Caractérisation des souches bactériennes par électrophorèse des protéines totales. Influence de l'étalonnage des gels sur le coefficient de proximité taxonomique

Characterization of bacterial strains by electrophoresis of total proteins. Influence of gel calibration on taxonomic proximity coefficient

Cemagref, division qualité des eaux, 14, avenue de Saint-Mandé, 75012 Paris

Résumé

La taxonomie numérique basée sur la séparation des protéines bactériennes en électrophorèse SDS requiert un étalonnage préalable des gels. Parmi les méthodes testées, la régression hyperbolique est la mieux adaptée à l'étalonnage des gels de protéines en SDS. Pour la comparaison des profils protéiques en taxonomie numérique, il semble important de pouvoir disposer d'un jeu de marqueurs étendu vers les hauts poids moléculaires.

Abstract

Numerical taxonomy based on SDS electrophoretic bacterial protein separation is a powerful tool for strain characterisation. Before any mathematical comparison, it is necessary to proceed at a gel calibration. Influence on similarity coefficient of three calibration modes calculated from every profile points: by segments, by Lagrange polynomial, by third order polynomial regression, have been compared. It appears that: calibration by segments and by third order polynomial regression are very efficient when used with a 5 standard protein reduce set. However, high molecular weight standard proteins appears to be influent on the similarity coefficient values. The different calibration curves obtained with standard proteins were compare. Calibration by segments and by Lagrange polynomial are less regular when the number of standards increase. With 10 standard proteins Lagrange polynomial is not any more usable. On the contrary, calibration by third order polynomial regression is getting more precise when used with an enlarge set of standards. Calibration curves obtained by robust modified hyperbola were compared. They appeared to be rather insensitive to the number of standard proteins used. This method present the lowest differences between known and calculated molecular weights. It seems to be the best calibration mode for SDS protein gel electrophoresis. However, it must be used with a large range molecular weight standard set including high molecular weight protein standards for numerical taxonomy purpose.