Relevance of Abnormal KCNN1 Expression and Osmotic Hypersensitivity in Ewing Sarcoma

Simple Summary The main oncogene in Ewing sarcoma directly drives a high expression of a previously unknown variant KCNN1 (encoding the KCa2.1 channel) that we also verified in samples from >200 patients. Yet, we found that the channel is not functional and does not modulate Ewing sarcoma cell behavior. We could explain this lack of functional impact by the surprising absence of any KCa2.1-carried K+ current in Ewing sarcoma cells. However, we show in a proof-of-principle study that the essential lack of a K+ conductance can be exploited by applying hypoosmotic stress and effectively and selectively killing the Ewing sarcoma cells. Abstract Ewing sarcoma (EwS) is a rare and highly malignant bone tumor occurring mainly in childhood and adolescence. Physiologically, the bone is a central hub for Ca2+ homeostasis, which is severely disturbed by osteolytic processes in EwS. Therefore, we aimed to investigate how ion transport proteins involved in Ca2+ homeostasis affect EwS pathophysiology. We characterized the expression of 22 candidate genes of Ca2+-permeable or Ca2+-regulated ion channels in three EwS cell lines and found the Ca2+-activated K+ channel KCa2.1 (KCNN1) to be exceptionally highly expressed. We revealed that KCNN1 expression is directly regulated by the disease-driving oncoprotein EWSR1-FL1. Due to its consistent overexpression in EwS, KCNN1 mRNA could be a prognostic marker in EwS. In a large cohort of EwS patients, however, KCNN1 mRNA quantity does not correlate with clinical parameters. Several functional studies including patch clamp electrophysiology revealed no evidence for KCa2.1 function in EwS cells. Thus, elevated KCNN1 expression is not translated to KCa2.1 channel activity in EwS cells. However, we found that the low K+ conductance of EwS cells renders them susceptible to hypoosmotic solutions. The absence of a relevant K+ conductance in EwS thereby provides an opportunity for hypoosmotic therapy that can be exploited during tumor surgery.


Introduction
Ewing sarcoma (EwS) is the second most common bone malignancy in children and adolescents, driven by a disease-defining translocation between chromosomes 11 and 22 (t(11;22)(q24;q12) [1]. In EwS, this translocation results in a chimeric fusion protein, mainly involving the EWSR1 (Ewing sarcoma breakpoint region 1 gene) and FLI1 (Friend leukemia integration 1 transcription factor), which can bind to GGAA microsatellites as well as to validated from previous publications when possible, or were designed de novo using Primer-BLAST [9].
RT-PCR was performed with the Red HS Master Mix (Biozym, Hessisch Oldendorf, Germany) using a Mastercycler ® (Eppendorf, Hamburg, Germany). Subsequently, gels of interest were cut out and DNA was extracted using the QIAquick ® Gel Extraction Kit (Qiagen, Hilden, Germany) following the manufacturer's protocol. Subsequently, capillary sequencing was performed using Mix2Seq overnight sequencing kits (Eurofins Scientific, Luxembourg, Luxembourg) with sequencing primers highlighted in Supplementary Data.
RT-qPCR was performed with the PowerUp™SYBR ® Green Mastermix (Thermo Fisher Scientific, Waltham, MA, USA) using a QuantStudio™ 3 cycler (Thermo Fisher Scientific, Waltham, MA, USA) and analyzed by the QuantStudio™ Design & Analysis Software (Thermo Fisher Scientific, Waltham, MA, USA). For RT-qPCR quantification, we normalized the gene expression of target genes to the mean expression of the housekeeper genes GAPDH and YWHAZ.

RNA In Situ Hybridization of Patient Tissue Microarrays
RNA in situ hybridization was performed on anonymized patient-derived TMAs derived from the cohorts Eicess92, E99 and EWING2008 [10,11], using custom-designed RNAScope™ probes (ACD Bio, Newark, CA, USA) and an RNAScope™ 2.5 HD Assay kit (ACD Bio, USA), according to the manufacturer s protocol. Briefly, paraffin-embedded TMAs were deparaffinized and treated with hydrogen peroxide before antigen retrieval. Next, tissue slices were treated with protease, followed by hybridization and signal amplification. Finally, staining was visualized using a brightfield and fluorescence microscope (Zeiss Axiovert 100, Zeiss, Oberkochen, Germany). Each TMA was scored in a blinded manner by three individual researchers from 0 to 4 with the scoring algorithm detailed in Supplementary Figure S2.
Afterwards, the mean scores were calculated and statistical analyses were performed with SAS. Correlation of dichotomized KCNN1 with clinical parameters were computed with contingency coefficient (r = C). To analyze bivariate correlations, Spearman, Pearson or exact Fisher Test correlation coefficients were calculated. For survival analysis, overall survival was described by Kaplan-Meier curves, and survival functions between the groups were analyzed by a log-rank test. Fischer's exact test was used to calculate p-values. p-values < 0.05 were considered as statistically significant. p-values were calculated from two-sided statistical tests, if not otherwise specified in the figure legends.

