Blind and naive classification of toxicity by fish chromatophores.

Cellular and molecular pathways involved in the ability of animals to change color have been studied previously as biosensors and cytosensors of active and toxic agents, but such studies generally have been limited to just a few standardized agents. Here we describe the performance of cultured chromatophore pigment cells from the fin tissue of Siamese fighting fish as sensors of toxic agents under blind sampling conditions at the September 2002 EILATox-Oregon Workshop. Detection was accomplished by monitoring motor protein-mediated movements of cellular pigment in chromatophores at both the gross population level as well as in singly imaged cells. Pigment responses were recorded both during the exposure of chromatophores to each blind sample as well as afterwards when the cells were examined for after-effects by challenging them with clonidine, an adrenergic drug that induces standardized pigment movements. After recording all results and upon breaking the key to reveal the identities of the toxic agents, it was found that all of the toxic samples in the study had been distinguished accurately from non-toxic controls that were included among the blind samples. Furthermore, it was revealed that most of the toxic agents detected had never before been tested or calibrated against chromatophores, demonstrating that detection can be achieved under naive conditions that have not been optimized for the analysis of any particular toxic agent. Finally, by organizing the results into categories of pigment responses, a binary classification tree was generated that distinguished each toxic agent as having a distinct response pattern from the others. Thus, chromatophore-based cytosensors can discover toxicity in the absence of prior knowledge of the agent in question, and the categories of responses of the cells can be used to distinguish one toxic agent from another.

Lifson-Roig nucleation for alpha-helices in trifluoroethanol: context has a strong effect on the helical propensity of amino acids.

We have investigated the effect of substituting each of 19 common amino acids (excluding P) at the X position in the peptide acetyl-Y-VAEAK-TSXSR-VAEAK-NH(2). This very different peptide is of interest because, in previous work, we showed that nucleation in the strong alpha-helix-forming pentamers VAEAK was unable to propagate the helix structure through the sequence TSDSR, which is neither a helix-forming sequence nor a breaker, but is indifferent to helix formation. Substitution in the center of the indifferent sequence reveals an interesting measure of the helical propensity for the 19 amino acids. CD spectra were measured in various mixtures of buffer and 2,2,2-trifluoroethanol (TFE), and then analyzed for helix propensity of the amino acids using the Lifson-Roig model. However, the nucleation parameter in the Lifson-Roig model has never been measured for TFE. We have empirically found that the nucleation parameter for a solvent can be determined from the data normally used to determine only the propagation parameters. The results of the analysis of the CD show that most amino acids are excellent or good helix formers in 90% TFE, while amino acids D, W, F and G are poor helix formers for the indifferent pentamer sequence. The helix propensity of the 19 amino acids is quite different from the helix propensity measured in other peptide sequences, demonstrating the context dependence of this property. The results as a function of alcohol concentration confirm that the relative order of helical propensity of amino acids changes with solvent environment. Clearly, the prediction of alpha-helical secondary structure from protein sequence requires more than a single helical propensity for each amino acid.