Introduction

Three-finger toxins (3FTxs) are a family of non-enzymatic proteins that are particularly widespread as components of venoms obtained from snakes in the Elapidae family1,2. Until relatively recently, high content of proteins with a three-finger structural motif (67–73%) was determined in the proteome of Naja ashei venom3,4. It is noteworthy that while many elapid venoms exhibit strong neurotoxicity, the venom of African spitting cobras, such as Naja ashei, primarily induces cytotoxicity, resulting in localized tissue damage, necrosis, and severe inflammation. These cytotoxic effects are attributed to the predominance of cytotoxic 3FTxs in their venom, which function by disrupting cell membranes rather than targeting neurotransmitter receptors5,6.

The presence of some common (repetitive) elements in the structure of three β-stranded loops make it possible to compare the properties of these amphipathic proteins1,7,8,9. Three finger toxins are made up of 70–75 residues10, some of which are made up of conservative sequences responsible for: (1) proper protein folding (for example, cysteine residues together with invariant residues such as Tyr-25 and Phe-27)7, (2) “polar belt” formation in the positively charged three-fingered proteins (Lys-5, 12, 23, 35, 44, 50 and Arg-36)11. The type of amino acids and the presence of exposed polar side groups have an impact on the specific chemical properties of the described subfamily of proteins, especially in connection with the ability of 3FTxs to interact with the hydrophilic part of membranes, and was proved by Dubovskii et al.11. Considering the experimental results published in the scientific works11,12,13,14,15, detailed information has been presented regarding the interaction between cytotoxins (particularly those belonging to the 3FTx group) and lipid membranes.

The results demonstrate, inter alia, the importance of the role of electrostatic forces and the participation of hydrophobic protein loops in the interaction of 3FTx with the membrane. In addition, specific phospholipids have been identified as favorable/neutrals for the interaction of toxins with the membrane11,13,14.

However, the effect that can be caused by the total charge presented by the lipid and protein composition of the membrane of an animal cell has not yet been determined. The knowledge of such data seems to be of particular interest in terms of the isolation of these toxic proteins due to their possible anti-tumor activity.

The search for new therapeutic solutions is heavily concentrated on cancer treatments. This is particularly relevant for conditions characterized by widespread systemic effects, such as leukemia16, and high mortality rates, like neuroblastoma (accounting for 15% of childhood tumor deaths17), where therapy implementation is hindered. For this reason, the aim of the study was a multidisciplinary search for the interaction between 3FTxs and the membrane, which would provide a concrete picture of the interaction of venom toxins with neuroblastoma and leukemia cells. The study attempts to answer how the surface charge presented by various 3FTx proteins affects the toxicity to different types of cancer cells and the properties of their lipid part of the membrane.

In details, these objectives were achieved by comparing the toxicity of various 3FTx fractions on SK-N-SH neuroblastoma cells and HL-60 leukemia cells as follows: (1) determination of cell viability (in vitro) upon exposure to different concentrations of the tested fractions and assessment of biochemical indicators (lactate dehydrogenase (LDH), malondialdehyde (MDA) level) which correlate with the damage of the membranes structure; (2) assessing the level of cell surface charge modification by adsorption of 3FTx proteins on membranes of tested cells and model membrane surfaces—testing the Zeta potential of cells and liposomes that structurally mimic the composition of the natural membranes (according to the tested cells); (3) study of the structural modification of the lipid part of the membranes with special emphasis on their mechanical properties, such as elasticity, fluidity, molecular composition density after 3FTx treatment study of Langmuir monolayers reflecting the multi-composition of specific cell types; (4) determination of those properties of the structure of lipid membranes that support the interaction with 3FTx—study of one-component Langmuir monolayers.

Results

Protein analysis

Chromatographic fraction G, containing 3FTx proteins from the initial Size-exclusion chromatography fractionation of Naja ashei venom, was used here for Ion exchange chromatography.

Figure 1 presents the chromatogram of fraction G separated on cation exchange resin (1a) and the corresponding LC–MS/MS analysis of its subfractions (1b). The obtained chromatogram showed that the apparently homogeneous fraction G (initially composed of nearly pure 3FTx proteins (99.8%) was separated into five subfractions based on the different charge distribution on the surface of the proteins. Peak shapes suggest the presence of multiple 3FTx proteoforms within certain fractions, differing not only in charge but also in other physicochemical and biological properties. However, this would not be particularly surprising given their abundant representation and protein diversity in the 3FTx group. The peak marked as G1 on the chromatogram stands as one of the most prominent in its intensity, while its appearance resembles the graphic representation that follows a normal distribution. The early retention time of this signal indicates that this fraction contains proteins with the lowest net positive charge. The next two signals resemble more asymmetric right-hand distributions, whereas the G4 and G5 fractions contain proteins with theoretically the highest net positive charges. The aforementioned fractions exhibited similar chromatographic behavior, as evidenced by Fig. 1a. It is worth noting that the calculated percentage of 3FTx proteins in the separated subfractions was higher than 98% in each case and reached 99.84% for the G4 fraction. The presence of proteins from other groups is marginal in each case and only in three fractions (PLA2 in G1, G2, and G3 fractions) exceeds 1% (Fig. 1b).

