Patterns in Protein Components Present in Rattlesnake Venom: A Meta-Analysis

The specificity and potency of venom components gives them a unique advantage in development of various pharmaceutical drugs. Though venom is a cocktail of proteins rarely is the synergy and association between various venom components studied. Understanding the relationship between various components is critical in medical research. Using meta-analysis, we found underlying patterns and associations in the appearance of the toxin families. For Crotalus, Dis has the most associations with the following toxins: PDE; BPP; CRL; CRiSP; LAAO; SVMP P-I & LAAO; SVMP P-III and LAAO. In Sistrurus venom CTL and NGF had most associations. These associations can be used to predict presence of proteins in novel venom and to understand synergies between venom components for enhanced bioactivity. Using this approach, the need to revisit classification of proteins as major components or minor components is highlighted. The revised classification of venom components needs to be based on ubiquity, bioactivity, number of associations and synergies. The revised classification will help in increased research on venom components such as NGF which have high medical importance.


Association between various venom components in Crotalus venom.
Using the frequent item-set data mining, we were able to identify total of 559 relationships between different venom components for Crotalus (Table S1) (first three rules did not identify the predictor protein and hence were discarded). In this study we highlight the top 20 associations ( Figure 2), e.g., Dis was associated with CTL with confidence =1, support = 0.667 (Table 2). Dis and CTL were expressed together 66.7% times in venom of all species of Crotalus, and if CTL was expressed in venom, then 100% times Dis was also expressed.  Table 2.

Venom constituents in Sistrurus venom
We identified compositional venom studies, through both transcriptomic and proteomic technologies, for 34 entries, including species and subspecies, within the genus Sistrurus. Few studies have focused on the Sistrurus subspecies' venom. 19 protein families were present in Sistrurus (Table 3). These protein families could be classified based on ubiquity or relationship with other proteins.

