Pictolysin-III, a Hemorrhagic Type-III Metalloproteinase Isolated from Bothrops pictus (Serpentes: Viperidae) Venom, Reduces Mitochondrial Respiration and Induces Cytokine Secretion in Epithelial and Stromal Cell Lines

From the venom of the Bothrops pictus snake, an endemic species from Peru, we recently have described toxins that inhibited platelet aggregation and cancer cell migration. In this work, we characterize a novel P-III class snake venom metalloproteinase, called pictolysin-III (Pic-III). It is a 62 kDa proteinase that hydrolyzes dimethyl casein, azocasein, gelatin, fibrinogen, and fibrin. The cations Mg2+ and Ca2+ enhanced its enzymatic activity, whereas Zn2+ inhibited it. In addition, EDTA and marimastat were also effective inhibitors. The amino acid sequence deduced from cDNA shows a multidomain structure that includes a proprotein, metalloproteinase, disintegrin-like, and cysteine-rich domains. Additionally, Pic-III reduces the convulxin- and thrombin-stimulated platelet aggregation and in vivo, it has hemorrhagic activity (DHM = 0.3 µg). In epithelial cell lines (MDA-MB-231 and Caco-2) and RMF-621 fibroblast, it triggers morphological changes that are accompanied by a decrease in mitochondrial respiration, glycolysis, and ATP levels, and an increase in NAD(P)H, mitochondrial ROS, and cytokine secretion. Moreover, Pic-III sensitizes to the cytotoxic BH3 mimetic drug ABT-199 (Venetoclax) in MDA-MB-231 cells. To our knowledge, Pic-III is the first SVMP reported with action on mitochondrial bioenergetics and may offer novel opportunities for promising lead compounds that inhibit platelet aggregation or ECM–cancer-cell interactions.

Peak 1 and peak 2 (10 µg) from the CM-Sepharose step were submitted to 12% SDS-PAGE, and the gel bands were transferred onto a nitrocellulose membrane (0.22 µm pore size) for 18 h at 30 V at 4 • C. Subsequently, the membrane was blocked with a blocking buffer (5% skimmed milk, 0.15 M NaCl, 0.05% tween-20, 0.02 M Tris-HCl, pH 7.5) for 2 h, with agitation at room temperature. Then, the membrane was washed three times with washing buffer and incubated with anti-Atr-III serum (provided by FUNED, Brazil; Atr-III: a P-III SVMP from B. atrox) (dilution 1:1000 in blocking buffer) for 2 h at room temperature. After incubation, three more washes were conducted and the membrane was incubated with anti-rabbit IgG-horseradish peroxidase (Sigma-Aldrich, diluted 1:5000 in blocking buffer) for 1 h at room temperature. Finally, after another three washes, the membrane was revealed with ECL reagent (Thermo Scientific, Waltham, MA, USA). The image was acquired using Chemidoc (Bio-Rad, Hercules, CA, USA). The uncropped Western blot is shown in the Supplementary Information.

cDNA-Encoding Pictolysin-III and Gene Expression Analysis
Total RNA was obtained according to Vivas-Ruiz et al. [29]. The gene of Pic-III was amplified with the kit Master Mix Platinum ® Taq DNA Polymerase (Invitrogen, Carlsbad, Pharmaceutics 2023, 15, 1533 4 of 31 CA, USA) according to the manufacturer's instructions using the external primers MpIIIF and MpIIIR and the internal primers MpIIIFi and MpIIIRi provided by Oliveira et al. [30]. The Pic-III gene expression level was analyzed according to Vivas-Ruiz et al. [28] using Verso 1-Step qPCR ROX Kit (ThermoFisher Scientific, Waltham, MA, USA). The master mix was prepared following the manufacturer's instructions. The assay was conducted in a 7500 Applied Biosystem Thermocycler (ThermoFisher Scientific, USA). Pictobin, Bpic-LAAO, and phospholipase A2 (PLA2) genes were used for comparison, and GAPDH was used as a housekeeping gene.

Cloning and Nucleotide Sequencing
The cDNA transcript encoding Pic-III was inserted into the pCR2.1-TOPO vector following the TOPO-TA cloning strategy and transformed into One Shot™ Escherichia coli TOP10 competent cells, according to the manufacturer's recommended protocol. Transformed E. coli cells were grown on Luria-Bertani (LB) agar plates overnight at 37 • C and spiked with ampicillin (50 µg/mL) for selection. Positive colonies were selected from agar plates by colony PCR amplification using M13 primers. Bacterial glycerol stocks were prepared for long-term storage at −80 • C. The identity of the Pic-III-encoding transcript was confirmed by sequencing on an ABI 3730 XL automated sequencer (Macrogen, Inc., Seoul, Republic of Korea).

In Silico Protein Analysis
The amino acid sequence of Pic-III was obtained from cDNA using the ORF finder program [31] and analyzed with NCBI BLAST [32]. The presence of a signal peptide and protein domains was performed on HMMER [33] and PROSITE [34] servers. The tertiary structure (excluding peptide signal and pro-peptide sequence) was built by homology modeling using the MODELLER program v. 10.0 [35] employing as a template catrocollastatin (P-III SVMP from Crotalus atrox, PDB: 2DW0), due to its high similarity (88.31%) and a high degree of resolution (2.1 Å). The three-dimensional structure was refined with GalaxiRefine server [36] and visualized with Pymol v.2.5. The model quality assessment was performed with PROCHECK [37] and for Ramachandran plot analysis, we used the SAVES tool (https://saves.mbi.ucla.edu/ accessed on 26 August 2022).

