Venom characterization of the Brazilian Pampa snake Bothrops pubescens by top-down and bottom-up proteomics

Abstract

The envenomation from the Bothrops genus is characterized by systemic and local effects caused by the main toxin families in the venom. In Bothrops pubescens venom we were able to identify 89 protein groups belonging to 13 toxin families with the bottom-up proteomics approach and 40 unique proteoforms belonging to 6 toxin families with the top-down proteomics approach. We also identified multi-proteoform complexes of dimeric L-amino acid oxidase using native top-down mass spectrometry.

The pit viper genus Bothrops is responsible for ~90% of snakebite envenoming in Brazil. The envenomation by Bothrops genus is generally characterized by local (hemorrhage, necrosis, vessel damage, and edema) and systemic effects (renal failure and coagulopathies) (Mamede et al., 2020; Gutierrez et al., 2010). These effects are caused by the main toxin families found in the venom, such as snake venom metalloproteinase (SVMP), snake venom serine protease (SVSP), and phospholipase A₂ (PLA2). SVMPs have fibrin(ogen)olytic activity, prothrombin activator action and are responsible for proteolytic degradation of the capillary basement membrane (Markland and Swenson, 2013). SVSPs are responsible for the cleavage of proteins involved in the clotting system and homeostasis leading to activation/inhibition of proteins involved in the coagulation system and homeostasis. (Sajevic et al., 2011). Additionally, PLA2 is accountable for neurotoxicity, myotoxicity, cardiotoxicity, and platelet aggregation/inhibition (Hiu and Yap, 2020)

Using snake venomics (Calvete et al., 2007) and other proteomics-based methods many Bothrops species have had their venom characterized in the past decade (Nery et al., 2016; Jorge et al., 2015; Mora-Obando et al., 2014, 2020; Mora-Obando et al., 2014, 2020; Gay et al., 2015; Amorim et al., 2018; Sanz et al., 2020a, 2020b; Goncalves-Machado et al., 2016; Ohler et al., 2010). These studies are essential for understanding the envenomation process, discovering potential pharmaceutical/biotechnological products (Waheed et al., 2017), and supporting phylogeny studies (Waheed et al., 2017; Segura et al., 2013). Bothrops pubescens is a venomous snake endemic to the Pampa biome in south Brazil (Machado et al., 2014). Although there are few phylogenetic studies about B. pubescens (Machado et al., 2014), the species is neglected regarding its venom composition. In this study, we present for the first time the venom composition of B. pubescens that was analyzed by top-down proteomics (TDP), native top-down mass spectrometry, and bottom-up proteomics (BUP).

B. pubescens venom was milked from 5 different adult specimens (2 males and 3 females) collected in the State of Rio Grande do Sul - Brazil. After extraction, the venom was pooled, lyophilized, and kept at −80 °C until further use. For the size exclusion chromatography (SEC), crude venom (500 μg) was diluted in 100 mM ammonium acetate and fractionated on a Yarra 3 μm SEC-3000 LC column 300 × 4.6 mm (Phenomenex) using an Agilent 1100 Series liquid chromatography system at a flow rate of 0.2 mL/min. Protein separation was monitored at 280 nm using Agilent OpenLab software. The fractions were subjected to SDS-PAGE, according to Laemmli (1970), and the gel was silver stained.

All the proteomics analysis was conducted as described in Melani et al. (2016). In brief, for BUP, 300 μg of B. pubescens crude venom was resuspended in 7 M urea, 2 M thiourea, and reduced using 30 mM DTT for 1 h at 37 °C. Cysteine residues were carboxamide methylated with 30 mM iodoacetamide for 1 h in the dark at room temperature. The sample was diluted to 1 M urea with Tris 100 mM pH 8.2. MS grade trypsin (Promega) was added (1:25, protease:substrate), and the sample was incubated overnight at 37 °C. Peptides were fractionated using an Ultimate 3000 nanoLC (Thermo Fisher Scientific) system on an in-house packed 2 cm × 150 μm i.d. trap-column, and 25 cm × 75 μm i.d. column (Jupiter C18, 3 μm particle size, 300 Å pore size, Phenomenex) coupled to a Q-Exactive HF (Thermo Fisher Scientific). Chromatography was performed at 300 nL/min flow rate with 95% water, 5% ACN, and 0.2% formic acid (FA) as mobile phase A and 95% ACN, 5% water, and 0.2% FA as phase B on a 120 min Gradient (5% B for 8 min, 5–10% B over 4 min, followed by 10–45% B over 88 min, 45%–94% B over 4 min, 95% for 4 min, 90–5% B for 2 min, and 5% for 10 min). Mass spectra were acquired by Xcalibur software operating in data-dependent acquisition mode, switching between full scan MS1 (60,000 resolution, 50 ms maximum injection time, AGC 3e⁶ charges, spectrum range from 300 to 1800 m/z) and MS2 (30,000 resolution, 100 ms maximum injection time, AGC 1e⁵ charges, spectrum range from 200 to 2,000 m/z). MS2 spectra were obtained by high-energy collision dissociation (HCD) fragmentation using 28% normalized collision energy. The peptides were analyzed in three technical replicates. Peptide search was performed using ProLuCID v1.3 search engine against sequences in the Serpente database, downloaded from UniprotKB on February 15, 2021. Carboxamidomethylation of cysteines and oxidation of methionine was set as fixed and variable modifications, respectively. Resultant peptides were processed and evaluated by Search Engine Processor (SEPro) (Reid et al., 2018) with the following parameters: 10 ppm deviation from theoretical peptide precursor, peptides longer than six amino acid residues, and a 1% estimated protein-level false discovery rate (FDR). Label-free protein quantitation was performed according to the normalized spectral abundance factor (NSAF).