Figure 1
figure 1

(a) Chromatogram of the fraction G obtained by ion exchange chromatography on the 6 ml Cytiva Resource™ S (strong cation exchanger) column. The colored lines on the chromatogram indicate the signal level on the spectrophotometric detectors (the blue line corresponds to the λ = 215 nm parameter, the blue-green corresponds to the λ = 255 nm parameter, the green corresponds to the λ = 280 nm parameter) and the conductometric detector (corresponds to the red line). The collected fractions are indicated by the black arrows; (b) qualitative and quantitative protein composition of the analyzed G1–G5 fractions. Experimental data were analyzed using MaxQuant and Perseus software. LAP (Low-abundant proteins)—groups of proteins whose total percentage share did not exceed 1% were included in the LAP group. Accurate qualitative and quantitative results for all identified proteins are provided in the Supplementary Materials (see Supplementary Materials 1).

Full size image

Human cell culture assays in treatment with 3FTx fractions

The cytotoxic effects of the 3FTx fractions were measured by exposing the human cell lines SK-N-SH and HL-60 to different concentrations of N. ashei three-finger toxins (10–240 ng/ml) (Fig. 2). Assay concentrations were determined based on the dose–response studies by Das et al.18, who exposed L-6 and HEK 293 cells to snake venom three-finger toxins (named Nk-3FTx) at 0–100 ng/ml. Above 100 ng/ml, no decrease in viability parameter was observed (cell viability remained constant even from a concentration of 50 ng/ml). Another indication for selecting concentrations for the study described in this work was the median lethal dose (LD50) data obtained for HL-60 cells in previous experiments by Dyba et al.13 (121.29 ± 1.42 mg/l/1 × 106 cells). The selection of 50% mortality as a reference point enabled this study to circumvent potential ambiguity when measuring extreme cases, thereby determining the estimated concentrations of the potential effects of the tested substances. Considering that the tested cell line HL-60 might react differently to toxin exposure, the lethal dose concentration was doubled to 240 ng/ml. To enable data comparison, the same concentrations were applied to second, neuronal cell line.

Figure 2
figure 2

The effects of 3FTx fractions (G1 vs. G5) on SK-N-SH and HL-60 cells, including cell viability, LDH release, and membrane lipid peroxidation, are shown as mean ± SD (n = 5). (a) Viability: the lines show SK-N-SH and HL-60 cells after treatment with 3FTx fractions. Significant differences (p ≤ 0.05) between 3FTx fractions are marked with single asterisks for HL-60 and double asterisks for SK-N-SH. (b) LDH Release: Significant differences (p ≤ 0.05) in the concentration-dependent LDH release are indicated by different letters: uppercase (G1) and lowercase (G5) for SK-N-SH, and uppercase Greek (G1) and lowercase Greek (G5) for HL-60. (c) Lipid peroxidation (shown by MDA content): different letters signify significant differences between 3FTx concentrations: latin alphabet for SK-N-SH and Greek alphabet for HL-60. Significant differences (p ≤ 0.05) between 3FTx fractions are marked with asterisks.

Full size image

In all biochemical experiments (presented in Fig. 2a–c), the G1 fraction appeared to stimulate cells more in altering the tested parameters. However, in the case of LDH levels measured in neuronal cells, the G5 fraction was found to be more stimulatory to changes.

Figure 2a illustrates a concentration-dependent decrease in cell viability for both cell lines, expressed as a percentage of the control. Notably, 10 ng/ml of G1 fraction stimulated a 4.8% increase in neuronal cell metabolism compared to the control. Cell metabolism progressively declined at doses ranging from 40 to 240 ng/ml. At the highest concentration, neuronal cell survival was approximately 76% (both G1 and G5 fractions), while immune cells showed 94% viability with G1 fraction and 86% with G5 fraction).

Figure 2b depicts changes in membrane integrity of the tested cells, as determined by LDH leakage. The data obtained from Fig. 2b revealed the harmful effects of both tested fractions. Even at the lowest concentration of 10 ng/ml, some degree of membrane damage was evident, as indicated by LDH efflux. In the HL-60 immune cell line, LDH values were strongly correlated with increasing doses of 3FTx (higher concentrations resulted in greater LDH release), and this trend was consistent for both fractions. However, differences in fraction effects were observed in neuronal cells (SK-N-SH): G1 slightly increased LDH efflux compared to G5 at a concentration of 40 ng/ml. For this type of tested cells, the LDH levels above this dose remained at a similar level (10.5–13.9% for G1 fraction; 7.9–9.6% for G5 fraction).

