ca 1.1.

Results
K Ca 1.1 is a prominent K + channel in the plasma membrane of glioblastoma cells. First, we validated the functional expression of the K Ca 1.1 in the membrane of GBM cells using the whole-cell patch-clamp technique. We expressed the currents as current density (J=pA/pF) where currents were normalized to the cell membrane capacitance to obtain a cell-size independent parameter. As seen in Fig. 1A, the voltage-gated K + current density in primary GBM cells is markedly increased in the presence of 1 µM intracellular Ca 2+ (N=3, n=5) compared to Ca 2+ −free intracellular solution (N=5, n=11) which is characteristic of K Ca 1.1 channels (−40 mV, p=0.52; −20 mV, p=0.0004; 0 mV and above: p<0.0001, Student's t tests). We also observed another hallmark of the K Ca 1.1 channel 12 that currents activate at much more negative membrane potentials in the presence of intracellular Ca 2+ . K + currents of the glioblastoma cell line U-87 MG (N=3, n=5) have a similar current-voltage (I-V) relationship as in primary GBM cells (Fig. 1B) (N=3, n=11); which confirms the suitability of the cell line as a model to study K Ca 1.1 in GBM. Also, whole-cell currents of primary GBM and U-87 MG cells are potently and irreversibly inhibited by applying 1 µM of the small-molecule blocker paxilline (Pax) (Fig. 1D). The remaining current fractions (RCF=I/I 0 , where I 0 and I are the peak currents in the absence and in the presence of the inhibitor, respectively) in the presence of 1 µM paxilline were RCF=0.38 ± 0.04, n=20, and RCF=0.13 ± 0.02, n=13 for GBM and U-87 MG cells, respectively (Fig. 1E). Using immunofluorescence labels against K Ca 1.1, we also detected a punctate membrane staining 13,14 on the membrane of GFAP positive GBM cells (Fig. 1C). These results support previous reports that K Ca 1.1 functions as a major K + channel in GBM 4,5,15,16 . β3 is the main auxiliary subunit associated with the K Ca 1.1 channel in glioblastoma. Auxiliary subunits of K Ca 1.1 are known to alter the biophysical characteristics of the channel 8 . However, only very limited information is available on β subunit expression in GBM. Thus, we aimed at demonstrating the expression of β subunits using molecular biology and confirming the association of K Ca 1.1 to its auxiliary β subunits with the combination of biophysical, pharmacological methods and genetic modulations.
First, we aimed to determine the functional expression of the β subunits in primary GBM cells and in the U-87 MG cell line using the patch-clamp technique (Fig. 2). Both lithocholic acid (LCA) and arachidonic acid (AA) activate K Ca 1.1 channels associated with the β1 subunit [17][18][19] , whereas AA also increases the current when K Ca 1.1 is assembled with the β2 or β3 subunits 20 . In primary GBM cells, AA approximately doubles whole-cell K Ca 1.1 currents compared to control (Fig. 2B, 1.84-fold increase, n=14, p=0.001, two-tailed Wilcoxon test), whereas LCA does not induce an increase of the K Ca 1.1 current ( Fig. 2A, 1.02-fold increase, n=16, p=0.63, two-tailed Wilcoxon test). In contrast, in U-87 MG cells, application of LCA and AA elicit a similar increase in the outward current compared to control (on average, 1.38-fold (n=16, p=0.0002) and 1.24-fold (n=16, p=0.008) current increase, respectively, two-tailed Wilcoxon test, Fig. 2A and 2B). K Ca 1.1 channels co-expressed with β2 subunits inactivate with an inactivation time constant of around ~20 ms 21 upon strong depolarization especially when the intracellular free Ca 2+ concentration is high (10 µM). At low cytosolic Ca 2+ concentrations and +100 mV depolarization, as in our study, the current would still significantly inactivate if β2 subunits are in complex with K Ca 1.1α 22,23 . Data in Fig. 2C do not support the co-expression of the β2 subunits with K Ca 1.1: whole-cell currents recorded from U-87 MG or primary GBM cells completely lack inactivation within 100 ms after the activation of the current. It is also known that K Ca 1.1 channels co-expressed with β4 subunits are resistant against the