Biological Activities and Proteomic Profile of the Venom of Vipera ursinii ssp., a very Rare Karst Viper from Croatia

The karst viper (Vipera ursinii ssp.) favours high-mountain dry grasslands in southern and south-eastern Croatia. It is medically less important than other Vipera species, because of its remote habitat and the very small amount of venom that it injects by its relatively short fangs. The scientific literature on Vipera ursinii deals mostly with the morphology, ecology and distribution range of this snake, due to the species’ conservation issues, while the toxinological aspects of its venom have not so far been investigated. Here we report on the composition and biological activity of the Vipera ursinii ssp. venom. Using a proteomics approach, we have identified 25 proteins in the venom that belong to seven protein families: snake venom metalloproteinase, serine protease, secreted phospholipase A2, cysteine-rich secretory protein, snake C-type lectin-like protein, serine protease inhibitor and nerve growth factor. The Vipera ursinii ssp. venom was found to be distinctively insecticidal. Its lethal toxicity towards crickets was more than five times greater than that of Vipera ammodytes ammodytes venom, while the opposite held in mice. Interestingly, the mode of dying after injecting a mouse with Vipera ursinii ssp. venom may suggest the presence of a neurotoxic component. Neurotoxic effects of European vipers have so far been ascribed exclusively to ammodytoxins and ammodytoxin-like basic secreted phospholipases A2. Structural and immunological analyses of the Vipera ursinii ssp. venom, however, confirmed that ammodytoxin-like proteins are not present in this venom.


Introduction
Orsini's viper, Vipera ursinii, is one of the most endangered snake species in Europe. Currently, four subspecies are officially recognized [1], but an additional one may be introduced based on a recent mitochondrial DNA analysis [2]. Three of them, V. u. ursinii, V. ursinii ssp. and V. u. macrops, prefer living at higher altitudes in mountain grasslands (karst vipers), while V. u. rakonsiensis and V. u. moldavica prefer lowlands (meadow vipers). The distribution area of V. ursinii is highly fragmented and covers south-western France, central Italy, Hungary, Romania as well as western and eastern parts of the Dinaric Mountains. Due to the threat of extinction, the Council of Europe adopted, in 2005, the V. ursinii protection plan [3].
V. ursinii ssp. (Figure 1) is a poorly described, novel subspecies that lives in the north-western part of the Dinarides (Croatia, Bosnia and Herzegovina). It is genetically separated from the other karst vipers that inhabit the Dinaric Mountains all the way to Albania, and is more like the West European subspecies [1,4]. The Croatian V. ursinii ssp. is a highly endangered viper that is found in five isolated localities: in southern Velebit, Poštak, Dinara, Troglav and the Kamešnica mountains, at altitudes above 1100 m [5,6]. It is the smallest venomous snake in Croatia, reaching an average size of 54 cm. Typically, its head is clearly separated from its neck and the body is short and smooth. The grey or brownish ground colour is usually lighter dorsally. In most cases, the animal is also characterized by a dark zigzag pattern along its back. The snake is entomophagous, i.e., its typical diet comprises insects (grasshoppers and crickets) [7]. People rarely encounter this snake, so accidents are very scarce. In addition, the snake has very short fangs (2-3 mm) through which it can inject only minute amounts of venom (1-4 mg). Accordingly, the symptoms reported in patients envenomed by the V. ursinii venom have been very mild and local, and were resolved spontaneously without medical intervention [8].
Toxins 2020, 12, x FOR PEER REVIEW 2 of 16 parts of the Dinaric Mountains. Due to the threat of extinction, the Council of Europe adopted, in 2005, the V. ursinii protection plan [3]. V. ursinii ssp. (Figure 1) is a poorly described, novel subspecies that lives in the north-western part of the Dinarides (Croatia, Bosnia and Herzegovina). It is genetically separate from the other karst vipers that inhabit the Dinaric Mountains all the way to Albania, and is more like the West European subspecies [1,4]. The Croatian V. ursinii ssp. is a highly endangered viper that is found in five isolated localities: in southern Velebit, Poštak, Dinara, Troglav and the Kamešnica mountains, at altitudes above 1100 m [5,6]. It is the smallest venomous snake in Croatia, reaching an average size of 54 cm. Typically, its head is clearly separated from its neck and the body is short and smooth. The grey or brownish ground colour is usually lighter dorsally. In most cases, the animal is also characterized by a dark zigzag pattern along its back. The snake is entomophagous, i.e., its typical diet comprises insects (grasshoppers and crickets) [7]. People rarely encounter this snake, so accidents are very scarce. In addition, the snake has very short fangs (2-3 mm) through which it can inject only minute amounts of venom (1-4 mg). Accordingly, the symptoms reported in patients envenomed by the V. ursinii venom have been very mild and local, and were resolved spontaneously without medical intervention [8]. The scientific literature on V. ursinii deals mostly with its morphology, ecology and distribution, addressing, in the first place, its preservation problems [1,3,9,10], while its venom composition and toxic activities have been only poorly investigated. Regarding the latter, the venom of V. ursinii was reported, in mice, to be haemorrhagic but not myotoxic [11]. The mRNA transcripts of two secreted phospholipases A2 (sPLA2s), ammodytin I1 (AtnI1) and ammodytin I2 (AtnI2), were identified in the V. ursinii venom glands [12].
To obtain an insight into the functioning at the molecular level of the venom of this rare and ecologically very distinctive snake, we report here a toxinological description of the venom along with its comprehensive proteomic analysis. This opened the way to unravelling a unique insecticidal activity, leading potentially to new pesticides.