Analysis of RNA-Sequencing and ChIP-Sequencing Data
All analyses were run using suitable standard values for the below tools, unless stated otherwise. RNA-sequencing datasets GSE61950, GSE94277 and GSE160898 were downloaded from the NCBI SRA (Sequence Read Archive) using sratoolkit, quantified using Salmon [12], and imported into R using tximport [13]. Differential gene expression was analyzed with quasi-likelihood F-tests based on generalized linear models using edgeR [14]. Graphics were generated using ggplot2 [15]. ChIP-sequencing data from GSE61953 and from GSE101646 [16] were downloaded from the NCBI SRA, trimmed and mapped to hg38 using bbtools [17]. MACS2 callpeak and bdgcmp [18] were used to call peaks and generate fold-enrichment tracks over input, bedtools intersect [19] to filter backlisted regions [20]. Locations of GGAA microsatellites were predicted using Tandem Repeats Finder [21,22]. Graphics were generated using SparK [23].

Patch-Clamp Electrophysiology
For electrophysiolocal recordings, cultured cells were seeded on sterile round coverslips. Whole-cell recordings were performed at a conventional patch-clamp setup equipped with an EPC-10 amplifier and PatchMaster software (HEKA Elektronik, Lamprecht, Germany). Recording pipettes were fabricated from borosilicate glass (G150TF-10, Clark Electromedical Instruments, Pangbourne, UK). Typical electrode and series resistances were 3-4 MΩ and 6-10 MΩ, respectively. Series resistance compensation of ≥30% was routinely applied. Once the whole-cell configuration was achieved, the pipette solution and intracellular milieu equilibrated within 1 min. Cells were recorded at RT, held at a potential of −40 mV and membrane outward currents were evoked using a ramp protocol (5 s duration) ranging from −80 to +60 mV. The current amplitudes at −50 mV of the depolarizing ramp were analyzed. A liquid junction of 21 mV was taken into consideration. Test substances were applied in close proximity to the recorded cell using a multibarrel application pipette with a diameter of about 100 µm. The following recording solutions were used: (1) Extracellular solution (in mM): 160 Na-gluconate, 5 KCl, 10 HEPES, 1 MgCl 2 , 1 CaCl 2 ; pH and osmolality were set to 7.3 and 310 mOsmol/kg, respectively. (2) Intracellular solution containing 5 µM free Ca 2+ (in mM): 145 K-gluconate, 10 HEPES, 1.3 EGTA, 1 MgCl 2 , 1.217 CaCl 2 , 1 Mg-ATP; pH and osmolality were set to 7.2 and 295 mOsmol/kg, respectively.

Intracellular Ca 2+ Measurement
Prior to the measurements, approximately 40,000 Ewing sarcoma cells were left to adhere on glass-bottom dishes (MatTek Life Sciences, Ashland, MA, USA) coated with 1 mg/mL fibronectin (Roche, Rotkreuz, Switzerland) at 37 • C and 5% CO 2 overnight. Next, cells were stained with 6 µM of the ratiometric Ca 2+ indicator dye Fura-2-AM (Cayman Chemical, Ann Arbor, MI, USA) at 37 • C for 30 min and subsequently washed with HEPES-buffered Ringer solution (NaCl: 122.5 mM, KCl: 5.4 mM, CaCl 2 : 1.2 mM, MgCl 2 : 0.8 mM, glucose: 10 mM, HEPES: 10 mM). Subsequently, an ionic imaging setup (Visitron, Puchheim, Germany) consisting of an inverted fluorescence microscope (Zeiss Axiovert 100, Zeiss, Oberkochen, Germany), an integrated high-speed shutter system, a heated perfusion system and a polychromator was used to monitor intracellular Ca 2+ dynamics at 37 • C. During the measurement, cells were superfused initially with HEPES-buffered Ringer solution, followed by Ca 2+ -free Ringer solution (NaCl: 122.5 mM, KCl: 5.4 mM, MgCl 2 : 0.8 mM, glucose: 10 mM, EGTA: 1 mM) to eliminate extracellular Ca 2+ . Then, Ca 2+free Ringer solution was supplemented with 1 µM thapsigargin (Böhringer, Ingelheim, Germany) to allow emptying of intracellular Ca 2+ stores. Next, cells were superfused with Ca 2+ containing HEPES-buffered Ringer solution with 1 µM thapsigargin to induce SOCE in the presence or absence of the K Ca 2 channel modulator 100 nM apamin. Lastly, [Ca 2+ ] i was calibrated using Ca 2+ -free Ringer solution supplemented with the Ca 2+ ionophore 1 µM ionomycin (AK Scientific, Union City, CA, USA), then 5 mM Ca 2+ -containing Ringer solution + 1 µM ionomycin. Images were acquired every 20 s using dual excitation with wavelengths of 340 nm (F 340 ) and 380 nm (F 380 ), corresponding to the excitation maxima of the Ca 2+ -bound and Ca 2+ -free form of the fluorophore, respectively, and emission was detected at 520 nm. F 340 /F 380 ratio was derived from these values in each cell over time and [Ca 2+ ] i was calculated as described by Grynkiewicz et al. [24].
Spheroid invasion assays with Ewing sarcoma and osteosarcoma spheroids were performed as previously described [25]. First, cells after trypsinization were counted and resuspended in a medium with 0.24% methyl cellulose (Sigma-Aldrich, St. Louis, MO, USA). Next, droplets of 50 µL, containing 3000 cells each, were allowed to form spheroids as hanging drops in a 10 cm Petri dish at 37 • C and 5% CO 2 for 48 h. Afterwards, spheroids were embedded into 100 µL droplets of collagen I-based extracellular matrix with the composition detailed above in 12.5 cm 2 flasks. After matrix polymerization at 37 • C and 5% CO 2 for 2 h, cell culture medium supplemented with 1µg/mL IGF-1 (Peprotech, Rocky Hill, KT, USA) to stimulate matrix invasion of sarcoma cells. Furthermore, medium was supplemented with either 0.1% DMSO for control or with 100 nM Apamin (Alomone, Jerusalem, Israel) and 10 µM Senicapoc (gifted by Prof. Dr. Heike Wulff, University of California Davis) for K Ca channel inhibition. After pH equilibration of the medium described above, spheroid invasion was recorded at 37 • C using phase-contrast microscopy over 48 h at a time interval of 10 min. Analysis was carried out using ImageJ. The spheroid's size over time was measured every hour for the whole observation period. In the first 8 h, the SOAS exercised a visible pull to the matrix, which was additionally measured every ten minutes.