Sequence-Based Analysis and Phylogenetic Analysis
The theoretical isoelectric point and molecular mass were estimated with the ExPASy ProtParam tool [38]. Multiple sequence alignment of Pic-III with atroxlysin-III (AQS99160), catrocollastatin (AAC59672.1), P-III SVMP from Bothrops jararaca (KAG5858165.1), and ADAM 12 from Homo sapiens (AAC08702.2) was performed by ClustalX 2.0 [39]. The Pic-III architecture was recognized based on alignment with catrocollastatin [40]. On the other hand, nucleotide sequences of P-III SVMP obtained from NCBI (https://www.ncbi. nlm.nih.gov/protein/ accessed on 27 August 2022) were aligned by ClustalX 2.0 and a phylogenetic tree was built using the maximum likelihood statistical method in the MEGA X program [41] with the Bootstrap method (2000 replications) and the General Time Reversible model with invariant sites (GTR + I).
2.6. Biochemical Characterization 2.6.1. Enzymatic Activity Azocasein was used as a substrate for the enzymatic characterization of Pic-III according to Gomes et al. [42], with some modifications: azocasein (1.5 mg/mL) in 20 mM Tris-HCl, 5 mM CaCl 2 was incubated with 5 µL of Pic-III (1 mg/mL) at 37 • C. After, 45 µL of trichloroacetic acid 20% (w/v) was added to stop the reaction. Microtubes were incubated at room temperature for 20 min, centrifuged at 5000× g for 15 min, and, finally, the supernatant was transferred to a 96-well plate. The absorbance of the supernatant was determined at 405 nm, using the ER-500 microplate reader (Sinnowa, Nanjing, China). The increase of 0.01 absorbance units at 405 nm was defined as one unit of enzymatic activity. Pic-III was incubated alone or with Pictobin (a thrombin-like enzyme from B. pictus) and/or α2-M (1:1:1 molar ratio) in 0.2 M Hepes buffer, pH 7.4 containing 0.15 M NaCl at 37 • C for 30 min. After that, 5 µL of the mixture was used to test the proteolytic activity using azocasein as a substrate following the method above mentioned.

Effect of Pictolysin-III Deglycosylation under no Reducing Conditions on Enzymatic Activity
To maintain the functional characteristics of the enzyme, Pic-III (50 µg) was incubated with 5 µL of recombinant PNGase F (500 U) in a mixture of 20 µL of 50 mM sodium phosphate buffer (pH 7.5) and 50 µL of H 2 O MilliQ. The incubation was carried out at 37 • C for 48 h. Subsequently, 5 µL of PNGase F (500 U) was added and incubated for an additional 48 h. Finally, the mixture was applied to an MLPC column (conditions above mentioned) to obtain homogenous Pic-III. The deglycosylated enzyme was evaluated for its azocaseinolytic activity in comparison with the native Pic-III (non-PNGase treated).

Proteolytic Activity upon Fibrinogen and Fibrin
Proteolytic activities were determined according to Oliveira et al. [30]. Human fibrinogen (H-Fg) was dissolved in 25 mM Tris-HCl, pH 7.4, 154 mM NaCl; samples with 100 µL of H-Fg (2.5 mg/mL) were incubated with 1 µg of Pic-III at 37 • C for 30 min; the reaction was stopped by adding 40 µL of denaturant buffer (SDS, β-mercaptoethanol, glycerol, and bromophenol blue). For fibrinolytic activity, fibrin was obtained using 2 NIH of human thrombin (Sigma-Aldrich) and 100 µL of H-Fg (2.5 mg/mL), incubated for 2 h at room temperature, then 1 µg of Pic-III was added to fibrin and incubated at 37 • C for 30 min. The reaction was stopped using a denaturant buffer. In both activities, the effects were observed by SDS-PAGE (14%).

Platelet Aggregation Assay
Human blood from two healthy volunteers (age/sex: 22/M and 23/F) was collected in acid-citrate-dextrose (ACD: 78 mM citric acid; 117 mM sodium citrate; 282 mM dextrose) [6:1, (v/v)], centrifuged at 200× g for 15 min to obtain platelet-rich plasma (PRP). Washed platelets were prepared as previously described by Oliveira et al. [30]. Platelet density was adjusted to 2.5 × 10 5 platelets/µL. Washed platelets (225 µL) were pre-incubated with Pic-III (from 2 to 32 µg) in Tyrode's solution pH 7.4, for 3 min before the addition of agonists: 6 µg/mL of convulxin (CVX) isolated from Crotalus durissus terrificus venom, 10 µg/mL of collagen-I, 1 U/mL thrombin or 5.5 µg/mL von Willebrand factor plus 0.5 mg/mL ristocetin. Platelet aggregation was conducted in a platelet aggregometer (AggRam Helena Laboratories, Beaumont, TX, USA) with stirring (600 rpm) at 37 • C. Light transmittance was recorded and the inhibition of platelet aggregation was measured at the maximum aggregation response. The hemorrhagic activity of fractions (5 µg), Pic-III, and the crude venom of B. pictus was analyzed by the skin assay procedure [43]. The assay was modified by using Swiss Webster mice. Each animal was inoculated intradermally. After 3 h, the dorsal skin was removed, and the area of the hemorrhagic halo was measured. PBS was used as a negative control. To determine the minimum hemorrhagic dose (MHD), three animals were used for each dose of Pic-III (0.5, 1.0, 2.5, 5.0, and 10.0 µg); crude venom (0.5, 0.8, 1.0, 1.5, and 4.0 µg), and PBS. The MHD corresponds to the dose of protein or venom that induced a hemorrhagic spot of 10 mm diameter and was calculated by extrapolation. This in vivo experiment was performed following the guidelines of the Brazilian College for Animal Experimentation and approved by the local Ethics Committee (Protocol number CEUA/Funed: 015/2019).

MTT Assay
The cell viability was determined using the MTT assay. Cells were incubated in 96-well plates at 7000 cells per well and incubated for 24 h. The cells were treated with Pic-III (1, 2.5, 5, 10, 20, and 50 µg/mL in PBS) for 48 h. After treatment, the culture medium was removed, and the cells were incubated with MTT for 1 h at 37 • C. Finally, 100 µL of DMSO was added and measured by spectrophotometry at 570 nm, as described by Córdova-Delgado et al. [44].

Determination of ATP and Mitochondrial ROS (mtROS) Levels
ATP levels were determined using a luciferin-luciferase assay system of the CellTiter-Glo Luminescent Cell Viability Assay kit (Promega, Madison, WI, USA) according to Urra et al. [45]. Cells (7000 cells/well) were incubated in 96-well plates for 24 h. The cells were treated with Pic-III (1, 2.5, 5, 10, 20, and 50 µg/mL) for 48 h. After treatment, the culture medium was removed, and then intracellular ATP levels were determined in lysed cells using a Varioskan Flash microplate reader (Thermo Scientific, USA) as described Córdova-Delgado et al. [44]. The mtROS levels were measured using MitoSOX ® Red Mitochondrial Superoxide Indicator (Invitrogen, Carlsbad, CA, USA). RMF-621 cells (10,000 cells/well) were incubated in 12-well plates for 24 h. The cells were treated with Vehicle (PBS) and Pic-III (50 µg/mL) for 48 h. Next, cells were incubated with MitoSOX Red ® (5 µM) for 30 min. Then, they were recollected, washed, and the fluorescence was detected by flow cytometry [45].