Association between various venom components in Sistrurus venom.
Using the frequent item-set data mining, we were able to identify eight relationship between different venom components for Sistrurus (Figure 4), e.g., NGF was associated with CTL with confidence =1, support = 0.75 (Table 4). NGF and CTL were expressed together 75% times in venom of all species of Sistrurus, and if NGF was expressed in venom, then 100% times CTL was also expressed. The rules are depicted by top twenty rules as stated in Table 2. and Phospholipases to enjoy more attention than other protein families such as Growth Factors.
However, it is by the combination of ubiquity, bioactivity, and relationship between the protein families that we can classify the venom components as major or minor.
In rattlesnakes MYO, PLA2, SVMP and SVSP are classified as major components based on medical importance and ubiquity [13,131], which is also confirmed by our analysis (Figure 1).
However, with a new approach of using both ubiquity and number of associations for each protein, we find that Dis, LAAO, CTL were all more ubiquitous and had more associations with other proteins in Crotalus species (Table 2). Similarly, in Sistrurus species, the SVMP inhibitor and NGF ( Figure 4) have most associations as compared to MYO which has only one association ( Table 3).
The few studies that have investigated venom proteins based upon their associations have highlighted the synergisms between toxins [76]. This synergism causes the joint effects of multiple toxins to assert greater effects than sum of individual potencies [76], making trace amount of snake venom to be highly efficient and effective [76,77]. Often, such combinations of venom proteins cause a variety of symptoms of bleedings, tissue degradation, necrosis, and further complications in prey and bite victims [72,176] and improve the lethality of whole crude venom in contrast to individual components [76,177].
Through mostly studies of predominant toxins, different general mechanisms for toxin synergisms have been proposed [76,178]: 1) Two or more toxins interacts with different targets on related biological pathways, resulting in synergistically increased toxicity 2) Two or more toxins recognize and interact with the same target in synergistic manner and produce the same effect, and is often called amplification Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 1 September 2020 doi:10.20944/preprints202009.0012.v1 3) One toxin (subunit) acts as chaperone to potentiate another one. The chaperone may expose active/functional site of the 2 nd toxin (subunit), or expose target sites, or increase affinity to target, or modify active surface of the other toxin (subunit). Such complexes usually dissociate after asserting their toxicity.
Synergisms has mostly been reported in major toxins in rattlesnake venoms [76]. Notable example of synergism through complex formation (mechanism 3) is crotoxin, a lethal neurotoxin from C. durissus terrificus, by 2 subunits: an acidic subunit component A (CA or crotapotin) and a basic subunit component B (CB) [112,118,119,179,180]. CB is identified as a basic PLA2 with phospholipase activities and low toxicity while the CA component is said to be small acidic, nonenzymatic, nontoxic subunit [76,180]. However, once combine non-covalently, CA improves the potency of CB by enabling CB to reach the specific crotoxin receptors at the neuromuscular junction as well as inhibits other CB functions such as catalytic and anticoagulant activities [118,180]. Thus, the resulting crotoxin complex is highly active, comparing to individual components, and showing the synergistic between 2 subunits at blocking acetylcholine release [179,180]. Similarly, in C.
scutulatus scutulatus, the Mojave toxin is also another PLA2 complex: one acidic and nonlethal subunit acts as a chaperon for the other basic subunit to improves lethality [44, 151,165]. Other examples of synergistic complexes have been found and reported in many species of Viperidae and Crotalidae [76]. Such interactions show the strong synergistic activities in rattlesnake venoms that have been studied intensively through previous endeavors.
Although the synergism referenced previously is studied through the scope of complex formation found in snake venoms, other synergisms were studied through co-administration of major toxins [76]. A prevalent example between major components are SVMP P-III and an acidic PLA2 in Bothrops alternatus called Baltergin and Ba SpII RP4 PLA2, respectively [181,182]. In which, Baltergin possess high edematogenic and myotoxic activities [181] while PLA2 has no myotoxicity, although it is the most abundant PLA2 in this specie [182]. When acting simultaneously, both are able to cause complete detachment of C2C12 myoblast cells, while none can achieve 50% of detachment on their own [183]. The analogous synergism has also been recorded in endothelial cells, SVMP's natural target [76,184]. The mechanism of synergism for such interaction is said to not based on PLA2 enzymatic activities but is proposed as through interactions with endothelial cells' membranes, free of catalysis [184]. Furthermore, this PLA2 does not target the extracellular matrix proteins like SVMP [181], indicating a mechanism 2 of synergism. Both of these enzymes are present in many rattlesnakes' venoms (Table 1 & 3) and their association is also reported through our analysis (Table S1) There are reports indicating the synergism between crotoxin and crotamine, a member of MYO toxins in Crotalus venoms, that facilitates the internalization of CB subunit, and increase the neuronal toxicity [76,185]. Unfortunately, these interactions are not found in the analysis ( Table 2) although are present in Crotalus venoms (Table 1), which could be due to the sparse reports on Crotalus' venoms with many species are still under-investigated as stated previously.
There are even fewer studies focusing on synergism between major and minor components: SVSP, a major toxin and BPPs, a minor toxin [76], indicating a biased approach in studying venom toxins produced by current major/minor toxin classification convention. BPPs, which are micromolecular hypotensive peptides in snake venoms, can inhibit angiotensin-converting enzyme and induce hypotensive action of bradykinin, accompanied with hyperpermeability of blood  [188,189]. On the other hand, many SVSPs show activities that are similar to kallikrein, a serine proteinases, with the specific and limited proteolytic functions that releases bradykinin [76,146,190]. Previous works indicated that BPPs can act synergistically with kallikrein-like SVSPs, which release bradykinin more effectively than endogenous kallikrein and produce potent hypotension and vascular shock in prey [76,98,146,[191][192][193] (mechanism 1). Such lack of studies regarding minor components shows the flaw in current classification of major and minor venom components, in which the current SVSP-BPP interaction that produces severe physiological consequences is lacking in number of dedicated researches with many aspects of the interaction remain unclear. Likewise, many components with substantial amount of associations with other toxins, like CTL or NGF, are not investigated at all for their potential synergisms. Therefore, there is a need to develop a deeper understanding of minor components in venom of rattlesnakes to discover more associations such as that of SVSP and BPP.
Another way to explain the associations of these toxins is through the evolution of toxins. One relationship that has been explored in previous studies is between SVMP P-III and Dis. Dis are small, non-enzymatic proteins that can bind to extracellular receptors (integrins) with many motifs and sizes, two of which are RGD and MVD motifs[9, 71,147,194,195]. While SVMP-PIII is a subclass of SVMP with a Dis-like domain [196,197]. Dis, especially the RGD/MVD motifs, is suggested to be produced from the rapid evolution occurred in the genes coding of SVMP-PIII [194,195,198]. The RGD/MVD motifs of Dis are presented in many Crotalus species [194] along with the presence of SVMP-PIII, represent as rule 17 (Table 2) can be explained through this evolution model, although the co-association with LAAO is still largely unknown.
Some associations may not need to be derived through their toxicity but could be explained through the proteins' house-keeping functions. The existence of SVMP inhibitor is thought to be a house-keeping molecule, despite its potential therapeutic activities, which helps neutralize the potent SVMP in the venom glands as a self-defense mechanism [199], yet not many studies have been invested in this family along with the lack of appearance in many rattlesnake venoms, where high amount of SVMP exists (Table 1), indicates a gap in knowledge that requires further efforts.
Likewise, NGF is known for its ability to inhibit SVMP proteolysis in Viperidae [200,201]. But growing works have suggested other mechanisms in which NGF can act as cytotoxic proapoptotic factors in tissues that don't have TrkA receptors [200,202,203]; or as ancillary functions, like Hya, to help with efficient absorption of venom component through release of granules molecules (histamine, serotonins, etc.) [178,200,204]. Such large release can also have its own impactful consequences (anaphylaxis, bronchoconstriction, vasodilation, etc.) [200,205]. However, not many Crotalus species have NGF as seen previously in Table 1, indicating yet another gap of knowledge in Crotalus venomics. Using SVMP as common targeting model to explain the association of NGF and SVMP inhibitor (rule 6, Table 4) is promising, but due to the insufficient amount of information provided, such explanation warrants further attempts in co-administration testing to confirm.

