Normal and Malignant Cells Exhibit Differential Responses to Calcium Electroporation

Calcium electroporation is a promising novel anticancer treatment in which electroporation induces high calcium influx into treated cells (1). Electroporation is a method where application of short, high voltage pulses induces a brief permeabilization of the plasma membrane letting ions and molecules enter and leave the cell cytosol (2–6). A more than 1,000-fold inward directed calcium gradient may lead to considerable influx of calcium that can reach cytotoxic levels if cells cannot extrude calcium efficiently. This novel treatment has previously been shown to induce cell death in vitro and tumor necrosis in vivo, associated with acute ATP depletion and thereby loss of energy for, among others, calcium pumps (1, 7). Because of the increasing use of electrochemotherapy, a treatment where a chemotherapeutic drug, for example, bleomycin is internalized by electroporation, electroporation equipment for clinical use is available in many cancer centers (8–12). We have previously shown similar electroporation parameters can be used for calcium electroporation as are used for electrochemotherapy (13). Thus, the clinically available electroporation equipment can also be used for calcium electroporation and indeed the first clinical trial for treatment of cutaneous metastases randomized between electrochemotherapy using bleomycin (a current standard treatment) and calcium electroporation (ClinicalTrials.gov ID: NCT01941901) is now completed.

Intracellular free calcium (Ca²⁺) is a very important second messenger involved in numerous intracellular processes from fertilization through development, differentiation, and proliferation, to cell death (14, 15). Thus, cells have to chelate, compartmentalize, or extrude calcium to maintain homeostasis such that free intracellular calcium is sub-micromolar. Cells extrude calcium from the cell by the plasma membrane calcium ATPase (PMCA) that exchanges protons for one Ca²⁺ per ATP hydrolyzed. PMCA is encoded by four genes (PMCA1–4) where PMCA-1 and 4 are expressed in all cells, whereas PMCA-2 and 3 are mainly expressed in the nervous system (16–19). In different colon and gastric cancer cell lines, lower PMCA-4 levels were observed compared with normal cell lines (20, 21). In addition, calcium is extruded from the cell cytosol by the ATP-dependent sarcoendoplasmic reticulum calcium ATPase (SERCA) and the ATP-independent sodium/calcium exchangers (NCX).

Interestingly, it has recently been shown that cancer cell lines are more sensitive to calcium electroporation than normal dermal fibroblasts when treated in spheroids, a 3D multicellular in vitro model. Intracellular ATP level was also determined in this study and showed ATP depletion in all cell spheroids, including the normal cell spheroids (22).

The question of sensitivity of normal and malignant cells to cancer therapy is absolutely fundamental, and must be addressed also for calcium electroporation. The aim of this study was to determine sensitivity to calcium electroporation in bladder, colon, lung, and breast cancer tumors in a mouse model, combined with examination of normal tissue response, along with dose response. Furthermore, finite element model calculations of electric fields in tumor and surrounding skin and muscle was performed, intracellular calcium level in cell lines and tumors after calcium electroporation was investigated, as were plasma membrane calcium ATPase levels.

Five different human cell lines (four cancer cell lines and normal primary fibroblasts) were used to represent different cancer histologies and the normal dermal fibroblasts were chosen since calcium electroporation has been tested on cutaneous metastases in a clinical trial. H69 (kindly provided by the Department of Oncology, Copenhagen University Hospital, Denmark in 2010) is a human small-cell lung carcinoma described in ref. 23, HT29 (ATCC #HTB-38; obtained in 2012) is a human colorectal adenocarcinoma, and these two cell lines were grown in RPMI1640 culture medium (Gibco, Invitrogen). MDA-MB-231 (ATCC #HTB-26; obtained in 2002) is a human breast adenocarcinoma, SW780 is a human bladder transitional cell carcinoma (kindly provided by Dr. Lars Dyrskjøt Andersen, Department of Molecular Medicine, Aarhus University Hospital, Skejby, Denmark in 2012) described in ref. 24, and HDF-n is normal neonatal primary dermal fibroblasts (kindly provided by Dr. Marie-Pierre Rols, Institute of Pharmacology and Structural Biology, IPBS, Toulouse, France in 2014) described in ref. 22, all grown in DMEM culture medium (Gibco, Invitrogen). All culture media were added 10% FCS (Gibco, Invitrogen), 100 U/mL penicillin, and 100 μg/mL streptomycin. Cells were maintained at 37°C and 5% CO₂. All cell lines were tested negative for mycoplasma (MycoAlert, Lonza). The cancer cell lines were without signs of infection when tested by rapid MAP27 panel (Taconic) and were authenticated in March 2015 by short tandem repeat (STR) profiling (LGC Standards) showing perfect match for H69, HT29, and SW780. The MDA-MB-231 cell line matched in 7 of 9 profile loci (loci D7 and VWA had lost a peak). Loss of heterozygosity (LOH) is a common feature of cancer cell lines and these two profile changes have been assessed not to influence the results of these experiments. Experiments with all cancer cell lines were performed within 2 months from thawing. Experiments with the normal cells were performed within 10 passages from collection.