Cell Viability Assays
Assessment of Ewing Sarcoma cell viability treated with K Ca channel inhibitors in the presence of chemotherapeutic drugs was performed using the MTT assay. Briefly, 27,500 cells were seeded into wells of a 96-well plate and after overnight incubation at 37 • C, 5% CO 2 , cells were treated with chemotherapeutic drugs routinely used in the treatment of Ewing sarcoma, namely, doxorubicin and/or vincristine (Sigma-Aldrich, St. Louis, MI, USA). Moreover, the effect of chemotherapy was investigated in the presence or absence of the K Ca 2.1 inhibitor apamin. Treatment duration was 72 h, after which cells were treated with 0.5 mg/mL MTT reagent (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide, Sigma-Aldrich, St. Louis, MI, USA) at 37 • C, 5% CO 2 for 1 h. Next, plates were centrifuged at 300 g for 5 min, and the solution was exchanged to 100 µL DMSO in each well. MTT absorbance was detected using a spectrophotometer (THERMOmax, Beckman Coulter, Brea, CA, USA) at wavelengths of 546 nm, as well as 650 nm as a reference wavelength. Data were normalized to the absorbance of the control treated with 0.1% DMSO. Each set of experiments was performed with technical triplicates.
Ewing sarcoma and osteosarcoma cell survival after hypoosmotic shock treatment was performed using the sulforhodamine B (SRB) assay. To examine the osmotic resistance of cells, a series dilution was performed by mixing cell culture medium with hypotonic solution (26.8 mOsm) containing 1.2 mM CaCl 2 , 0.8 mM MgCl 2, 5.4 mM KCl, and 10 mM HEPES. For the nominally zero mOsm solution, dH 2 O was used. Prior to the experiment, 20,000 cells were seeded into wells of a 96-well plate coated with 25 µg/mL fibronectin at 37 • C, 5% CO 2 for 24 h. Next, cell culture medium was exchanged to solutions with different osmolarities, namely 279 mOsm, 153 mOsm, 111 mOsm, 77 mOsm, 55 mOsm, 27 mOsm and 0 mOsm, for 5 min. After 5 min, wells were washed once with the same solution to eliminate dead cells that lost adherence. Afterwards, SRB assay was performed as described by Orellana and Kasinki [27]. Briefly, cells in each well were fixed using trichloracetate at 4 • C for 60 min. Then, the wells were washed four times with dH 2 O, after which fixed cells were stained with 0.04% SRB (Sigma-Aldrich, St. Louis, MO, USA) solved in 1% acetic acid at RT for 1 h. Next, wells were washed four times with 1% acetic acid, then were left to air dry for an additional 30 min. Lastly, SRB-protein complexes in each well were dissolved using Tris-Base-solution (10 mM, pH 10.5). SRB absorbance was detected with the same settings as MTT absorbance (see above). Data were normalized to the absorbance of control treated with the isosmotic (279 mOsm) solution. Each set of experiments was performed with technical triplicates.

Cell Volume Measurements
For cell volume measurements by fluorescence exclusion microscopy [28], cells were seeded into Luer µ-Slides (Ibidi, Gräfelfing, Germany) coated with Fibronectin (25 µg/mL; Roche, Rotkreuz, Switzerland) and left to adhere in HEPES-buffered RPMI 1640 medium at 37 • C overnight. Next, the Luer µ-Slides were superfused with solutions containing 400 µg/mL 70 kDa-FITC-dextran using a syringe and visualized using a confocal microscope (Lecia DMI 6000, Leica, Mannheim, Germany). Because of the high molecular mass, FITC-dextran does not penetrate the cell membrane, so we obtained negative images of the cells. To obtain a good approximation of cellular volume, z-stacks of 0.3 µm resolution were taken, and measured cell volume was calculated by performing a volume rendering with Amira 2019.2 (Thermo Fisher Scientific, Waltham, MA, USA). The measured volumes were multiplied by 0.83 because of an aberration induced by the different refractive indices of the immersion oil and the Ringer solution [29]. During the measurement, cells were firstly incubated in isosmotic HEPES-buffered Ringer solution (271.8 mOsm, composition see above), and after 5 min, medium was exchanged with an isosmotic solution where NaCl was partly replaced by mannitol. By means of the mannitol equilibration, it was ensured that cell volume changes were not induced by single electrolyte concentrations, but only by the osmolarity. Next, a hypoosmotic solution without mannitol (171.8 mOsm or 221.8 mOsm) was used to induce cell swelling and subsequently RVD in the presence or absence of the K Ca 2.1 inhibitor 100 nM apamin. Z-stacks were acquired every 2 min for 20 min. Analysis was carried out using the Amira software. Cell volumes of every cell were normalized to the initial cell volumes in the mannitol-equilibrated isosmotic solution.

