Profiling cytotoxicity of nanofractionated elapid snake venoms in human cell lines representing different tissues

Abstract

Elapid snakebites cause severe toxicity, predominantly neurotoxicity and general cytotoxicity. However, the specific cellular impacts of individual venom toxins remain largely underexplored. This study developed a high-throughput platform for profiling cytotoxicity from elapid venoms, focusing on nanofractionation analytics to enhance selectivity and toxin identification. Elapid Venoms were tested on four human cell lines, representing kidney (RPTEC/TERT1), liver (HepaRG), endothelial (iPSC-EC), and skin (HaCaT) tissues. Cytotoxic effects were assessed through cell coverage, viability, and metabolic assays in both crude and nanofractionated venom samples. Nanofractionation revealed selective cytotoxicity in venom components, notably phospholipases A₂ (PLA₂s) and three-finger toxins (3FTxs), which impaired membrane integrity and cellular metabolism. Crude B. multicinctus venom displayed specific cytotoxicity toward liver and skin cells but not kidney or endothelial cells. Cytotoxicity of nanofractionated B. multicinctus venom was lost, likely due to denaturing conditions of the reversed-phase separation. Fractionation after size exclusion chromatography (SEC) for post-column bioassaying to avoid toxin denaturation yielded bioactive fractions, with 3FTxs, PLA₂s, and Kunitz-type serine protease (KUNs) likely responsible for the observed cell permeability disruption, extracellular matrix (ECM) degradation, and metabolic loss. This integrated analytical workflow, combining nanofractionation with high-throughput cytotoxicity assays and venomics, enabled rapid identification of venom components with cell type-specific toxicity. Our findings contribute to understanding elapid venom toxicity and can aid in developing targeted snakebite treatments focusing on cytotoxicity responsible for tissue-specific damage.

Graphical abstract

Highlights

• •High-throughput analytical platform for profiling selective cytotoxicity of elapid venoms toxins.
• •Nanofractionation analytics to enable cytotoxin identification in venoms.
• •Naja venoms show dose-dependent cytotoxic effects, while nanofractionated toxins reveal selective toxicity across cell lines.
• •Supports research on snakebite treatments by providing insights into mechanisms of action of toxins in various cell types.
• •The method developed gives a strategy for rapid analysis of natural products from a snake venom perspective.

1. Introduction

In 2017, snakebite envenoming was categorised as a priority neglected tropical diseases [1]. Snakebites cause more than 100,000 deaths and 400,000 disabilities per year, especially in the developing world [2]. The main pathological effects of snakebite envenoming can be classified as hemotoxicity, neurotoxicity, and cytotoxicity [3], and which is the result of the action of diverse, medically relevant, toxin families such as the phospholipase A₂s (PLA₂s), metalloproteases (SVMPs), serine proteases (SVSPs), and three-finger toxins (3FTxs) [4]. Snake venom analysis is challenging due to the considerable variation in toxin composition found in the venoms of different snake species, with factors such as geographical location, age, sex, and habitat associated with variable toxin variability. Differences in venom composition can be the result of differences in genes, transcription, translation and/or post-translational modifications (PTMs) [5].

Current snakebite treatment relies on antivenoms consisting of immunoglobulins purified from the serum or plasma of animals immunized by snake venom [6]. Antivenoms are highly specific in neutralizing only one or several venoms from species of certain geographical locations used in the immunization process [7], though in some cases also neutralize the venom of related species with similar venom compositions [8]. With such complex variations in venom composition and drawbacks in current antivenom treatment, new-generation candidate snakebite treatment strategies are needed [9]. To develop new snakebite treatment options and to better understand the pathophysiology of snakebite envenoming, it is critical to robustly characterize both the composition and biological activities of venom toxins from medically important snake species.

Compared to the systemic neurotoxic and hemotoxic effects that present following snakebite, venom-induced cytotoxicity is much less studied, but can be highly relevant for both local envenoming pathology (e.g., tissue damage near the bite site [10]), as well as systemically relevant effects (e.g., disruption of capillary blood vessels [11] or direct action on kidney function [12]). In vitro cellular assays are valuable primary assays for the screening of venom cytotoxicity due to their high-throughput potential, straightforward assay readout, ethical considerations, and availability and affordability compared to ex vivo or in vivo assays. Examples of the use of in vitro cytotoxicity assays in venom research include plate reader-based and flow cytometric analysis for cytotoxicity assessment. Examples of such assays are cytotoxicity analysis towards human umbilical vein endothelial cells [13] and keratinocytes [14] and analysis of cell proliferation using rat skeletal muscle cells [15]. Cell metabolism measurement of a murine skeletal muscle cell line against Lachesis muta muta venom and integrity assessment of the cell membrane of normal human keratinocytes stained by Sytox Green analyzed by fluorescence microscopy [16] are other examples. Many studies focused on the measurement of the cytotoxicity of crude venoms, while the actual toxins that contributed to the overall cytotoxicity observed were not investigated.

To investigate the cytotoxic properties of individual toxins in crude venoms and to simultaneously identify these toxins, separation of the individual toxins in these venoms is required prior to studying the purified toxins. A high-throughput analytical methodology to achieve this is nanofractionation analytics. Nanofractionation analytics is an analytical method in which the eluate of chromatographic separation of snake venom is fractionated in high resolution on a 96, 384, or 1536 well plate for subsequent bioassaying [17]. Crude venom is separated by reversed-phase high-performance liquid chromatography (RP-HPLC) followed by a flow split into two parts of which 90% of the eluate is fractionated and 10% is directed to mass spectrometry (MS), for parallel accurate toxin mass determination. After vacuum-centrifuge drying of the well plates with collected eluent with fractionated toxins, the dried toxins remain in the wells of the well plates and subsequently, a bioassay of choice can be applied. Examples of this analytical methodology in which different bioassays were used previously include coagulation bioassaying [18], PLA₂ enzyme activity bioassaying [19], and hemotoxicity profiling for toxins causing erythrocyte lysis [20]. Toxin proteomics is also required for actual toxin identification. Proteomics on venom toxins is called venomics and was initially introduced by Juárez et al. [21] in 2004. In this study, a high throughput (HT) venomics methodology recently introduced by Slagboom et al. [22] was integrated into the nanofractionation analytics workflow. For this, instead of performing a bioassay on a well plate with nanofractionated venom toxins, HT venomics was applied.

Toxicity assessment using in vitro mammalian cellular assays is in demand, especially within the nanofractionation analytics workflow to study toxicity in different mammalian cell lines such as human umbilical vein endothelial cells, renal tubular epithelial cells, hepatoma cells, keratinocytes, and erythrocytes [23,24] to assess the possible selectivity of venom toxins. Recently, Bittenbinder et al. [25] described a mammalian cytotoxicity assay within nanofractionation analytics in which the immortalized human renal proximal tubular cell line (RPTEC/TERT1) was used. As snake venom toxins can exert cytotoxicity on different cell types in many ways, there is a desire to expand to other cell lines to acquire a more complete overview of the cell-specific cytotoxic effects of cytotoxins in snake venoms at the cellular level. In this study, the nanofractionation approach was coupled with cytotoxicity assessments against four cell types, namely kidney (RPTEC/TERT1), liver (HepaRG), endothelial (iPSC-EC), and skin (HaCaT) cell lines. Integrating fluorescence-based cytotoxicity assays with HT venomics enabled characterization of cytotoxins in venoms. This integrated analytical approach may aid in advancing the development of next-generation snakebite treatments by identifying the key pathologically relevant toxins present in snake venoms.