All in vivo experiments were conducted in accordance with European Convention for the Protection of Vertebrate Animals used for Experimentation and with approval from the Danish Animal Experiments Inspectorate (# 2012-15-2934-00091) and we adhere to the 3 Rs (replacement, reduction, and refinement). Experiments were performed as described previously (1). Briefly, 5.0 × 10⁶ cells in 100-μL PBS (except for the MDA-MB-231 cells where 2.5 × 10⁶ cells in 100-μL PBS were used) were injected subcutaneously or transplanted in the flank of 9–19 weeks old NMRI-Foxn1^nu mice (males and females, 21–43 g at injection/transplantation time; Harlan and bred at Department of Oncology, Herlev Hospital, Denmark). When tumor volume reached 85 mm³, mice were randomized into groups, as described below. Tumor volume was calculated as a x b² x π/6, where a is the largest diameter and b is the longest diameter perpendicular to a.

Tumors treated with electroporation were subjected to 8 pulses of 1,000 V/cm (applied voltage to electrode distance ratio), 100 μs, and 1 Hz using a square wave electroporator (Cliniporator, IGEA). Plate electrodes of 10 mm in width and 6 mm between the electrodes were placed in close connection to the skin over the tumor, after application of conductive gel to ensure contact. Before treatment, mice were anesthetized (also untreated controls) by intraperitoneal injection of hypnorm and midazolam. Tumors were electroporated immediately after (app. 20 seconds) injection of drug.

HT29, MDA-MB-231, and SW780 cell lines were used for this experiment. H69 cells have previously been tested to investigate tumor necrosis after calcium electroporation (1).

Tumors were treated with 168 mmol/L CaCl₂ injected in a volume equivalent to half the tumor volume (as previously used; refs. 1, 25) and electroporation. This calcium concentration did not show any signs of hypercalcemia in rats (25), and we chose to use this concentration even though we treated mice and used a slightly higher injection volume, but this study also included testing of a wider range of injection volumes and concentrations (see below). Tumors were removed before treatment (n = 2) and 2 hours, 1 day, 2 days, and 6 days after treatment (n = 4). Histology was performed as described below.

The MDA-MB-231 cell line was used for this experiment, as calcium electroporation induced around 65% tumor necrosis in MDA-MB-231 tumors, in the experiment described above, thus when testing dose–response in this experiment, both higher and lower fraction of necrosis could be shown. Mice were randomized into 18 groups (n = 3 for untreated and sham treated; n = 3–5 for treated). Untreated, sham treated, and physiologic saline injection with or without electroporation were used as controls. In treatment groups, four different injection volumes equivalent to 20%, 40%, 60%, and 80% of the tumor volume of 168 mmol/L CaCl₂ with or without electroporation, and three different CaCl₂ concentrations (100 mmol/L, 220 mmol/L and 500 mmol/L) in a volume equivalent to 50% of the tumor volume with or without electroporation were used. Two days after treatment, tumors were removed as well as the skin above and the muscle below the tumor. Histology was performed as described below.

NMRI-Foxn1^nu mice were also used for this experiment. An area of 6 × 10 mm (equivalent to the size of the electrode) was marked on the flank of the mouse. The area was injected with 15-μL physiologic saline or 168 mmol/L CaCl₂ in the skin and the same volume in the muscle tissue below the skin (a volume equivalent to 50% of the tissue volume) followed by electroporation or not, as described above. Two days after the treatment the treated skin and muscle tissue was removed (n = 5–6). Untreated skin and muscle was removed from the opposite flank. Histology was performed as described below.

Tissue samples were fixed in formalin (10% neutrally buffered), and paraffin embedded. Tissue sections (3 μm) were stained with hematoxylin and eosin and evaluated by a pathologist blinded with respect to treatment status, using a light microscope (Leica DM 2000). In tumor sections, fraction of necrosis was estimated by stereological point counting.

In addition, in normal tissue sections, changes were evaluated semiquantitatively. In skin sections, presence of inflammation in dermis and subcutis was scored from 1–3 (1 = minimal, 2 = moderate, and 3 = severe). Edema in the dermis and extravasations of erythrocytes was noted if present. In muscle sections, presence of necrosis was scored by counting numbers of necrotic myocytes per 10 high-power fields (HPF), the area visible under maximum magnification (×400). Counting was performed in the area with the most obvious and worst morphologic changes. The results were grouped into 3 groups: 0 necrotic myocytes/10 HPF (no necrosis), 1–4 necrotic myocytes/10 HPF (scattered, solitary necrotic myocytes), or ≥5 necrotic myocytes/10 HPF (focal areas of coagulation necrosis). Interstitial inflammation, internalization of myocyte nuclei, and extravasation of erythrocytes was noted if present. Samples were not oriented, and hence location in relation to the electrodes not known. This also contributed to the pathologist being blinded with respect to treatment/no treatment.

A simulation of the electric field distribution in tumor, skin above, and muscle below, like the in vivo setting, was done using finite element method software, COMSOL Multiphysics 4.2a (Comsol AB), on a 64-bit Windows platform. The simulations was based on the (electrostatic) continuity equation, −∇•(σ•∇ϕ) = 0, where σ is the tissue conductivity and ϕ is the electric potential. This was solved with boundary conditions ϕ = V₀ and ϕ = 0 (ground) applied to the two electrode surfaces, respectively. Boundaries of the model were set as electrical insulating (−n•J = 0). The finite element method mesh was defined as physics-controlled and element size fine.