Extracellular Acidification Rate in Real Time
To analyze extracellular acidification rate (ECAR) in cell lines, MDA-MB-231, Caco-2, and RMF-621 (10,000 cells/well) were seeded on XFe96 V3-PS multi-well plates and kept for 48 h at 37 • C in 5% CO 2 with DMEM culture medium supplemented with FBS. After 48 h, cells were stimulated by 10, 20, and 50 µg/mL Pic-III for 8 h, and then the culture medium was replaced with assay media (unbuffered DMEM without red phenol, with 4 mM glutamine, and 10 mM glucose, pH = 7.4) 1 h before the assay. Glycolysis was evaluated by the sequential injection of 10 mM glucose, 1 µM oligomycin, and 100 mM 2-desoxi-D-glucose (2-DG), and ECAR was analyzed in real-time in the Agilent Seahorse XFe96 Analyzer (Seahorse Agilent, Santa Clara, CA, USA) with specific excitation and emission wavelengths of protons (470/530 nm) [45,47]. Each experiment was run at least in triplicate.
Automated image analysis was performed in Fiji using custom-written macros (NIH, Bethesda, MD, USA, http://rsb.info.nih.gov/ij accessed on 26 August 2022). In brief, images were equalized to the full 16-bit range, and local contrast enhancement and median filter were applied. Binary masks of approximate cell outlines were generated by the Otsu algorithm, binary operations to smoothen outlines and fill holes, and water-shedding to separate touching cells. Binary masks of cytoplasm were determined by band-pass filtering of raw cell mask images, followed by splitting of touching cells using Voronoi diagrams from binary DAPI segmentation images. For determining the morphological characteristics of a single cell, a batch process was performed on a set of experimental images obtaining area, aspect ratio, length, and circularity using the MorphoLibJ-plugin (Plugins-MorphoLibJ-Label Images) [48,49] Cytokine secretion (TNFα, IL-8, IL-10, and IL-1β) by Caco-2 and RMF-621 cells (7.000 cells/well) treated with PBS (vehicle) or non-cytotoxic concentrations of Pic-III (Caco-2: 50 µg/mL and RMF-621: 20 µg/mL) was detected at 48 h using a Human Inflammatory Cytokines Kit (BD Biosciences, San Jose, CA, USA) following the manufacturer's instructions. Briefly, a supernatant of cells exposed to treatments was incubated with a mixture of anti-cytokine capture antibodies-conjugated beads and PE-conjugated detector antibodies for 3 h at room temperature in the dark, subsequently washed with 1× wash buffer, and centrifuged at 200× g for 5 min at room temperature. Data was acquired using a FACS CantoTM II flow cytometer (BD) and analyzed using BD Cytometric Bead Array software (Version 1.4) (BD) as described [50].

Statistical Analysis
All statistical analyses were performed using GraphPad Prism 5.0 (GraphPad Software, San Diego, CA, USA). Statistical analysis was performed using one-way ANOVA with Bonferroni's post-test for pairwise comparisons. The data were considered statistically significant when p < 0.05.

Pictolysin-III Purification and Autoproteolysis Identification
The enzyme was purified to homogeneity using successive chromatography on Sephacryl S-200, DEAE Sepharose CL6B, and CM Sepharose CL-6B columns, as shown in Figure 1. First, the venom (682 mg) was separated into five peaks (P1 to P5) by the S-200 column at pH 8.0 ( Figure 1A). Proteolytic fractions were concentrated in P1 and between P2 and P3. These fractions were dialyzed against distilled water and lyophilized (155 mg). For the second step, the active pool was applied to a DEAE Sepharose CL-6B column with a linear salt gradient from 0-0.3 M NaCl, and five main peaks (P1 to P5) were obtained ( Figure 2B). The enzymatic activity was detected in three peaks that correspond to P-III, P-II, and P-I, respectively. Thus, the enzyme corresponding fractions (before peak 2) were pooled (31 mg). In the third step, the pool from the preceding step was applied to a CM Sepharose CL-6B column ( Figure 2C); by this procedure, 20 mg of the active enzyme was obtained, which was termed Pictolysin-III (Pic-III). A peculiar result was observed in this third purification step since two peaks were detected, but only peak 1 presented activity ( Figure 1C). The SDS-PAGE analysis of peak 1 ( Figure 1D) evidenced two bands, 62 and 33 kDa, both under reducing and non-reducing conditions. Western blot analysis ( Figure 1E) showed that both were recognized by Anti-Atroxlysin-III (SVMP-III from B. atrox venom). The 33 kDa band was only present in peak 2 ( Figure 1E). We deduced that it would be a process of autoproteolysis. An additional step using an MPLC system was conducted to separate the complete active enzyme from the proteolyzed fragments. A homogeneous 62 kDa band was obtained, which was reduced to a 53 kDa band after PNGase F treatment. ( Figure 1F) In addition, Pic-III represents almost 22% of all transcripts in fresh venom ( Figure 1G). This amount is not significantly different to Bpic-LAAO, but it is lower than PLA2 and higher than Pictobin, which are important toxic components of the B. pictus venom [27,28].

Functional Characterization of Pictolysin-III
The pH and temperature dependences of the enzymatic activity of Pic-III were measured ( Figure 2). Values of similar proteolytic activity higher than 10 azocasein units were observed at pH 6.0 to 10.0, being the maximum value close to pH 8.0, with a drastic fall of activity at pH lower than 6.0 ( Figure 2A). Pic-III exhibited high activity below 25 • C, while maintaining moderate activity between 35 • C and 55 • C. At 75 • C, the enzyme lost 80% of its proteolytic activity ( Figure 2B). Likewise, the hydrolysis of azocasein by Pic-III was moderately inhibited by the reducing agent 2β-mercaptoethanol (47%) and marimastat (45%) but drastically inhibited by DTT (100%) and the chelating agent EDTA (95.8%). In contrast, the enzyme was not affected by PMSF ( Figure 2C). The pre-incubation of Pic-III with Ca 2+ and Mg 2+ cations increased the proteolytic activity but Zn 2+ decreased it significantly ( Figure 2C). Additionally, the neutralization of Pic-III was evaluated using half, one, and two neutralizing doses of polyvalent antibotropic serum (PAS) and the residual activity was 72, 63, and 53%, respectively ( Figure 2D). We evaluated if the proteolytic activity of Pic-III on azocasein depended on glycosylation. As Figure 2E,F shows, the enzymatic activity decreased to approximately 40% after its deglycosylation with PNGase F.