Figure 2c illustrates malondialdehyde (MDA) levels, a marker of lipid peroxidation. The G5 fraction significantly increases peroxidation compared to G1 in both cancer cell membranes. Nerve cell membranes exhibit consistent peroxidation across all concentrations, with MDA secretion ranging from 0.38 to 0.48 for G1 fraction and 0.67 µM to 0.79 µM for G5 fraction compared to the control. In HL-60 cells, peroxidation occurs rapidly and is dose-dependent, with values ranging from 0.07 µM to 0.58 µM for G1 and 0.28 µM to 1.21 µM for G5.

Influence of the 3FTx surface charge on the properties of model membranes in SK-N-SH and HL-60 cancer cell lines

The zeta potential results presented in Table 1 were obtained following the research model proposed by Bondar et al.19, and Wilson et al.20, which applied dynamic light scattering to measure membranes from SK-N-SH and HL-60 cancer cell lines. Additionally, a parameter ζ was determined for liposomes imitating the membranes of tested cells.

Table 1 Zeta potential (ζ) of the cell lines and liposomes measured at pH 7.4 during exposure to 3FTx fractions—G1 and G5.
Full size table

Liposomes treated with 3FTx protein fractions: The zeta potential was more negative in HL-60 cells than in SK-N-SH cells. G5 fraction treatment significantly reduced zeta potential in neuroblastoma cells, while G1 fraction had a stronger effect in lowering zeta potential in HL-60 cells compared to G5.

Cancer cell lines treated with 3FTx protein fractions: The zeta potential was less negative in HL-60 cells compared to SK-N-SK cells. Only the G5 fraction induced a statistically significant decrease in zeta potential (2.3 mV) of neuroblastoma cells. In HL-60 cells, the G1 fraction (40 ng/ml) had a greater effect in reducing the zeta potential to almost 7.4 mV, while the G5 fraction (at a concentration of 40 ng/ml) reduced the ζ parameter by approximately 4.1 mV compared to cells in physiological solution.

3FTx fractions influence mechanical properties that characterize model membranes

The static compression modulus (Cs−1) quantifies membrane mechanical properties. We investigated changes in membrane mechanical properties induced by G1 and G5, finding evidence of multi-component Langmuir monolayers (Supplementary Fig. S2.2). Altered π-A isotherm shapes indicated changes in physicochemical parameters due to venom proteins. Cs−1 values were calculated from isotherms at a surface pressure of 30 mN/m, corresponding to in-cell conditions21. The percentage change in the parameter, compared to the venom-free system (100%), is presented in Fig. 3.

Figure 3
figure 3

The percentage of changes calculated for static compression modulus (Cs−1) obtained for monolayers mimicking SK-N-SH cell lines and HL- 60 cell lines. Values represent the mean value as ± SD calculated from 3–5 sets based on independently registered isotherms (n = 5). There were significant differences (p ≤ 0.05) between the applied concentrations for the SK-N-SH cell line, where the designation is given in the Latin alphabet and for HL-60, the designation is given in the Greek alphabet. Asterisks indicate significant difference between the 3FTx fractions tested, while differences between fraction G1 and G5 for a given concentration were also marked with an asterisk.

Full size image

In the neuroblastoma cell model, a significant change was observed only at 40 ng/ml concentration, with a decrease of 14% for G1 fraction and 10.3% for G5 fraction (Fig. 3). In HL-60 cell monolayers, G1 fraction had a stronger effect on membrane mechanical properties compared to SK-N-SH cells, showing a decrease of 27.9% at 10 ng/ml, while G5 reduced it to 9.5%. When G1 fraction was applied at 40 ng/ml, it decreased the Cs−1 value by almost half compared to control (46.5%), whereas G5 fraction decreased it by a third (33.3%). The main reason for measuring single-component monolayers was to indicate interaction between differently charged 3FTx fractions (isotherms are included in supplementary material 2 (see Fig. S2.3—as an example) as: (1) presented charge on the polar part of the lipid and also (2) on the lipid hydrophobic part. From the above, choline lipids that differ in their hydrophobic parts (18:1 and 16:0) were used to demonstrate the importance of the non-polar part. In order to check the participation of the polar part in the tested interaction, phospholipids that have the same fatty acids (16:0) but different hydrophobic parts—choline—DPPC and glycerol—DPPG were chosen. The polar parts of the presented lipids at pH = 7.4 show a relative charge of 0 (positive–negative polar part) within the polar heads of DPPC and − 1 (possessing the negatively charged polar part) within the polar heads of DPPG molecules22.

Figure 4 reveals a uniform trend of decreasing static compression modulus for all membrane systems investigated.. G1 fraction had a more significant influence on membrane mechanics compared to G5 fraction, especially at higher concentrations. G5 exhibited dominance over G1 at 40 ng/ml, while both fractions acted similarly at 10 ng/ml.