peptide inhibitor iberiotoxin (IbTx) 24 . We tested the inhibition of the whole cell currents in primary GBM and U-87 MG cells at 23 nM IbTx concentration which is ~2-10-fold the IC 50 for K Ca 1.1 inhibition 25 and compared the RCF to that determined using 1 µM paxilline, which is expected to block fully the KCa1.1 current when applied at negative holding potentials (IC 50~1 2 nM). The use of ~ μM paxilline to define the K Ca 1.1 current component of whole-cell currents in native cells is a commonly used strategy 26,27 . Fig. 2D shows that in case of primary GBM the RCF values are similar upon 23 nM IbTx (RCF = 0.47 ± 0.04 (n=18)) and 1 µM paxilline (RCF=0.38 ± 0.04, n=20, p=0.12, Student's t tests) application. Similarly, the RCF values were statistically the same for the inhibition of the whole-cell currents in U-87 MG by 23 nM IbTx (RCF= 0.13 ± 0.03 (n=8) and 1 µM paxilline (RCF=0.13 ± 0.02, n=13 for, p=0.99, Student's t tests) (Fig. 2D). The similar extent of current inhibition for the peptide and nonpeptide blocker precludes the presence of K Ca 1.1 β4 subunits in the channel complex. In summary, the effect of pharmacological modulators on the β subunit-associated channels is consistent with the presence of the β1, β2, and β3 subunits in the K Ca 1.1 complex, whereas the existence of the β4 subunit/K Ca  www.nature.com/scientificreports/ membrane seems to be highly unlikely. This, combined with the lack of current inactivation (under conditions above) reduces the β subunit repertoire to β1 and β3 to combine with K Ca 1.1 in GBM or U-87 MG. Next, we aimed at supporting the functional data using molecular biology techniques. As indicated in Fig. 3A, several K Ca 1.1 auxiliary subunit mRNAs are expressed in the U-87 MG cell line. Using RT-qPCR, we found that β1 and β2 subunits are expressed at low levels in both the primary tumor and U-87 MG cells, whereas the β3 subunit shows the highest expression (Supp. Fig. 1 for primary GBM).
As small-molecule pharmacological modulators such as LCA and AA activate multiple ion channels and signaling pathways 28,29 , we applied genetic modulation of U-87 MG cells using siRNAs targeting different K Ca 1.1 β subunits to correlate β subunit expression and LCA/AA effects. Western blots in Fig. 3B demonstrate that the visible bands identified by anti-K Ca 1.1 β2 (KCNMB2) or β3 (KCNMB1) antibodies cannot be detected after the application of the corresponding siRNAs for 48 h (Fig. 3B). As the anti-K Ca 1.1 β1 (KCNMB1) antibody failed to identify any protein bands on the gel we applied this antibody to Chinese hamster ovary (CHO) cells were a   30 ). The functional consequence of β subunit silencing was tested using the pharmacological modulators LCA and AA. Lithocholic acid has a similar effect on whole-cell currents after K Ca 1.1 β1 silencing compared to the scrambled siRNA control (p=0.63, Student's t-test, N silence =3) (Fig. 3C). In contrast, both K Ca 1.1 β2 and β3 siRNAtreatment decreases the response of U-87 MG cell to arachidonic acid compared to scrambled RNA silencing (p=0.083 and p=0.038, respectively) ( Fig. 3D and 3E). Silencing of the β3 auxiliary subunit did not affect the sensitivity of the current to paxilline (Supp. Fig. 2). Taken together the higher sensitivity of the β3 silencing on the AA response (Fig. 3E) and the larger expression of the β3 RNA compared to other auxiliary subunits To ascertain the functional presence of the K Ca 1.1 β3 auxiliary subunit in GBM cells, we performed biophysical studies with specialized patch-clamp protocols 31,32 . We increased the intracellular free Ca 2+ concentration to 10 µM and depolarized the membrane to a more positive test potential (+ 180 mV). Under such conditions, K Ca 1.1 channels associated to β3 subunits would be more likely to undergo a rapid and incomplete inactivation 31 . Indeed, as depicted in Fig. 4, the inactivation of the whole-cell current of U-87 MG cells follows this phenotype, . Whole-cell currents were recorded as in Fig. 1D, peak currents were measured and normalized peak currents were