The system was constructed using three tissue conductivity domains (Supplementary Table S1). The natural conductivity of the cornified layer (20 μm) of the epidermis is very low (0.0005 S/m; ref. 26), but the conductivity has been reported to increase dramatically (up to three orders of magnitude) just a few microseconds from the onset of a permeabilizing voltage pulse (27). The conductivity remains increased during the remaining pulse and can therefore safely be modeled as constant (27). The skin was modeled as one physical layer, the epidermis/dermis, with a total thickness of 500 μm. Conductivities of 0.2 S/m and 0.5 S/m, respectively, were applied to simulate the expected value (0.5 S/m) and a worst-case value (0.2 S/m). The conductivities of the tumor with 168 mmol/L CaCl₂ were measured in vivo to 0.48 S/m. The conductivity of physiologic saline solution was measured to be roughly 30% lower than for the 168 mmol/L CaCl₂ solution, indicating that the control condition of tumor with physiologic saline solution is reasonably comparable in terms of the electric field intensity. The tumor was modeled as an ellipsoid. Conductivity of muscle was set to 0.6 S/m, which is three times the mean of the values found in ref. 28 for the nonelectroporated muscle. The increase in conductivity of muscle by a factor of 3 during electroporation has previously been modeled (29). The overall geometry was constructed as an extruded 2D geometry and central slices were regarded representative.

In this experiment, cells were electroporated in the presence and absence of calcium as well as treated without electroporation in the presence of calcium. During the experiment, intracellular calcium concentration was followed with calcium fluorophore. SW780 and HDF-n cells were used for this experiment. Before experiments 30,000 cells in 2-mL medium were plated in a Willco dish (WillCo Wells BV) and incubated overnight. Attached cells treated with electroporation were subjected to 8 pulses of 1.2 kV/cm (applied voltage to electrode distance ratio), 100 μs, and 1 Hz using a square wave electroporator (Cliniporator, IGEA) and a custom-made contact copper electrode with 8 mm between the electrodes. Electroporation parameters were optimized for both cell lines for high permeabilization (electroporation in buffer containing Lucifer yellow) and high viability (Trypan blue staining after electroporation).

Before electroporation, medium was removed from attached cells and they were washed twice with 1 mL Krebs Ringer buffer (120 mmol/L NaCl, 25 mmol/L sodium gluconate, 1 mmol/L MgCl₂·6H₂O, 0.4 mmol/L KH₂PO₄, 1.6 mmol/L K₂HPO₄·3H₂O, 1.5 mmol/L CaCl₂·2H₂O, 10 mmol/L glucose·H₂O in MiliQ water, pH 7.4). Then, 375-μL Krebs Ringer buffer containing 6 μmol/L Fluo-4-AM (Invitrogen) was added to the dish and cells were incubated at room temperature for 28 minutes. Cells were then washed with Krebs Ringer buffer with EGTA to chelate calcium (115 mmol/L NaCl, 25 mmol/L sodium gluconate, 1 mmol/L MgCl₂·6H₂O, 0.4 mmol/L KH₂PO₄, 1.6 mmol/L K₂HPO₄·3H₂O, 10 mmol/L HEPES, 5 mmol/L EGTA, 5 mmol/L glucose·H₂O in MiliQ water, pH 7.4). Finally, 400-μL Krebs Ringer buffer with EGTA without calcium (0 mmol/L) or with calcium such that the final concentration would be 0.25 mmol/L. Calcium concentration needed to reach 0.25 mmol/L calcium in the used buffer was calculated using http://maxchelator.stanford.edu/CaMgATPEGTA-TS.htm (temperature was set to 20 degrees, pH 7.4, 5 mmol/L EGTA, 1 mmol/L MgCl₂, and ionic strength 155 mmol/L in the present buffer). A calcium concentration of 0.25 mmol/L was used to limit immediate detachment and cell death. Cells were treated with 0.25 mmol/L calcium (no electroporation), 0.25 mmol/L calcium with electroporation, and 0 mmol/L calcium with electroporation. After 15 minutes, the electrode was removed, to avoid disturbance of cells during resealing and calcium was added to a final concentration of 1 mmol/L to all experimental conditions. The whole procedure was performed at room temperature and was stopped after 80 minutes where cells (also untreated) were observed not to be at optimal conditions. Images were taken from 20 minutes before treatment to 50 minutes after treatment (corresponds to 10–80 minutes after adding Fluo-4-AM; at time point 10, 15, 25, 31, 32, 34, 36, 38, 40, 50, 60, 70, 80; one image per time point) using a Leica DMI6000B microscope with 20× objective and connected to a Leica DFC450C camera. Mean fluorescence intensity in cells was calculated using Image J software (NIH, Bethesda, MD) by marking all cells in each image from each experiment (n = 5–6). Microscope settings were the same for all experiments; images were thresholded and followed by measuring the mean light intensity within the cells.