Multiple Ca 2+ -Modulated Ion Channels Are Functional in Ewing Sarcoma
Osteolytic growth of EwS is coupled with marked local Ca 2+ release from the degraded bone. We hypothesized that tumor cells adapt to the increased [Ca 2+ ] e by expressing Ca 2+regulated ion channels ( Figure 1A). Multiple Ca 2+ -modulated ion channels are involved directly or indirectly in store-operated entry (SOCE). Intracellular Ca 2+ homeostasis as well as the relevant Ca 2+ regulatory ion channels in EwS have not yet been studied. We conducted intracellular Ca 2+ measurements in cells from two established EWS cell lines ( Figure 1B-D), where we found that both SK-ES-1 and A673 cells have a resting [Ca 2+ ] i ranging from 100 to 200 nM ( Figure 1C; SK-ES-1: 174 ± 9 nM Ca 2+ ,N = 4, n = 108; A673: 132 ± 9 nM Ca 2+ , N = 4, n = 111). Moreover, EwS cells contain functional Ca 2+ stores and utilize SOCE triggered by thapsigargin (TG) stimulation ( Figure 1D, peak [Ca 2+ ] i concentrations of SK-ES-1: 576 ± 26 nM Ca 2+ , N = 4, n = 108, p < 0.0001; A673: 266 ± 18 nM Ca 2+ , N = 4, n = 111, p < 0.0001).  To gain a better understanding of the potential molecular players of Ca 2+ signaling in EwS cells, we inspected the expression of [Ca 2+ ] i regulating ion channel genes in a published EwS RNAseq surfaceome database [30]. We selected 22 highly expressed genes from the TRP, Stim/Orai, K Ca , Ca v and Piezo families of ion channels and confirmed their expression using RT-qPCR in three EwS cell lines, namely, SK-ES-1, TC-71 and A673 ( Figure 1E, expression relative to GAPDH and YWHAZ, N = 3, n = 6). Seven of the twentytwo genes are expressed at a high level in all three EwS cell lines, namely TRPM7, TRPC1, STIM2, ORAI1, ORAI2, KCNN1 and PIEZO1. Taken together, we found that multiple ion channels possibly involved in intracellular Ca 2+ homeostasis and SOCE are present in EwS cells. Based on its particularly high abundance and its unique biophysical and pharmacological properties we decided to focus on KCNN1 and its protein product, the K Ca 2.1 channel, in this study.

EWSR1-FLI1 Drives the Expression of an Alternative KCNN1 Transcript Variant in Ewing Sarcoma
To investigate whether KCNN1 has a specificity for EwS over the alternative primary bone sarcoma in this age group, osteosarcoma, we performed RT-qPCR examining a larger panel of EwS cell lines (SK-ES-1, A673, TC-71, 5838, TTC-466 and CADO) and compared KCNN1 expression to the osteosarcoma cell lines HOS, SAOS and ZK-58 ( Figure 2A). We found that osteosarcoma cell lines have negligible KCNN1 expression compared to EwS cell lines (expression relative to GAPDH and YWHAZ: EwS: 0.02 ± 0.003, N = 3, n = 48; OS: 0.0001 ± 0.00002, N = 3, n = 22). Furthermore, we found out that KCNN1 is expressed in tumor cells but not in the surrounding healthy bone tissue using RNA in situ hybridization (RNAish) ( Figure 2B, N = three whole EwS tissue slides). Overall, these findings indicate that KCNN1 has a considerable tissue specificity in EwS compared to osteosarcoma and healthy bone tissue.
Next, we analyzed whether EWSR1, when fused to ETS family transcription factors such as FLI1, drives the expression of KCNN1. As a proof of principle, we focused on the fusion oncogene EWSR1-FLI1, and evaluated whether it directly or indirectly affects KCNN1 expression. RNAseq datasets (GSE27524 and GSE103837, respectively) from two independent reports by Bilke et al. [31] and by Zöllner et al. [32] show that KCNN1 is downregulated upon EWSR1-FLI1 knockdown. We reanalyzed the ChiPseq datasets GSE101646 and GSE61953 [16,33] and found that EWS-FLI1 establishes an enhancer at the KCNN1 locus. H3K4me1 (mono-methylation at the fourth lysine residue of the histone H3 protein) marks two poised enhancers at the KCNN1 locus that are inactive in HEK293 cells. FLI1 directly binds to an alternative enhancer in the third intron of KCNN1 ("Enhancer B" in Figure 2C), which consists of GGAA microsatellites. This leads to a strong activation of this enhancer in EwS cells, as evidenced by increased H3K27ac (acetylation of the lysine residue at N-terminal position 27 of histone H3). Moreover, around the GGAA microsatellite, the enhancer ("Enhancer B") signal becomes much broader. Activation of this enhancer leads to transcriptional activation of the KCNN1 gene (as visualized by H3K4me3 (tri-methylation at the fourth lysine residue of the histone H3 protein)) in the EwS cell lines A673 and SKMNC but not in HEK293 lacking FLI1 activity. Overall, the differential methylation and acetylation patterns indicate that the usage of an alternative enhancer facilitates the transcription of KCNN1. As a further link between EWSR1-FLI1 and KCNN1, we found a more than two-fold higher expression of KCNN1 in EWSR1-FLI1-transfected HEK293 cells compared to control transfection ( Figure 2D, expression relative ctrl transfection, ctrl vs. EWSR1-FLI1 transfection: 1.00 ± 0.07, N = 3, n = 12; vs. 2.17 ± 0.24, N = 3, n = 9, p < 0.0001). Published datasets from GSE160898 [34] confirm that overexpression of EWSR1-FLI1 induces KCNN1 expression in HEK293 cells ( Figure 2E). In contrast, when EWSR1-FLI1 is either inhibited pharmacologically or knocked down in A673 and SKMNC, the expression of KCNN1 drops significantly ( Figure 2E, data from GSE94277 and GSE61950, respectively [33,35]).  To verify that the above results are not due to the lack of primer specificity, we amplified longer sequences of canonical and alternative transcripts ( Figure 3C Surprisingly, we also found a short transcript when amplifying canonical exons 1-8. To assess the sequence of the short transcript, and to confirm the sequence of the other transcripts, we resorted to capillary sequencing ( Figure 3D, Suppplementary Data). This confirmed the presence of the canonical transcript (NM_002248) and the alternative transcript variant 2 (NM_001386974). We termed the previously not characterized transcript variant 2 of KCNN1 driven by EWSR1-FLI1 KCNN1 EWS . As exon 3 is upstream of the open reading frame (ORF), the protein sequence resulting from KCNN1 EWS translation does not change compared to the canonical K Ca 2.1. Sequencing also revealed that the short transcript is due to splicing out exons 5-6. This results in the lack of the S2 and S3 transmembrane domains of the resulting K Ca 2.1 channel protein, likely rendering the product non-functional.