2. Materials and methods

2.1. Reagents

Chemicals were of analytical grade. Acetonitrile (ACN) and trifluoroacetic acid (TFA) of ultra-high performance liquid chromatography-mass spectrometry (UHPLC/MS) grade were purchased from Biosolve (Valkenswaard, The Netherlands). Water used came from a Milli-Q plus system (Millipore, Amsterdam, The Netherlands). 10 mg/mL Hoechst 33342 was from Invitrogen (Amsterdam, The Netherlands). From Sigma-Aldrich were 10 mg/mL propidium iodide (PI) (Sigma p4170), dimethyl sulfoxide (DMSO; ≥99.9 %), Dulbecco's phosphate-buffered saline (DPBS), insulin (Sigma I9278), hydrocortisone hemisuccinate (Sigma H2270), Penicillin/Streptomycin (Pen/Strep; Sigma P4333), fetal calf serum (FCS; Sigma-Aldrich F6765), egg yolk, iodoacetamide, β-mercaptoethanol and ammonium bicarbonate (Sigma-Aldrich, Zwijndrecht, The Netherlands). Glutamax (32551–020) and resazurin were from Thermo Fisher Scientific (Amsterdam, The Netherlands). Recombinant trypsin (EMS0006–4X 100UG) was from Promega Benelux B.V. (Leiden, The Netherlands) and 96 well plates (F-bottom, with lid, sterile, Cat-No. 655,180, CELLSTAR@) were from Greiner Bio One (Amsterdam, The Netherlands).

2.2. Venoms

Lyophilized snake venoms from Naja mossambica (Mozambique spitting cobra, captive bred), Naja naja (Indian cobra, captive bred), and Naja haje (Egyptian cobra, Uganda) were provided by the Centre for Snakebite Research and Interventions Herpetarium (Liverpool School of Tropical Medicine, UK). Bungarus multicinctus (many-banded krait) was provided by the University of Singapore. The lyophilized venoms were stored long-term at −80 °C. Stock solution of crude venom (1 ± 0.1 mg/mL) was prepared in water before analysis, and then aliquoted and stored at −80 °C until use.

2.3. Cell culture and assays

2.3.1. Cell culture

Cells were cultured at 37 °C in a 5% CO₂ humidified atmosphere routinely in 10 cm dishes and medium was replaced every 2–3 days. Cytotoxicity experiments cells were cultured and treated in 96 well plates. The human proximal tubular cell line, RPTEC/TERT1 (Evercyte GmbH, Vienna [26]), was cultured in a mixture of Dulbecco's modified Eagle's medium (DMEM) F-12 supplemented with glutamax, epithelial growth factor (EGF), insulin-transferrin-selenium (ITS), hydrocortisone, and Pen/Strep (as previously described [27]). For treatment, differentiated, contact-inhibited cells were used. HepaRG cells (Biopredic International, St Grégoire, France) were maintained in Williams' E medium supplemented glutamax, ITS, hydrocortisone, and 0% FCS. HepaRG was differentiated by adding 1.7% DMSO for 4 days. For experiments, medium without DMSO and without FCS were used [28]. Induced pluripotent stem cell derived endothelial cells (iPSC-ECs) were differentiated from human iPSC as previously described [29]. The protocol uses a 3-step differentiation over 6 days, with the last step including forskolin and vascular endothelial growth factor (VEGF) stimulation. We deviated from the protocol by purifying the cells with CD31 (PECAM) magnetic bead sorting instead of CD144 (VE-cadherin, Amsterdam, The Netherlands). CD31-positive cells were cultured on fibronectin-coated plates and maintained in VascuLife endothelial medium complete kit (LL-0003, LifeLine Cell Technology, Amsterdam, The Netherlands) containing VEGF165 (StemCell, Amsterdam, The Netherlands), SB-431542 (10 μM, TGF-beta inhibitor, M1794, Amsterdam, The Netherlands) and 10% FCS and Pen/Strep. Cells were not allowed to reach confluence. The immortalized keratinocyte cell line (HaCaT, Vrije Universiteit, Amsterdam, The Netherlands) was cultured in DMEM, glutamax, 10% FCS, and Pen/Strep.

2.3.2. Cellular assays

2.3.2.1. Live cell count ratio

It was generated by adding PI (0.5 μg/mL) and Hoechst 33342 (1 μg/mL) to the incubation medium [30]. Cells were imaged using an operetta CLS high content imager (PerkinElmer, Waltham, MA, USA) and analyzed with Harmony software 4.8 (PerkinElmer, Waltham, MA, USA). The number of PI-positive cells was subtracted from the Hoechst positive count (total cells), to give the live cell number. This was converted to a control ratio by dividing with the untreated control value.

2.3.2.2. Cell area

In the above assay, nuclei will be counted only if the cells are adhered to the plastic. Thus, if a treatment specifically alters cell or monolayer adherence (e.g. by extracellular matrix (ECM) degradation), the live cell count ratio will not reflect the true situation. Thus, bright field imaging was used to determine the amount of cell coverage on the well. Harmony software was used to quantify the total surface area of the attached using a learn-by-example approach [25].

2.3.2.3. Resazurin reduction assay

At the end of the treatment, the medium was replaced with 44 μM resazurin solution (dissolved in DPBS) in each well and incubated for 1 h at 37 °C. Resorufin formation was assayed at 540 nm excitation and 590 nm emission using a CLARIOstar Plus reader (BMG Labtech, Ortenberg, Germany) to measure the signal of all wells at 540 nm excitation and 590 nm emission. Fluorescent values were converted to ratio of untreated cells. Cell counting is a high-throughput identification method based on morphological assessment. This assessment can be further validated using the resazurin reduction assay. Only when both assays yield consistent results, this will be considered as the observed effects are indicative of cytotoxicity.

2.3.3. Cell treatment

Differentiated cells were incubated with the venoms at 100 μL per well for 24 h in a cell culture incubator. Plates were imaged for fluorescence and bright field in the high content imager and the resazurin reduction assay was carried out in the same wells. Three wells were used per determination. 0.5% Triton X-100 was employed as a positive control in the cytotoxicity screening assays due to its well-documented ability to disrupt cellular membranes, leading to cell lysis and death. As a non-ionic detergent, Triton X-100 effectively solubilizes membrane lipids and proteins, resulting in the complete breakdown of cell integrity. This characteristic makes it an ideal agent for establishing a baseline for maximum cytotoxicity within the assay. For crude venom, 0, 1.1, 3.3, 11.1, 33.3, and 100 μg/mL concentrations were used. In addition, treatment was also conducted in the presence of 1.75 mg/mL egg yolk. All raw cytotoxicity data was normalized with the data from the negative control. The ratios of the three readouts were used to generate the cytotoxicity chromatograms. Cytotoxicity is validated by using both cell count and metabolic assays, with a threshold larger than 20% compared to the control to be considered indicative of cytotoxicity.

All cellular bioassay data are plotted as means ± standard error of the mean (SEM) unless otherwise specified. Statistical analysis was performed using GraphPad Prism. Intergroup differences for continuous variables were assessed by two-way analysis of variance (ANOVA), using Dunnett's post-test to determine the significance of differences between control and treated groups.

