This peptide class shows clear similarity with members of the GAST (giberellic acid stimulated transcript) and GASA (giberellic acid stimulated in Arabidopsis) protein families from Arabidopsis. In this conjuncture, both have been classified as members of the snakin/GASA family [3] and [22]. Mature snakin-1, from potatoes, is composed of 63 amino acid residues including 12 cysteine ones, which are involved in the formation

of six disulfide bonds [29]. Nevertheless, no information about the three-dimensional structure or their cysteine bonding pattern has been provided until now. The lack of structural confirmation of plant bactericidal peptides prevents Ganetespib in vitro more detailed classification of plant AMPs [6] and [22]. Furthermore, this structural knowledge can help us to avoid errors in AMP classification as was observed for plant defensins, which were classified as a subclass of thionins before their structural characterization [6] and [22]. Bearing this in mind, this paper describes the prediction of the three-dimensional structure of snakin-1 through the combination of ab initio and comparative molecular modeling together with a disulfide bond predictor. The snakin-1 sequence was taken from the UniProt database (UniProt: Q948Z4) and the mature sequence Metformin mouse was extracted according to the annotation (residues 26–88). The mature sequence

was used as a seed for searching against UniProt, through PHI-BLAST [1] and the pattern “CX3CX3CX7,11CX3CX2CCX2CX1,3CX11CX1,2CX11,14KCP” [31], where ‘X’ indicates a wild card, which can be filled up by any

of 20 natural amino acid residues, and the numbers between brackets indicate the number of repetitions of the prior character (i.e. ‘X7,11’ means that ‘X’ can be repeated seven to eleven times). The mature sequences from retrieved sequences were taken according to the annotation. The multiple sequence alignment was done in ClustalW 2 [33]. The snakin-1 mature sequence was submitted to the QUARK ab initio molecular modeling server [35] in order to create an initial model. Then the cysteine connectivity was predicted as follows: the cysteine residues involved in disulfide bonds in the initial model were replaced by serine residues and then this modified NADPH-cytochrome-c2 reductase sequence was submitted to the DiANNA 1.1 server [10], in order to predict the remaining cysteine pairs. The final model was constructed with MODELLER 9.10 [9]. The ab initio model was used as a template and the disulfide bonds were included using the method patch from the automodel class. Thus, 100 molecular models were constructed, and the final model was selected according to the discrete optimized protein energy (DOPE) scores. This score assesses the energy of the model and indicates the most probable structures.