KCNN1 Expression Does Not Lead to K Ca 2.1 Function in Ewing Sarcoma
Based on the above results ( Figure 2B), we tested whether KCNN1 could be a feasible diagnostic or prognostic marker in EwS. As available antibodies against K Ca 2.1 are not specific (Supplementary Figure S4, Supplementary Method), we used a human KCNN1-specific 20ZZ probe (targeting nucleotides 1153-2659 recognizing all three described transcripts) for RNAish in tissue microarrays of 217 EwS patients from cohorts Eicess92, E99 and EWING2008 [36][37][38] (Figure 4A).

KCNN1 Expression Does Not Lead to KCa2.1 Function in Ewing Sarcoma
Based on the above results ( Figure 2B), we tested whether KCNN1 could be a feasible diagnostic or prognostic marker in EwS. As available antibodies against KCa2.1 are not specific (Supplementary Figure S4, Supplementary Method), we used a human KCNN1specific 20ZZ probe (targeting nucleotides 1153-2659 recognizing all three described transcripts) for RNAish in tissue microarrays of 217 EwS patients from cohorts Eicess92, E99 and EWING2008 [36][37][38] (Figure 4A). There was a substantial heterogeneity in the RNAish signal of EwS patients and we clustered 78% of the patients into the KCNN1 high group, whereas RNAish signal was scarce in 22% of EwS patients (KCNN1 low group) ( Figure 4A,B). In a double-blinded experiment, we correlated RNAish signal intensity with various aspects of clinical outcome such as overall survival ( Figure 4C). We found no correlation with overall survival (hazard ratio and 95%CI: 0.85 (0.45-1.61), p = 0.62) and with other clinical parameters such as tumor volume, age at diagnosis, presence of primary metastases, therapy response, and time to first relapse (summarized in Supplementary Table S2). In summary, quantitative KCNN1 expression assessed using RNAish is not a suitable prognostic marker. There was a substantial heterogeneity in the RNAish signal of EwS patients a clustered 78% of the patients into the KCNN1 high group, whereas RNAish signal was in 22% of EwS patients (KCNN1 low group) ( Figure 4A,B). In a double-blinded exper we correlated RNAish signal intensity with various aspects of clinical outcome s overall survival ( Figure 4C). We found no correlation with overall survival (hazar and 95%CI: 0.85 (0.45-1.61), p = 0.62) and with other clinical parameters such as volume, age at diagnosis, presence of primary metastases, therapy response, and first relapse (summarized in Supplementary Table S2). In summary, quantitative K expression assessed using RNAish is not a suitable prognostic marker.

K Ca 2.1 Is Non-Functional in EWS Cells
Having ruled out the possibility of using KCNN1 as a prognostic marker, we tested whether the druggability of the protein product of KCNN1 could be potentially exploited for EwS therapy ( Figure 5). K Ca 2.1 can be selectively inhibited by the bee toxin apamin and has a very restricted tissue distribution [39], making it an attractive therapeutical target. We therefore addressed the impact of K Ca 2.1 function on numerous other aspects of EwS cell behavior ( Figure 5A). Our cell migration studies on a collagen-coated two-dimensional substrate ( Figure 5B We considered the possibility that K Ca 2.1 might be expressed intracellularly, e.g., in mitochondria, similar to other K Ca channels [40]. Therefore, we inhibited K Ca 2.1 also with the membrane-permeable inhibitor NS8593 (Supplementary Figure S5) [41]. We found that 1 µM NS8593 does not affect EwS cell migration (migration velocities of control vs. NS8593; A673: 0.31 ± 0.05 µm/min vs.  To test whether K Ca 2.1 is involved in EwS cell migration in a more complex system, we created spheroids from SK-ES-1 cells, embedded them into a complex collagen-based extracellular matrix similar to the bone [5] and assessed the motility of EwS cells within these structures ( Figure 5C). Inhibition of both K Ca 2.1 and K Ca 3.1 by using apamin and senicapoc in combination failed to attenuate EwS cell motility out of the spheroid (relative SK-ES-1 spheroid area @48 h control vs. apamin+senicapoc: 2.14 ± 0.13 vs. 2.50 ± 0.26 µm/min, N = 4, n = 8, p = 0.25). Generally, we found that SK-ES-1 cells are not particularly motile in a three-dimensional environment (Supplementary Video S1). For comparison, we quantified their motility along with cells from an osteosarcoma cell line (SAOS, Figure 5D,E, Supplementary Video S2). SAOS osteosarcoma cells are vividly emigrating out of the spheroid into their surroundings (relative spheroid area @48 h SK-ES-1 vs. SAOS: 2.14 ± 0.13 vs. 3.30 ± 0.54 µm/min, N = 4, n SK-ES-1 = 8, n SAOS = 7, p = 0.03). In summary, EwS cell migration in a two-dimensional as well as in a three-dimensional system both remain unaltered upon K Ca 2.1 modulation.
To further confirm the lack of K Ca 2.1 function in EwS cells, we tested whether 100 nM apamin has any effects on Ca 2+ homeostasis ( Figure 5H). As expected, we found that apamin has no effect on SOCE in both A673 and SK-ES-1 cells ( Figure 5I Overall, the results from a variety of functional in vitro assays ( Figure 5) consistently argue against any functional role of K Ca 2.1 in EwS cell lines despite its overexpression. This is consistent with the fact that KCNN1 expression does not correlate with the clinical parameters of EwS (Figure 4).