2.4. HPLC separation, nanofractionation, and mass spectrometry

Venom solution (50 μL) was injected with a Shimadzu SIL-20AC autosampler (Shimadzu Benelux, s-Hertogenbosch, The Netherlands), and then separated using a Waters Xbridge Peptide BEH 300 C18 analytical column (100 mm × 4.6 mm, 5-μm particle size, 300 Å pore size; Waters, Etten-Leur, The Netherlands) at 30 °C in a Shimadzu CTD-10AC column oven. The eluate (a flow rate of 0.5 mL/min was used for separation) was sent to a Shimadzu SPD-20A (variable wavelength detector; wavelength set at 220 nm) prior to being directed to a flow splitter. The analytical mobile phase gradient was controlled by two Shimadzu LC-20AB pumps. Mobile phase A was composed of 98% H₂O, 2% ACN, and 0.1% TFA, while mobile phase B was 98% ACN, 2% H₂O, and 0.1% TFA. All the settings of the system were controlled with Shimadzu Lab Solutions software (v. 5.117, Amsterdam, The Netherlands). The HPLC gradient used was optimized for each venom included in the study (HPLC gradient optimization data is not shown in this study). For N. mossambica, a linear increase of the mobile phase B from 0% to 13% was done in 5 min, followed by going from 13% to 50% in 30 min, which was accelerated by going from 50% to 90% in 3 min. This was kept for 7 min, followed by a decrease from 90% to 0% in 1 min, which was finally followed by equilibration for 5 min. For N. naja, a linear increase of mobile phase B was from 0% to 14% in 5 min, from 14% to 50% in 30 min, from 50% to 90% in 3 min, 7 min at 90%, a decrease from 90% to 0% in 1 min followed by 5 min at 0%. For N. haje, a linear increase of mobile phase B was from 0% to 20% in 5 min, from 20% to 37% in 26 min, from 37% to 90% in 3 min, 7 min at 90%, a decrease from 90% to 0% in 1 min followed by equilibration in 5 min. For B. multicinctus, a linear increase of mobile phase B was from 0% to 19 % in 5 min, from 19% to 33% in 30 min, from 33% to 90% in 3 min, 7 min at 90%, a decrease from 90% to 0% in 1 min followed by 5 min equilibration. After separation, the elute was split into a 1:9 (v/v) ratio, and the 90% portion was transferred to a FractioMate™ nanofractionator (Vrije Universiteit, Amsterdam, The Netherlands) controlled by FractioMator software (v.1.0, Vrije Universiteit Amsterdam, The Netherlands). The fractions were collected at a resolution of 20 s/well onto 96-well plates. All the collected toxins in the plates were subsequently vacuum centrifuged to dryness overnight using a Christ Rotational Vacuum Concentrator (RVC 2–33CD plus, Martin Christ, Osterode am Harz, Germany) operated with a −80 °C cooling trap. The dried plates were stored at −20 °C until bioassaying or HT Venomics. Low concentrations of TFA and ACN at a sufficient concentration during separation can denature toxins, leading to a loss of cytotoxicity. To assess these possible effects, we conducted tests using crude venom dissolved in Milli-Q water, along with tests where crude venom was dissolved in solutions with different concentrations of ACN and TFA. Next, these solvents were vacuum-centrifuged to dryness after which the dried venoms were tested in the bioassays. For 3FTx and PLA₂ toxins, it was found that they were very resistant to both TFA and ACN and from there we continued with our study involving venom separation for post-column bioassaying. It also has to be noted that the composition of the mobile phase to which venom toxins are exposed during separation is dependent on the elution and retention times of each toxin, as gradient LC is employed in the process. The 10% portion of the elute was sent to MS detection (maXis QTOF, Bruker Daltonics, Germany) operated with electrospray ionization (ESI) in positive-ion mode at the range between 800 and 5500 m/z. The ESI source parameters were capillary voltage of 3.5 kV, source temperature of 200 °C, nebulizer at 0.8 Bar, and dry gas flow at 6.0 L/min. In-source collision-induced dissociation (CID) was set at 200 eV, and 1 average spectrum was stored per s. Bruker Compass software (v. 3.0, Bruker Daltonics, Germany) was used for instrument control and data analysis.

For MS data processing, the total-ion current (TIC) was plotted from the recorded MS data. Extracted-ion currents (EICs) were extracted from the TIC by plotting the EIC of the most abundant charge state of each toxin. By matching EICs with bioactivity peaks in the bioactivity chromatograms through retention time and peak shape matching, those EICs that matched were assigned as the probable accurate masses of the bioactive toxins.

Since there are small time differences of eluted toxins for arriving at the ultraviolet (UV) and the MS detector and for fractionation, the UV, MS and fractionation times needed to be aligned. This was done by measuring the retention time of the thrombin inhibitor argatroban in UV and MS, and by fractionating the inhibitor on a well plate followed by performing a coagulation bioassay as described by Slagboom et al. [18]. The bioassay chromatogram resulting from the coagulation bioassay will give a negative peak for argatroban from which a retention time can be measured. By measuring the differences in retention time between UV, MS, and the coagulation bioassay, the delay times could be calculated, which were then used to align the UV, MS, and bioassay data presented in this study.

2.5. HT venomics to characterize venom toxins

HT venomics was developed by Slagboom et al. [22] and applied in this study. In short, 120 μL fractions from the LC separations of the venoms included in this study were collected in wells of 96 well plates using the FractioMate in the same manner as performed for collecting fractions for cytotoxicity bioassaying, and which were then transferred to 384 well plates. Next, the 384 well plates were vacuum centrifuged to dryness overnight and then stored at −20 °C until HT venomics analysis. For this, nanofractionated venom toxins in the appropriate wells were mixed with 25 μL of 25 mM ammonium bicarbonate and 0.05% β-mercaptoethanol (pH 8.2) and then heated at 95 °C for 15 min for reduction. After that, 10 μL 12.5 mM iodoacetamide was added and well plates were incubated for 30 min at room temperature in the dark for alkylation. Subsequently, a trypsin solution was made by diluting a trypsin stock solution (1 μg/μL trypsin in 50 mM acetic acid) 100 times in 25 mM ammonium bicarbonate. Of this solution, 10 μL trypsin solution was added to each well and then the plates were incubated overnight at 37 °C. Next, the plates were centrifuged at 1000 rpm for 1 min in an Eppendorf Centrifuge 5810 R (Amsterdam, The Netherlands), followed by the addition of 10 μL of 1.25% formic acid (FA) to the well plates. All pipetting steps for the tryptic digestion procedure were performed with a ThermoFisher Multidrop (ThermoFisher Scientific, Ermelo, The Netherlands). Finally, the well plates were ready for analysis using nanoLC-MS/MS. For this, the digested samples were separated by an UltiMate 3000RSLCnano system (Thermo Fisher Scientific, Ermelo, The Netherlands) with an Acclaim PepMap 100C18 HPLC Column (150 mm × 75 μm, 2 μm particle size, 100 Å pores) in combination with an Acclaim PepMap 100C18 trapping column (5 mm × 0.3 mm, 5 μm particle size, 100 Å pores, ThermoFisher Scientific Ermelo, The Netherlands). The columns were kept in a column oven at 45 °C. The autosampler was run in full-loop injection mode with 1 μL injection volume. Mobile phase A consisted of 98% H₂O, 2% ACN, and 0.1% FA, and mobile phase B consisted of 98% ACN, 2% H₂O, and 0.1% FA. The gradient started with a 3 min isocratic separation at 1% B, linear increase to 40% B in 7.5 min followed by linear increase to 85% B in 0.1 min, 0.7 min isocratic separation at 85% B, linear decrease to 1% B in 0.2 min, and equilibration at 1% B for 3.7 min. Detection was performed using a Bruker maXis qTOF mass spectrometer (Bruker Daltonics, Bremen, Germany) equipped with a Bruker Captivespray source operating in positive-ion mode. The source parameters were: source temperature, 150 °C; capillary voltage, 1.2 kV; gas flow, 3 L/min. Spectral data were stored at a rate of 1 average spectrum/s in the range of 50–3000 m/z. MS/MS spectra were obtained using CID in data-dependent mode using 10 eV collision energy. Bruker Compass software version 3.0 was used for instrument control and data analysis. The mgf files for each well were generated by using the ProcessWithMethod function which is ready for Mascot (Matrix Science, London, UK) searches using UniProt database containing only Serpentes accessions. R packages [31] were used for high throughput data processing in which all relevant parameters were script-controlled extracted from the Mascot files (i.e., toxin family, protein intensity, composition) for each well and then deposited into a single Excel file. Using this Excel sheet, so-called protein score chromatograms (PSCs) were plotted for each toxin found in each venom for which the retention time of toxin fractionation was plotted on the x-axis versus the protein score of each toxin on the y-axis. More details on this procedure, and the R-scripts used, can be found in the study of Slagboom et al. [22].