calculated as in Fig. 2A  www.nature.com/scientificreports/ as the current inactivation is incomplete, the I 10ms /I peak ratio, corresponding to the current at 10 ms following the start of the depolarization (I 10ms ) over the peak current (I peak ), is decreased to 0.86 ± 0.03 (n = 6; Fig. 4C). Moreover, the inactivation kinetics is very rapid, the inactivation time constant (tau) is characteristically short (2.5 ± 0.43 ms, n = 5; Fig. 4D) for the KCa1.1/β3 complex under these experimental conditions 31 . Association of K Ca 1.1 α subunits with β3 confers sensitivity of the complex to the reducing agent dithiothreitol (DTT) 32 . The outward rectification of the complex is abolished upon exposure to DTT as a consequence of the disruption of the disulfide links between the extracellular parts of the β3 subunits 21 . We exploited this phenomenon and showed that DTT treatment resulted in the appearance of the instantaneous current (i.e. tail current) when returning to a negative membrane potential ( Fig. 4E and Supp. Fig. 3), whereas the tail current is absent without DTT application ( Fig. 4A and Supp. Fig. 3). The peak currents at -100 mV were −487 ± 66 pA (n = 3) and 82 ± 18 pA (n = 3) respectively (Fig. 4F , p = 0.004, Student's t tests). The deactivation time constant of the tail current in K Ca 1.1 β3 is involved in U-87 MG Ca 2+ signaling but shows no cell cycle dependence. To investigate whether the K Ca 1.1 β3 subunit is involved in downstream mechanisms of K Ca 1.1 function, we studied the intracellular Ca 2+ signaling evoked by the acetylcholine (ACh) analogue, carbachol 16,33 (Fig. 5). Fig. 5A left panel shows that U-87 MG cells respond to the application of 10 µM carbachol with a marked increase in the cytosolic free Ca 2+ concentration, as reported by the increased F 340 /F 380 ratio. The Ca 2+ response of the cells is inhibited by the simultaneous administration of carbachol and 1 µM paxilline (Fig. 5A, right panel). The heat maps in Fig. 6B highlight that approximately 50% of the cells in each population (59 of 120 cells in the control group, 49 of 91 cells in the carbachol + paxilline-treated group) respond to cholinergic stimulation by more than 20% increase in the F 340 /F 380 ratio compared to their initial value. The statistical analysis (Fig. 5C) clearly indicates that paxilline www.nature.com/scientificreports/ (n=49 cells; peak F 340 /F 380 = 1.6 ± 0.06) inhibits the Ca 2+ response of the U-87 MG cells to carbachol (n= 59 cells; peak F 340 /F 380 = 2.1 ± 0.07; p<0.0001). Interestingly, the peak of the carbachol-induced Ca 2+ signal (Fig. 5D) at t = 220s is inhibited by β3 silencing (KCNMB3 siRNA transfected, n=30; p=0.0004) similarly to silencing of the pore forming subunit of K Ca 1.1 (KCNMA1 siRNA transfected, n=51; p=0.025). The F 340 /F 380 , ratios at t = 220 s were 1.1 ± 0.06 (n=39) for the siGAPDH treatment, 0.8 ± 0.04 (n=51) for the siKCNMA1 (siKCNMA1), and 0.9 ± 0.04 for siKCNMB3 treatments (n=30), (Fig 5D). Based on our results we conclude that K Ca 1.1 co-expressed with the β3 subunit mediates Ca 2+ -signaling in response to carbachol in U-87 MG cells. As cytosolic Ca 2+ fluctuates during the cell cycle and K + channels are expressed in a cell cycle-dependent manner 34,35 , we aimed at testing whether β3 subunit-dependent modulation of the K Ca 1.1 current is influenced by the cell cycle of glioblastoma cells. To this end, we synchronized U-87 MG cells in M phase using colchicine and in G 0 phase using serum starvation (Fig. 6). Fig. 6A shows the flow cytometry data of the synchronized cells. The histograms and Supp. Table 3 show that 36 ± 3 % of the cells were in G 2 /M phase 24 h after a 10 µM colchicine treatment (N=3, n=3) as compared to 16 ± 1% in the untreated group (N=3, n=5). Upon serum starvation, a high percentage of cells reside in the G 0 /G 1 phase (57 ± 3% for untreated, 78 ± 1% for serum starvation, N=3, n=5 and N=2, n=2, respectively) (Supp. Table 3). Interestingly, we observed a marked increase in the magnitude of the whole-cell currents in M phase