KCNN1 Transcripts in EwS Are Non-Conductive
Our results outline a discrepancy between very robust KCNN1 expression derived from molecular biology (Figures 1-3) and a lack of ion channel function, as derived from our clinical prognostic and functional cellular analyses (Figures 4 and 5). To resolve the discrepancy, we hypothesized that KCNN1 transcripts in EwS conduct no K + and thereby carry no conductive role in EwS cell physiology. To confirm this idea, we utilized the gold-standard assay for ion channels, whole-cell patch-clamp ( Figure 5A-C). In SK-ES-1 and TC-71 cells, membrane currents were generally small and revealed a rather linear current-voltage relationship reversing at around −33 mV, thereby indicating a deficiency in functional K + channels. As expected, we found that 100 nM apamin has no effect on ionic currents in EwS cells ( Figure 6A,C; currents at −50 mV control vs. 100 nM apamin; SK-ES-1: −35 ± 10 pA, versus −50 ± 14 pA, N = 5, n = 12, p = 0.38; TC-71: −44 ± 18 pA versus −38 ± 14 pA, N = 5, n = 12, p = 0.81). Strikingly, EwS cells lack outward current even at −50 mV, indicating that they lack any K Ca function. In addition, the current slope is largely linear, indicating a deficiency in K V channels and, generally, K + conductance. Notably, we found that the reversal potential of EwS cell lines is ≈ −33 mV at the given ionic composition of the medium. Thus, the background ionic conductance of the EwS cell membrane is very likely due to Cl − , that has an equilibrium potential of −30 mV.

Lack of KCa2.1 Channel Function Sensitizes EwS Cells to Hypoosmotic Stress
Throughout our study, the absence of any detectable KCa2.1 channel activity, and more broadly, K + channel function in EwS cells was perplexing. Despite the negative results, we saw this as an opportunity to selectively target EwS cells. Generally, K + conductance has a vital role in cell volume homeostasis. Thus, we hypothesized that the absence of K + channel function hampers the defensive mechanisms of EwS cells to hypoosmotic stress. As a proof of concept, we applied hypoosmotic shock (170 mOsm) on EwS cells Next, we tested whether the absence of K Ca 2.1 function can possibly be explained by a dominant negative effect of the short KCNN1 transcript or KCNN1 EWS . We hypothesized that these variants, similarly to K Ca 3.1, could form heteromers with K Ca 2.1 [44], rendering the channel non-functional. To this end, we transfected HEK293 cells with the canonical KCNN1 and/or EWSR1-FLI1 ( Figure 6B). HEK293 cells, transiently transfected with KCNN1, expressed apamin-sensitive K Ca 2.1 currents, verifying our earlier methodology (currents at −50 mV HEK-KCNN1 control vs. 100 nM apamin: 435 ± 134 pA versus 39 ± 30 pA, N = 3, n = 7, p = 0.01). In contrast, HEK293 cells, transfected solely with EWSR1-FLI1, expressed no apamin-sensitive K Ca 2.1 currents (currents at −50 mV HEK-EWSR1-FLI1 control vs. 100 nM apamin: −46 ± 9 pA, −45 ± 9 pA, N = 3, n = 7, p = 0.95), even though they expressed KCNN1 ( Figure 2D). When co-transfecting HEK293 cells with EWSR1-FLI1 and KCNN1, we observed similar currents as with single KCNN1 transfection (currents at −50 mV HEK-KCNN1 vs. HEK-EWSR1-FLI1-KCNN1: 435 ± 134 pA versus 360 ± 56 pA, N = 3, n = 7, p = 0.63). This evidence argues against a dominant negative effect of alternative KCNN1 transcripts in EWS. During these experiments, we recorded the resting membrane potential (RMP) of the measured cells ( Figure 6D). EwS cell lines have a markedly depolarized membrane potential near the equilibrium potential of Cl − (SK-ES-1: −24 ± 3 mV, TC−71: −23 ± 3 mV), which is a logical consequence of their absent K + conductance. In contrast, the resting membrane potential of HEK293 cells transfected with the canonical K Ca 2.1 is largely hyperpolarized (HEK-KCNN1: −78 ± 3 mV, HEK-EWSR1-FLI1-KCNN1: −67 ± 9 mV). In summary, EWSR1-FLI1-induced pathognomonic KCNN1 variants are non-functional transcripts that likely do not exert a dominant negative effect on canonical K Ca 2.1 channel function.