3. Results and discussion

The objective of this study was to establish and apply a methodology for profiling the toxicity of elapid snake venoms (i.e., N. mossambica, N. haje, N. naja, and B. multicinctus) in cells representing different tissue phenotypes. These were renal proximal tubule cells, liver hepatocytes, endothelial cells and dermal keratinocytes using RPTEC/TERT1, HepaRG, iPSC-EC, and HaCaT cells, respectively. The analytical workflow included crude venom cytotoxicity assessment, venom separation, cytotoxicity assessment of separated toxins for the human cells, toxin characterization by accurate mass assessment, and HT venomics for toxin identification. The assaying workflow described by Bittenbinder et al. [25] was applied here as the basis using three readouts, live cell ratio (PI/Hoechst), cell area (bright field), and cellular metabolism (resazurin reduction). Since saturated fatty acids such as phospholipids can facilitate indirect hemolytic activity in red blood cells (RBC) for PLA₂ toxins [20,32], experiments were conducted also with the addition of egg yolk emulsion to provide external phospholipids. The egg yolk was included as a source of exogenous phospholipids to enhance and reveal the catalytic activity of PLA₂s via membrane interactions, based on previous studies in the venom field [25,33]. While egg yolk is primarily composed of phospholipids and lipoproteins, we cannot entirely exclude the possibility that other venom components might non-specifically interact with egg yolk constituents. However, PLA₂s are uniquely capable of enzymatically hydrolyzing phospholipids, and the increase in cytotoxicity we observed in the presence of egg yolk was consistent with known PLA₂-driven indirect cytotoxic effects (e.g., enhanced membrane permeability and metabolic disruption). Moreover, the effect was selective to venoms known to contain active PLA₂s, supporting the interpretation that PLA₂ activity is the main contributor. Still, minor interactions by other toxin families cannot be entirely ruled out. By investigating different tissue-damaging representative cell lines to assess cell-type-specific cytotoxins in elapid venoms with parallel toxin characterization by HT venomics, the composition of these venoms can be qualitatively evaluated together with pinpointing the cytotoxins and their cell-type specific toxicities.

3.1. Cytotoxicity profiling of crude venoms

3.1.1. Renal proximal tubule RPTEC/TERT1 cells

An elaborate description of the results is given in Supplementary material 1.1. In RPTEC/TERT1 cells, the live cell number only got drastically reduced at a venom concentration of 100 μg/mL for N. haje (Fig. 1A), the strongest toxicity was induced by N. mossambica (Fig. 1B) and N. naja (Fig. 1C). There were no morphology and/or quantitative statistical changes when the cells were exposed to B. multicinctus at any of the concentrations tested (Fig. 1D). Live cell counts and resazurin reduction were observed at lower venom concentrations than found for N. haje (i.e., 11.1 μg/mL). Higher cytotoxicity was observed for the three Naja venoms in the presence of egg yolk (N. haje (Fig. 1E), N. mossambica (Fig. 1F), N. naja (Fig. 1G)). No cell permeability and cell metabolism changes were observed for B. multicinctus venom in the indirect cytotoxicity (Fig. 1H). A similar effect has been reported in previous research on hemolytic activity [34]. Two Naja venoms (N. naja and N. mossambica) caused more substantial cytotoxicity effects in the presence of egg yolk by disrupting cell membrane permeability and cell metabolism from lower venom concentrations compared to the results of the direct cytotoxicity effects. Notable disruption of the cell membrane from a venom concentration of 3.3 μg/mL and slight cell metabolism loss was observed for N. haje venom in the presence of egg yolk. B. multicinctus venom is known to mainly cause neurotoxicity symptoms in envenoming cases [35] and also did not show cytotoxicity in a recent study on cytotoxicity of snake venoms by Bittenbinder et al. [25]. No observed cell area changes either by direct or indirect cytotoxicity at any venom concentration indicated no ECM degradation for all elapid venoms tested in this cell line.

Fig. 1.

Effects of crude venoms in renal cells. RPTEC/TERT1 cells were differentiated in 96 well plates and treated with crude venoms at concentrations of 0, 1.1, 3.3, 11.1, 33.3, and 100 μg/mL for 24 h. Black group: live cell number staining ratio, blue group: cell area ratio, orange group: resazurin viability vs. control ratio. Bars represent the mean ratio plus standard error. PC is the positive toxicity control (0.5% Triton-X-100). Statistical significance was calculated with a two-way analysis of variance (ANOVA), n = 3, ^∗P < 0.05, ^∗∗P < 0.005, ^∗∗∗P < 0.0005. (A, E) The cell readouts of Najahaje. (B, F) The cell readouts of Najamossambica. (C, G) The cell readouts of Najanaja. (D, H) The cell readouts of Bungarusmulticinctus. A–D without egg yolk, E–H with egg yolk. All the bar charts share the same y-axis units, representing the ratio percentage (%) between the tested wells and the negative control wells.

3.1.2. Liver hepatocyte HepaRG cells

The same procedure as described for the RPTEC/TERT1 experiments was applied to the HepaRG cell line. An elaborate description of the results is given in Supplementary material 1.2. To summarize, the venoms of the three Naja species (N. haje (Fig. 2A), N. mossambica (Fig. 2B), N. naja (Fig. 2C)) showed similar cytotoxicity effects both for direct and indirect cytotoxicity with the loss of live cell number and decrease in resazurin reduction, highlighting compromised cell permeability disruption and cell metabolism decrease. As shown in Fig. 2D, B. multicinctus venom exhibited significant cytotoxicity, inducing cell permeability disruption, ECM degradation, and loss of cell metabolic activity. Moreover, enhanced cytotoxicity for three Naja species was notably observed for the indirect cytotoxicity assessment (i.e., assaying in the presence of egg yolk), while no ECM degradation was observed as deduced from no cell area changes either by direct or indirect cytotoxicity (N. haje (Fig. 2E), N. mossambica (Fig. 2F), andand N. naja (Fig. 2G)). These effects associated with cell permeability disruption were similar to the results for the RPTEC/TERT1 cell line, with the main difference being that cell metabolism loss was observed in the HepaRG cell line and not in the RPTEC/TERT1 cell line. Indirect cytotoxicity intensified the effects that were observed in the direct cytotoxicity in three Naja species instead of adding new bioactivities. For B. multicinctus venom with extra egg yolk (Fig. 2H), notable and distinct cytotoxicity was observed for which venom-induced cell permeability disruption, ECM degradation, and cell metabolism loss were measured. This notable cytotoxicity difference of B. multicinctus venom between RPTEC/TERT1 and HepaRG cells is apparently attributed to a stronger and/or selective cytotoxicity for the HepaRG cell line than when RPTEC/TERT1cells were exposed to this venom.

Fig. 2.

Effects of crude venoms in liver cells. HepaRG cells were differentiated in 96 well plates and treated with crude venoms at concentrations of 0, 1.1, 3.3, 11.1, 33.3, and 100 μg/mL for 24 h. Black group: live cell number staining ratio, blue group: cell area ratio, orange group: resazurin viability vs. control ratio. Bars represent the mean ratio plus standard error. PC is the positive toxicity control (0.5 % Triton-X-100). Statistical significance was calculated with a two-way analysis of variance (ANOVA), n = 3, ^∗P < 0.05, ^∗∗P < 0.005, ^∗∗∗P < 0.0005. (A, E) The cell readouts of Najahaje. (B, F) The cell readouts of Najamossambica. (C, G) The cell readouts of Najanaja. (D, H) the cell readouts of Bungarusmulticinctus. A–D without egg yolk, E–H with egg yolk. All the bar charts share the same y-axis units, representing the ratio percentage (%) between the tested wells and the negative control wells.