For the TDP approach, the pooled venom was resuspended in mobile phase A and then directly analyzed by LC-MS/MS. NanoRPLC analysis was performed using an Ultimate 3000 nanoLC (Thermo Fisher Scientific) system on an in-house packed polymeric reverse phase resin (PLRP-S, 5 μm particle size, 1,000 Å pore size, Agilent Technologies) for the trap-column (2 cm × 150 μm i.d.) and analytical column (20 cm × 75 μm i.d.). The sample was analyzed using a 120 min Gradient (5% B for 10 min, 5–15% B over 2 min, followed by 15–50% B over 88 min, 50%–95% B over 2 min, 95% for 5 min, followed by 90–5% B over 3 min, and 5% for 10 min). Mass spectra were acquired by Xcalibur. MS1 spectra were acquired at 120,000 resolving power (AGC 1e⁶ charges, 50 ms maximum injection time, spectrum range 400–2,000 m/z) using 4 μscans. Data-dependent MS2 were acquired at 60,000 resolving power (AGC 1 e⁶ charges, maximum injection time 800 ms, spectrum range 200–2,000 m/z) using 4 μscans on a Q Exactive HF. MS2 spectra were obtained by HCD fragmentation using an isolation window of 4 m/z and 21%, 22%, and 25% stepped normalized collision energy. The raw data was searched using TDPortal against the Serpente database downloaded from UniProtKB on March 8, 2020, using 2.2 Da precursor window tolerance, 10 ppm fragment tolerance, and 1% FDR.

For the native top-down characterization, the fractions from SEC were individually analyzed into a Q-Exactive UHMR (Thermo Fisher Scientific) mass spectrometer using a Nanospray Flex Ion Source (Thermo Fisher Scientific), spray voltages between 1,500 and 2,000 V, and the ion transfer tube was set to 310 °C. The mass spectrometer was run in positive mode, data collected from 700 to 15,000 m/z with 20 μscans, resolution of 17,500 (at 200 m/z), and a target AGC of 1e⁶ charges. The S-lens RF level was set to 200% and extended trapping set to 100 V. MS2 spectra were obtained by HCD fragmentation using 21% normalized collision energy. Raw spectra from each experiment were summed across scans, and mass deconvolution was performed using UniDec (Melani et al., 2016). All data are available via ProteomeXchange with identifier PXD027650.

We were able to identify 89 proteins (Supplementary Table 1) by BUP that were assigned to 13 toxin families. We estimated the relative abundance of the venom composition using label-free quantification, and SVMP was the most abundant toxin family corresponding to 39% of the total (Fig. 1A). The venom composition of B. pubescens showed to be similar to other Bothrops species, presenting around 80–100 proteins belonging to characteristic toxin families (Fig. 1B) (Gay et al., 2015; Amorim et al., 2018; Alape-Girón et al., 2008; Galizio et al., 2018). When comparing the amount of L-amino acid oxidase (LAAO) among different Bothrops venoms, B. pubescens (21%) has a slightly higher level than B. moojeni (17.7%), and superior amounts than venoms from B. asper (9.2%), B. jararaca (7.9%), and B. diporus (7.4%). This could indicate that B. pubescens has a stronger myonecrosis, edema-forming, hemorrhage-promoting, platelet-aggregating and/or -inhibiting effect than other Bothrops venoms (Tan et al., 2018; Ribeiro et al., 2016). Furthermore, B. pubescens venom showed a lower content of SVSP (6%) compared to B. moojeni (19.6%), B. asper (18.2%), and B. jararaca (26.65%). Such a low amount is comparable to B. diporus (7.2%), which belongs to a phylogenetic sister group of B. pubescens (Machado et al., 2014).

Fig. 1. Relative protein composition of B. pubescens venom.