Lack of K Ca 2.1 Channel Function Sensitizes EwS Cells to Hypoosmotic Stress
Throughout our study, the absence of any detectable K Ca 2.1 channel activity, and more broadly, K + channel function in EwS cells was perplexing. Despite the negative results, we saw this as an opportunity to selectively target EwS cells. Generally, K + conductance has a vital role in cell volume homeostasis. Thus, we hypothesized that the absence of K + channel function hampers the defensive mechanisms of EwS cells to hypoosmotic stress. As a proof of concept, we applied hypoosmotic shock (170 mOsm) on EwS cells and observed a regulatory volume decrease (RVD) ( Figure 7A,B) [8]. Commonly, eukaryotic cells can completely regenerate their original volume within 10-20 min after mild hypotonic shock. However, we found that TC-71 cells barely regulate their volume, as represented by a low overall RVD even after 20 min ( Figure 7C). Moreover, we tested whether the inhibition or genetic knockdown of the K Ca 2.1 affects the volume regulation of TC-71 cells. We found no significant differences in the cell volume between KCNN1-silenced cells and cells treated with scrambled control siRNA ( Figure 7C  . When a cell is exposed to a hypotonic solution, it swells due to the osmotic gradient. The subsequent RVD is a counter-regulatory mechanism which is mainly ensured by ion-channel-mediated K + -and Clefflux through the plasma membrane, which is followed by a water loss. (B) Representative confocal images of TC-71 EwS (top) SAOS osteosarcoma (bottom) cells (black) which were exposed to a hypotonic solution (220 mOsm) containing high-molecular-weight FITC-dextran (green). Cells first swell, which could be observed at t = 2 min. Thereafter, cells undergo RVD aiming to regenerate their original volume. Scale bar (lines)=50 µm.(C) RVD performance of TC-71 cells treated with siKCNN1 (red, N = 4, n = 32) or with scrambled siRNA (black, N = 4, n = 38), indicated by changes in their relative cell volume . When a cell is exposed to a hypotonic solution, it swells due to the osmotic gradient. The subsequent RVD is a counter-regulatory mechanism which is mainly ensured by ion-channel-mediated K + -and Cl − efflux through the plasma membrane, which is followed by a water loss. (B) Representative confocal images of TC-71 EwS (top) SAOS osteosarcoma (bottom) cells (black) which were exposed to a hypotonic solution (220 mOsm) containing high-molecular-weight FITC-dextran (green). Cells first swell, which could be observed at t = 2 min. Thereafter, cells undergo RVD aiming to regenerate their original volume. Scale bar (lines)=50 µm. To verify whether this poor volume regulation of the EwS cells results in a vulnerability for hypotonic stress, EwS and osteosarcoma cells were exposed to different hypotonic solutions for 15 min, and cell survival was subsequently measured with a sulforhodamine B (SRB)-assay ( Figure 7E). After incubation with hypotonic solutions of 77 mOsm or less, TC-71 EwS cell survival was considerably lower than that of SAOS osteosarcoma cells (relative SRB absorbance at 77 mOsm, TC-71 versus SAOS, 0.57 ± 0.07 versus 0.96 ± 0.03, N = 3, n = 9, p < 0.0001). The IC 50 value of osmolarity-dependent survival of TC-71 was 65.7 mOsm. In sharp contrast, SAOS cells were still viable even after 15 min in distilled water (relative SRB absorbance after incubation in distilled water, TC-71 (Supplementary Video S3) versus SAOS (Supplementary Video S4) (0.01 ± 0.01 versus 0.59 ± 0.03, N = 3, n = 9, p < 0.0001).