3.1.3. Human umbilical vein endothelial iPSC-EC cells

iPSC-EC cells have been used to investigate the toxicities of many endogenous molecules, xenobiotics, and cancer drug candidates [36]. As some snake venom toxins can exert their cytotoxicity to capillary blood vessels, this cell line was included in this study. The same assaying procedures were used and a detailed description of the results for this cell line is provided in the Supplementary material 1.3. The three Naja venoms showed similar cytotoxicity effects, with decreases of live cell numbers and resazurin reduction, indicating disruption of cell membrane permeability and cell metabolism, but not showing ECM degradation (N. haje (Fig. 3A), N. mossambica (Fig. 3B), and N. naja (Fig. 3C)). For B. multicinctus venom, there were no significant differences measured in the readouts of live cell number, cell area, and resazurin reduction compared with the control group (Fig. 3D). Surprisingly, a drastic enhanced effect in cytotoxicity was observed for this cell line which was different from the previous cell lines investigated (N. haje (Fig. 3E), N. mossambica (Fig. 3F), N. naja (Fig. 3G)). For the iPSC-EC cell line, B. multicinctus did not show any cytotoxicity at any venom concentration tested in the indirect effects (Fig. 3H).

Fig. 3.

Effects of crude venoms in human umbilical vein cells. iPSC-EC cells were differentiated in 96 well plates and treated with crude venoms at concentrations of 0, 1.1, 3.3, 11.1, 33.3, and 100 μg/mL for 24 h. Black group: live cell number staining ratio, blue group: cell area ratio, orange group: resazurin viability vs control ratio. Bars represent the mean ratio plus standard error. PC is the positive toxicity control (0.5 % Triton-X-100). Statistical significance was calculated with a two-way analysis of variance (ANOVA), n = 3, ^∗P < 0.05, ^∗∗P < 0.005, ^∗∗∗P < 0.0005. (A, E) The cell readouts of Najahaje. (B, F) The cell readouts of Najamossambica. (C, G) The cell readouts of Najanaja. (D, H) The cell readouts of Bungarusmulticinctus. A–D without egg yolk, E–H with egg yolk. All the bar charts share the same y-axis units, representing the ratio percentage (%) between the tested wells and the negative control wells.

3.1.4. Dermal keratinocyte HaCaT cells

The cytotoxicity profiles of the venoms of N. haje, N. mossambica, N. naja, and B. multicinctus were finally assessed for both direct and indirect cytotoxicity on HaCaT keratinocytes. An elaborate description of the results is given in the Supplementary material 1.4. To summarize these results, the three Naja venoms (i.e., N. haje, N. mossambica, and N. naja) showed similar cytotoxicity profiles but to a different extent for the direct cytotoxicity measurements (N. haje (Fig. 4A), N. mossambica (Fig. 4B), N. naja (Fig. 4C)). Stronger cytotoxic effects were observed with N. mossambica and N. naja venoms, as both the live cell number loss and decrease in resazurin reduction were observed at lower concentrations (i.e., starting from 11.1 μg/mL) than observed for N. haje venom, which only caused a slight disruption of cell permeability at the highest concentration tested. For B. multicinctus venom, a notable cytotoxicity was observed that could be deduced to result from induced cell permeability disruption, ECM degradation, and cell metabolism, in a similar manner as observed for the HepaRG cell line (Fig. 4D). For the screening of indirect cytotoxicity, cell metabolism loss for N. haje (Fig. 4E), assessed by resazurin reduction, was observed at the highest concentration tested, which was slightly increased compared to the results found for direct testing. no increased cytotoxicity was shown for N. mossambica venom (Fig. 4F), while a slightly increased cytotoxic effect was also seen for N. naja (Fig. 4G). No cell area changing was observed for the indirect cytotoxicity tests indicated that none of the three Naja venoms caused ECM degradation at the doses tested. No significant difference in cytotoxicity of B. multicinctus venom was found for the indirect cytotoxicity tests compared to the direct cytotoxicity tests (Fig. 4H).

Fig. 4.

Effects of crude venoms in epidermis tissues cells. HaCaT cells were differentiated in 96 well plates and treated with crude venoms at concentrations of 0, 1.1, 3.3, 11.1, 33.3, and 100 μg/mL for 24 h. Black group: live cell number staining ratio, blue group: cell area ratio, orange group: resazurin viability vs control ratio. Bars represent the mean ratio plus standard error. PC is the positive toxicity control (0.5 % Triton-X-100). Statistical significance was calculated with a two-way analysis of variance (ANOVA), n = 3, ^∗P < 0.05, ^∗∗P < 0.005, ^∗∗∗P < 0.0005. (A, E) The cell readouts of Najahaje. (B, F) The cell readouts of Najamossambica. (C, G) The cell readouts of Najanaja. (D, H) The cell readouts of Bungarusmulticinctus. A– D without egg yolk, E–H with egg yolk. All the bar charts share the same y-axis units, representing the ratio percentage (%) between the tested wells and the negative control wells.

Based on the screening results of the three Naja venoms with the four cell lines (Table S1), cytotoxicity was characterized by increased PI permeability and decreased resazurin reduction as opposed to cell detachment via ECM degradation. While differences in cytotoxicity potency among the venoms of the different species tested were observed, N. mossambica venom contributed to the strongest overall cytotoxicity for all the cell lines investigated, followed by N. naja venom. These two Naja snake species are well known to have cytotoxic venoms and our results are in line with this [36,37]. N. haje had the weakest cytotoxicity which only showed an effect at a very high venom concentration tested (i.e., 100 μg/mL) in three cell lines (RPTEC/TERT1, HepaRG, and HaCaT) while slightly stronger cytotoxic effects were observed in the iPSC-EC cell line (from 33.3 μg/mL). A previous investigation by Omran et al. [38] demonstrated that N. haje caused cytotoxicity starting at 10 μg/mL in the human embryonic kidney cells (293T) and the mouse myoblast cell line (C2C12). Cytotoxicity was measured by the activity of lactate dehydrogenase or creatine kinase activity measured in the supernatant media, which are characteristics of injury to the plasma membrane.

For B. multicinctus, the neurotoxic properties of its venom have been thoroughly studied and reported in the literature [35]. However, this is not the case for its potential cytotoxic properties. Our results on crude B. multicinctus venom cytotoxicity profiling with the four different cell lines showed specific cytotoxicity to only two of the tested cell lines (i.e., HepaRG and HaCaT). In other words, selective cytotoxicity for different cell lines was observed, which was not the case for the three cobra venoms (Table S1). However, similar selective cytotoxicity results have been previously described, as Lu et al. [39] isolated a cytotoxic l-amino acid oxidase from B. multicinctus venom that showed cytotoxicity to tumor cells (MGC-803, SMMC-7721, and PC3), and much less cytotoxicity to normal hepatocarcinoma cells (MCF-10A). The potential reason for this might be the different susceptibility to snake venom toxins of different cell membranes and/or specific receptors (i.e., ion channels) present in cell membranes.

3.2. Cytotoxicity profiling of nanofractionated venom toxins with integrated reversed-phase liquid chromatography-mass spectrometry (RPLC-MS) and HT venomics

The same four cell lines were then utilized in cytotoxicity profiling experiments for the venoms of N. mossambica and N. naja after nanofractionation analytics using RP-HPLC separations. Based on the prior crude venom cytotoxicity profiling, the venom of N. haje was only examined in the RPTEC/TERT1 cell line after nanofractionation, while the venom of B. multicinctus was examined with two separation methods (high-performance reversed-phase liquid chromatography (HP-RPLC) and size-exclusion chromatography (SEC)) before nanofractionation, with cytotoxicity analysis performed following HP-RPLC against the HepaRG and HaCat cell lines. After chromatographic separation of toxins, followed by nanofractionation in serpentine fashion into a 96-well plate, the cytotoxicity of each well was analyzed using the same three readouts as used for crude venom analyses. By plotting these three readouts on the y-axis for each well vs. retention time of fractionation on the x-axis, three so-called bioactivity chromatograms were produced. In parallel, via a post-column flow split, LC-MS data was acquired from which EICs of venom toxins and the TIC were plotted. In addition to nanofractionation for bioassaying, each venom was also one-time nanofractionated for HT venomics. From the HT venomics data, Mascot protein scores for each toxin retrieved in each well were plotted on the y-axis vs. retention time of fractionation in each well on the x-axis and labeled as Uniprot Identifier codes thereby producing PSCs for identification of the cytotoxins. This was done by matching peak shape and retention time of bioactivity chromatogram peaks with PSCs.