Figure 4
figure 4

The percentage of changes calculated for static compression modulus (Cs−1) obtained for single component monolayers. Values represent the mean value as ± SD calculated from 3 to 5 sets based on independently registered isotherms (n = 5). There were significant differences (p ≤ 0.05) between the applied concentrations, which were marked with an asterisk. 100% were described as a Cs−1 obtained for the control (lipid monolayer without toxins).

Full size image

Influence of the tested fractions on the hydrophobic part: significant reductions in the Cs−1 parameter were noted for DPPC compared to DOPC, with G1 fraction causing an 83.2% reduction for DPPC and 47.5% for DOPC at 40 ng/ml. G5 fraction reduced the parameter by 51.2% for DPPC, with no significant changes for DOPC.

Influence of the tested fractions on the hydrophilic part: At a concentration of 40 ng/ml, the G1 fraction induced a more pronounced decrease in static compression modulus than the G5 fraction in both monolayer types (Fig. 4). However, lower Cs−1 values were observed for phosphatidylcholine 16:0 (DPPC) than for phosphatidylglycerol (DPPG).

Discussion

Cells within the body interact with chemical substances in the extracellular space in various ways. From a physicochemical perspective, understanding the mechanisms of action of toxins provides insights into how the lipid-protein composition of membranes and the structure of toxins influence their interactions, ultimately resulting in biochemical and physiological effects. Previous studies (Dyba et al.13) have shown significant differences in the lipid membranes of U-937 and HL-60 cells. These differences were responsible for distinct reactions to the presence of Naja ashei venom proteins in the extracellular space, even though both cell types belonging to the immune system.

In the analysis of selected cancer cell lines representing immune and neuronal cells differences in their lipid composition were observed. In tested cells, the polar groups exposed by membrane phospholipids and the side chains of amino acids building membrane proteins contribute to differences in the surface charge of both compared cell lines. Specifically, in the cell membranes of the SK-N-SH (neuroblastoma) line23, the following observations were made: (1) a significant content of the main type of membrane lipids—phosphatidylcholine (PC; 64% compared to HL-60 cells (leukemia cell)24, which contain 48%), (2) a lower amount of lipid-cholesterol fractions (SK-N-SH; Chol: PL 21%:79%) compared to HL-60 (Chol: PL 35%:65%), and (3) a higher proportion of saturated fatty acids (48%) compared to HL-60 cells, which contain 34.6%.

These differences made the fractions 3FTx with different surface charges (G1 and G5) an intriguing subject for observing electrostatic changes using these two types of membranes. The protein fraction G1, which eluted from the chromatographic column first, exhibits a lower positive charge than the protein fraction G5, which exited the column towards the end of the separation process.

Previous studies have documented the cytotoxic effects of 3FTx proteins from various venoms on diverse cell types. It was observed that the cytotoxicity and apoptosis of HL-60 cells occurred due to changes in the permeability of their membranes, as evidenced by the action of cytotoxins extracted from snake venoms: Naja oxiana, Naja kaouthia, and Naja haje25, as well as following administration of the total fraction containing 95% 3FTx obtained from Naja ashei venom13. In the case of neuroblastoma cells, apoptosis has been evidenced (CTX4 and CTX5; Naja naja atra venom26), as well as the promotion of neurotoxicity, involving components of snake venoms from the genus Naja, such as Naja mossambica (CVC, cardiotoxins)27.

The distinct actions of G1 and G5 were verified using three biochemical assays, and the direction of changes was consistent with those reported in the literature. The first indicator employed was cell viability (MTT assay), the second was the assessment of membrane lipid peroxidation levels (measured using MDA), and the third was the degree of cell membrane damage under the influence of the tested 3FTx fractions (LDH assay). Both tested fractions elicited toxic, lethal effects on the investigated cells. Biochemical tests revealed SK-N-SH cells to be more sensitive than HL-60 cells.For the lowest doses of the tested fractions, both cell death and processes leading to membrane lipid damage, and consequently, excessive membrane permeability, were more pronounced.

Lipid oxidation results in increased membrane stiffness, reduced thickness, enhanced permeability, and impaired membrane function28,29,30.Interestingly, higher levels of polyunsaturated fatty acids in membrane lipids were observed in the cell membranes of the HL-60 line (11.45%)24 compared to the second investigated line (3.4%)23. Due to the greater availability of polyunsaturated phospholipids, 3FTx fractions in the HL-60 cell line, through their toxic action, indirectly promote processes inducing lipid oxidation in cell membranes (which is dose-dependent). Biochemical analysis revealed that 3FTx protein fractions with varying positive charges independently induce effects at both cellular and membrane levels. To more precisely analyze membrane alterations and the effects of tested fractions on neuroblastoma and leukemia cells, research focused on measuring changes in total membrane surface charge needs to be performed. Typically, the surface charge represented by the outermost cell membrane originates from the hydrophilic portions of lipids, as well as from polar groups belonging to proteins. Liposomes with a more negative charge (compared to cell lines) exhibit a charge generated solely by phospholipids (where polar groups are predominantly ionically neutral or negative) and do not encompass the charge exposed by other components of the cell membrane (e.g., proteins and polysaccharide residues), as described by Bondar et al.19.