synchronized cells as compared to control (non-synchronized, untreated) and G 0 synchronized ones (Fig. 6B-C). The increase in the peak currents becomes evident at depolarizations to +40 mV or above (Mann-Whitney test, Fig. 6C). As cell volume and cell surface can also change during the cell cycle 36 , we also determined current density (J, see above, section "KCa1.1 is a prominent K+channel in the plasma membrane of glioblastoma cells"). Similar to the peak currents, the current density was significantly larger in the M phase synchronized cells as compared to control and G 0 phase synchronized ones at depolarized test potentials (above +20 mV, Mann-Whitney test, Fig. 6D). To ensure that the main component of the whole-cell currents remains K Ca 1.1 in these cells, we applied paxilline (1 µM) to each synchronized and control cell population (Fig. 6E). We found a pronounced inhibition of the whole cell K + current by paxilline in all cell cycle phases, especially in the M phase, the average RCF in colchicine-and serum starvation-treated cells were 0.05 ± 0.01 (n=9) and 0.17 ± 0.02 (n=10), respectively (p=0.007, Kruskal-Wallis test, Fig. 6E). This confirms that the major current component is K Ca 1.1 in colchicine-synchronized cells. On the other hand, the increase in the whole-cell current induced by 30 µM AA was similar in all groups (current increase: 1.26 ± 0.09 (n=16), 1.1 ± 0.08 (n=10) and 1.09 ± 0.03 (n=10) for the untreated, colchicine-and serum starvation treated cells respectively, p=0.46, Kruskal-Wallis test, Fig. 6F). Together, these results indicate that K Ca 1.1 function is increased after M phase synchronization, without alterations in the β3 subunit-dependent K + current modulation.

Discussion
In this study, we showed that the Ca 2+ -and voltage dependent K + channel K Ca 1.1 functions in the plasma membrane of patient-derived primary glioblastoma cells as well as the U-87 MG cell line in association with its auxiliary β3 subunit (Figs. 1, 2, 3 and 4). This is relevant for cellular Ca 2+ signaling (Fig. 5) but not for cell cycle progression (Fig. 6). Even though K Ca 1.1 is ubiquitously expressed in many tissues in the human body, auxiliary β subunits have a much more restricted tissue expression. Particularly, the β3 subunit is rarely found in healthy tissues 37 , and has only been described to date in the testes, pancreas and spleen. This may be relevant in diagnosis and/or therapy, especially since the increased expression of K Ca 1.1 β3-encoding gene KCNMB3 correlates with poor survival of GBM patients 10 and increased β3 auxiliary subunit expression was described in fibroblast-like synoviocites in rheumatoid arthritis 9 . Also, an intriguing prospect of KCNMB3 expression in high-grade gliomas 10 is that based on our findings K Ca 1.1 β3 can be a cell surface prognostic marker in GBM. Therefore, generating auxiliary subunit-specific probes is warranted in a future study.
Based on the following pieces of evidence we argue for the presence and functional activity of β3/K Ca 1.1 α complexes on glioblastoma cells: (i) the transcript for β3 is highly expressed, its relative expression is similar to that of the α subunit in U-87 MG cells (Fig. 3A); (ii) the whole cell current was augmented by AA application (Fig. 2B), and this effect was diminished following silencing of the β3 auxiliary subunit (Fig. 3E); (iii) fast and incomplete inactivation of the whole cell current was recorded at high, 10 µM intracellular free Ca 2+ concentration and depolarization to +180 mV (Fig. 4A-D); (iv) inward tail currents were recorded at −100 mV upon exposure of the cells to the reducing agent DTT (Fig. 4D-E). Moreover, sensitivity of the whole cell current to 23 nM IbTx is consistent with the pharmacology of the K Ca 1.1/β3 complex 38 . Nevertheless, the findings listed in i.-iv. should be collectively interpreted and used as an argument for the presence of the β3 auxiliary subunit/K Ca 1.1 α complex as many of these characteristics are shared with other β subunit/K Ca 1.1 α complexes, as discussed below.