Toxinological Characterisation of V. ursinii ssp. Venom
The lethal toxicity of V. ursinii ssp. venom was determined in mice and crickets. V. ursinii ssp. venom was less toxic than V. a. ammodytes venom in mice-its average LD50 was more than 4 times higher than that of V. a. ammodytes venom (Table 1). Clinical pictures of mice dying from V. ursinii ssp. or V. a. ammodytes venom were similar. After i.v. application of venom, animals rapidly showed signs of anxiety and severe dyspnoea, followed by convulsions and ataxia with spasms. Finally, they The scientific literature on V. ursinii deals mostly with its morphology, ecology and distribution, addressing, in the first place, its preservation problems [1,3,9,10], while its venom composition and toxic activities have been only poorly investigated. Regarding the latter, the venom of V. ursinii was reported, in mice, to be haemorrhagic but not myotoxic [11]. The mRNA transcripts of two secreted phospholipases A 2 (sPLA 2 s), ammodytin I 1 (AtnI 1 ) and ammodytin I 2 (AtnI 2 ), were identified in the V. ursinii venom glands [12].
To obtain an insight into the functioning at the molecular level of the venom of this rare and ecologically very distinctive snake, we report here a toxinological description of the venom along with its comprehensive proteomic analysis. This opened the way to unravelling a unique insecticidal activity, leading potentially to new pesticides.

Toxinological Characterisation of V. ursinii ssp. Venom
The lethal toxicity of V. ursinii ssp. venom was determined in mice and crickets. V. ursinii ssp. venom was less toxic than V. a. ammodytes venom in mice-its average LD 50 was more than 4 times higher than that of V. a. ammodytes venom (Table 1). Clinical pictures of mice dying from V. ursinii ssp. or V. a. ammodytes venom were similar. After i.v. application of venom, animals rapidly showed signs of anxiety and severe dyspnoea, followed by convulsions and ataxia with spasms. Finally, they died. Such symptoms are characteristic of neurotoxins or stroke. The two venoms, however, differ distinctly in that V. ursinii ssp. venom induced sudden and sporadic, dose-unrelated deaths when injected in experimental animals, while V. a. ammodytes venom did not. Both venoms were also lethally toxic for crickets. In contrast to the results in mice, V. ursinii ssp. venom displayed higher toxicity in crickets than V. a. ammodytes venom. The average LD 50 of the first was more than 5-fold lower than that of the latter (Table 1). Interestingly, for both types of animals, the LD 50 values are in the same µg range. The species selectivity of a particular venom is clear -comparing mass-normalized LD 50 s, V. a. ammodytes venom was almost 114-fold more lethal for mice than for crickets, while V. ursinii ssp. venom was only 5-fold ( Table 1).
The haemorrhagic activities of the two venoms in rats were similar (Table 1).