The membrane of HL-60 cells exhibits a significantly more positive surface charge compared to SK-N-SH cells (which have a more negative surface charge). Consequently, the investigated fractions preferentially interact electrostatically with neuroblastoma cells. However, referring to the findings of Bozek et al.31 and Sastry32, that neural cells are particularly enriched in lipids and maintain a more diverse lipid composition than other tissues, it can be presumed that there are more lipids in the membranes of neuroblastoma cells than in HL-60 membranes This correlation aligns with the greater sensitivity of SK-N-SH cells to the effects of the tested toxins, as demonstrated by biochemical tests.

Liposome zeta potential measurements conclusively demonstrate that lipid molecules are direct targets for the studied proteins, enabling interaction with 3FTx fractions regardless of their varying positive charges. Changes in the surface charge of liposomes after treatment with 3FTx fractions are evidenced by the hydrophilic and hydrophobic regions of proteins with the three-finger motif, which may also promote interactions with amphiphilic lipids. The effect of altering the surface charge occurs in both liposomal systems prepared as models of the analyzed cell types. Positively charged proteins interacted significantly more with the negatively charged HL-60 liposome model, consistent with electrostatic principles. The interactions of three-fingered proteins with micellar systems have been described by Batenburg et al.33, who highlighted in their work that one of the three-fingered proteins (cardiotoxin) can induce fusion of lipid vesicles and also induce the formation of inverted micellar structures. In the case analyzed in the article, for proteins of fractions G1 and G5, the more positively charged regions of the three-finger toxins are oriented towards the negative surface of the micelles, while the remaining areas of the molecule, representing negative polar groups, are oriented outward.

The conclusions drawn regarding changes on the surfaces of cells and liposomes prompted a detailed investigation into the interaction of the lipid component—3FTxs in the studied model Langmuir membranes. Membrane dynamics and reorganization are critical for cancer cell survival, proliferation, and drug resistance, emphasizing the importance of understanding these processes34. The flexibility of the lipid part of the membrane, on the other hand, contributes to increased mechanical stability during rapid cell divisions, characteristic of cancer cells (reducing shear force during cell separation)35.

Measurements of the rigidity and elasticity parameter (Cs-1) conducted for monolayers with defined lipid composition of both investigated cancer cells showed that the administration of proteins from fractions G1 and G5 caused membrane softening in both studied cancer cell lines. Despite containing more saturated phospholipids, SK-N-SH membranes exhibited greater stiffness than HL-60 membranes following treatment with proteins G1 and G5.It is worth mentioning that at a higher tested protein concentration (40 ng/ml), the elasticity of HL-60 cell membranes increased almost twofold compared to untreated membranes.

The accuracy of the Langmuir monolayer modeling technique allowed for the supplementation of studies regarding the interaction of the investigated proteins with specific membrane lipids. The results suggest potential for targeted therapies as phosphatidylcholine (PC), a primary membrane phospholipid in both HL-60 and SK-N-SH cells, is a potential target. Phosphatidylserine and phosphatidylethanolamine, which under physiological conditions are mainly present on the inner side of cell membranes, have increased surface expression on the outer membrane of cancer cells36. Considering also the data from Konshina et al.14, cytotoxins may preferentially interact with cancer cells because they prefer negatively charged polar parts of PS. The investigated 3FTX fractions, due to their amphiphilic nature, interact much more readily with these lipids that present a positively-negative membrane charge, represented by choline (PC).

Supplementary to the studies elucidating the role of charge in protein-lipid interaction are the measurement effects obtained for another lipid—phosphatidylglycerol—with a pI = (− 1). A single-component monolayer formed for a lipid with a negatively charged polar part, representing the polar group—glycerol, still retained its rigidity, whereas the lipid PC exhibited increased elasticity. Saturated fatty acid bonds promote closer phospholipid packing in the liquid-crystalline membrane, reducing monolayer dynamics and responsiveness to mechanical changes37. When analyzing the interactions of the investigated proteins with the non-polar part of phospholipids, structural changes in the membrane will be observed more rapidly for membranes containing more phospholipids with saturated fatty acids.

Conclusions

  • The correlation between the obtained results and the membrane structure, along with its physicochemical properties such as surface charge of cells and liposomes (Fig. 5), suggests that the mechanism of action of 3FTx is linked to the modification of the lipid part of the membrane.

    Figure 5
    figure 5

    Graphical summary—the stages of the analyzed toxicity mechanisms.