For example, the β1 subunit, prominently expressed in smooth muscle cells, prolongs activation kinetics of K Ca 1.1 and has a well characterized pharmacological activation by bile acids 8,17,39,40 . Even though ionic currents of primary U-87 MG cells, and to a smaller extent in GBM, are activated by 75 μM lithocholic acid, two factors argue against the presence of K Ca 1.1 β1 in glioblastoma cells. First, we could not detect K Ca 1.1 β1 in RT-qPCR (Fig. 3A). In line with this, we observed similar pharmacological response in KCNMB1-silenced cells as in the scrambled siRNA-treated controls, as if the effect of LCA on the currents was oblivious to the presence or absence of the β1 subunit. Knowing that bile acids activate a multitude of other ion channels, e.g. bile acid sensitive ion channels (BASIC) 41 , it is much more likely that LCA acts on different ion channels than K Ca 1.1 in glioblastoma cells. K Ca 1.1 in complex with β1 is also quite resistant to IbTx inhibition (IC50~between ~65 nM 24 to ~370 nM 38 , which is in contrast to our finding in Fig. 2D). Furthermore, it is known that β1 associated K Ca 1.1 channels have instantaneous tail currents at negative membrane potentials 42 , and based on our recordings, we have only seen tail currents at negative membrane potentials when we treated the cells with 20 mM DTT (Fig. 4E and F).
Regarding the β2 subunit of K Ca 1.1, we found that it is present in the cells on mRNA level as determined by RT-qPCR. β2 subunits are similar to β3 subunits in a manner that arachidonic acid increases the K Ca 1.  20,29 . Moreover, KCNMB2-silenced U-87 MG cells show less AA-dependent response compared to control-silenced cells, indicating that KCa1.1 may be in complex with this auxiliary subunit. However, β2 subunits lead to a complete inactivation of K Ca 1.1-mediated currents, even at low cytosolic Ca 2+ concentration and modest depolarization to +100 mV, which was used to obtain data in Fig 2C 22, 23 . Complete inactivation of the current was rarely (n=2 out of n=51 primary GBM cells) observed in our settings, with most whole-cell currents showing no inactivation during 200 ms (Fig. 2C). We did not see the complete inactivation of the current either when the recording conditions were optimal to see the inactivation induced by the β2 subunits, i.e., when the free Ca 2+ concentration in the pipette filling solution was 10 µM (Fig. 4). The lack of inactivation -characteristic to the presence of β2 -can be attributed to the variability in the stoichiometry between the different β subunits associated with to the α subunit of K Ca 1.1: as four possible β subunits can simultaneously bind to one functional channel 43 , the ratio of different β subunits associated with the channel may become very important, as indicated previously 44,45 . In GBM, a biological consequence of altered subunit stoichiometry is easily possible: more β2 subunits linked to one K Ca 1.1 channel would mean complete channel inactivation, thus less driving force for Ca 2+ signals, whereas more β3 subunits would lead to a rapid but incomplete channel inactivation (as can be seen in Fig. 4) and a prolonged Ca 2+ influx. Therefore, a thorough assessment is warranted in a further study to prove this concept in GBM. The K Ca 1.1 β4 subunit is known to be expressed in the central nervous system 37,46 . In transfected model cells, K Ca 1.1 channels coupled to β4 subunit become resistant to inhibition by the peptide toxin IbTx 24,47 . Bicistronic expression experiments confirmed that when β4 is present in saturating stoichiometry the β4/K Ca 1.1 complexes are insensitive to IbTx-mediated inhibition 38 . In contrast, we observed that IbTx inhibits whole-cell currents potently (Fig. 2D). Thus, it is likely that the β4 subunit is not associated with K Ca 1.1 in glioblastoma. Lastly, K Ca 1.1 gamma subunits, are unlikely in GBM cells: as K Ca 1.1 channels associated with γ subunits already open at very negative membrane potentials of -150 mV 42 . In comparison, K Ca 1.1 starts to conduct at a more positive membrane potential in primary GBM as well as in U-87 MG cells (Fig. 1A). In summary, besides the evident association of K Ca 1.1 channels to β3 in the plasma membrane of GBM cells, it is likely that a minority of the channels may be coupled to β2.