Interaction with a TLE (Pictobin) and Plasma Inhibitor α2-Macroglobulin (α2-M)
The α2-M is a human plasma inhibitor that interacts and blocks virtually any protease [51], including some snake toxins, as we previously reported [28]. Therefore, we evaluated whether α2-M inhibited the proteolytic activity of Pic-III and Pictobin, a thrombin-like enzyme isolated from B. pictus venom [28]. Pic-III was significantly inhibited by α2-M (50%), but this inhibition was reduced when Pic-III was co-incubated with Pictobin and α2-M (only 20% of enzymatic inhibition). No significant changes in the azocasein hydrolysis were observed when only Pic-III and Pictobin were pre-incubated. Little or no activity upon the substrate was observed with only Pictobin or α2-M, respectively. These results suggest that there is a greater affinity between Pictobin and α2-M than for Pic-III, demonstrating a possible functional synergy between both toxins.

Pictolysin-III Acts as a GPVI Antagonist That Produces Hemorrhage In Vivo
We analyzed the fibrino(geno) activity of Pic-III on H-Fg (plasminogen-free) and fibrin ( Figure 3). Under our experimental conditions, Pic-III digested the α chains of H-Fg and fibrin in a time-dependent manner at 37 • C ( Figure 3A,B). No significant alterations in the electrophoretic mobility of the Bβ and γ chains of H-Fg were detected even at 60 min incubation period. Likewise, the effect of EDTA on H-Fg ( Figure 3C) and bovine fibrinogen ( Figure 3D) activity of Pic-III was evaluated. In both cases, the pattern of hydrolysis of the Aα chain was inhibited by the chelating agent. Since SVMP toxins act by provoking hemorrhage and affect hemostasis and thrombosis [8], the effect of Pic-III on platelet aggregation stimulated by different agonists was evaluated. For this, platelets were pretreated for 3 min with increasing concentrations of Pic-III. We observed that the enzyme inhibited the thrombin (Thr)-and convulxin (CVX)-induced aggregation ( Figure 3E,F) but did not interfere with collagen (Col)-and von Willebrand Factor + ristocetin (vWF + Ris)-induced aggregation ( Figure 3G,H). In addition, B. pictus venom and Pic-III showed hemorrhagic activity in vivo with a MHD = 0.6 µg and 0.3 µg, respectively ( Figure 3I,J).

Pictolysin-III Acts as a GPVI Antagonist That Produces Hemorrhage In Vivo
We analyzed the fibrino(geno) activity of Pic-III on H-Fg (plasminogen-free) and fibrin ( Figure 3). Under our experimental conditions, Pic-III digested the α chains of H-Fg and fibrin in a time-dependent manner at 37 °C ( Figure 3A,B). No significant alterations in the electrophoretic mobility of the Bβ and γ chains of H-Fg were detected even at 60 min incubation period. Likewise, the effect of EDTA on H-Fg ( Figure 3C) and bovine fibrinogen ( Figure 3D) activity of Pic-III was evaluated. In both cases, the pattern of hydrolysis of the Aα chain was inhibited by the chelating agent. Since SVMP toxins act by provoking hemorrhage and affect hemostasis and thrombosis [8], the effect of Pic-III on platelet aggregation stimulated by different agonists was evaluated. For this, platelets were pre-treated for 3 min with increasing concentrations of Pic-III. We observed that the enzyme inhibited the thrombin (Thr)-and convulxin (CVX)-induced aggregation ( Figure  3E  A total of half, one, and two neutralizing doses were tested. The enzyme (25 µg) was incubated with PAS for 30 min at 37 • C before the activity. The effect was described as % residual activity. (E) Pic-III (1 mg/mL), in the presence or absence of PNGase F (100 U), was incubated with azocasein to assess its proteolytic activity. (F) Effect of PNGase F on Pic-III proteolytic activity at 1 h of incubation. (G) Interaction between Pic-III and Pictobin (Picb) (1:1:1 molar ratio) with the endogenous α2-M inhibitor. Pic-III is not inhibited when it interacts with the serine protease, demonstrating a possible functional synergy. The data shown are the mean ± SD of three independent experiments. * p < 0.05, ** p < 0.01, *** p < 0.001 vs. control (Pic-III); n.s., not significant. In (A,B), vs. proteolytic activity at pH = 4 and 5 • C, respectively.

Structural Characterization of Pictolysin-III
The primary sequence of Pic-III has 610 amino acid residues and presents a signal peptide (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20) and pro-peptide domain . The predicted functional sequence consists of 421 amino acid residues and presents a metalloproteinase domain (M domain), a disintegrinlike domain (D domain), and a cysteine-rich domain (C domain) (Figure 4). The predicted functional sequence shows a theoretical pI = 5.13 and Mr = 46771.7 Da and also shows 35 cysteine residues. The Pic-III architecture is similar to other SVMP P-III members of the Reprolysin family ( Figure 5). The Ca 2+ -binding sites are conserved and are found in the M and D domains. Likewise, the canonical zinc-binding motif (HEXXHXXGXXH) and Metturn are found in the D domain of Pic-III. The alignment also shows that Pic-III exhibits the cysteine residues conserved in other SVMP P-III toxins. The predicted three-dimensional structure of Pic-III harbors the classical MDC domain architecture (metalloproteinase domain, disintregin-like, cysteine-rich) ( Figure 6A and Supplementary Figure S1), structure prototype of mammalian ADAM proteins [40]. In the N-terminal region, Pic-III presents a globular M domain, followed by the D domain, which is divided into two structurally distinct subsegments, the "shoulder" segment (Ds) and the "arm" segment (Da). The C domain, which is in the C-terminal region, has the "hypervariable region" (HVR). Moreover, the phylogenetic tree revealed that Pic-III is evolutionary close to other SVMP P-III from the Viperidae species, which form a distinct clade to the two composed of Elapidae and some rear-fanged species [52,53] (Figure 6B). Finally, an N-glycosylation site was found in Asn183.