3.2.1. Nanofractionation analytics and cytotoxicity profiling of N. haje venom

The liquid chromatography-ultraviolet (LC-UV) trace of separated N. haje venom (220 nm) is shown in Fig. 5A, while Figs. 5B and C display the corresponding TIC and EIC from online MS measurements, respectively. In Fig. 5D, no significant direct cytotoxicity is shown for nanofractionated N. haje venom for the RPTEC/TERT1 cell line. At the same time, slight negative peaks were observed at 22 min for the indirect cytotoxicity chromatographic results (Fig. 5E). Multiple toxins (3SA3, 3SA2, 3SA5, 3SA6, 3SA7, and 3SAA) were closely coeluted in this time frame based on the PSCs found for this venom (Fig. 5F). These exact masses were then correlated with the accurate masses acquired from the LC-MS measurements. 3SA3_NAJHA (3FTx) could be tentatively assigned to the accurate mass of 6812.11 Da (calculated from an m/z value of 1705.5389⁴⁺), while toxins 3SA2, 3SA5, 3SA6, and 3SA7 might have been derived from the same accurate mass found of 6843.2460 Da (1713.0742⁴⁺) as their calculated exact masses were 6845.24, 6843.28, 6844.26, and 6843.28 Da, respectively (Table S2). Although N. haje is considered as a snake with non-cytotoxic venom, this study showed significant crude venom cytotoxicity although at a relatively high venom concentration in the RPTEC/TERT1 cell line. In another study, El Hakim et al. [40] separated N. haje venom by gel filtration on a Sephadex G-50 column (2.6 cm × 50 cm; Pharmacia LKB, Uppsala, Sweden), and the main cytotoxic fraction was then applied to ion-exchange chromatography on a CM-Sephadex C-25 column (0.65 cm × 25 cm) for further purification. The purified fraction (NHV-1c) presented with cytotoxicity against lymphoblastic leukemia (1301) cells, hepatocellular carcinoma (Hep-G2) cells, colon carcinoma (HCT-116) cells, and cervical carcinoma (HeLa) cells.

Fig. 5.

Identification of cytotoxins in Najahaje venom by correlating reversed-phase liquid chromatography-mass spectrometry (RPLC-MS) data, high throughput (HT) venomics data, and cytotoxicity data. (A) Liquid chromatogram-ultraviolet (LC-UV) trace of separated snake venom at 220 nm. (B) Total-ion current (TIC) acquired by online MS measurement. (C) Extracted-ion currents (EICs) acquired by online MS measurement. (D, E) Cytotoxicity bioactivity chromatograms RPTEC/TERT1. Without egg yolk (D) and with egg yolk (E). Negative peaks indicate cytotoxicity, black is live cell count, blue is cell area and orange is viability (resazurin reduction). (F) HT venomics chromatograms plot protein scores chromatographically for identified toxins (via bottom-up proteomics and Mascot searches using Swiss-Prot databases). For plotting chromatographic bioassay readouts from the nanofractionated toxins analyses, the y-axis data points consisted of the average bioassay signals of three replicates from three readouts with standard deviations represented on the y-axis versus retention time of fractionation for each well on the y-axis. All the cytotoxicity chromatograms share the same y-axis units, representing the ratio percentage (%) between the tested wells and the negative control wells, as well as the same x-axis units given in the legend and in the figure, representing retention time in minutes.

3.2.2. Nanofractionation analytics and cytotoxicity profiling of N. mossambica venom

The LC-UV trace of separated N. mossambica venom (220 nm) is shown in Fig. 6A, while Fig. 6B and C display the corresponding TIC and EIC from online MS measurements, respectively. The chromatographic cytotoxicity results for each of the four cell lines (RPTEC/TERT1, HepaRG, iPSC-EC, and HaCaT) investigated are shown in Figs. 6D–G, respectivey, with toxins eluting at 20 min corresponding with cytotoxicity observed in RPTEC/TERT1, iPSC-EC, and HaCaT cell lines, those eluting at 22.5 min caused cytotoxicity in all four cell lines, those at 25 min caused cytotoxicity only in the iPSC-EC cell line, and those eluting at 27.5 min caused cytotoxicity in RPTEC/TERT1 and HepaRG cell lines. Thus, these results show different profiles in cytotoxicity peaks against different cell lines, implying selectivity in cytotoxicity for at least some of the cytotoxins in this venom. An elaborate description of these results is given in the Supplementary material 2.2, and details can be found in Table S2. As outlined above, four chromatographic bioactivity peaks (i.e., 20, 22.5, 25.5, and 27.5 min) were responsible for the main cytotoxic effects observed and the bioactivity peak at 20 min could be correlated to three candidate 3FTxs (i.e., 3SA2, 3SA4, and 3SA5) which induced cytotoxicity in the RPTEC/TERT1, iPSC-EC, and HaCaT cell lines (Fig. 6H). The cytotoxic bioactivity peak around 22.5 was related to non-specific cytotoxicity for all the cell lines investigated and tentatively resulted from PLA₂s (likely PA2B2, PA2B3, and PA2A1). Those PLA₂s (PA2A1, PA2B2, and PA2B3) have been purified and identified as CM-I, CM-II, and CM-III in another study [41]. A slight bioactivity peak at 25.5 min was only observed for the screening of the iPSC-EC cell line, from which the 3FTx, 3SA4, and 3SA5 were tentatively identified from the HT venomics results, next to co-eluting PA2B2. The cytotoxicity peak at 27.5 min was observed with the RPTEC/TERT1 and HepaRG cell lines and not with the iPSC-EC and Hacat cell lines. For this cytotoxicity peak, closely co-eluting 3FTxs 3SA1 and 3SA3 were tentatively assigned. For the indirect cytotoxicity measurements, clearly enhanced cytotoxicity was observed for the RPTEC/TERT1 and HepaRG cell lines, which were already tentatively identified with the direct cytotoxicity assessment as a PLA₂ and a P-type-cytotoxin, respectively. This enhanced effect was consistent with the results from the crude venom screening.

Fig. 6.

Identification of cytotoxins in Najamossambica venom by correlating reversed-phase liquid chromatography-mass spectrometry (RPLC-MS) data, high throughput (HT) venomics data, and cytotoxicity data. (A) Liquid chromatogram-ultraviolet (LC-UV) trace of separated snake venom at 220 nm. (B) Total-ion current (TIC) acquired by online MS measurement. (C) Extracted-ion currents (EICs) acquired by online MS measurement. (D–G) Direct cytotoxicity bioactivity chromatograms obtained by plotting the results of three cytotoxicity assays in RPTEC/TERT1 (D), HepaRG (E), iPSC-EC (F), and HaCaT (G), respectively. Negative peaks indicate cytotoxicity. (H) HT venomics chromatograms plot protein scores chromatographically for identified toxins (via bottom-up proteomics and Mascot searches using Swiss-Prot databases). All the cytotoxicity chromatograms share the same y-axis units, representing the ratio percentage (%) between the tested wells and the negative control wells, as well as the same x-axis units given in the legend and in the figure, representing retention time in minutes.