Discussion
Our study aimed to better understand the complex Ca 2+ homeostasis of an osteolytic tumor, EwS, and specifically to outline the "Ca 2+ toolkit" necessary for EwS cells to cope with increased environmental [Ca 2+ ]. Our work sheds first light on the expression of ion channel genes that may coordinate Ca 2+ homeostasis in EwS. Among them, we found the expression of genes that encode channels permeable to Ca 2+ , such as SOCE-mediating channels, Piezo 1 and 2, as well as voltage-gated Ca 2+ channels (e.g., Ca V 3.1) and TRP channels (including TRPC1, TRPM7, and TRPV4). We also found prominent gene expression in the case of K + channels that are activated by a rise in the intracellular Ca 2+ concentration, specifically in KCNN1 encoding K Ca 2.1. To understand why KCNN1 expression is prominent, we assessed the epigenetic landscape of the locus. Previously, it has been described that the promoters of multiple voltage-gated K + channels' genes are differentially methylated in EwS when compared with nonmalignant adult tissues [45]. Additionally, differential promoter methylation also causes epigenetic dysregulation in the KCNN4 locus in lung cancer [46]. Here, we demonstrated a differential histone methylation and acetylation pattern (Figure 2) in the KCNN1 locus with the consequence of activating an alternative enhancer repressed in healthy tissue and ultimately driving the expression of the alternative channel transcript that we termed KCNN1 EWS . Whether other Ca 2+ (regulated) ion channel genes outlined in our work could also have cryptic enhancers that become activated by the oncoprotein EWSR1-FLI1 remains to be studied.
Throughout our study, we were confronted with the discrepancy that high KCNN1 expression at the mRNA level in all the EwS lines studied as well as tumor cells in EwS patient material is by no means reflected in the functional assays and in the clinical data. The seemingly simple, and at the same time intriguing, explanation is that there is no K Ca 2.1 channel-mediated current in Ewing sarcoma cells and, consequently, none of the functional assays revealed an impact of the channel on EwS cell behavior. We addressed this conundrum in multiple ways. In most assays, we used the bee venom apamin to inhibit K Ca 2.1. Because apamin is a polypeptide acting from the extracellular side [47], we considered that the reduced effectiveness of apamin can be due to K Ca 2.1 being localized intracellularly (in analogy to mitochondrial K Ca 2.2 and K Ca 2.3 channels [48,49]). However, the channel is likely not active in intracellular membranes such as mitochondria in EwS cells, because the membrane-permeable inhibitor NS8593 fails to affect cell migration (Supplementary Figure S5). Similar results for non-functional K Ca channels have already been described [50]. The Sontheimer group detected expression of all K Ca channel genes in glioma cells, but they could only measure currents of the K Ca 1.1 channel across the cell membrane. Neither K Ca 2-nor K Ca 3.1-carried membrane currents could be measured. Apparently, cancer cells functionally express only a very limited array of the potentially available ion channels.
This raises the question of whether the KCNN1 mRNA is translated into K Ca 2.1 protein in EwS tumor cells at all. Naturally, we aimed to assess protein expression using Western blot, but we found every commercially available antibody we tested to be unspecific for K Ca 2.1 (Supplementary Figure S4). Ion channel proteins often have non-conductive properties, e.g., behaving as membrane anchors [51]. There is, however, no information on non-conducting properties of the K Ca 2.1 channel, likely because of the lack of a specific probe. By using siRNA against KCNN1, however, we could not observe any difference in cell volume homeostasis and RVD ( Figure 7C). This evidence argues against a nonconducting channel function. Taken together, we presume that the KCNN1 mRNA and the K Ca 2.1 protein-if translated-is most likely degraded or sequestered. A potential mechanism for this could be post-transcriptional mRNA editing that has already been described for ion channels [52][53][54]. To study the events that happen after translation and to better understand the K Ca 2.1 channel in general, a novel specific K Ca 2.1 antibody would need to be designed. Another plausible approach would be to conjugate the bee toxin apamin with a fluorophore, similarly to scorpion toxins for detecting K V channels [55].
We detected multiple transcript variants of KCNN1 in EwS. Among these, the transcript variant 2, termed KCNN1 EWS, is intriguing, as only untranslated exons are altered compared to the canonical transcript variant. Whether this affects channel expression in EwS cells or in a model system still remains to be investigated in more detail. The truncated ∆S2-S3 isoform of KCNN1 raises the possibility of a dominant negative effect, as earlier described for KCNN3 [56]. However, we failed to confirm the dominant negative effect when heterologously expressing EWSR1-FLI1 together with KCNN1, which argues against a possible dominant negative effect. In addition, due to the several transcripts found in EwS, we detected exons preserved in every transcript variant (Exons 7-12) in our RNAish study. Whether there is a spatial pattern in the expression of the different transcript variants or whether any specific transcript variant correlates with the clinical features of EwS is an open question to be pursued in future studies.
A surprising finding was the generally low K + conductivity of the EwS cell membrane. Biophysically, this would make EwS cells ideal models for K + channel transfection. From a physiological point of view, this feature leaves them vulnerable to sudden changes in environmental osmolarity. The question is whether the vulnerability of EwS cells to osmotic stress can also be exploited therapeutically, for example in the form of lavage with distilled water during surgery to lyse any tumor cells remaining after resection. Such a form of tumor therapy is already established, for example, in ruptured hepatocellular carcinoma, where peritoneal lavage with distilled water significantly improves disease-free survival as well as overall survival [57,58]. However, in EwS, it is unlikely that local control and prognosis can be further improved by hypotonic lavage of the surgical site, since tumors located in the extremities can be surgically removed by wide resection-a key component of standard therapy. Moreover, since EwS, unlike hepatocellular carcinoma, is radiosensitive, unresectable local disease can mostly be controlled by radiotherapy. Prognosis limiting in EwS is rather the presence of hematogenous (micro-)metastases, causing distant relapses even after effective local tumor therapy and not easily amenable to osmotic manipulation. Therefore, it would be vital to design a systemic therapy that induces osmotic stress in vivo by selectively targeting tumor cells.

Conclusions
In summary, numerous Ca 2+ (regulated) ion channel genes are expressed in EwS that may protect cells from the elevated environmental [Ca 2+ ] upon tumor-induced osteolysis. From these genes, we followed up on the prominent expression of KCNN1, a channel that is normally only expressed in excitable cells. We elucidated in detail that the oncoprotein EWSR1-FLI1 directly activates an enhancer in the KCNN1 locus by directly binding to GGAA microsatellites. The activation of the enhancer leads to the transcription of a novel transcript variant KCNN1 EWS . In a large EwS patient cohort, we found that KCNN1 mRNA does not correlate with any of the investigated clinical parameters. In hindsight, this is not surprising, as we found that K Ca 2.1, the channel encoded by KCNN1, elicits no K + currents in EwS cells and has no measurable implication in numerous aspects of EwS cell physiology. Lastly, we found that the generally low K + conductance of EwS cells leads to an ineffective volume homeostasis, which we could then exploit by inducing EwS cell death by hypoosmotic stress.