Immunological
Cross-Reactivity between V. ursinii ssp. and V. a. ammodytes Venoms Rabbit antisera against the whole V. a. ammodytes venom (anti-Vaa) and its various components, the pool of V. a. ammodytes haemorrhagins (anti-H), V. a. ammodytes neurotoxic sPLA 2 Atx (anti-Atx) and V. a. ammodytes non-toxic sPLA 2 AtnI 2 (anti-AtnI 2 ) were used to probe cross-reactivity between the V. ursinii ssp. venom components, using WB and ELISA assays. Anti-Vaa, anti-H and anti-AtnI 2 antibodies proved to be paraspecific in ELISA tests. On the other hand, anti-Atx antibodies did not recognize the V. ursinii ssp. venom-coated ELISA plate at all ( Figure 2). ELISA results were confirmed by WB analysis. As shown in Figure 3, Atx-like sPLA 2 s were not detected using Atx-specific antibodies in V. ursinii ssp. venom.     To analyse the V. ursinii ssp. venom proteome, the crude venom collected from the wild snakes was separated by 1-DE in the molecular mass range 10 to 80 kDa ( Figure 4A), and by 2-DE in the isoelectric point (pI) range of 3 to 10 and a molecular mass range of 3.5 to 110 kDa ( Figure 4B). V. ursinii ssp. venom was resolved into seven and eight distinct bands on non-reducing and reducing 1-DE, respectively. ursinii ssp. venom was estimated from intensities of the 2-DE spots in which they were detected. SVMP, snake venom metalloproteinase; CRISP, Cys-rich secretory protein; sPLA 2 , secreted phospholipase A 2 ; SVSP, snake venom serine protease.
More than 50 protein spots were observed on the 2-DE gel. We assumed that proteins from the most intense bands on the 1-DE gel were those appearing as the major spots in the 2-DE gel at the same molecular masses, e.g., at~50,~35,~30 and~15 kDa. Proteins from the 1-DE Bands 3 and 4 thus constituted mostly 2-DE Spots 11-20. The major protein from Band 5 was located in 2-DE Spots 27 to 29, the one from Band 6 in 2-DE Spots 34 to 37, and proteins from Band 7 mostly in 2-DE Spots 42 and 45. For the MS analysis, dominant protein bands were thus excised from the non-reduced 1-DE gel (7 bands labelled on the gel in Figure 4A). Less abundant protein bands were excised from the reduced 1-DE gel (3 bands labelled 8 to 10 on the gel in Figure 4A) and picked from the 2-DE gel (31 spots designated by red numbers in Figure 4B).
Again, we assumed that the main protein of 1-DE Band 9 at~18 kDa is the same as the protein in 2-DE Spot 39, the most intense spot at the same molecular mass. The proteins were trypsinized in-gel and the resulting peptides extracted and analysed by LC-ESI-MS/MS. Altogether, in selected bands and spots, 25 proteins were identified. They can be grouped into seven protein families: snake venom metalloproteinase (SVMP), snake venom serine protease (SVSP), sPLA 2 , cysteine-rich secretory protein (CRISP), snake C-type lectin-like protein (snaclec), Kunitz-type serine protease inhibitor (KSPI) and venom nerve growth factor (VNGF) family (Table S1). No proteins could be identified in Spots 4,6,8,9,40,49 and 50, due either to too low amounts of material or to the lack of similar/identical sequences in the protein database used to search the MS spectra. The most abundant were proteins of 50 kDa belonging to the SVMPs that constituted 1-DE Bands 1 to 4 and 8 (corresponding to 2-DE Spots 11 to 20), and 2-DE Spots 1 to 5, 10 and 23 to 25. The second most abundant were CRISPs, which were identified as the major components of 1-DE Band 6 with apparent molecular masses of~27 kDa (corresponding to 2-DE Spots 34 to 37). CRISP was also detected as a minor constituent of the 1-DE bands at higher apparent molecular masses of 43 kDa (Band 4) and 35 kDa (Band 5), as well as 50 kDa For conservation and research purposes, some V. ursinii ssp. specimens are kept in captivity, where they experience a food regime different from that in the wild. Namely, captive animals were fed on mice, while, in their natural habitat, they mostly eat insects. In order to examine how different diets influence the venom composition, venoms collected from the snakes in captivity (VuCro-c) and those obtained from wild animals (VuCro) were analysed comparatively by 2-DE ( Figure 5A,B). Differences were observed mainly in the acidic part of the 2-DE gels where, in the case of the VuCro-c venom, certain spots appeared more intense than those in VuCro venom. In addition, in this area, several additional spots are present in the case of VuCro-c venom (red encircled Spots 1 to 7 in Figure 5A). Proteins identified in these spots are listed in Table S2. Spots of the increased intensity in VuCro-c regarding to the VuCro venom ( Figure 5A,B) correspond to the Spots 30, 38, 39, 40 and 42 in Figure 4B. In these spots, proteins belonging to SVSPs, sPLA 2 s, SVMPs, snaclecs, VNGFs and KSPIs were identified (Table  S1). Of these, sPLA 2 s appeared much more highly abundant in the venom of animals in captivity (VuCro-c) than in the venom of those living in the wild (VuCro). In Table 2 the proteins are listed whose abundance is distinctively different in VuCro-c and VuCro venoms.  Table S2. In spots encircled black no proteins were identified. Table 2. Proteins with a higher or lower abundance in the venom of the Croatian V. ursinii ssp. snakes living in captivity (VuCro-c) than in the venom of the snakes living in the wild (VuCro). Abbreviations: AtnI, ammodytin I; CRISP, cysteine-rich secretory protein; KSPI, Kunitz-type serine protease inhibitor; sPLA 2 , secreted phospholipase A 2 ; SVMP, snake venom metalloproteinase; SVSP, snake venom serine protease; VNGF, venom nerve growth factor; Vaa, V. a ammodytes; Vu, V. ursinii.

Higher Abundance
Protein Family

Comparative
Analysis of V. ursinii ssp. and V. a. ammodytes Venoms Comparative 2-DE analysis of the crude V. ursinii ssp. and V. a. ammodytes venoms exposed large differences in their composition with respect to distribution and intensity of most of the protein spots ( Figure 5B,C). The V. ursinii ssp. proteome was compared with the recently published proteome of the V. a. ammodytes venom [15]. In general, there are far fewer basic proteins in the V. ursinii ssp. venom than in the V. a. ammodytes venom. Further, in the V. ursinii ssp. venom, high molecular mass proteins prevail over those with low molecular masses and vice versa in the case of the V. a. ammodytes venom. Thus, high molecular mass SVMPs are the major constituents of the V. ursinii ssp. venom, while low molecular mass sPLA 2 s and snaclecs dominate in the V. a. ammodytes venom. The latter are scarce in the V. ursinii ssp. venom, which is also completely devoid of basic sPLA 2 s. SVSPs are found in many spots in both venoms, but in much higher quantities in the V. a. ammodytes venom, where they are also much more diverse regarding their pIs. In contrast, the CRISPs are more abundant in the V. ursinii ssp. than in the V. a. ammodytes venom. LAAOs were not detected in the V. ursinii ssp. venom. LAAOs are usually the components that colour snake venoms yellow. In accordance with our results, the colour of V. ursinii ssp. venom is only slightly yellow.

