*2.6. Individual Heterogeneity in Venom-Induced Lethality*

Finally, to link functional variation to toxicity to specific prey, we decided to investigate the ability of the venoms from each habitat to kill rodents (mice) or birds (chicks). We also quantify individual variation in venom-induced lethality within the same habitat. Venoms from five specimens from each habitat were injected into groups of six mice or six chicks, and the time of death of the animals was monitored over 48 h (Figure 6).

**Figure 6.** Time of death induced by individual *B. atrox* venoms in avian and murine models. Single doses of selected venom (200 μg) were injected i.p. into groups of six animals (mice or chicks). Next, the deaths were followed up every hour until 6 h after, then at 12, 24, and 48 h after venom injection. The data are representative of two independent experiments.

We observed a great deal of variation in the lethal activity of venoms from individual venoms from all groups analyzed. In venoms from the forest, degraded area, and pasture areas, mortality of experimental animals was observed from 1 to 2 h after venom injection, while in a number of cases, several animals survived through 48 h of observation. Similar results were observed for both the mice and birds, which matches the generalist diet preferences of *B. atrox*. There were some quite interesting patterns of toxicity in floodplain venoms. These venoms had similar activity to venoms from the forest, pasture, and degraded area when injected in mice. However, some venoms from the floodplain habitat (V8 and V16) induced deaths only a few minutes after its inoculation in chicks, a pattern that was not observed in mice. In addition, one venom (V12) was noticeably more lethal to chicks, killing half the animals in the first two hours and all animals of the group within 24 h, whereas in mice, the same venom induced the death of only two animals, 12 and 24 h after inoculation. However, not all floodplain venoms were highly lethal to birds. The V10 snake venom was much more lethal to rodents than to birds, killing all mice within 24 h, and only two chicks until the end of the experiment. The venom of snake V13 killed the same number of chicks and mice (only 2), although at different time intervals. Similar to the variations observed in the venom composition, the differential lethality to chicks observed with floodplain venoms was not due to the two distant floodplain spots of snake collection. Venoms that preferentially killed chicks were from snakes collected at Santarém (ATXV 8 and 16) or Oriximiná (ATXV 12 and 13).

#### **3. Discussion**

In this study, we analyzed the influence of different environments on the individual variability of venom samples obtained from *B. atrox* snakes captured in four different habitats of the Brazilian Amazon: forest, pasture, degraded area, and floodplain. Our analyses revealed clear differences in both venom composition and its biological activities among snakes from different habitats. However, major differences were also observed among venom samples collected from snakes in the floodplain habitat, a dynamic environment, subject to periods of annual drought and floods.

In a previous report involving *B. atrox* snakes, the variability in venom composition was attributed mainly to low-abundance proteins that would be a genetic reservoir for quick adaptive changes [2], while the most abundant isoforms from each toxin family were considered as "core toxins", conserved in venoms of most specimens and responsible for the major activities of the venom [2]. A similar observation was reported by Margres and collaborators [23], showing that, in *Crotalus adamanteus*, *Sistrurus miliarius*, and *Agkistrodon piscivorus* snakes, the differences in the expression levels were present mostly in the low-expression proteins. However, in the present study, both rare (low expression) and abundant (high expression) proteins show contribute to venom variation within and between snakes from different habitats.

For example, our data also showed that Peak 23 was composed mostly of Batroxrhagin, a P-III class SVMP from *B. atrox* venom isolated by our group [24], was widely conserved in all the 37 venoms analyzed, within and across all groups; this confirms a previous observation [2] that Batroxrhagin would act as a "core toxin" in *B. atrox* venoms, acting on important physiological targets of diverse prey types. In contrast, other abundant venom proteins showed less conservation and could represent more specific or even "adaptive" variation in these proteins. Clear examples of abundant fractions differentially expressed between the groups are fraction 21, which contains predominantly a PI-class SVMP and had lower expression in floodplain venoms; fraction 10, which contains mostly SVSPs and had higher expression in floodplain venoms; and fraction 20, which contains predominantly a PI-class SVMP and CTLs and is higher expressed in venoms from pasture compared to the other areas. Other fractions rich in SVSPs and CLTs were also were variable. These proteins participate to the hemorrhagic and coagulant activity of *B. atrox* venom and the observed differences in expression of the isoforms may alter procoagulant activity and

to promote differential lethality in rodents and birds, suggesting an important role of the environment in selecting for venom variation.

Individual variation in the abundance of SVMPs, SVSPs, and PLA2s could have arisen as a result of environmental differences between habitats. In our previous study [15], we compared the functional profiles of pools of venoms from *B. atrox* snakes from the same habitats and observed a less hemorrhagic and more procoagulant phenotype in the pool of venoms from floodplain snakes. The pool of venoms from the floodplain snakes was also significantly faster to induce clotting in several types of plasmas, including avian plasma [25]. Here, we used similar experimental approaches to investigate the individual variability within each group of snakes, introducing modifications in the tests of DC50, which were performed in the presence of calcium, to ensure the detection of coagulotoxins, dependent or not on this cofactor. In addition, our tests were performed with avian plasma as previously reported [26,27], which allowed us to construct better dose–response curves to assess the procoagulant effects of the individual venoms.