Conclusions
Through this meta-analysis we have discovered myriad of associations many of which are yet to be described, but they do provide promising potential synergistic effects that are worth investigating. For example, using rules 2 and 13 in Crotalus venoms (Table 2) protein/glycoprotein that specifically binds to carbohydrate moieties and glycoconjugate, can target and interact with platelet receptors and blood coagulation factors [206], which are also targets for Dis [207], indicate their potential synergisms with antiplatelet toxins and assert the hypotensive results along with many other toxin groups like SVSP [76,98,146,[191][192][193]. Thus, there is a need to characterize toxin components and their associations within the toxicology studies as well as for pharmaceutical interests [72]. With an increase amount of characterization studies, novel families may also be correctly added into the venom profiles such as three-finger toxins (3FTx), which often is present in elapids and a few occasions in rattlesnakes' genome and transcriptome [14-16, 168,170].
Furthermore, similar studies for Sistrurus species are still less abundant than their relative Crotalus species, indicating an underdeveloped field of study that warrant further interests. Additionally, this work also address the problem of conventional classification of venom toxins as major or minor based on importance and ubiquity, which are often MYO, PLA2, SVMP and SVSP [13,131], as the cause of much more attention on these dominant toxin families and overlooking other protein families such as Growth Factors. Therefore, we highlight the importance of studying venom components not only as individual components but also in understanding the relationship between them. It is by the combination of ubiquity, bioactivity, and relationship between the protein families that we can classify the venom components as major or minor.

Materials and Methods
We collected articles and abstracts on venom for each Crotalus and Sistrurus species through the following: databases (PubMed, ScienceDirect, Scopus, Google Scholar, Web of Science), journal's databases (BMC Genomics, Journal of Proteome Research, Journal of Proteomics, Toxicon, Toxicology, Toxins), publisher databases (Wiley Online Library, MDPI, Elsevier). We used "venom" OR "proteomic" OR "venomic" OR "transcriptome" OR "proteome" AND "Name of the species" as keywords for conducting our search. We also, examined references in studies produced from the search results for any additional information. Collected records were earliest obtainable records to those that are published in January of 2020.
From collected records, any article that did not contain information regarding venom composition and components of any Crotalus and Sistrurus species was not used in the current analysis. Otherwise, the articles' full-text version would be further assessed with the following inclusion and exclusion criteria.
For the article to be included in the current analysis, it had to fulfill one of the following criteria: (1) report proteome or transcriptome profile of the venom of any corresponding species; (2) report at detect toxin families/components. From the studies that fulfilled our inclusion criteria and did not meet any exclusion criteria we collected and compiled all venom constituents that are reported for each species in the genus Crotalus and Sistrurus in Table 1 and Table 3, respectively. The compiled data was cross-checked by authors for correctness and confirmations.
Using the data from Table 1  that if x and y occur, then 60% of times z will also occur. An association rule is valid only if lift is greater than 1. Higher the value of the lift, the higher is the validity of the rule. Since, many studies associated with rattlesnake venom concentrated on highly abundant species or species containing more "major components", this affects the performance of the statistical models due to presence of null values. Since, we did only used presence absence data for toxin families from individual studies the chances of bias from individual studies affecting our results were low. With increase in venom composition and variation data, the associations produced by Frequent Item-set data mining analysis will be more informative. All analysis was performed using the software R (R Core Team, 2019).

Supplementary Materials:
The following is available in the supplementary material attachment. Table S1