Discussion
V. ursinii ssp. is one of the rarest snakes in Europe. Endangered by extinction, it can be found only in certain mountain regions in Croatia and Bosnia and Herzegovina, at altitudes exceeding 1000 m. Due to the difficulty in encountering this snake in the wild, its venom remained practically unstudied until this work.
Due to its extreme habitat, V. ursinii ssp. has a special diet consisting mostly of insects. The adaptation of the snake venom to this prevailing food is clear. Compared to V. a. ammodytes, which feeds predominantly on rodents, the venom of V. ursinii ssp. displayed a more than 5-fold higher lethal toxicity in crickets but more than 4-fold lower toxicity in mice (Table 1). Such an adaptation of venom efficacy to diet has been reported before. For example, the venoms of vipers that are not entomophagous (V. berus and V. nikoloski) have a lower toxicity for crickets than those of pronouncedly entomophagous vipers (V. renardi and V. lotievi) [7]. In addition, the average mass-normalized LD 50 values that we determined in crickets for V. ursinii ssp. and V. a. ammodytes venoms (Table 1) were comparable with those reported in [7] for entomophagous and non-entomophagous snakes,~10 and 30 µg/g, respectively.
The lethal toxicities of both V. ursinii ssp. and V. a. ammodytes venoms were higher in mice than in crickets (Table 1). Higher lethality for mice than for arthropods has also been observed in the case of other viper venoms, e.g., those of the Pelias group of snakes [7] and Echis spp. [16,17]. This points to the fact that, during evolution, rodents have been the principal food of vipers, with other species like insects or scorpions being secondary. Snakes of other families may have developed other food preferences and, correspondingly, venoms with other prey specificities. For example, the venom of a colubrid, the brown tree snake (Boiga irregularis), is more toxic to birds and lizards than to mammals [18]. However, the mass-adjusted LD 50 of the V. a. ammodytes venom was more than 100-fold lower in mice than in crickets, while that of the V. ursinii ssp. venom was just 5-fold lower. This shows that V. ursinii ssp. adapted to their new living environment in which insects prevail over mice as the available food. The snake, therefore, has been forced to turn from being a mouse-hunting to an insect-hunting animal.
Gel-based proteomic analysis of the venom of the wild V. ursinii ssp. species revealed a proteome much simpler than that of the recently published V. a. ammodytes venom proteome [15]. Almost two times fewer proteins were identified in the V. ursinii ssp. than in the V. a. ammodytes venom. While the proteins in the venom of V. ursinii ssp. represent seven different protein families, representatives of 16 protein families were found in the venom of V. a. ammodytes. Both venoms contain SVMPs, SVSPs, sPLA 2 s, CRISPs, snaclecs, VNGFs and KSPIs, which are typical of most viperid snake venoms [19,20]. In general, the number of homologues, representatives of the most abundant enzyme families, SVMPs, SVSPs and PLA 2 s, is much lower in the V. ursinii ssp. than in the V. a. ammodytes venom. Furthermore, snaclecs appear only as covalently bound SVMP subunits in the V. ursinii ssp. venom. A similar observation was reported in the venomic study of V. b. berus [21]. In the V. a. ammodytes venom, however, snaclecs are also present as an abundant and structurally diversified group of disulphide-linked heterodimeric proteins. Members of the two protein families that are usually present in viperid venoms, LAAOs and disintegrins, were not identified in the V. ursinii ssp. venom. So far, of the genus Vipera, only V. anatolica, has been reported to be devoid of LAAOs and V. nikolskii of disintegrins [20].
The dominant components of V. ursinii ssp. venom are SVMPs, representing 55% of the total venom proteins ( Figure 4C). Among snakes of subfamily Viperinae, to which V. ursinii ssp. belongs, a similar prevalence of SVMPs was also observed in venoms of genera Cerastes and Echis, and M. lebetina from Tunisia [20]. SVMPs are structurally and functionally highly diverse zinc-dependent enzymes [22,23]. The V. ursinii ssp. venom displays high haemorrhagic activity. Its MHD in rat is similar to that displayed by the V. a. ammodytes venom ( Table 1). The predominance of the P-III class SVMPs in the V. ursinii ssp. venom explains its haemorrhagic nature. The high haemorrhagic potential of the P-III SVMPs is based on the presence of non-catalytic domains (disintegrin-like and Cys-rich) in these molecules that mediate their binding to specific substrates in the microvasculature [24]. Haemorrhagic SVMPs namely cause blood microvessel damage by hydrolysing the key proteins in the basement membrane of the capillary wall particularly type IV collagen. This leads to weakening and disruption of the wall due to the hemodynamic pressure. In accordance with the high degree of immunological cross-reactivity of SVMPs from V. ursinii ssp. and V. a. ammodytes venoms (Figure 2), we have identified V. ursinii ssp. haemorrhagins, mostly as homologues of those described in the V. a. ammodytes venom, VaH3, Vaa-MPIII-1, Vaa-MPIII-3 and Vaa-MPIII-4 (Table S1). Of these, VaH3 has been characterized as one of the main haemorrhagic factors of the V. a. ammodytes venom [25]. As an N-glycosylated homodimer, it belongs to the P-IIIc subclass of SVMPs. In vitro, VaH3 also markedly prolonged blood coagulation, due to its fibrinogenolytic activity, but only weakly affected platelet aggregation. Vaa-MPIII-1 and Vaa-MPIII-4 are present in trace amounts in the V. a. ammodytes venom [15]. In contrast, in the V. ursinii ssp. venom, these two SVMPs have been identified in several 1-DE bands and 2-DE spots (Figure 4, Table S1). Moreover, Vaa-MPIII-1 appears to be the major SVMP of the V. ursinii ssp. venom. Comparison of the Vaa-MPIII-1 and Vaa-MPIII-4 primary structures with those of the other SVMPs whose activities are known, revealed their highest identities (~75%) with the haemorrhagic SVMPs from viperid venoms. This suggests that Vaa-MPIII-1 and Vaa-MPIII-4 also possess haemorrhagic activity. Vaa-MPIII-3 is a member of the recently introduced P-IIIe SVMP subclass. That subclass evolved from the classical P-III SVMPs by loss of the MP domain [15], thus consisting of only a complete or a partial disintegrin-like domain and a Cys-rich domain. Interestingly, the V. ursinii ssp. Vaa-MPIII-3 peptides were found in the high molecular mass protein spots (48 to 120 kDa), suggesting the existence of a Vaa-MPIII-3-like molecule also containing the MP domain.
The pathologic effect of haemorrhagins can be exacerbated by the action of other venom components that interfere with the processes of blood coagulation and/or platelet aggregation [22]. For example, the extensive and uncontrolled activation or degradation of coagulation factors by SVMPs and SVSPs leads to a consumptive coagulopathy [26]. Potentially having such a role, we found, in the V. ursinii ssp. venom, homologues of two SVMP activators of the coagulation factor X (FX), VLFXA from M. lebetina (AAQ17467) and RVV-X from D. russelii. The corresponding proteins in the V. a. ammodytes venom are VAFXAs [27]. All the aforementioned FX activators belong to the P-IIId subclass, in which a dimeric snaclec (light chain) is disulphide-linked to the Cys-rich domain of a P-III SVMP (heavy chain). The snaclec subunit serves to bind the protease to the Gla-domain of FX [28], thus positioning it correctly for effective proteolytic activation of the factor.