    Full size image

  • 3FTx fractions interact electrostatically with various types of membranes, potentially resulting in differences in the intensity of changes observed in the studied biological and physicochemical parameters (Fig. 5).

  • SK-N-SH membranes, enriched in phosphatidylcholine (PC) and exhibiting higher saturation, are more susceptible to 3FTx interactions than HL-60 membranes.

  • HL-60 cells membrane demonstrates a notably greater elasticity response to interactions with 3FTx compared to SK-N-SH cells.

  • Individual membrane components, particularly phosphatidylcholine (PC), play a significant role in mediating the interaction between 3FTx proteins and cells.

  • Fraction G1 (with a lower positive charge) exerts a more pronounced effect on viability parameters and lipid peroxidation levels than fraction G5 (with a higher positive charge).

  • In the case of the neuroblastoma cell line, interaction with fraction G5 positively affects the electrostatic level, likely due to the more negative total charge of SK-N-SH membranes compared to HL-60.

  • Understanding the interaction between model and native membranes of tested cancer cells offer valuable insights for further therapeutic research, particularly in cell biology and oncology.

Materials and methods

Biological material

Proteins separation

The starting material was a chromatographic fraction containing a mixture of 3FTx proteins, which was the product of the preliminary fractionation of Naja ashei venom by Size Exclusion Chromatography (SEC). Chromatographic separation of crude venom sample using SEC technique was described in detail in our previous report13.

To separate 3FTx with similar physicochemical properties (especially regarding their relative charge), we have applied an additional chromatographic step in the form of the Ion Exchange Chromatography (IEX). First, the fraction G after SEC chromatography was desalted and mixed with 50 mM acetate buffer (pH 5.0) in appropriate proportions to obtain samples containing 2.5 mg of proteins in 500 µl of buffer. The sample was applied to a strong cation exchange column (Cytiva Resource™ S column, 6 ml; GE Healthcare, Little Chalfont, UK) using the NGC Chromatography System (Bio-Rad, Hercules, USA). 50 mM acetate buffer (pH 5.0) serves as the mobile phase, whereas bounded proteins were eluted gradually with increasing concentration of the eluent (50 mM acetate buffer containing 1 M NaCl). The chromatographic separation of the proteins was monitored at 215, 255, and 280 nm, with 1 ml fractions being collected manually. Obtained samples collected from the same chromatographic peak were pooled and concentrated on Vivaspin 2 centrifugal filters with 3000 MWCO PES membrane (Sartorius Stedim Lab Ltd., Stonehouse, UK). In order to obtain enough material for further studies, five identical chromatographic separations were performed. Protein concentration in pooled fractions was measured using the Pierce™ BCA Protein Assay Kit according to the manufacturer's instructions.

Identification of proteins by LC–MS/MS

The protein composition of the purified Ion Exchange Chromatography fractions was analysed with the use of the Liquid Chromatography with tandem mass spectrometry (LC–MS/MS) experiment. Prior to analysis, the samples were subjected to several preparation steps where the protein content in the samples was enriched and purified by acetone precipitation and its re-dissolution in 0.1% (v/v) solution of RapiGest (Waters, Milford, MA, USA) in 50mM ammonium bicarbonate pH 8. While proteins reduction and alkylation were performed with dithiothreitol and iodoacetamide solutions, respectively. A trypsin solution was used for the protein digestion and the resulting peptides were further purified on StageTips C18 resin according to the procedure described by Wiśniewski et al.38 A detailed description of the sample’s preparation steps was included in our previous report39.

The detailed procedure of the analysis of the peptides obtained by LC–MS/MS method is elaborated in more detail in the full mass spectroscopy data processing protocol in our previous report3. Briefly, 0.86 μg of digested peptides were collected for LC–MS/MS analysis each time. Peptide separation was performed on a Dionex Ultimate 3000 RSLC NanoLC system (Thermo Fisher Scientific, Waltham, MA, USA) using Acclaim PepMap RSLC nanoViper C18 column (75 μm × 25 cm; 2 μm granulation (Thermo Fisher Scientific, Waltham, MA, USA)) with 180 min ACN gradient (from 4 to 60%; in 0.1% formic acid). Ion signals were detected on a Q Exactive Orbitrap mass spectrometer (Thermo Scientific, Waltham, MA, USA) operating on-line with the LC system during data-dependent acquisition (DDA) mode.

The acquired MS/MS raw data files were analysed using MaxQuant software (ver. 1.6.7.0). Protein identification was performed according to the UniProtKB Serpentes database (release 9/2019). The fixed modification was as follows: carbamidomethylation of cysteines; variable modifications oxidation of methionine and protein N-terminal acetylation. The MS spectra were analyzed using the following database search strategyas follows: 20 ppm; main MS search tol.: 4.5 ppm; MS/MS fragment ions tol.: 20 ppm. Enzyme properties: Trypsin with full specificity; missed cleavages: max. 2. PSM and protein False Discovery Rate (FDR): 1%. iBAQ (intensity-based absolute quantification) values of razor and unique peptides were used for the calculation of the amount of particular protein in the sample.