The HT29 cell line was used for this experiment as these tumors were well-delineated and it was necessary to be able to easily and consistently separate tumor and normal tissue without microscopy. Mice were treated with 168 mmol/L CaCl₂ injection (50% of tumor volume) with or without electroporation. After treatment (15, 60, and 240 minutes) tumor, skin above, and muscle below were removed, cut in smaller pieces, weighed (immediately after removal from animal), and washed twice for 30 minutes in 1.5-mL Krebs Ringer buffer (115 mmol/L NaCl, 25 mmol/L sodium gluconate, 1 mmol/L MgCl₂·6H₂O, 0.4 mmol/L KH₂PO₄, 1.6 mmol/L K₂HPO₄·3H₂O, 10 mmol/L HEPES, 5 mmol/L glucose·H₂O in MiliQ water, pH 7.4). Tissue that was collected only strictly from treated area (n = 4–9, tumors weighing 113.1 ± 61.9 mg (mean ± SD), skin tissue 134.7 ± 80.7 mg, and muscle tissue 48.0 ± 24.6 mg) was put in 5% ice cold trichloroacetic acid (tumor and muscle in 250 μL and skin in 400 μL) overnight, lysed 4 minutes at 30 Hz using a Tissue Lyser (Qiagen), and centrifuged at 10,000 × g for 2 minutes. Supernatant was kept in −80°C until analysis. Samples were centrifuged at 16,000 × g for 3 minutes and supernatant was analyzed for total calcium content using VITROS 5,1 FS Chemistry System (Ortho Clinical Diagnostics). Samples that were measured to <0.25 mmol/L (the detection limit of the machine) was set to 0 mmol/L for further analyses.

All five cell lines (H69, HT29, MDA-MB-231, SW780, and HDF-n) were examined. Harvested cells (n = 3) were lysed and protein content was measured in the supernatant and a BSA standard using BCA Protein Assay (Pierce). Protein extract (10 μg) mixed with Sample load buffer (Invitrogen) and reducing agent (Invitrogen) was run on a NuPAGE Tris-Acetate Mini Gel (Novex) by applying 100 V for 90 minutes. Gels were transferred to a polyvinylidene fluoride (PVDF) membrane (Invitrogen) for 105 minutes at 90 mA. Membranes were blocked in a 2:1 mixture of blocking diluent A and B (Invitrogen) for 60 minutes, and incubated with primary antibody (1:1,000) for 16 hours in a buffer of blocking diluent A and B (2:3). The primary antibodies used were mouse-anti-PMCA antibody 5F10 (Thermo Fisher Scientific), mouse anti-PMCA-4 antibody JA9 (Thermo Fisher Scientific), and rabbit monoclonal anti-PMCA-1 antibody EPR12029 (Abcam). The PVDF membranes were washed thrice in wash buffer for 5 minutes followed by incubation for 60 minutes with secondary horseradish peroxidase (HRP)-conjugated antibody (1:10,000; Invitrogen) in the same buffer used for primary antibody incubation. The PVDF membranes were washed thrice in wash buffer for 5 minutes and briefly rinsed with water prior to incubation with Novex ECL Chemiluminiscent Substrate Reagent Kit (Invitrogen). Chemiluminiscent signals were elicited using ImageQuant (LAS-4000 Chemiluminiscence and Fluorescence Imaging System, Fujitsu Life Science).

Total protein content was estimated by staining the dry PVDF membrane (not differing from staining the gel; ref. 30) with SimplyBlue SafeStain (Invitrogen), that is, soaking the membranes for 30 minutes followed by destaining with 20% methanol and 10% acetic acid for 2 × 10 minutes (Supplementary Fig. S1). The Coomassie-stained PVDF membrane was scanned and total protein content was determined by marking the area of each lane, hence including all the bands, and measuring the total intensity of this area using LAS-3000 Imaging System (Fuji). PMCA expression was normalized to total protein content as it has previously been shown that the most common housekeeping genes can vary significantly between different tissue types as well as between normal and cancer tissue (31).

All five cell lines were used for this experiment (H69, HT29, MDA-MB-231, SW780, and HDF-n). RNA was isolated by homogenizing the cell pellet (n = 3) in TRIzol reagent (Ambion) as described in manufacturer's manual. RNA to DNA concentration was measured and degree of DNA contamination where absorbance at 260 nm/280 nm > 1.8 was used as standard cut-off criterion. Isolated RNA was used as template for cDNA synthesis using qScript cDNA SuperMix (Quanta Bioscience) as described in manufacturer's protocol. For qPCR, first-strand cDNA (2–4 μL) was mixed with Brilliant II SYBR Green qPCR Master Mix (Agilent Technologies) together with a final concentration of primers of 500 nmol/L each. Fragments specific to, respectively, all four PMCA, PMCA-1, and PMCA-4 were amplified using the following primers: PMCA(1–4) sense (5′cgcgaattcacncaygtnatggargg), PMCA(1–4) antisense (5′cgcgaattcgcngtngcrttncccat), PMCA-1 sense (5′ttcaacgaaataaatgcccgg), PMCA-1 antisense (5′agggtggaggactggagttacg), PMCA-4 sense (5′caactcccgaaagatccatg), and PMCA-4 antisense (5′tggttaacacagcagctgac; DNA Technologies; refs. 32, 33). The mixed qPCR samples were incubated for 2 minutes at 94°C followed by 35 cycles of 94°C for 10 seconds, 55°C for 30 seconds, and 72°C for 60 seconds, finalized with 7-minute incubation at 72°C. The qPCR products were quantified using the threshold cycle (C_t) values and normalized to the C_t of the reference gene (β-actin).