We have also identified several members of the SVSP family in the V. ursinii ssp. venom that may have pro-and anti-coagulant activities. Those being potentially procoagulant are V. ursinii ssp. homologues of RVV-V from the Daboia siamensis venom [29] and Vaa-SP-VX from the V. a. ammodytes venom (Latinović et al., unpublished), which activate FV. In contrast to the SVMP activators of blood coagulation factors, the SVSP activators are usually single-chain molecules with three-dimensional structures that resemble the structure of physiological activators, e.g., the heavy chain of the activated blood coagulation factor X (FXa) and thrombin, though with sequence replacements that confer high substrate specificity and resistance towards the natural inhibitors, the serpins [30]. The most abundant V. ursinii ssp. SVSP is a homologue of an anticoagulant V. a. ammodytes protein, Vaa-SPH-1 (KT148824). The latter is an enzymatically inactive SVSP that inhibits the activity of the intrinsic tenase complex by binding to FVIIIa in place of FIXa [31].
Like SVSPs, venom sPLA 2 s can also affect haemostasis in a catalytic activity-independent and/or dependent way [22]. Ammodytoxins (Atx), basic sPLA 2 s from the V. a. ammodytes venom, inhibit blood coagulation by binding to FXa, thereby preventing formation of the prothrombinase complex. They have not been detected in the V. ursinii ssp. venom. Ammodytins (Atn), acidic sPLA 2 s from the V. a. ammodytes venom, induce anticoagulant and antiplatelet effects by hydrolysing phospholipids [32]. AtnI 1 and AtnI 2 are present in the V. ursinii ssp. venom, AtnI 1 (CAE47156) being the major isoform. AtnI 1 proteins from the V. ursinii ssp. and V. a. aspis (AAN59986) venoms are practically identical (99% sequence identity). The latter was characterized as being indirectly haemolytic and weakly anticoagulant [33], and the same activities can be expected in the case of the V. ursinii ssp. enzyme. On the other hand, V. ursinii ssp. AtnI 2 is probably endowed with a potent haemostatic activity based on 96% sequence identity with Vur-PL2 (ADG86231) from the closely related Russian snake V. renardi (former V. ursinii renardi). Vur-PL2 was namely described as a strongly anticoagulant and platelet aggregation inhibitory enzyme [34]. Although only weakly toxic in mice (LD 50 = 10.7 mg/kg), one third of experimental animals died from internal bleeding in various organs at an LD 50 dose of Vur-PL2.
Although the V. ursinii ssp. venom was much less toxic to mice than the V. a. ammodytes venom, the mode of dying of animals after they received a lethal dose of one or other of the venoms was frequently similar, suggesting neurotoxicity as the cause of death. Unusually, the neurotoxic-like effects did not take place in every animal envenomed by the V. ursinii ssp. venom, and, when it did take place, they were apparently independent of the received dose of the venom. Such phenomena have never been observed when poisoning mice with the V. a. ammodytes venom, which suggests different molecular backgrounds of the effects induced by the two venoms. In the case of V. a. ammodytes and related European viper venoms, the neurotoxicity stems from basic sPLA 2 s such as Atx [35]. The latter were not found in the V. ursinii ssp. venom when using alternative analytical approaches (Figures 2, 3A  and 5). A similar situation, i.e., sporadic induction of symptoms considered as neuropathological (e.g., paraesthesia) in spite of the absence of basic sPLA 2 s in the venom, was also reported for envenomation by venom from V. u. ursinii in south-eastern France [12]. If the effects were indeed neurotoxic then they cannot be induced by neurotoxic sPLA 2 s. Sudden and sporadic, dose-unrelated deaths of animals injected with V. ursinii ssp. venom, however, offered also another possible explanation-a stroke, induced by coagulopathic venom components following the i.v. application of the venom.
The non-enzymatic proteins found in the V. ursinii ssp. venom are members of the CRISP, VNGF and KSPI protein families, all non-lethal constituents of snake venoms. The CRISP family is represented by homologues of Vaa-CRISP-1 (KT148819) and Dr-CRPK from the D. russelii venom (ACE73567). These two proteins share 90% sequence identity but neither has a known biological function. Although CRISPs are widespread in snake venoms, knowledge concerning their biological targets and activities is limited. Many so-far characterised CRISPs either modulate the conductivity of various ion channels [36] and induce inflammation [37,38] or have antiprotozoal [39] or antiangiogenic [40] activity.
VNGFs are common components of snake venoms [41] and the V. ursinii ssp. venom is not an exception. These molecules are either covalent or non-covalent homodimers expressing nerve growth-promoting activity. Some of them, like the non-covalent homodimer from M. lebetina venom (P25428), are glycoproteins. V. ursini's VNGF (AEH59582) shares 97% sequence identity with the M. lebetina VNGF and has a conserved N-glycosylation site at Asn148. The apparent monomer mass of both V. ursinii ssp. and M. lebetina VNGF proteins is~18 kDa ( Figure 3A, Table S1). This suggests that V. ursinii ssp. VNGF is also glycosylated.
Venoms of Viperidae and Elapidae snakes contain~60 amino acid-long Kunitz-type SPIs (KSPIs). We have also identified such molecules in the V. ursinii ssp. venom as homologs of VaaChi, a KSPI from the V. a. ammodytes venom [42], and Vur-KIn, a KSPI from the V. renardi venom [12]. The structural fold of KSPIs resembles that of the bovine pancreatic trypsin inhibitor [43]. Three KSPIs from the V. a. ammodytes venom have already been characterised, two as potent inhibitors of trypsin (VaaTi) and as the inhibitor of chymotrypsin (VaaChi) [42]. VaaTi also inhibited plasmin and plasma kallikrein, while VaaChi was shown to form a non-covalent complex with Atx, increasing its toxicity [44]. The activity of Vur-KIn has not yet been determined but it shows the highest sequence identity (≥80%) with trypsin inhibitors from Viperinae snakes, VaaTi from V. a. ammodytes (P00991), PPTI from Pseudocerastes persicus (C0HLB2) and a KSPI from Eristicophis macmahoni (P24541). Further supporting the trypsin inhibitory activity of Vur-KIn is the preservation of its trypsin-interaction site at the N-terminal part of the molecule, with Lys17 at the P1 position. Many trypsin-inhibiting KSPIs interact with coagulation factors, for example thrombin, FXa and FXIa, and therefore express antithrombotic and anticoagulant activities [43].
We observed extensive differences in venom proteomes between V. ursinii ssp. snakes living in the wild and those living in captivity ( Figure 5; Table 2 and Table S2). This finding was, however, expected given that the animals in captivity had exclusively been fed on mice, as opposed to the primarily insectivorous diet of wild V. ursinii ssp. A shift in the venom composition due to a diet change has been described before [45]. In our case, the most prominent difference between the VuCro and VuCro-c venom proteomes lay in the content of sPLA 2 s. This finding supports the hypothesis that the acquisition of diverse venom sPLA 2 isozymes by accelerated evolution [46] represents a strong selective advantage for diverse physiological functions, including rapid adaptation to available prey through changes in gene expression [47].