During the measurements, we observed selective cytotoxicity in different cell lines. This is based on relative toxicity and from there we found that certain fractions exhibited cytotoxicity in specific cell lines while others did not. For instance, 3SA2, 3SA4, and 3SA5 in N. mossambica venom at 20 min were found as causing selective cytotoxicity. We thus define selective cytotoxicity as a relative phenomenon. All nanofractionated venom plates were identically prepared and processed and subsequently tested in parallel on separate cell lines. This workflow ensures comparability across assays, allowing us to ascribe cell-type-specific effects to specific toxin fractions. For instance, certain fractions of N. mossambica venom containing 3FTxs (e.g., 3SA2, 3SA4, and 3SA5) were cytotoxic in RPTEC/TERT1, iPSC-EC, and HaCaT cells but not in HepaRG cells. This pattern points to true cell-type-selective cytotoxicity, but does not give absolute quantitative variations in potency. Such findings provide a valuable starting point for investigating toxin specificity in more detail.

3.2.3. Nanofractionation analytics and cytotoxicity profiling of N. naja venom

The LC-UV trace of separated N. naja venom (220 nm) is shown in Fig. 7A, while Figs. 7B and C display the corresponding TIC and EIC from online MS measurements, respectively. Similar to N. mossambica, N. naja venom showed differences in cytotoxicity against the different cell lines following chromatographic separation, with cytotoxins eluting at 25 min corresponding to cytotoxicity in RPTEC/TERT1, iPSC-EC, and HaCaT cell lines, those at 27.5 min caused cytotoxicity in HepaRG, iPSC-EC, and HaCaT cell lines, while those eluting at the 30.0 min only affected the HaCaT cell line (Figs. 7D–G). All cytotoxicity peaks were observed with the loss of live cell numbers and resazurin reduction, indicating disruption of cell permeability and metabolism. An elaborate description of the results is given in Supplementary material 2.3, and details can be found in Table S2. The peak at 25 min was related to non-selective cytotoxicity and from the HT venomics data (Fig. 7H), the co-eluting toxins 3SA1 (P01447), 3SA2 (P01440), 3SA3 (P24780), 3SA7 (P86382) and/or PA2A2 (P15445, acidic phospholipase A₂) could be correlated to this cytotoxicity peak. Suzuki-Matsubara et al. [42] gave a detailed document about the cytotoxicity of 3FTxs in N. naja venom in other cell lines. This study showed that some purified 3FTxs had significant cytotoxicity against human erythrocytes, and showed selective cytotoxicity of some 3FTxs when exposed to human erythrocyte cells and four kinds of established cell lines, NIH/3T3 (mouse embryonic fibroblast cell line), L1210 (mouse lymphocytic leukemia cell line), A-431 (human epidermoid carcinoma cell line), and A-172 (human glioblastoma cell line). At a retention time of 27.5 min, there was a notable cytotoxicity peak observed in two cell lines (HepaRG, iPSC-EC, and HaCaT), which corresponded to five tentatively assigned 3FTxs (3SA1, 3SA2, 3SA7, 3SA8, and 3SA9). Only slight cytotoxicity was observed in the HaCaT cell line at 30 min, which was correlated to co-eluting 3FTxs (i.e., 3SA1, 3SA2, 3SA2A, 3SA3, and 3SA7). For the indirect cytotoxicity, clearly enhanced cytotoxicity was observed for the RPTEC/TERT1 and HepaRG cell lines.

Fig. 7.

Identification of cytotoxins in N. naja venom by correlating reversed-phase liquid chromatography-mass spectrometry (RPLC-MS) data, high throughput (HT) venomics data, and cytotoxicity data. (A) Liquid chromatogram-ultraviolet (LC-UV) trace of separated snake venom at 220 nm. (B) Total-ion current (TIC) acquired by online MS measurement. (C) Extracted-ion currents (EICs) acquired by online MS measurement. (D–G) Direct cytotoxicity bioactivity chromatograms obtained by plotting the results of three cytotoxicity assays in RPTEC/TERT1 (D), HepaRG (E), iPSC-EC (F), and HaCaT (G), respectively. Negative peaks indicate cytotoxicity. (H) HT venomics chromatograms plot protein scores chromatographically for identified toxins (via bottom-up proteomics and Mascot searches using Swiss-Prot databases). All the cytotoxicity chromatograms share the same y-axis units, representing the ratio percentage (%) between the tested wells and the negative control wells, as well as the same x-axis units given in the legend and in the figure, representing retention time in minutes.

3.2.4. Nanofractionation analytics and cytotoxicity profiling of B. multicinctus venom

Fig. 8 shows all superimposed chromatographic data obtained from analysing B. multicinctus venom within nanofractionation analytics. The LC-UV trace of separated B. multicinctus venom (220 nm) is shown in Fig. 8A, while Fig. 8B and C display the corresponding TIC and EIC from online MS measurements, respectively. An elaborate description of the results is provided in Supplementary material 2.4. B. multicinctus venom is known to be able to cause neuromuscular blockage, which may result in paralysis of the respiratory muscles leading to death [35]. The venom of B. multicinctus venom is not characterized as being cytotoxic and no significant cytotoxicity was observed for either the direct or indirect measurements of crude venom in the RPTEC/TERT1 and iPSC-EC cell lines. For the HaCaT and HepaRG cell lines, however, cytotoxicity was observed in our study. As most significant cytotoxicity of crude B. multicinctus venom was observed in the HaCaT cell line, we also profiled nanofractionated venom toxins for both direct and indirect cytotoxicity using this cell line. After nanofractionation analytics, however, no cytotoxicity peaks were observed (Fig. 8D and E). Fig. 8F is the PSCs for identified toxins through HT venomics. To evaluate if the mobile phases used for the reversed-phase separation denatured the responsible cytotoxin(s), crude venom was dissolved in different compositions of the mobile phases A and B. For this, venom was dissolved in mobile phase A, in B, in 50% A/50% B, and in MQ. Subsequent vacuum centrifugation of the dissolved venom in the different solvents followed by reconstitution in assay medium was then performed following cytotoxicity screening. It was found that only the venom dissolved in MQ was still cytotoxic to the HaCaT cells. The mobile phases A and B both contained 0.1% TFA as mobile phase acidifier, which was most likely the denaturing component. Other studies also have shown cytotoxic activity to be lost in acidic conditions, and Lu et al. [39] previously showed that B. multicinctus venom l-amino acid oxidase (LAAO) induced cytotoxicity was lost when the pH is below 2. We also tested different mobile phases with ammonium formate, which also resulted in a loss of cytotoxicity. This would imply that a low pH in the organic solvent ACN also denatured the cytotoxin(s).

Fig. 8.

Identification of cytotoxins in B. multicinctus venom by correlating reversed-phase liquid chromatography-mass spectrometry (RPLC-MS) data, high throughput (HT) venomics data, and cytotoxicity data. (A) Liquid chromatogram-ultraviolet (LC-UV) trace of separated snake venom at 220 nm. (B) Total-ion current (TIC) acquired by online MS measurement. (C) Eextracted-ion currents (EICs) acquired by online MS measurement. (D) Direct cytotoxicity bioactivity chromatograms RPTEC/TERT1. (E) Indirect cytotoxicity bioactivity chromatograms RPTEC/TERT1. Negative peaks indicate cytotoxicity. (F) HT venomics chromatograms plot protein scores chromatographically for identified toxins (via bottom-up proteomics and Mascot searches using Swiss-Prot databases). All the cytotoxicity chromatograms share the same y-axis units, representing the ratio percentage (%) between the tested wells and the negative control wells, as well as the same x-axis units given in the legend and in the figure, representing retention time in minutes.