Statistical analyses were performed using SAS software (version 9.2). Differences in fraction of necrosis in tumors and differences in intracellular calcium level in vitro were assessed by two-way ANOVA with post least-squares-means test with Bonferroni correction. Calcium concentrations were log transformed before analysis. Differences in PMCA protein level were assessed by one-way ANOVA with post least-squares-means test with Bonferroni correction and log transformed before analysis. Linear regression of the fraction of necrosis, level of inflammation, and regenerative changes related to the calcium dose was calculated using GraphPad Prism (version 6.04). Differences in calcium content in vivo were assessed by two-way ANOVA with post least-squares-means test with Bonferroni correction using SPSS software (version 19) where calcium content were square root transformed (due to some values being 0) and due to violation in the assumption of homogeneity of variance in the muscle samples Welch adjustment was used. Data are shown as mean with SD or as median with first and third quartiles.

To test the effect of calcium electroporation on different tumor types in vivo, we treated three different tumor types; colon, breast, and bladder cancer, with calcium electroporation and estimated fraction of necrosis after treatment (Fig. 1; ref. 1). The results are shown together with similar results of a small-cell lung cancer tumor (H69) from a previous study shown here for a reference (1). Calcium electroporation induced tumor necrosis in all treated tumors, with a significantly higher fraction of necrosis in two of the tumor types 2 days after treatment compared to before treatment (H69, P < 0.0001; MDA-MB-231, P < 0.05). Thus, calcium electroporation induced tumor necrosis in several tumor types although a difference in sensitivity was seen. In three of the tumor types (HT29, MDA-MB-231, and SW780), a decrease in the fraction of necrosis was seen 6 days after treatment compared with 2 days after treatment. This could be due to regrowth of tumor tissue, and/or healing of the treated area with replacement by connective tissue after necrosis of the tumor cells (Fig. 1; Supplementary Fig. S2).

In addition to the effect of calcium electroporation on different tumor types a most pertinent question was the response of tumors to calcium electroporation using different doses of calcium, as well as the response of normal tissue to calcium electroporation. We tested the effect of different doses of calcium electroporation on MDA-MB-231 (breast cancer) tumors (Fig. 2A), both with regards to injected volume and concentration used. This showed that the tested injection volumes (equivalent to 20%–80% of tumor volume) of 168 mmol/L CaCl₂ as well as the tested CaCl₂ concentrations (100–500 mmol/L) in an injection volume equivalent to 50% of tumor volume (Fig. 2A) were effective in inducing tumor necrosis compared with control tumors treated with NaCl [CaCl₂ ± EP vs. NaCl ± EP (P < 0.05) except for 60% CaCl₂ ± EP (P = 0.0894)]. This also shows that calcium injection alone also has some effect on tumors compared with control tumors treated with NaCl. When comparing the fraction of necrosis in tumors treated with calcium alone and calcium electroporation for each of the different injection volumes and concentrations used we did not see a significant difference. In this study, the acute effects (2 days after treatment) of calcium electroporation and calcium injection alone was investigated and the acute effect of calcium injection alone might be due to a hyperosmotic shock induced by the injection of the high calcium concentrations and/or calcium toxicity. It would be interesting to further investigate whether other hyperosmotic solutions could induce similar effects. Calcium injection alone has some acute effects on tumor tissue; however, we have previously shown no long-term effects on tumor volume of calcium injection alone (1). Thus, due to the difference in the long-term effects between calcium electroporation and calcium injection alone we do not believe treatment with calcium injection alone is sufficient as anticancer treatment.

When calculating linear regression of the fraction of necrosis related to the calcium dose used (Supplementary Fig. S3), the fraction of tumor necrosis appeared to increase with increasing calcium dose; however, this was not statistically significant, both for tumors treated with calcium alone and calcium electroporation. Thus, the effect of calcium electroporation seems to be similar within the wide range of doses tested (injection volumes and concentrations).

As described the effect of calcium electroporation on normal surrounding tissue was also tested when treating MDA-MB-231 (breast cancer) tumors with different doses of calcium electroporation, both with regards to injected volume and concentration used. This was followed by analyses of tumor as described above (Fig. 2A) as well as the skin above (Supplementary Fig. S4) and muscle below the tumor (Fig. 2B; Supplementary Fig. S5) two days after treatment. As the calcium concentration in the surrounding tissue must be assumed to be lower than in the tumor where calcium is injected, we also tested the effect on normal tissue two days after the normal skin (Fig. 3) and muscle (Fig. 3; Supplementary Fig. S6) were treated directly with calcium electroporation.