Conclusions
V. ursinii ssp. is, ecologically, a very special viper. Its reliance on insect rather than rodent prey is also reflected in the distinctively higher toxicity of its venom for insects than for mice as compared with venom of a closely related, typical mouse-feeding snake, V. a. ammodytes. V. ursinii ssp. venom is, however, as haemorrhagic to mice as venom from V. a. ammodytes. The V. ursinii ssp. venom proteome is also much less complex than that of V. a. ammodytes, containing representatives of only seven different protein families compared to 16 in V. a. ammodytes. In addition, most of these families contain fewer protein isoforms in V. ursinii ssp. compared to V. a. ammodytes. SVMPs, which are the most abundant proteins in the V. ursinii ssp. venom, are most likely responsible for its haemorrhagic property. The insecticidal activity also suggests the existence of insect-specific toxins in the V. ursinii ssp. venom, which may have potential as bioinsecticides.

Snake Venoms and Antivenoms
A pooled sample of V. ursinii ssp. venom was obtained by milking more than 10 adult snakes caught at an isolated locality of the southern Velebit, identified by an expert herpetologist and then released. A small amount of venom was also collected from 4 adult V. ursinii ssp. snakes kept in captivity for research purposes at Zagreb Zoo Research and Conservation Centre, and designated as VuCro-c. Two of these (male and female) were captured 6 years ago, and, since then, fed exclusively on mice. Two were however born in captivity and have never been fed on insects. Crude venom of V. a. ammodytes was collected by milking snakes kept at the Institute of Immunology Inc., Zagreb, Croatia. All V. ursinii ssp. and V. a. ammodytes venom samples were air dried at room temperature and stored in the dark at +4 • C until use.
Rabbit serum against the whole V. a. ammodytes venom (anti-Vaa) and against V. a. ammodytes venom's components: haemorrhagins (anti-H), Atx (anti-Atx) and AtnI 2 (anti-AtnI 2 ) were produced on a small scale according to the in house immunisation scheme [48,49]. Mice and rats used for in vivo assays were bred at the Institute of Immunology Inc., Croatia. Mice used for the assay of lethal toxicity were of the NIH Ola/Hsd strain. Rats of the Lewis strain were used to assess haemorrhagic activity. Animals were housed under a 12 h light/dark cycle at a temperature of 23 ± 3 • C. A standard mouse/rat diet (Mucedola srl, Milano, Italy) and water were supplied ad libitum during the whole period of experiments.