The venom was analyzed by LC-MS/MS. Peptide search was performed on ProLuCID v1.3, and peptides were filtered by Search Engine Processor (SEPro). (A) Relative protein composition (%) of B. pubescens venom was estimated by label-free protein quantitation performed according to the normalized spectral abundance factor (NSAF). (B) Comparison between B. pubescens, B. moojeni (Melani et al., 2016), B. asper (Laemmli), B. jararaca (Mora-Obando et al., 2020), and B. diporus (Amorim et al., 2018) venom composition. SVMP - snake venom metalloproteinase, LAAO - L-aminoacidic oxidase, CTL - C-type lectin, PLA₂ - phospholipase A₂, SVSP - snake venom serine protease. Toxin families with less than 2% in the venom (others) include VEGF – vascular endothelial growth factor, BPP - bradykinin potentiating peptides, CRISP - cysteine-rich secretory protein, NGF - nerve growth factor, PDE - phosphodiesterase, GPC - glutamine-cyclotransferase, PLB - phospholipase B, NT - nucleotidases.

With the TDP approach, we identified 40 proteoforms belonging to 6 toxin families (Supplementary Table 2). From the total, 12 proteoforms mapped to SVMP protein family, 13 to PLA2, 7 to bradykinin potentiating peptide (BPP), 6 to LAAO, 1 to SVSP, and 1 to SVSP inhibitor. All proteoforms and the SVSP inhibitor family were only identified by TDP, even without performing multiple fractionation steps before LC-MS/MS and not having genomic data on B. pubescens. Further, we were able to identify 5 alpha-amino acetylated residues at the N-terminus of SVSPs, SVMPs, PLA2, and SVSP inhibitor families. TDP applied to snake venoms is a growing field that allows the analysis and characterization of intact toxin proteoforms compared to identifying peptides by BUP (Melani et al., 2017). However, few snake venoms have been characterized by this technique (Carvalho et al., 2012; Gocmen et al., 2015; Petras et al., 2015, 2016; Calderon-Celis et al., 2016; Zhou et al., 2020; Ainsworth et al., 2018; Kazandjian et al., 2021; Calvete et al., 2021), and the method still has limitations in identifying proteins larger than 30 kDa. In summary, TDP we were able to locate PTMs, identify unique proteoforms and a toxin family not identified by de BUP approach. This indicates the potential of the technique since a lot of information is lost in the BUP approach.

On the other hand, native top-down analysis maintains the non-covalent associations of protein complexes, while this information is inferred in BUP and denaturing TDP (Zhou et al., 2020). For native TDP, we collected 13 fractions using SEC (Fig. 2A), and they were submitted to SDS-PAGE (Fig. 2B) to check their molecular composition. In fraction 5, we observed a cluster of peaks in the charge states 19–22+ (Fig. 2C), which presented intact masses from 113.3 to 115.5 kDa after deconvolution (Fig. 2D). Subsequently, a whole charge state was fragmented using HCD, and an amino acid sequence was partially obtained. The sequence tag “PYQFQHFSEALTA” was identified by BLASTP against the protein sequences characterized in the BUP approach as LAAO with 100% sequence identity to B5AR80, Q6TGQ9, P0CC17, and P56742. The observed intact masses correspond to multi-proteoform complexes (MPC) of two monomers of LAAO, a protein complex known to occur in snake venom (Gay et al., 2015; Amorim et al., 2018; Ohler et al., 2010; Alape-Girón et al., 2008; Galizio et al., 2018).

Fig. 2. Identification of dimeric L-amino acid oxidase (LAAO) in B. pubescens venom.

(A) Chromatogram trace at 280 nm of the venom proteins highlighting the 13 fractions collected by size-exclusion chromatography (SEC). (B) Silver-stained SDS-PAGE analysis of the 13 SEC fractions. (C) Spectrum obtained using native top-down proteomics showing in red the charge state distribution of different LAAO multiproteoform complexes (MPC). (D) Deconvoluted MS1 spectrum and the intact mass of 8 different LAAO MPCs (a–h). The red squares indicate the fraction containing LAAO, and the red circle the LAAO bands region in the SDS-PAGE gel.

In summary, this work shows the first characterization of the B. pubescens venom, a neglected species from the southern cone, using multiple proteomics approaches. BUP and TDP combined were able to identify 14 toxin families and demonstrate that B. pubescens venom is a classical Bothrops venom, although it presents some peculiarities. Combining complementary BUP and TDP approaches allowed us to understand the B. pubescens venom composition, including toxin families, toxin proteoforms, and posttranslational modifications (Laemmli, 1970). Also, with the application of native top-down mass spectrometry, we have a better view of the isoforms we have in the venom composition (Gocmen et al., 2015). Furthermore, TDP was used to analyze a bothropic venom for the first time.

Data availability

No data was used for the research described in the article.