Evaluation of antiproliferative activity of exudates of Salvia species by means of the Helianthus tuberosus classical in vitro growth model system.

Salvia species are used as antitumor agents in folk medicine in various parts of the world [1-3] and various extracts and isolated compounds have been shown to possess cytotoxic and antiproliferative activity [4-7]. In this work we evaluated the antiproliferative activity of the secretion material of several Salvia species, namely Salvia elegans Vahl. (1), S. buchananii Hedge (2), S. miniata Fernald (3), S. corrugata Vahl. (4), S. somalensis Vatke (5), S. namaensis Schinz (6), S. wagneriana Polak (7), S. confertiflora Pohl. (8), S. x jamensis J.Compton (9), S. chamaedryoides Cav. (10), S. cacaliaefolia Benth. (11), S. fallax Fernald (12), S. aurea L. (13), S. ianthina Otto & A.Dietr. (14), S. cinnabarina Mart. & Galeotti (15), S. scabra Linn.fil. (16), S. indica Heyne ex Roem. & Schult (17), by means of a plant callus growth model. As a matter of fact, long-term in vitro cultures of Helianthus tuberosus dormant parenchyma explants constitute a classical growth model that can be used to evaluate the proliferative or antiproliferative and/or cytotoxic effects of different mixtures of chemicals [8]. Explants of dormant tubers were cultured in vitro with 10 g/ml of each secretion mixure suitably diluted in CH3OH or DMSO. Each test was repeated with the same medium added with 10 M 2,4-D. After four weeks of culture, the explants were collected and weighed (fresh weight and dry weight). The obtained data were analyzed by a method based on an unsupervised learning approach, in which Self-Organizing Maps (SOMs), or Kohonen Networks [9], have been used, and clustered thanks to the k-means methodology. The SOMs realize a projection of the prototypes, representing the data set, from the d-dimensional input space onto a low-dimensional (usually 2-dimensional) grid. The grid, i.e. “the map”, can be considered as a convenient visualization surface to illustrate the different features of the data set [10, 11]. The algorithm is versatile and flexible, because it does not depend on the data set content, but on the structure of the input data. A set of replicates and features for each species is necessary, in order to have the correct data structure for the analysis procedure and to have a significant quantity of data input. The ranking of the antiproliferative activity of the various species of Salvia has been carried out through the definition of a new index, the BGP index. The results indicated that only one species, S. cacaliaefolia, showed antiproliferative acitivity also in the presence of the synthetic auxin able to induce breaking of dormancy and cell proliferation