Pictolysin-III Promotes Actin Network Disruption, Reducing the Viability of MDA-MB-231 and RMF-621 Cells
To determine the effect of Pic-III on the morphology of cell lines, a time-lapse experiment was conducted. As Figure 7A shows, before treatment with Pic-III, Caco-2 cells exhibited projections of the plasma membrane with an epithelial morphology. At three hours of treatment, Pic-III blocked the formation of cellular projections ( Figure 7A), producing disruptions in the actin network ( Figure 7B-D) and reducing the cell spreading ( Figure  7E). These morphological changes were characterized by an aggregation of actin (an increased density of F-actin, Figure 7F), increased circularity (values were close to 1 in the presence of the Pic-III treatments, Figure 7G), and loss of cell polarization (reduction of elongation, Figure 7H) in a concentration-dependent manner.

Pictolysin-III Promotes Actin Network Disruption, Reducing the Viability of MDA-MB-231 and RMF-621 Cells
To determine the effect of Pic-III on the morphology of cell lines, a time-lapse experiment was conducted. As Figure 7A shows, before treatment with Pic-III, Caco-2 cells exhibited projections of the plasma membrane with an epithelial morphology. At three hours of treatment, Pic-III blocked the formation of cellular projections ( Figure 7A), producing disruptions in the actin network ( Figure 7B-D) and reducing the cell spreading ( Figure 7E). These morphological changes were characterized by an aggregation of actin (an increased density of F-actin, Figure 7F), increased circularity (values were close to 1 in the presence of the Pic-III treatments, Figure 7G), and loss of cell polarization (reduction of elongation, Figure 7H) in a concentration-dependent manner. Since the ECM-cell interaction is relevant to adapt the metabolic requirement for promoting cell viability [22], we evaluated the effect of Pic-III on the viability in epithelial (MDA-MB-231 and Caco-2) and stromal (RMF-621) cells. Our results indicate that Pic-III reduces the viability in MDA-MB-231 and RMF-621 cells at 50 µg/mL and Caco-2 cells were unaffected (Figure 8). Since the ECM-cell interaction is relevant to adapt the metabolic requirement for promoting cell viability [22], we evaluated the effect of Pic-III on the viability in epithelial (MDA-MB-231 and Caco-2) and stromal (RMF-621) cells. Our results indicate that Pic-III reduces the viability in MDA-MB-231 and RMF-621 cells at 50 µg/mL and Caco-2 cells were unaffected (Figure 8).

Pictolysin-III Inhibits the Basal and Maximal Mitochondrial Respiration Cell Lines
Based on previous results, we evaluated the effect of Pic-III on mitochondrial respiration and glycolysis in MDA-MB-231, Caco-2, and RMF-621 cells, using real-time measurements of oxygen consumption rate (OCR) and extracellular acidification rate (ECAR), respectively. As Figure 9 shows, changes in the profile of respiration of MDA-MB-231, Caco-2, and RMF-621 were observed when stimulated with Pic-III for 8 h. MDA-MB-231 and RMF-621 cells decreased the basal OCR and maximal OCR (which represent the maximal electron transport flux) in a concentration-dependent manner ( Figure 9B,F). In contrast, Caco-2 cells decreased the basal OCR in all concentrations of Pic-III; however, maximal OCR was only reduced at 50 µg/mL ( Figure 9C,D). Therefore, these results indicate that Pic-III decreases the electron transport chain in MDA-MB-231, Caco-2, and RMF-621 cells, an event previously unreported for some SVMP-III. Notably, 50 µg/mL Pic-III completely reduced the ATP-linked respiration (oligomycin-sensitive OCR) in RMF-621, and it stimulated the proton-leak-driven respiration in Caco-2 cells. In MDA-MB-231 cells, this concentration of Pic-III lacked an effect on oligomycin-sensitive respiration, suggesting differential effects on OXPHOS coupling in the three cell lines ( Figure 9A,C,E). Notably, non-mitochondrial OCR was reduced in the three cell lines evaluated, suggesting that other cellular oxidative reactions not linked to energy metabolism may be inhibited by Pic-III (Supplementary Figure S2).

Pictolysin-III Reduces Glycolysis in MDA-MB-231 and RMF-621, but Increases It in Caco-2 Cells
Inhibition of mitochondrial respiration and, consequently, the ATP synthesis, promotes a metabolic shift toward glycolysis. We speculate that the inhibitory effect of mitochondrial respiration induced by Pic-III may promote a compensatory increase in glycolysis to maintain the intracellular ATP levels as described [54]. After 8 h of treatment, effects on glycolysis were different between the three cell lines (Figure 10). RMF-621 and MDA-MB-231 cells reduced the glycolysis and glycolytic capacity at 50 µg/mL Pic-III ( Figure  10A,B,E,F). No changes in the glycolytic reserve were observed (Supplementary Figure  S3). On the other hand, Caco-2 cells exhibited an increase in glycolysis, suggesting a possible adaptive metabolic shift ( Figure 10C,D). Collectively, these results suggest that Pic-III differentially affects glycolysis in the three cell lines.

Pictolysin-III Inhibits the Basal and Maximal Mitochondrial Respiration Cell Lines
Based on previous results, we evaluated the effect of Pic-III on mitochondrial respiration and glycolysis in MDA-MB-231, Caco-2, and RMF-621 cells, using real-time measurements of oxygen consumption rate (OCR) and extracellular acidification rate (ECAR), respectively. As Figure 9 shows, changes in the profile of respiration of MDA-MB-231, Caco-2, and RMF-621 were observed when stimulated with Pic-III for 8 h. MDA-MB-231 and RMF-621 cells decreased the basal OCR and maximal OCR (which represent the maximal electron transport flux) in a concentration-dependent manner ( Figure 9B,F). In contrast, Caco-2 cells decreased the basal OCR in all concentrations of Pic-III; however, maximal OCR was only reduced at 50 µg/mL ( Figure 9C,D). Therefore, these results indicate that Pic-III decreases the electron transport chain in MDA-MB-231, Caco-2, and RMF-621 cells, an event previously unreported for some SVMP-III. Notably, 50 µg/mL Pic-III completely reduced the ATP-linked respiration (oligomycin-sensitive OCR) in RMF-621, and it stimulated the proton-leak-driven respiration in Caco-2 cells. In MDA-MB-231 cells, this concentration of Pic-III lacked an effect on oligomycin-sensitive respiration, suggesting differential effects on OXPHOS coupling in the three cell lines ( Figure 9A,C,E). Notably, non-mitochondrial OCR was reduced in the three cell lines evaluated, suggesting that other cellular oxidative reactions not linked to energy metabolism may be inhibited by Pic-III (Supplementary Figure S2).