As expected, calcium electroporation induced tumor necrosis; however, the normal surrounding tissue was far less affected than the tumor tissue. We observed no necrotic myocytes per 10 high-power fields (HPF) in 80% of the muscle tissue sections. In 13% of muscle sections, minimal necrosis was observed (1–4 myocytes/10 HPF), and in only 6% of the samples, a more pronounced necrotic reaction was seen (≥5 myocytes/10 HPF; Fig. 2B). In the normal muscle tissue treated directly with calcium electroporation (Supplementary Fig. S6), we observed slightly more necrotic myocytes per 10 HPF, where 60% of the samples showed pronounced necrotic reaction (≥5 myocytes/10 HPF), but 40% of the samples showed no necrosis and when looking at the fraction of necrosis, only 10% of the muscle tissue was necrotic, which is much lower than what was observed in treated tumor tissue. In line with this, an in vitro study of normal muscle cells and muscle fibrosarcoma cells showed that calcium electroporation was cytotoxic to the fibrosarcoma cells while the normal muscle cells were less affected (34). An in vivo study testing calcium electroporation as a possible use for contingency elimination of transgenic protein expression in muscle tissue showed some diffuse damages to the muscle tissue 24 hours after treatment and signs of regeneration 24 hours after treatment (25). In these two studies, the normal muscle cells and tissue were also directly treated with calcium electroporation.

In the skin tissue samples surrounding the treated tumors, inflammation and extravasation of erythrocytes were scored (Supplementary Fig. S4). Inflammation was seen in the skin dermis and the subcutis in all samples, both treated and untreated, and was dominated by mast cells with only few mononuclear cells and granulocytes. Edema was also seen in dermis in all samples, and the epidermis was generally displaying reactive changes. In the skin tissue directly treated with calcium electroporation (Fig. 3), the same degree of inflammation was observed as in untreated skin. No edema was observed in these samples, indicating that edema was due to the presence of the tumor and not the treatment.

Taken together, limited effects were seen in the muscle and especially skin tissue samples when comparing treated samples with untreated and sham-treated controls, which indicates that the changes seen in normal tissue were caused by the presence of the tumor. However, we observed minor changes (more necrosis and inflammation) in the muscle tissue directly treated with calcium electroporation compared with untreated muscle tissue. The normal and tumor tissue used in this study was of different tissue types, and indeed in the clinical situation the typical scenario would be, for example, breast cancer metastases surrounded by normal skin tissue. However, it would be interesting to compare normal and tumor tissues of the same type in a future study, for example, normal muscle cells with sarcoma cells. The osmolarity and conductivity of physiologic saline was determined to be around 40% and 30% lower than the calcium solution, respectively. This will result in a slightly lower electric field in the calcium electroporated tumors compared with tumors treated with saline and electroporation. In the normal tissue treated directly with calcium electroporation, the electric field intensity will be slightly higher than in the controls treated with saline and electroporation; however, we still only observed limited effects in the normal tissue.

The fact that a range of concentrations of calcium are efficacious in tumors is useful knowledge when it comes to clinical trials. Thus, when injecting calcium into a tumor there will be concentration gradients and loss of calcium to the circulation, therefore it is good news that there is a wider window of efficient doses, and at the same time that doses within this range have limited effect on normal tissues.

The in vivo data were supplemented by electric field simulations that showed the electric field intensity is higher in the portion of the skin facing the electrodes as compared with the tumor and the muscle tissue underneath (Fig. 4). This is true even when the conductivities of the skin and tumor are at comparable levels (0.5 S/m versus 0.48 S/m), an indication of the robustness of the calculation. The electric field intensities in the portion of the skin facing the electrodes are in the range 1,000–1,500 V/cm with a skin conductivity of 0.2 S/m, and 650–750 V/cm with a skin conductivity of 0.5 S/m. Corresponding values for the tumor are 450–550 V/cm (skin conductivity = 0.2 S/m) and 550–650 V/cm (skin conductivity = 0.5 S/m). The muscle/normal tissue electric field intensity is comparable with that of the tumor (only slightly lower). Thus, the electric field distribution is similar in the tumor and muscle tissue and may be slightly higher in the skin tissue; therefore, the electric field distribution is not expected to cause the differences seen in the effect of calcium electroporation on malignant and normal tissue.

Simulations are based on tissue conductivity estimates found in the literature and subject to some uncertainty. Although absolute conductivity values are uncertain, relative conductivities of the tissues are less uncertain, and as electric field calculation is governed by relative conductivities (voltage clamped) these are reliable. It is, for example, a reasonable assumption that the conductivity of the electroporated skin is lower than the conductivity of the tumor and muscle/normal tissue.

Interestingly, a recent finding is that normal cells perform faster repair of cell membranes than do malignant cells (35), so that even if the field is high over the skin, the cell repair is possibly faster.