Venoms from *B. atrox* specimens from the floodplain showed low SVMP activity and higher SVSP catalytic activities. However, the hemorrhagic and procoagulant phenotypes observed in the pool of venoms were not consistent across all individual venoms collected from this environment. Some individual venoms such as from V5 and V8 snakes, showed both a potent procoagulant action on avian plasma and high hemorrhagic activities in mice. Moreover, the higher DC50 value observed, which corresponds to the less procoagulant venom, was precisely from a floodplain snake venom (V10), demonstrating that not all snakes from the floodplain have venoms that induce potent clotting of avian plasma, but the range of clotting activity was the highest in this environment.

Bernardoni and collaborators [3] analyzed *Bothrops neuwiedi* venom and found SVMPs isoforms that are functionally different and capable of affecting different targets in the hemostatic systems of birds, rodents, and humans. They suggested that some SVMPs are less selective, guaranteeing the action of the venom on different targets, while other isoforms are more selective, modulating the action of the venom for specific prey. More recently, we also investigated the effects of these SVMP isoforms on amphibian plasma (*Rinella marina*), demonstrating the coagulotoxic effects of these toxins on different types of animal plasmas [18]. In the present study, some fractions rich in SVMPs had a very variable distribution in the *B. atrox* venoms of all habitats and may be involved in the variability observed in the coagulotoxic activity observed in these individual venoms. In contrast, the major SVMP eluted in Peak 23 was very conserved in the venoms of all 37 specimens analyzed, confirming the previous assumption [2] that Batroxhagin is the core SVMP of *B. atrox* venom. On the other hand, the strong hemorrhagic action observed in some floodplain venoms (V5, V8, and V13), is compatible with a prominent increase of the fractions represent by Peaks 21 and 22 that shows height/area comparable to the ones observed in venoms from the other habitats. These peaks contain Atroxlisin-Ia, a PI-class SVMP that induces hemorrhage comparable to the P-III class enzymes [28].

During envenomation, various components present in snake venoms can act synergistically to cause the prey's organism to collapse, which usually results in rapid death. The speed to kill/immobilize prey is crucial for food acquisition of terrestrial viperids during hunting [29,30]. Previous works also report that lethality profiles of snake venoms can be variable in different animal models, such as mammals, birds, reptiles, and amphibians [5,31]. For this reason, we included the functional characterization of the individual venoms in two different animal models, birds (chicks) and mammals (mice). We found substantial differences in venom activities in both, the number of test animals dying, and the time of deaths in the two models. Forest venoms showed similar lethality profiles in rodents and birds across individual venom samples, while venoms from pasture and degraded areas were slightly more lethal to rodents. Of particular note, venoms from floodplain snakes were highly variable according to their ability to prey on mammals or birds. For example, the venoms of the snakes V8 and V16 were more toxic to chicks, inducing deaths within few minutes in birds, even more quickly than the venom of *B. insularis*, a snake with a diet specialized in birds [32]. The venom of the snake V10 however was more lethal to rodents. Interestingly, the venoms with the highest procoagulant activity are more lethal to birds while V10 venom, the weakest procoagulant, killed predominantly mice. The limitations imposed by the small available quantities of venom prevented using the same individual venoms in all the functional tests performed or to carry out dose– response analyses, hindering a more direct assessment of the relationship between lethality and the other toxic activities evaluated.

Nevertheless, our findings strongly suggest that the procoagulant activity of the floodplain venoms contributes significantly to their lethal effects in rodents and birds. The most striking and differential characteristics of the floodplain venoms are its potent effects on coagulation in different types of plasma. The venom of the snake V10 (from floodplain habitat) was more efficient in killing rodents than birds, and this same venom had the highest EC50 value on avian plasma. A possible explanation for the remarkable action of some floodplain venoms on hemostasis and velocity to induce death in birds could be directly linked to the floodplain environment, where these snakes were captured. Floodplains are periodically flooded by the lateral overflow of waters rich in sediments from the Amazon River. The floodplain regions alternate cyclical periods of drought and flood, remaining flooded for a few months during the year, this seasonal variation being driven by flood and precipitation seasons [33]. Balancing selection pressures could result in greater venom heterogeneity, as evidenced by the differential lethality to rodents or birds observed in some floodplain venoms, while the snake venoms from the other habitats presented a more homogeneous lethality pattern.

In line with our findings, Smiley-Walters et al. [31] showed that variation in the venom of *Sistrurus miliarius* at the population level has an adaptive role in terms of toxicity to prey. Testing adaptive hypotheses requires careful analysis of phenotypic characters whose variation has clear functional consequences. Venoms directly affect the ability of an individual snake to immobilize and kill its prey [31,34–37]. For *B. atrox*, a snake with a generalist diet that includes arthropods, frogs, lizards, birds, and small mammals [38,39], the functional diversity of the venoms could represent a local adaptation of individual venoms in each population to different sets of prey. Within an ecological context, the nature of the interactions between species, including prey–predator relationships, may change among populations [40,41], especially for species with a wide geographical distribution. of the functional versatility and diversity of *B. atrox* venoms may explain its wide distribution throughout all Amazon regions [7], reflectinghow this species adapts to prey communities in different habitats.

#### **4. Conclusions**

Individual venom variation of *B. atrox* snakes, captured in different habitats within the Brazilian Amazon, support the idea that dynamic environments may select for more variable venoms. Such differential distribution of protein isoforms leads to differential function and toxic effects on different prey. The fact that the greater heterogeneity in terms of composition/toxicity of the venom was found in snakes from the floodplain habitat suggests that balancing selection for expression of different isoforms could be enacted by the drastic seasonal changes (drought/flood) present in this type of environment. Variable selective pressures in different habitats may likely exert localized impacts on the venom phenotype in a species with a wide geographic distribution, such as *B. atrox*.