Pictolysin-III Reduces Glycolysis in MDA-MB-231 and RMF-621, but Increases It in Caco-2 Cells
Inhibition of mitochondrial respiration and, consequently, the ATP synthesis, promotes a metabolic shift toward glycolysis. We speculate that the inhibitory effect of mitochondrial respiration induced by Pic-III may promote a compensatory increase in glycolysis to maintain the intracellular ATP levels as described [54]. After 8 h of treatment, effects on glycolysis were different between the three cell lines (Figure 10). RMF-621 and MDA-MB-231 cells reduced the glycolysis and glycolytic capacity at 50 µg/mL Pic-III ( Figure 10A,B,E,F). No changes in the glycolytic reserve were observed (Supplementary Figure S3). On the other hand, Caco-2 cells exhibited an increase in glycolysis, suggesting a possible adaptive metabolic shift ( Figure 10C,D). Collectively, these results suggest that Pic-III differentially affects glycolysis in the three cell lines.

Pictolysin-III Reduces Glycolysis in MDA-MB-231 and RMF-621, but Increases It in Caco-2 Cells
Inhibition of mitochondrial respiration and, consequently, the ATP synthesis, promotes a metabolic shift toward glycolysis. We speculate that the inhibitory effect of mitochondrial respiration induced by Pic-III may promote a compensatory increase in glycolysis to maintain the intracellular ATP levels as described [54]. After 8 h of treatment, effects on glycolysis were different between the three cell lines (Figure 10). RMF-621 and MDA-   . On the other hand, Caco-2 cells exhibited an increase in glycolysis, suggesting a possible adaptive metabolic shift ( Figure 10C,D). Collectively, these results suggest that Pic-III differentially affects glycolysis in the three cell lines.

Pictolysin-III Alters the Intracellular ATP, NAD(P)H, and Mitochondrial ROS Levels
To evaluate if the mitochondrial respiration and glycolysis inhibition by Pic-III promotes a metabolic dysfunction, we evaluated the effect of this toxin on the intracellular ATP, NAD(P)H, and mitochondrial ROS (mtROS) levels. Pic-III reduced the intracellular ATP levels in the three cell lines ( Figure 11A-C), with MDA-MB-231 cells more affected in

Pictolysin-III Alters the Intracellular ATP, NAD(P)H, and Mitochondrial ROS Levels
To evaluate if the mitochondrial respiration and glycolysis inhibition by Pic-III promotes a metabolic dysfunction, we evaluated the effect of this toxin on the intracellular ATP, NAD(P)H, and mitochondrial ROS (mtROS) levels. Pic-III reduced the intracellular ATP levels in the three cell lines ( Figure 11A-C), with MDA-MB-231 cells more affected in a concentration-dependent manner ( Figure 11A). In Caco-2 cells, a significant increase in ATP levels was observed at 10 µg/mL ( Figure 11B), which correlated with the metabolic remodeling toward enhanced glycolysis ( Figure 10D). a concentration-dependent manner ( Figure 11A). In Caco-2 cells, a significant increase in ATP levels was observed at 10 µg/mL ( Figure 11B), which correlated with the metabolic remodeling toward enhanced glycolysis ( Figure 10D). Since inhibition of maximal OCR produced by Pic-III might imply a reduction of respiratory complexes activity [55], producing a disruption in the NAD/NADH ratio, we evaluated the effect of Pic-III on NAD(P)H in the epithelial and stromal cell lines. As Figure 11D-F shows, FCCP (a protonophore agent that uncouples OXPHOS) and antimycin Since inhibition of maximal OCR produced by Pic-III might imply a reduction of respiratory complexes activity [55], producing a disruption in the NAD/NADH ratio, we evaluated the effect of Pic-III on NAD(P)H in the epithelial and stromal cell lines. As Figure 11D-F shows, FCCP (a protonophore agent that uncouples OXPHOS) and antimycin A (a respiratory complex III inhibitor) produce a decrease and increase of NAD(P)H levels, respectively, and Pic-III (50 µg/mL) significantly increases the NAD(P)H close to two folds of the control in the three cell lines.
The respiratory complex I contributes about 40% of the proton motive force required for mitochondrial ATP synthesis, through the oxidation of NADH and producing superoxide [56,57]. Interestingly, mtROS levels were reduced in RMF-621 cells, but in Caco-2 cells, mtROS was increased after 48 h of treatment ( Figure 11G,H). Finally, we evaluated whether mitochondrial membrane potential (∆ψm) is required for the effect of Pic-III. For this, RMF-621 cells were pre-treated with FCCP (1 µM) for 1 h and then, were exposed to Pic-III for 48 h ( Figure 11I). FCCP reduced the viability to 0.50 ± 0.06 folds of the control (p < 0.001 vs. control), Pic-III to 0.09 ± 0.01 folds of the control (p < 0.001 vs. control), and a combination of FCCP + Pic-III to 0.15 ± 0.03 folds of the control (p < 0.001 vs. control). No significant differences were observed between Pic-III treatment and FCCP + Pic-III, suggesting that ∆ψm is not a determinant of the effect of Pic-III. Collectively, these results suggest that Pic-III reduces cellular metabolism, affecting energy production, and being more active in MDA-MB-231 breast cancer cells.

Pictolysin-III Increases the Secretion of Cytokines in Caco-2 and RMF-621 Cells
Previously, we described that fibroblast RMF-621 under metabolic stress conditions changes the expression of pro-inflammatory genes [58]. Based on this, the effect of Pic-III on the expression of IL1β and TNFα genes was evaluated. At 8 h of treatment, Pic-III increased the mRNA levels for IL1β without effects on TNFα gene expression ( Figure 12A,B). As it has been recognized that mitochondrial dysfunction can promote the secretion of cytokines [59][60][61], we determined the effect of Pic-III on cytokine secretion in RMF-621 and Caco-2 cells at non-cytotoxic concentrations for 48 h. Pic-III generated changes in the cytokine secretion profile of RMF-621 cells, increasing the IL-8, IL-1β, and TNF levels with no changes in IL-10 secretion ( Figure 12C-F). Pic-III induced the production of IL-8 and IL-1β, decreasing IL-10 in Caco-2 cells ( Figure 12G-I). TNFα was not induced by Pic-III in this cell line ( Figure 12J).