As calcium electroporation seems less effective in normal than malignant tissue, we wanted to investigate whether there was any difference in calcium uptake between normal and cancer cells. For this purpose, we monitored the intracellular calcium (Ca²⁺) concentration using calcium fluorophore Fluo-4 after calcium electroporation in vitro in SW780 bladder cancer cells and HDF-n normal primary fibroblasts (Fig. 5). Fluorescence intensity of Fluo-4 was normalized to the intensity detected 25 minutes after loading cells with Fluo-4-AM, 3 minutes before treatment. Cells were incubated in Ringer solutions with zero calcium or modest calcium concentration (0.25 mmol/L). Higher concentrations of calcium during electroporation were not used to prevent cell detachments during imaging. Note that there still would be more than 1,000-fold electrochemical gradient to drive calcium into the cell. Calcium electroporation caused significantly increased Fluo-4 intensity (3.2-fold and 2.9-fold in the bladder cancer cells and normal fibroblasts, respectively; P < 0.01 and P < 0.05), and the intensity stayed within the range of, respectively, 2.3–3.3 fold and 1.9–3.2 fold higher than before electroporation treatment, for up to 50 minutes after treatment. No significant difference in fluorescence intensity was detected over time in the two cell lines, indicating that similar levels of calcium enter the cells after calcium electroporation. Addition of calcium alone (without electroporation) seems to not change the intracellular calcium concentration. Electroporation of cells in presence of the calcium chelator EGTA caused a reduction in fluorescence intensity in both cell lines (38%–48% in bladder cancer cells and 19%–26% in normal fibroblasts). In the last step of the experiment, when calcium (1 mmol/L final concentration) was added to the buffer 15 minutes after treatment, the Fluo-4 intensity increased 1.9–2.6 fold and 2.0–2.2 fold in the bladder cancer cells and normal fibroblasts, respectively, compared with intensity before treatment. It has previously been shown that the plasma membrane reseals slowly when cells were electroporated in the absence of calcium (36). Thus, it is likely that the plasma membrane was not resealed after treatment, and when calcium was added to the cells it entered the cytosol.

Within the time frame of the experiment, normal and cancer cells did not recover to extrude calcium but longer protocols (i.e., 4 hours) were not feasible on cells bathed in Ringer solution.

Calcium content was also investigated in vivo in tumors treated with calcium electroporation, as well as in the skin above and muscle below. Samples were removed 15, 60, and 240 minutes after calcium electroporation, washed in buffer to dilute the extracellular calcium content before lysing the cells, and measuring total calcium content (Fig. 6).

In tumors, calcium content increased 3.9-fold 15 minutes after calcium injection alone and over time decreased to a level only 1.5-fold higher than untreated controls (no significant difference in tumors injected with calcium alone at any of the time points investigated compared with untreated control). In tumors treated with calcium electroporation, calcium content increased 4.5-fold after 15 minutes (P = 0.057) and was significantly higher 60 and 240 minutes after treatment where calcium content was 8.8- and 5.7-fold higher, respectively (P < 0.05).

In the skin above the treated tumors, the intracellular calcium content appeared to increase after injection with calcium alone followed by a decrease; however, none of these measurements reached significant level (apparent 4.9-fold 15 minutes after injection, P = 0.154; 3.5-fold after 60 minutes, P = 0.842 and 2.1-fold after 240 minutes; P = 1). In the skin above tumors treated with calcium electroporation, calcium content was significantly higher 15 and 60 minutes after treatment compared with untreated controls (5.6-fold and 7.3-fold, respectively, P < 0.05). Interestingly, the calcium content decreased 240 minutes after treatment to a level comparable with untreated controls (2.4-fold, P = 1). This indicates that cells in the skin are able to remove calcium from the cell cytosol by pumping calcium out of the cell after calcium electroporation.

No significant increase in calcium content was seen in the muscle tissue below tumors treated with neither calcium alone nor calcium electroporation. Muscle cells have very efficient mechanism to maintain calcium homeostasis, as they have to cope with much higher calcium levels necessary for contractile mechanisms (37).

This in vivo experiment and the in vitro experiment described above show that calcium enters cells after calcium electroporation. Even though a similar increase in calcium content was observed in tumor and skin there might be a difference in the ability of the cells to cope with this, as calcium electroporation induces tumor necrosis but only has limited effect on the skin. As previously suggested (1, 22), the increased intracellular calcium concentration might affect many cellular processes such as opening of permeability transition pores (PTP) in the mitochondria, increased concentration of reactive oxygen species, increased lipase and protease activity, as well as affecting the cytoskeleton and membranes differently in malignant and normal cells. In this study, we have chosen to investigate whether the ability of cells to remove the intracellular calcium may be linked to expression levels of plasma membrane calcium ATPases.

To investigate whether the expression level of PMCA could account for the observed difference in intracellular calcium content and presumably Ca²⁺ concentration in skin and tumor tissue after calcium electroporation, we estimated the PMCA protein and mRNA level in four human cancer cell lines (small-cell lung cancer, colon cancer, breast cancer, and bladder cancer), as well as in normal human dermal fibroblasts (Fig. 7). Total PMCA level in the normal fibroblasts was significantly higher than in the cancer cells (1.8- to 7.9-fold, P < 0.01). For further investigation of this difference in PMCA level, the level of two of the PMCA isoforms was determined. As described previously, there are four PMCA isoforms, where PMCA-1 and 4 are expressed in all cells. Therefore, we estimated PMCA-1 and PMCA-4 levels in the cell lines. PMCA-1 protein level was lower in the normal fibroblasts compared with three of the cancer cell lines (significantly for H69 P < 0.01 but not significantly for HT29 and MDA-MB-231); thus, another isoform must dominate the total PMCA pool. PMCA-4 was likely the isoform that dominated the pool of total PMCA, as PMCA-4 in the normal fibroblasts was significantly higher (4- to 36-fold, P < 0.0001) compared with the four cancer cell lines. The lower expression level of PMCA in the malignant cells compared with normal fibroblasts correlates with the calcium content in tumor and skin tissue after calcium electroporation, indicating that the tumor cells have a reduced ability to remove intracellular calcium after calcium electroporation.