To our knowledge, we are the first to describe that K Ca 1.1, coupled to its auxiliary β3 subunit, modulates the Ca 2+ signal upon ACh receptor stimulation (Fig. 5). The reduced Ca 2+ signal in the presence of the K Ca 1.1, inhibitor paxilline is consistent with a model where K Ca 1.1-dependent membrane potential alterations provide the electrical driving force for Ca 2+ entry 48 . Generally, the function of ancillary ion channel subunits is to finetune the expression and biophysical properties of the pore-forming (here K Ca 1.1) α subunit 8,49,50 . It has been recently proposed that ACh-induced signals, in a Ca 2+ -dependent manner, induce matrix metalloprotease 9, which ultimately increases glioblastoma cell invasiveness 33 .
Interestingly, the K Ca 1.1 channel together with the β3 subunit is also functional in fibroblast-like synoviocytes in rheumatoid arthritis 9 . One explanation for this can be that both glioblastoma and rheumatoid arthritis are accompanied by pronounced inflammation altering multiple parameters such as pH and the mechanical environment [51][52][53] . For example, it has been described that mechanosensitivity of K Ca 1.1 is conferred by the auxiliary β1 subunit in vascular smooth muscle 54 . Whether alterations in the microenvironment indeed modify auxiliary subunit expression via a common mechanism in diseases involving inflammation remains to be elucidated.

Conclusion
We found that K Ca 1.1 channels are coupled primarily to the auxiliary β3 subunit in the cell membrane of glioblastoma and U-87 MG cells with functional consequences on Ca 2+ -signaling of GBM cells upon muscarinic acetylcholine receptor activation. The β3 subunit expression of GBM cells may allow either specific targeting of the tumor cells using β3 subunit specific inhibitors and/or allow diagnostic tools to be developed based on β3 subunit expression. In conclusion, we propose that the β3 subunit of the channel acts as a membrane-localized marker for glioblastoma cells to be exploited for diagnostic or therapeutic approaches.

Materials and methods
Glioblastoma cell isolation. Experiments on patient-derived GBM tissue samples were carried out under the approval of the Hungarian Research Ethical Committee (ETT-TUKEB, IV/186-1 /2022/EKU). Diagnosis was established according to WHO criteria 55 by a neuropathologist (T.H.). Informed consent was obtained for all human subjects involved in this study. Tumor samples were collected from anonymized adult patients during the surgical removal of the glioblastoma and transported for further processing in HBSS (Hank's Balanced Salt Solution) on ice. Next, tissue samples were digested for 30 min in Collagenase type I (Sigma Aldrich, Burlington, MA, USA), and eventually homogenized using a tissue homogenizer and Pasteur pipettes, as modified from Souza et al. 56 . Lastly, single cell suspension was achieved using a 70 µm cell strainer (Corning, Corning, NY, USA). Single cells were left to adhere in DMEM + 10% FCS at 37 °C and 5% CO 2 for 2 h, then washed 3× with PBS. Cells were incubated in DMEM medium including 10% FCS, 1% glutamate, 1% penicillin-streptomycin and 1% non-essential amino acids for a maximum of three passages. Glioblastoma cell purity was routinely assessed using GFAP immunocytochemistry 57 ) were diluted to ≤0.1% V/V in the bath solution. Solution exchange was achieved by using a gravity-flow perfusion system (flow rate: 2 ml/min) with continuous excess fluid removal. For the biophysical characterization of the K Ca 1.1 currents, the cells were depolarized from a holding potential of −100 mV to +100 mV in +20 mV increments. We used a 100 or 200 ms depolarization protocol from −100 mV to +100 mV for testing of K Ca 1.1 channel modulators and 20 ms-long test pulses to +180 mV from −180 mV holding potential to study inactivation, and measured the instantaneous currents at −100 mV. All experiments were carried out at room temperature. Voltage-clamp data were acquired with pClamp10 (Molecular Devices, CA, USA). In general, currents were low-pass-filtered using the built-in analog four-pole Bessel filters of the amplifiers and sampled at 20 and 50 kHz. Whole-cell current traces were digitally filtered (five-point boxcar smoothing) before analysis. Clampfit 10.7 (Molecular Devices, CA, USA) and GraphPad Prism 7 (GraphPad, CA, USA) were used for data display and analysis.