Assay of Lethal Toxicity in Mice
Lethal toxicity, expressed as the median lethal dose (LD 50 ), was determined in adult male NIH Ola/Hsd mice according to the method of Theakston and Reid [51] and European Pharmacopeia (Ph.Eur.01/2008:0145). Groups of six animals (18-20 g each) were injected i.v. with 250 µL of the respective venom solution in saline. Venom doses ranged from 5.25 to 25 µg per mouse. Mortality was recorded after 48 h. For each venom sample, the assay was repeated at least twice and up to five times. The median lethal dose was calculated by Spearman-Kärber analysis. The results were expressed as the mean of n determinations ± standard error (SE).

Assay of Lethal Toxicity in Crickets
The entomotoxic effects of V. ursinii ssp. and V. a. ammodytes venoms were investigated on crickets (Gryllus assimilis) as described [7]. Crickets (0.50-0.95 g each), in groups of 4, were injected with 5 µL of solutions in water for injection (WFI) of the respective venom. Venom doses ranged from 1.56 to 100 µg per cricket. An equal volume of WFI was injected into a control group. Mortality was recorded after 48 h. The median lethal dose was calculated by Spearman-Kärber analysis. The results were given as the mean of n determinations ± SEM.

Assay for Assessing Haemorrhagic Activity
Haemorrhagic activity was estimated according to Theakston and Reid [51], with experimental details as described in [52]. Briefly, Lewis rats (250 g) were shaved on the dorsal skin and injected i.d. with 100 µL of the venom solution in saline. Six doses, spanning from 1.28 to 50 µg, were tested on each animal. The haemorrhagic lesions on the inner surface of the removed skin were observed 24 h later and their perpendicular diameters measured, from which the average value for each venom dose was calculated. The minimal haemorrhagic dose (MHD) of the venom is defined as the amount of venom (µg dry mass) which, when injected intradermally, 24 h later results in a haemorrhagic lesion of 10 mm in diameter. Results are given as means of n determinations ± SEM.