The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier: PXD033540 (https://doi.org/10.6019/PXD033540).

Cell cultures

The human neuroblastoma cell line SK-N-SH was purchased from the European Collection of Authenticated Cell Cultures (ECACC) and the HL-60 cell line from the American Type Culture Collection (ATCC). SK-N-SH cells were cultured with Dulbecco's Modified Eagle's Medium (DMEM), and HL-60 cells were grown in RPMI 1640 medium, in the presence of 10% fetal bovine serum (FBS). Both cell cultures were supplemented with 0.01% penicillin–streptomycin (purchased from CytoGen GmbH) at 37 °C in a humidified atmosphere containing 5% CO2.

Measurement protocols for cancer cell lines

The G fraction was separated based on differences in protein surface charge. As a result of the size-exclusion chromatography + ion exchange chromatography, two of the five obtained fractions were selected for further research: G1 with the smallest positive charge and G5 with the highest positive charge. For the other 3FTx fractions, intermediated charges—G2, G3, G4 were also obtained and some tests (such as static compression modulus) were performed, which are presented in Supplementary materials 2 (see Fig. S2.1).

  1. (a)

    Measurement of Cell Viability (MTT Assay).

The cells were cultured in 96-well plates in an amount of 104 cells in a volume of 0.2 ml per well for 24 h in full medium. The MTT assay was used to measure the mitochondrial function of the cells after a 24-h incubation with two 3FTx fractions: G1 and G5 (prepared separately one at a time, at a concentration from 0 to 240 ng/ml) at 37 °C. The colorimetric reaction of the reduction of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide was measured at 570 nm using a microliter plate reader (BioTek Instruments, VT, USA). After incubation time, a volume of 50 µL of the MTT solution (5 mg/l) was added to each well. Subsequently, the cells were incubated at 37 °C for 2 h prior to the addition of 0.4 ml of DMSO to each well. The mixtures were centrifuged after incubation (5 min) and the absorbance was detected; the values obtained represented the number of viable cells. Cell viability after each treatment was calculated and reported according to the method described by Czyżowska et al.40.

  1. (b)

    Measurements of membrane integrity using LDH assay.

The lactate dehydrogenase (LDH) assay was used as a marker of cell membrane integrity. Cells in the number of 1 × 104 cells per well were seeded in a 96-well plate for 24 h and then were incubated in the presence of G1 and G5 fractions of 3FTx for a 24-h incubation (prepared separately one at a time). After this time, the cells were centrifuged and the supernatant (150 µl) containing a pre-warmed mixture (at 37 °C for 10min) consisting of 0.5 ml of 0.75 mM sodium pyruvate and 10 µl NADH (140 µM) was added to the tubes. Subsequently, after 30 min, 0.5 ml of 2,4-dinitrophenylhydrazine was added to the samples. LDH release from cells in the presence of G1 and G5 fractions was measured spectrophotometrically by detecting the absorbance at 450 nm compared with their controls.

  1. (c)

    Assessment of membrane peroxidation by measuring the malondialdehyde concentration (MDA).

Cells were seeded into 48-well plates (SK-N-SH in amount of 5 × 104 cells per in 0.5 ml; HL-60 in amount 2 × 107 cells per well in 0.5 ml) and cultured to full confluention. Cells were incubated with G1 and G5 fractions (at concentrations ranging from 0 to 240 ng/ml) for 24 h. Subsequently, the cell suspension was collected and centrifuged at 1000 × g for 5 min. Subsequently, 0.5 ml of 0.5% trichloroacetic acid (TCA) was added to the supernatants obtained after centrifugation and then lysed by sonication (15 kHz) for 5 min. After repeated centrifugation at 10,000 × g for 10 min, 0.4 ml of supernatant was added to 1.25 ml of 20% TCA with 0.5% TBA. After that, samples were heated in a dry thermoblock (100 °C) for 30 min and after cooling, the absorbance of the MDA-TBA complex was measured at 532 nm. The molar extinction coefficient of MDA (ε = 155 mM) and lipid solutions to model membranes were used for concentration calculations.