Next, the non-denaturing chromatographic SEC approach was used in an attempt to overcome toxin denaturation. For this, we switched to SEC separation implemented within nanofractionation analytics and also here included HT venomics for toxin identification. In our SEC separation, nanofractionation, and HT venomics analytical setup, LC-UV data was collected from SEC (Fig. 9A) and we observed cytotoxicity (Fig. 9B). which in cell permeability disruption, ECM degradation, and loss of cell metabolism. Fig. 9B shows a significant negative peak in the cytotoxicity readouts at the retention time frame of 8.5–10.5 min. The toxins matching the negative peaks observed in the SEC bioactivity chromatograms with HT venomics data were members of the 3FTx, PLA₂, and KUN families (Figs. 9C–F). As SEC is an intrinsically low-resolution separation technique, also at an analytical scale, and it separates based on toxin size, all 3FTxs and PLA₂s will elute in two non-baseline separated peaks. Larger toxins will elute before the 3FTxs and PLA₂s as a broad peak. This suggests that it will be difficult to identify the toxins responsible for the cytotoxicity observed using SEC alone for separation. It can, however, be assessed into what mass range the cytotoxin(s) observed fall, and as such, possible cytotoxin toxin families can be suggested and/or excluded. From the results obtained with the SEC separation, it was found that KUNs eluted from 8 to 9.5 min followed by PLA₂s and 3FTxs from 9.5 to 10.2 min. Interestingly, we found no evidence of cytotoxicity in fractions matching retention times consistent with the molecular mass of the previously described LAAO toxins.

Fig. 9.

Identification of cytotoxins in Bungarusmulticinctus venom by correlating size exclusion chromatography (SEC) data, high throughput (HT) venomics data, and cytotoxicity data. (A) SEC-ultraviolet (UV) trace of separated snake venom at 220 nm. (B) Direct cytotoxicity bioactivity chromatograms HaCaT. (C) HT venomics chromatograms plot protein scores chromatographically for identified toxins (via bottom-up proteomics and Mascot searches using Swiss-Prot databases). (D) Extracted plot protein scores chromatogram from HT venomics chromatograms that belong to 3FTx (three-finger toxins) class. (E) Extracted plot protein scores chromatogram from HT venomics chromatograms that belong to phospholipases A₂ (PLA₂) class. (F) Extracted plot protein scores chromatogram from HT venomics chromatograms that belong to Kunitz-type serine protease (KUNs) class. All the cytotoxicity chromatograms share the same y-axis units, representing the ratio percentage (%) between the tested wells and the negative control wells, as well as the same x-axis units given in the legend and in the figure, representing retention time in minutes.

Recent studies in snake venom analytics have been promising, though there are still some limitations to address. Carregari et al. [43] studied snake venom extracellular vesicles (SVEVs) using proteomics identification and bioactivity assessment. However, they did not correlate and define the SEC-separated fractions with their bioactivity. Similarly, Rodrigues et al. [44] conducted biological and proteomic characterization of Crotalus durissus venom. They used HPLC-sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) to separate venom proteins, followed by identification of the digested gel bands using matrix-assisted laser desorption/ionization-time of flight (MALDI-TOF/TOF). For the bioactivity assessment, they tested the cytotoxicity of the crude venom on kidney cells but could not identify the specific toxins responsible for the cytotoxic effects. The study by Damnn et al. [45] focused primarily on the compositional deconvolution of snake venom. Despite achieving good separation, they did not fully ensure the preservation of bioactivities and did not look at selective cytotoxicity as only one cell line was investigated. Lopes-de-Souza et al. [46] conducted a similar study on Bothrops atrox venom, aiming to identify related toxins and their families. They collected five fractions from SEC, which may have led to insufficient separation for identification of the individual cytotoxins. Using our platform by integrating high-resolution nanofractionation with four parallel cellular cytotoxicity readouts and high-throughput venomics, allowed us to maintain toxin bioactivity post-separation, assign cytotoxic effects to specific venom fractions, and directly correlate those effects with identified toxin families and isoforms. Based on concentrations of venoms tested in the crude venom analyses, the venom concentrations and injection volume for nanofractionation were selected. From these injected venom concentrations and amounts, we see that similar total cytotoxicity was found. In this regard, we found back and focused on the main cytotoxic toxins in the venoms under study, which are expected to be most important for pathologies in snakebite victims. We did not use higher venom concentrations for nanofractionation analysis as this will result in poorer separation resolution and will result in excessively broad cytotoxicity peaks making their correlation with MS and proteomics data more difficult to even impossible. Additionally, our study is based on cytotoxicity profiling of elapid snake venoms focusing on the major, pathology relevant, cytotoxins. We thereby demonstrated toxin-level resolution in combination with selective cytotoxicity assessment. Our approach enabled a robust profiling of venom cytotoxicity across four distinct human cell lines representing different tissues. The approach is broadly applicable to other complex toxin mixtures and can be adapted to explore different bioactivities (e.g., neurotoxicity, procoagulant activity), making it a valuable contribution to the field of venom-based toxicology and pharmacology, and potentially therapeutic discovery.

4. Conclusions

Crude Naja venoms (N. mossambica, N. haje, and N. naja) induced concentration-dependent toxicity in all four tested cell lines, though with some differences in potency across cells and venoms. After nanofractionation analytics towards post-column cytotoxicity profiling, PLA₂ and 3FTxs were the prominent toxins causing cytotoxicity in the venoms of N. mossambica and N. naja, consistent with previous findings from the literature with other cobra venoms [40,41]. However, in some cases, different toxin isoforms appear to exert distinct cytotoxic potencies against the different cell lines investigated. While our study focuses on cytotoxicity profiling and toxin identification, the downstream molecular mechanisms by which PLA₂s and 3FTxs exert their effects merit further investigation [47]. PLA₂s are known to hydrolyze membrane phospholipids, leading to membrane destabilization, leakage, and initiation of pro-inflammatory signaling cascades through arachidonic acid metabolites [48]. In contrast, 3FTxs often interact directly with membrane components or specific receptors, disrupting cell integrity via pore formation or receptor-mediated effects [49]. Certain cytotoxic 3FTxs have also been shown to induce apoptosis or necrosis depending on cell type and concentration [50]. However, detailed mechanistic studies (such as analyses of receptor interactions, signaling pathway activation, or lipid bilayer permeability) would require purified toxins, targeted biochemical assays, and molecular biology approaches. These are beyond the current study's scope but comprise interesting next steps to deepen mechanistic understanding of venom-induced cytotoxicity. The venom of B. multicinctus showed selective cytotoxicity specific to HepaRG and HaCaT cell lines. After nanofractionation by RP-HPLC, the cytotoxicity was lost, implying on-column denaturation. By using SEC as an alternative separation to prevent toxin denaturation, cytotoxic peaks in the cytotoxicity chromatographic bioassays were observed and it could be concluded that 3FTxs, PLA₂s and KUNs contributed to the cytotoxicity observed. In our proof-of-concept study, we included four representative cell lines for measuring toxin-induced cytotoxicity. Further studies could be strengthened by including additional representative cell lines for mimicking other toxin-specific cytotoxicities, such as neuronal cell lines. Through this research, we developed and validated a cytotoxicity-assessing analytical platform that can be used to rapidly study cytotoxic venoms to identify their cytotoxins and to investigate their potentially selective cytotoxic effects against different tissues, with the long-term goal of informing snakebite pathophysiology and priority targets for therapeutic neutralization.

Besides analytically investigating snake venom toxins in snake venoms for selective and non-selective cytotoxicity concerning pathologies in snakebite victims, the platform can also be utilized for the rapid screening of toxins that exhibit selective and specific cytotoxicity toward cancer cells. This research can thereby aid in getting a more comprehensive understanding of the underlying mechanisms in selective cancer cell cytotoxicity and as such aid in finding toxins in snake venoms with a potential as cancer therapy candidates. As such, the presented approach could facilitate the profiling of novel drug candidates, biomarkers, and diagnostic compounds in snake venoms.

CRediT authorship contribution statement

Haifeng Xu: Writing – review & editing, Writing – original draft, Validation, Funding acquisition, Formal analysis, Data curation, Conceptualization. Mátyás A. Bittenbinder: Writing – review & editing, Validation, Methodology, Conceptualization. Julien Slagboom: Writing – review & editing, Validation, Methodology. Nicholas R. Casewell: Writing – review & editing, Validation. Paul Jennings: Writing – review & editing. Jeroen Kool: Writing – review & editing, Validation, Supervision, Project administration, Funding acquisition, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following are the Supplementary data to this article: