3.2 Mambaquaretins form a distinct group of Kunitz-type snake toxins
Amino acid sequences of Kunitz peptides were aligned according to MQ1 with representative sequences of dendrotoxins and three BPTI sequences (P00974, P00975 and P04815, Fig. 3A). This alignment was used to reconstruct the phylogenetic relationships of Kunitz-type snake toxins. The phylogram showed that all nine MQs constitute a well-supported monophyletic group, sister to ion-channel modulators isolated from mambas (so-called dendrotoxins). MQs are organized into two sub-groups: MQ1 to MQ6 are very closely-related to each other, whereas MQ7 to MQ9 are more divergent (Fig. 3B, red background). MQ8 is a natural variant of the dendro-B (P00983, named here MQ7), a weak trypsin inhibitor (Strydom and Joubert, 1981). Both diverge from MQ1 by 19 mutations. MQ9 is the most original sequence with 21 mutations compared to MQ1 and a one residue shorter loop 1 (Fig. 3A). All MQs are a sister group to a branch composed of seven dendrotoxins (Fig. 3B, blue background). The peptides DTX-R55 (Q7LZS8) and DTX-E (P00984) have never been characterized and are probably active on serine proteases as they possess the dyad Lys15-Ala16. The five other dendrotoxins are potassium and calcium channel inhibitors, which are not active on V2R up to 10 µM (Fig. 2D).
The next group is composed of two cobra toxins (Fig. 3B, green background). P19859 is an uncharacterized toxin from the Indian cobra Naja naja (Shafqat et al., 1990). Q5ZPJ7 is a weak chymotrypsin inhibitor from the Chinese cobra Naja atra (Zhou et al., 2004). The synthetic versions of P19859 and Q5ZPJ7 (Supplementary Fig. 4) bind V2R (Fig. 2D) and antagonize cAMP production (Fig. 2E, Table 1) with Ki and Kinact equivalent to those of other MQs. Injected in rats, both cobra toxins raise diuresis (2-fold increase) with a concomitant decrease of urine osmolality, demonstrating an aquaretic effect (Table 2). The other sequences used to build the phylogram belong for the most part to viper or Australian snakes. Among them, we checked the activity of five selected toxins (E7FL11, B4ESA3, D5J9Q8, F8J2F6 and C1IC51, Supplementary Fig. 4) distributed all along the dendrogram and having a dyad composition compatible with V2R activity (Asn15-Gly16, Asn15-Ala16 or His15-Gly16). None of them were active on V2R up to 10 µM (Fig. 1D).
The nine MQs constitute a new phylogenic Kunitz group associated with a V2R activity. The comparison between the 11 V2R-active toxin sequences and the non-active V2R Kunitz peptides highlighted 5 positions that might play positive roles in binding to the V2R (Fig. 3A, supplementary Fig. 5). The positions 17, 28, and 44 are strictly conserved in V2R-active Kunitz and highly variable in non-active V2R Kunitz. The position 34 is occupied by a threonine residue for 10 toxins and by an asparagine for one. It is highly variable in non-active V2R Kunitz. Finally, the position 10 is basic (Lys) or neutral (Ala, Ser) in V2R-active Kunitz whereas it is mostly acidic in other Kunitz groups (Fig. 3A, supplementary Fig. 5).