Pictolysin-III Produces Sensitization to BH3 Mimetic ABT-199 (Venetoclax) in MDA-MB-231 Cells
The inhibition of mitochondrial bioenergetics promotes an early induction of intrinsic pathway (mitochondrial pathway) apoptosis by ABT-199 [62], a BCL2-selective inhibitor with cytotoxic action in several cancer cells [63], including triple-negative breast cancer cells [64]. Therefore, we speculated that disruption of mitochondrial respiration by Pic-III may sensitize the cytotoxic effect of ABT-199. As Figure 13 shows, ABT-199 and Pic III reduced the viability to 0.65 ± 0.11 (p < 0.01 vs. control) and 0.45 ± 0.08 folds of the control (p < 0.001 vs. control), respectively, at 48 h of treatment. Notably, the combination Pic-III + ABT-199 reduced the viability to 0.15 ± 0.05 folds of the control (p < 0.001 vs. control), suggesting that Pic-III-induced mitochondrial dysfunction produces sensitization to BH3 mimetic ABT-199 (Venetoclax) in MDA-MB-231 cells.

Discussion
Snake venom metalloproteinases (SVMPs) are the primary factors responsible for snake-venom-induced local and systemic hemorrhage [8,43]. Particularly, the largest SVMPs, categorized into the P-III class, present a high structural complexity associated with a high functional diversity of their M, D, and C domains [17]. In this work, we describe the purification and characterization of Pictolysin-III (Pic-III), a novel hemorrhagic P-III metalloproteinase isolated from B. pictus, affecting mitochondrial bioenergetics of cell lines. Our results can contribute valuable knowledge to the structure-function relationship of this protein family, and the recently explored mechanisms of venom on cancer biology.
Our purification protocol shows that Pic-III undergoes autoproteolysis ( Figure 1D). This event previously reported for other SVMPs could occur due to alkaline conditions, Ca 2+ absence, and slightly elevated temperature promoting autolysis between the M and DC domains. Although the presence of other proteases that promote the separation of these domains that are removed during the purification process cannot be ruled out [65].
Pic-III has a proteolytic activity dependent on the pH and temperature, which was optimal at pH 8.0 and 25 • C, respectively. These values are close to the experimental data obtained in other SVMP-III such as leucurolysin [66] and bothropoidin [42]. Interestingly, it was observed that at lower temperatures (5 • C and 10 • C), the proteolytic activity was high. Possibly, this is because autoproteolysis is lower at these temperatures, maintaining more enzyme units than at physiological temperature. We regard this fact as an exclusive event of the in vitro condition and not representative of an enzymatic characteristic involved in the biological action of Pic-III. In the presence of Ca 2+ and Mg 2+ , Pic-III significantly increases its proteolytic activity in line with SVMPs [30,42,67].
SVMPs generally exhibit fibrin(ogen)olytic and extracellular-matrix-degrading (hemorrhagic) activities [68]. Pic-III is a venom alfa fibrin(ogen)olytic enzyme that breaks down fibrin clots and avoids clot formation by degrading fibrinogen, actions which allow the spreading of venom compounds [69] and alteration of the homeostatic system [70]. In addition, it has been well-documented that many SVMPs inhibit platelet aggregation induced by different agonists [71]. Pic-III does not affect the collagen-or vWF-induced aggregation but inhibits aggregation induced by thrombin and CVX, a very potent GPVI agonist isolated from rattlesnake (Crotalus durissus terrificus) venom [72,73]. Collagen induces platelet aggregation by interacting with GPVI and α2β1 integrins, whereas vWF binds to the GPIb complex [74,75]. Our results suggest that Pic-III acts as a GPVI antagonist but does not interfere with the GPIb pathway. On the other hand, human α2-M is a plasmatic inhibitor blocking the active site of many SVMPs and some serineproteases such as Pictobin (thrombin-like enzyme from B. pictus) [28,76,77]. Since a greater affinity between Pictobin and α2-M than for Pic-III was observed in our experiments, a potential Pic-III-Pictobin synergistic interaction may be involved in the hemorrhagic effect of Bothrops pictus venom [6]. The gene structure of all SVMP presents a signal peptide and a pro-peptide domain before the M domain, cleaved by proteolytic processing before being secreted from snake venom glands [78]. Mature SVMP P-III has thirty-five strictly conserved cysteine residues among the Reprolysin family with three, eight, and six disulfide bridges formed in the M, D, and C domains, respectively [78]. However, a cysteine residue (Cys189 of Pictolysin-III) does not form a disulfide bridge because its side chain is embedded in the hydrophobic core and only is conserved in SVMP P-III.
Pic-III contains a protein motif conserved in zinc-dependent metalloproteases (HEXXHXXGXXH), where histidine residues play a fundamental role in the interactions with cofactors for enzymatic activity [79]. Moreover, this conserved sequence and the Met-turn are characteristics of members of the metzincins [80]. The disintegrin-like domain contains a motif that interacts with integrins: ECD, a characteristic region of SVMP P-III and ADAM proteins, that replaces the RGD domain of the disintegrins [81]. The predicted molecular weight of Pic-III (46771.7 Da) can be different from SDS-PAGE due to potential protein glycosylation. Although the molecular basis is not understood, SVMP P-III is more hemorrhagic and tends to have long carbohydrate chains [82]. The predicted glycosylation of Pic-III (Asn183) was close to the methionine turn, whose extrinsic flexibility is proposed as a critical factor in the hemorrhage [82,83]. Similar to other snake proteases (e.g., [28][29][30]), our results demonstrate that Pic-III activity is dependent on N-linked carbohydrates, as its activity was reduced by up to 45% after deglycosylation with PNGase F treatment.
Alternatively, Pic-III contains all the calcium-binding residues for three calcium ions: site I in the M domain, which is coordinated by oxygen atoms from the side chain of Glu12, Asn203, and Asp96, and the carbonyl group of Cys200; site II in the D domain, coordinated by oxygen atoms from Val215, Leu220, Glu222, and Asp228 residues; and site III, also in the D domain, which is coordinated by oxygen atoms of Asp279 (from the ECD motif), Glu282, Asp294, and Val 295 [84].