ELISA Assay
ELISA assays for detecting haemorrhagin-, Atx-or Atn-like proteins in V. ursinii ssp. venom were performed as described [53]. Briefly, microtiter plates were coated with 100 µL/well of the venom solution (1 µg/mL) in 50 mM carbonate buffer, pH 9.6, and left overnight at room temperature (RT). After washing and blocking with 0.5% (m/v) BSA in PBS/T (0.05% (v/v) Tween 20 in PBS) buffer (200 µL/well) at 37 • C for 2 h. The rabbit antisera against whole V. a. ammodytes venom (anti-Vaa) or its components (anti-H, anti-Atx or anti-AtnI 2 ) were added in 2-fold serial dilutions (100 µL/well) in duplicate and left overnight at RT. In the subsequent steps, incubation with HRP-anti-rabbit IgG (100 µL/well of a 10,000-fold dilution) at 37 • C for 2 h took place, followed by the addition of 100 µL/well of OPD solution (0.6 mg/mL with 0.015% H 2 O 2 (v/v) in citrate-phosphate buffer, pH 5.0). After 30 min incubation in the dark, the enzymatic reaction was stopped with 1 M H 2 SO 4 (50 µL/well) and the absorbance at 492 nm measured.

One-Dimensional SDS-PAGE (1-DE) and Western Blot
1-DE analysis of V. ursinii ssp. and V. a. ammodytes venoms was performed on 4-12% Bis-Tris precast gels with MES as running buffer, under both reducing and non-reducing conditions at 180 V for 50 min in an Xcell SureLock Mini-Cell, according to the manufacturer's instructions (Invitrogen, Carlsbad, CA, USA). A total of 40 µg of respective venom was applied in each well. Molecular mass standards were from Invitrogen. Protein staining was carried out with CBB R250. For Western blots (WB) following SDS-PAGE analysis, venoms were electro-blotted to the PVDF membrane (GE Healthcare, Buckinghamshire, UK) in an Xcell Sure Lock Mini Cell according to the manufacturer's procedure (Invitrogen, Carlsbad, CA, USA). The blocking was performed with 2% (m/v) BSA in PBS/T buffer at +4 • C overnight. The blotted membrane was incubated first with anti-Atx serum (diluted 20,000-fold) and then with HRP-anti-rabbit IgG (diluted 10,000-fold) at 37 • C for 1 h. The Enhanced ChemiLuminescence (ECL) plus Western Blotting Detection System was used for detection, according to the manufacturer's instructions (GE Healthcare, Buckinghamshire, UK).

Two-Dimensional Gel Electrophoresis (2-DE)
In the first dimension, isoelectric focusing of V. a. ammodytes and V. ursinii ssp. venoms was performed in a ZOOM IPG Runner Mini-Cell, using immobilised pH gradient (IPG) strips (7 cm long, non-linear pH 3-10) previously rehydrated with the protein sample (250 µg), according to the manufacturer's procedure (Invitrogen, Carlsbad, CA, USA). The following voltage protocol was applied: 200 V for 20 min, 450 V for 15 min, 750 V for 15 min and 2000 V for 5 h. The focused IPG strips were first reduced with 20 mM DTT, then alkylated with 125 mM IAA. Cys residues in proteins are namely not stable, which hinders their precise analysis. Reaction of Cys residues with IAA derivatizes them to S-carbamidomethyl-Cys residues, which are, however, stable. Each step was performed for 15 min at RT. In the second dimension, SDS-PAGE was performed as described under 5.7.2. For spot detection and quantification, the gel image obtained by an Image Scanner using LabScan 5 software was analysed by Image Master 2D Platinum 6.0 software (GE Healthcare, Amersham Biosciences, Amersham, UK). The 2-DE spot intensity was calculated by integrating the optical density over the spot area.

Mass Spectrometry (MS) Analysis
V. ursinii ssp. venom proteins, separated by 1-DE or 2-DE, were excised from the gel and digested with trypsin. The resulting peptides were extracted from the gel and analysed on an ESI-IT mass spectrometer, 1200 series HPLC-Chip-LC/MSD Trap XCT Ultra (Agilent Technologies, Waldbronn, Germany), as described in [54]. Prior to MS analysis, peptides were purified on self-prepared Stage-Tips. Proteins were identified using licensed software MASCOT version 2.1 (Matrix Science, London, UK), searching the "Snakes" (taxid 8750; 159,187 entries) protein database extracted from the non-redundant NCBI (National Centre for Biotechnology Information) protein databank in December 2017 and supplemented with the V. a. ammodytes venom gland cDNA transcripts library [55]. The following search parameters were used: parent ion error tolerance of 1.2 Da, fragment error tolerance of 0.60 Da, maximum of 2 missed cleavages allowed, peptide charges +2 and +3, methionine oxidation as variable and carboxamidomethyl cysteine as fixed modification, and an automatic decoy database search. The results were validated using Scaffold software (version 2, Proteome Software, Inc., Portland, OR, USA) with the following parameters: protein threshold of 95%, minimum one peptide per protein and a peptide threshold of 90%. The estimated Prophet false discovery rate was 0.2% for proteins and 5.6% for peptides. Additional peptide matches to identified proteins were obtained by performing an error-tolerant Mascot search and manual validation of the obtained results.
Supplementary Materials: The following are available online at http://www.mdpi.com/2072-6651/12/3/187/s1, Table S1: Identification of the structures of the Croatian V. ursinii ssp. venom proteins. Table S2: Identification of proteins with higher abundance in the venom of the Croatian V. ursinii ssp. snakes living in captivity (VuCro-c) than in the venom of those living in the wild (VuCro).