These results, showing a lower PMCA level in the tested cancer cells compared with the normal fibroblasts, support previous findings (20, 21). The dissociation constant (K_d) for ATP of PMCA-4 is higher than the K_d for PMCA-1, hence, PMCA-4 can cope with higher calcium levels (16, 38). As the cancer cells displayed a significantly reduced expression of PMCA-4, these data may indicate that cancer cells have reduced ability to handle increases in intracellular calcium concentration induced by calcium electroporation. However, to test whether PMCA expression level correlates with sensitivity to calcium electroporation, further investigations, such as estimation of the activity of PMCA in each cell line, are needed.

We also estimated the mRNA level for all PMCA, PMCA-1, and PMCA-4. The results showed no correlation between PMCA protein level and mRNA level (Supplementary Fig. S7), indicating that the reduced PMCA protein level in the cancer cell lines are caused after mRNA synthesis, for example, due to changes in mRNA stability and translation efficacy or breakdown rate (39). A further research question could be whether calcium electroporation can induce changes in PMCA expression provided that the cell is not immediately on the path to necrosis.

There are many perspectives of this potential novel anticancer treatment as electroporation-based treatments are increasingly used for treatment of cutaneous metastases (9, 10, 12) and under investigations for use in internal organs such as brain, liver, and bone metastases (40–43). Treatment with electric pulses using other parameters than described in this study is also increasingly used, such as irreversible electroporation (44–46) and nanosecond pulse electric fields (47–49) where addition of calcium might also increase the effect and/or treatment area. A recent study indicates that calcium electroporation as well as electrochemotherapy stimulates the immune system, as previously treated mice did not develop tumors when rechallenged with the same tumor cells (50).

In conclusion, we have shown that calcium electroporation induced significantly higher fraction of necrosis in tumors compared with control tumors treated with NaCl ± EP two days after treatment. We also observed an acute effect of calcium injection alone, which might be caused by hyperosmotic shock due to the calcium concentrations used or calcium toxicity. In the normal tissue surrounding the treated tumors and in normal tissue directly treated with calcium electroporation only limited effects were observed. Minor effects were observed in muscle tissue directly treated with calcium electroporation. This difference in sensitivity between normal and malignant tissue (of different tissue types) might be caused by more efficient calcium recovery systems in the normal tissue, as calcium entry seems similar in both normal and cancer cells. We propose that this could be explained by a significantly lower PMCA protein levels in malignant cells. We have also shown that calcium electroporation induced tumor necrosis in several cancer types in vivo and within a range of different calcium concentrations (100–500 mmol/L) and injection volumes (equivalent to 20%–80% of the tumor volume).

This study indicates that calcium electroporation, similarly to electrochemotherapy, induces cell death in tumors of different histology while sparing surrounding normal tissue. We have seen a difference in sensitivity to calcium electroporation of the tested tumor types, which might be caused by difference in PMCA expression and/or activity, resilience to ATP depletion, and potentially also other factors. Further investigations are needed to elucidate the mechanism behind calcium electroporation and to study the possibility and effect of retreatment of tumors less sensitive to the calcium electroporation. This promising novel anticancer treatment, where equipment is already available and in clinical use, is a simple, inexpensive local treatment that also has potential as anticancer treatment in low-income countries.

A patent has been submitted and licensed - PCT/DK2012/050496 (co-inventors: S.K. Frandsen and J. Gehl). T. Tramm holds a patent for a gene signature associated with efficacy of radiotherapy in invasive breast cancer (international patent publication no. WO 2013/132354 A2). The patent is not related to the present work. No potential conflicts of interest were disclosed by the other authors.

Conception and design: S.K. Frandsen, I. Novak, J. Gehl

Development of methodology: S.K. Frandsen, M.B. Krüger, I. Novak, J. Gehl

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S.K. Frandsen, M.B. Krüger, U.M. Mangalanathan, T. Tramm, I. Novak

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S.K. Frandsen, M.B. Krüger, U.M. Mangalanathan, F. Mahmood, J. Gehl

Writing, review, and/or revision of the manuscript: S.K. Frandsen, M.B. Krüger, U.M. Mangalanathan, T. Tramm, F. Mahmood, I. Novak, J. Gehl

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S.K. Frandsen, J. Gehl

Study supervision: I. Novak, J. Gehl

The authors thank Anne Boye, Lone Christensen, Marianne Fregil, Mogens Jøns Johannsen, Maria Sarfraz, and Asma Richhall for providing excellent technical assistance.

This work was supported by grants from The University of Copenhagen (S.K. Frandsen), The Danish Council for Independent Research Natural Sciences (DFF-4002-00162 to I. Novak), and FP7 Marie Curie ITN “IonTraC” (289648 to I. Novak), and The Danish Cancer Society (R110-A6996 to J. Gehl).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.