Immunocytochemistry. We followed standard immunofluorescence protocol as described in 13  Lastly, after washing coverslips were mounted onto slides using DAKO mounting medium (Agilent, Santa Clara, CA, USA). Acquisition and qualitative assessment of the stainings were performed at 40× magnification using a confocal microscope (Olympos FV1000). Cells were labelled GFAP positive when the intracellular staining had a typical filamentary phenotype, and K Ca 1.1 staining was considered positive when it had a punctate membrane staining pattern typical of ion channels 13,14 .
Scientific Reports | (2022) 12:22023 | https://doi.org/10.1038/s41598-022-26196-w www.nature.com/scientificreports/ CO 2 for 24 h. To measure the efficacy of synchronization, colchicine-synchronized as well as the untreated U-87 MG cells were fixed and permeabilized with 80% ethanol at room temperature for 20 min and stained with 2 µg/ ml propidium-iodide (PI) at room temperature for 10 min for flow cytometry measurements 58 . The data were acquired with BD FacsAria III Cell Sorter (BD Biosciences, NJ, USA). 561 nm excitation laser and 610/20 nm emission filter with 600 nm long-pass dichroic mirrors were used for event detection. Data was subsequently evaluated with Flowjo V10 software (BD, Franklin Lakes, NJ, USA).
Intracellular Ca 2+ measurements. U-87 MG cells were loaded with 3 µM Fura-2-AM (Invitrogen, Waltham, MA, USA)-containing HEPES-buffered Ringer`s solution with glucose (140 mM NaCl, 5.4 mM KCl, 1.2 mM CaCl 2 , 0.8 mM MgCl 2 , 5.5 mM D-glucose, and 10 mM HEPES, titrated to pH = 7.4, titrated with NaOH) for 20 min at 37 °C. Next, cells were washed twice with fresh Ringer`s solution and were then visualized using an imaging setup composed of a Zeiss AxioVert 100 inverted fluorescence microscope (Zeiss, Oberkochen, Germany), a high-speed shutter, a polychromator (Visitron Systems, Puchheim, Germany) and a 37 °C acquisition cabin. Fura-2 excitation wavelengths were 340 nm and 380 nm, corresponding to the Ca 2+ -loaded and Ca 2+ -free excitation optima, respectively. Fluorescence emission was recorded at a wavelength of 510 nm. The ratio of the fluorescence intensities emitted upon 340 nm and 380 nm excitation (F 340 /F 380 ) is directly proportional to the intracellular Ca 2+ concentrations and was used in this study to report the cytosolic free Ca 2+ concentration 59 . The cells were kept at 37 °C during the whole measurement. During the acquisition, cells were initially superfused with the control solution (0.1% DMSO in Ringer´s solution) for 2 min, followed by 5 min with either only 10 µM acetylcholine (Ach)-receptor agonist carbachol (carbamylcholine chloride; Sigma Aldrich, Burlington, MA, USA)-containing Ringer´s solution to elicit a Ca 2+ -signal (previously described by 16,33 ), or with 10 µM carbachol + 1 µM paxilline containing Ringer´s solution to simultaneously inhibit K Ca 1.1. Ratios were evaluated with the Visiview 3.0 software (Visitron Systems, Puchheim, Germany), and ultimately, individual F 340 /F 380curves were visualized using R 60 .
Statistical analysis. Data are presented as mean ± SEM. Statistical analysis was carried out using GraphPad Prism 7. Following a D' Agostino-Pearson normality test, unpaired Student's t tests or one-way ANOVA were performed with Tukey's post hoc test, in other cases Mann-Whitney or Kruskal-Wallis tests were used. To assess the effects of the channel modulators we performed Wilcoxon signed-rank tests. Statistical significance was assumed when p < 0.05.