The predicted model of Pictolysin-III presents the classic architecture of the MDC domains (metalloproteinase, disintegrin-like, and cysteine-rich domains), a structure that is the prototype of mammalian ADAM proteins [40]. The M domain shows an α/β scaffold forming four helices (H1, H2, H3, and H4), a five-stranded parallel β-sheet (strands S1, S2, S3, S4, and S6), and an anti-parallel strand (S5), a type of secondary structure corroborated in other SVMPs P-III [40,84]. This domain is separated from the DC domains with the sequence EPLGTDIISP [85] and has a loop conformation in Pic-III. The Ds segment protrudes from the M domain, opposes the catalytic site, and is close to the Ca 2+ -binding site [40]. Domain D together with domain C, form a "C"-shape with its surface concave towards domain M. This structure is maintained by cysteine residues and Ca 2+ -binding residues, which are strictly conserved in most ADAM proteins [78].
Mitochondrion-driven innate immunity involves the ability of mitochondrial DAMPs release (e.g., mtDNA, cardiolipin, formyl-methionine-labeled peptides, and cytochrome c) to activate pattern recognition receptors (e.g., Toll-like receptors) and trigger a proinflammatory cascade [86][87][88]. Recently, it has been shown that the venom of B. laceolatus releases mtDAMPs from cardiomyocytes, and in a murine model, B. asper venom induces the release of mtDNA and cytochrome c in the circulation [89], suggesting that mitochondria would mediate inflammatory signals in the envenomed tissue environment [89][90][91]. Notably, the inhibition of mitochondrial respiration stimulated by glutamate plus malate (complex I's substrates), succinate (complex II's substrate), and TMPD (an artificial complex IV's substrate) plus ascorbate is a previous event for mtDAMPs production [90]; however, the toxin classes involved remain uncertain. Accordingly, the crude venom of viper Macrovipera lebetina obtusa affects respiration in the Vero monkey kidney epithelial cell line, with this effect sensitive to metalloproteinase and phospholipase inhibitors [21]. To our knowledge, there are no studies that identify the action of an isolated SVMP on mitochondrial respiration and its implication for the disruptive action of the cell-ECM interaction.
Our results suggest that Pic-III produces morphological changes in cells, decreases lamellipodia formation, and induces actin network disruption, reducing mitochondrial bioenergetics in epithelial (MDA-MB-231 and Caco-2) and stromal (RMF-621 fibroblast) cell lines. It has recently been described that some PLA2 from snake venom can interact with respiratory complexes, inhibiting their activity [92]; however, we have not been able to establish whether Pic-III interacts directly with any respiratory complex in the mitochondrion. In MDA-MB-231 and RMF-621 cells, Pic-III blocked the glycolysis, with the former cell line more affected, damaging the ability to remodel the metabolism for maintaining ATP levels. Under mitochondrial respiration inhibition, mtROS production can drive pro-inflammatory cytokine production [61,93] and trigger NLRP3 inflammasome activation [94]. Although it has been extensively reported that SVMPs induce the secretion of pro-inflammatory mediators [95,96], our results suggest an unexpected link between mitochondrial respiration inhibition and IL-1β production by Pic-III in epithelial and stromal cells that requires further studies.
Mitochondrial bioenergetics is driven by ECM composition for supporting the initial steps of metastasis and chemoresistance [22,97]. The ECM-cancer-cell interaction requires the ability of these cells to adhere to ECM components and migrate through them [98]. Integrins activate intracellular signaling controlling cytoskeleton organization, cell polarity, leading-edge formation, and mitochondrial bioenergetics of migrating cancer cells [22,99,100]; therefore, disrupting the ECM-integrin-metabolism axis is an attractive anti-cancer target for new antagonist molecules based on SVMP scaffolds. Accordingly, some SVMPs P-III inhibit proliferation [101] and induce apoptosis [102,103] in cancer cells. Interestingly, our results indicate that the triple-negative breast cancer (TNBC) cell line MDA-MB-231, which exhibits high metabolic plasticity for supporting migration [45,104], was more sensitive to the Pic-III treatment, reducing ATP levels and mitochondrial respiration compared to breast fibroblast RMF-621.
The BCL-2-selective inhibitor ABT-199 (Venetoclax) is a potent and orally bioavailable anti-cancer drug in several cancer cells [63], including TNBC cells [64]. In leukemic cells, ABT-199 has high efficacy in inducing apoptosis in nanomolar and sub-micromolar concentrations [62]; however, it induces intrinsic apoptosis at higher concentrations in breast cancer cells [105]. Reports suggest that IDH1/2 mutations [106], inhibition of glutaminolysis [107], and mitochondrial cristae remodeling [108], and OXPHOS uncoupling [62] increase the anticancer efficacy of ABT-199. Notably, our results indicate that Pic-III sensitizes to ABT-199 in MDA-MB-231 cells, which may be mediated by inhibition of the mitochondrial respiration. Nonetheless, the cell signaling triggered by Pic-III-induced mitochondrial dysfunction in cancer cells remains to be elucidated. Finally, Pic-III is the first of the SVMPs reported with action on mitochondrial bioenergetics. Emergent evidence suggests that several snake venom toxin classes (e.g., phospholipases [92,109], thrombin-like enzymes [28], three-finger toxins [110,111]) act on the mitochondrial bioenergetics [3], exhibiting unique and novel mechanisms that are not yet fully understood. This highlights the mitochondrion as an essential player in the snake venom action that requires more studies.

Conclusions
We characterize a novel type-III snake venom metalloprotease, called Pictolysin-III (Pic-III), isolated from Bothrops pictus venom. Pic-III is a 62.5 kDa proteinase and has an optimum temperature and pH of 40 • C and 7.5, respectively. It hydrolyzes dimethyl casein, azocasein, gelatin, fibrinogen, and fibrin. In human platelets, it inhibits the CVX-and thrombin-induced platelet aggregation, acting as a GPVI antagonist and in vivo, exhibits hemorrhagic action. Notably, Pic-III induces actin network disruption, reduces the mitochondrial respiration and ATP levels in epithelial (MDA-MB-231 and Caco-2) and stromal (RMF-621) cells, promoting pro-inflammatory cytokine secretion and sensibilization to the cytotoxic BH3 mimetic ABT-199 (Venetoclax). Pic-III is the first SVMP reported with action on mitochondrial bioenergetics. Importantly, since several SVMPs are promising lead compounds that inhibit platelet or ECM-cancer-cell interactions, it is necessary to further understand their mechanisms of action to reduce undesirable toxic effects.