Composition of model membranes

Cell membrane structure can vary between normal and cancer cells due to genetic alterations, culture conditions, and differentiation status. Based on the literature data, the compositions of the lipid model membranes were prepared as lipid multicomponent solutions. A similar procedure was previously used by Kreczmer et al.41. The high purity, synthetic lipids (Avanti Polar Lipids. Alabaster, AL USA) were selected to prepare model lipid systems as follows: 1,2-dipalmitoyl-sn-glycero-3-phosphocholine 16:0 (DPPC), 1-oleoyl-2-palmitoyl-sn-glycero-3-phosphocholine (PC Brain), 1,2-dioleoyl-sn-glycero-3-phosphocholine 18:1 (DOPC), lysophosphatidylcholine (lyso PC), 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine 18:1 (DOPE), 1,2-dipalmitoyl-sn-glycero-3-phospho-l-serine 16:0 (DPPS), 1,2-dioleoyl-sn-glycero-3-phosphocholine; and l-α-phosphatidylethanolamine (PE Brain); sphingomyelin (SM Brain) and Cholesterol (Chol). DOPC, DPPC, DOPE, PE Brain, SM and Chol lipids were dissolved in chloroform and DPPS were prepared in chloroform solution in 9:1 v/v methanol solution. The solvents (chloroform, methanol) of chemical purity were procured from POCH (Poland).

Lipids were mixed in proportions matching the literature-reported lipid composition of the target cell membranes. The following concentrations were prepared for the HL-60 model membrane system24,42,43: 15,96 mol% DPPS, 18.64 mol% DPPC, 32.42 mol% DOPC, and 32.98 mol% DOPE. For SK-N-SH, the membrane solution was prepared as follows23,44,45: 64% PC (mixed in a ratio of 18:1–16:0 PC, lyso PC and PC Brain), 19% PE brain, and 18% SM Brain. Cholesterol and phospholipid (PL) were mixed for SK-N-SH: in a ratio of 79% PL, 21% Chol and for HL- 60 in a ratio of 65% PL, 35% Chol. The ratio of saturated (S) to unsaturated (US) fatty acids in solutions was from 48 to 52% for SK-N-SH and for HL-60 it represented a ratio of 34,6% for S to 65,4% for US fatty acids.

Measurement protocols for monolayers

  1. (a)

    Zeta potential measurement.

Liposomes were prepared according to the standard procedure described earlier in literature46,47. The composition of the lipid mixture used for the liposomes was composed to mimic the SK-N-SH (64% PC, 19% PE, 18% SM, and 21% Chol) and HL-60 (48% PC, 31% PE, 15% PS, and Chol 35%), the same mixtures were applied to the investigated monolayers.

A thin layer of lipid was formed (under an argon stream) by evaporating the lipid solution (0.4 ml of 1 mg/ml chloroform mixture) by wetting the walls of a round-bottom glass tube. After this procedure, a defined amount of pure water was added to the cell with a lipid film, and the whole mixture was subjected to ultrasonification and vortexing. SK-N-SH (trypsinized) and HL-60 cells were mixed with a physiological 0.9% NaCl solution and centrifuged at 1000 × g for 5 min to remove proteins present in the medium. The zeta potential of the cell suspension (0.5 × 106 cells/ml) was recorded according to the method described in the article of Bondar et al.19.

Freshly prepared liposomes and cell suspensions were used for determination of mean size and zeta potential measurement. Various concentrations (10 and 40 ng/ml) of G1 and G5 fractions from the 3FTx family were added directly to the cell suspension (0.5 × 106 cells/ml) or the liposomes suspension.

Zeta potential was measured for both cell suspension and liposomes in a U-shaped cell with gold-plated electrodes at pH 7.4 and at the temperature of 25 °C using a Dynamic Light Scattering method with a Malvern Zetasizer Nano ZS analyser (Malvern Instruments, Great Britain). The zeta potential was automatically calculated from electrophoretic mobility based on the Smoluchowski equation. The results represent the zeta potential over 10 replicate measurements.

  1. (b)

    Calculation of static compression modulus.

Using the Langmuir technique, containing a mixture of lipids were prepared from the solutions mimicking the membranes of the tested cancer cell lines40. To ensure a high reproducibility of the obtained π-Å isotherms for each monolayer, the measurements (controlled by using KSV NIMA software) were repeated 3–5 times (± 0.1–0.3 Å2). The standard deviation for surface tension measurements was ± 0.1 mM/m. The obtained data were transferred into Sigma Plot to calculate static compression modulus (Cs−1) according to the formula: Cs−1 = − (dπ/dln Am), where π is surface pressure, and Am—is a value characterizing the surface area occupied by a molecule in the membrane. Based on the obtained diagrams, the parameter at 30 mN/m was determined.

Statistical analysis

The standard deviation (SD) was calculated for the obtained physicochemical and biochemical measurements. Duncan’s multiple range test (from PC SAS 8.0 software) was used for static analysis between different treatments. The assumed significance level was p < 0.05. Statistical tests were performed using the STATISTICA 13.3 software package.

For the entire article, statistically significant differences between 3FTx concentrations (e.g., 10 ng/ml and 40 ng/ml) were marked in a standardized system and named as SK-N-SH cell line taken from the Latin alphabet, and the HL-60 cell line taken from the Greek alphabet; asterisks indicate a significant difference between 3FTx fractions (e.g., G1 and G5).

Ethics approval and consent to participate

This manuscript was prepared in accordance with international ethical guidelines.