Résumé : There is an urgent need to develop reliable and cost-saving methods to select pre-clinically new drug candidates with original mechanism for cancer therapy. Previous results have shown that IR spectra of cancer cells exposed to various drugs provided a global signature of all the metabolic changes induced by the treatments. In this thesis, we attempted to develop a selection criterion – based on FTIR spectroscopy – for potential antitumor compounds according to their mechanism of action.

In chapter III, it was demonstrated that spectral variations in IR spectra of cancer cells induced by a treatment can be correlated to the mechanism of the drug. Human prostate cancer PC-3 cells were exposed to 7 well-described anticancer drugs belonging to 3 distinct classes. Each class is characterized by a unique mode of action. Drugs known to induce similar types of metabolic disturbances appear to cluster when spectrum shapes are analyzed. Chapter IV generalized the results obtained on PC-3 cells with six other cell lines. We showed that the spectral signatures of drug effects are mainly independent of the cell line. Chapter V indicated that, while the cell cycle phase influence IR spectra of cells, the drug spectral signature was dominated by global metabolic modifications and not much by the cell cycle perturbations due to this drug.

Chapter VI and VII focused on lipids. While the precise identification of particular molecules is particularly complex with IR spectroscopy, we attempted to extract more precise information and to assign spectral variations to specific changes in lipids. IR spectra of lipids contain very interesting details on their nature and structure. We achieved to build a tool which quantifies five major lipid classes in complex mixtures such as total lipid cell extracts. However, based on this tool, the treatments used do not induce any variation in the lipid cell composition (for five classes).

Finally, in chapter VIII, we applied the method developed previously on a new potential class of anticancer molecules: the polyphenols. A global method was particularly interesting as the development of therapy using these compounds is hampered by the complexity of the multiple anticarcinogenic mechanisms of these molecules. We have noticed the similarities and discrepancies among 3 very close synthetic molecules and the observations were coherent with previous biological data. We also compared them with 3 natural molecules already in clinical phase for treatment of various cancers.

In conclusion, we developed an objective classifier for potential anticancer drugs based on their global effects on cancer cells. Applied to a larger scale, this method could constitute a first step in the screening method to select drugs with original mode of action.

There is an urgent need to develop reliable and cost-saving methods to select pre-clinically new drug candidates with original mechanism for cancer therapy. Previous results have shown that IR spectra of cancer cells exposed to various drugs provided a global signature of all the metabolic changes induced by the treatments. In this thesis, we attempted to develop a selection criterion – based on FTIR spectroscopy – for potential antitumor compounds according to their mechanism of action.

In chapter III, it was demonstrated that spectral variations in IR spectra of cancer cells induced by a treatment can be correlated to the mechanism of the drug. Human prostate cancer PC-3 cells were exposed to 7 well-described anticancer drugs belonging to 3 distinct classes. Each class is characterized by a unique mode of action. Drugs known to induce similar types of metabolic disturbances appear to cluster when spectrum shapes are analyzed. Chapter IV generalized the results obtained on PC-3 cells with six other cell lines. We showed that the spectral signatures of drug effects are mainly independent of the cell line. Chapter V indicated that, while the cell cycle phase influence IR spectra of cells, the drug spectral signature was dominated by global metabolic modifications and not much by the cell cycle perturbations due to this drug.

Chapter VI and VII focused on lipids. While the precise identification of particular molecules is particularly complex with IR spectroscopy, we attempted to extract more precise information and to assign spectral variations to specific changes in lipids. IR spectra of lipids contain very interesting details on their nature and structure. We achieved to build a tool which quantifies five major lipid classes in complex mixtures such as total lipid cell extracts. However, based on this tool, the treatments used do not induce any variation in the lipid cell composition (for five classes).

Finally, in chapter VIII, we applied the method developed previously on a new potential class of anticancer molecules: the polyphenols. A global method was particularly interesting as the development of therapy using these compounds is hampered by the complexity of the multiple anticarcinogenic mechanisms of these molecules. We have noticed the similarities and discrepancies among 3 very close synthetic molecules and the observations were coherent with previous biological data. We also compared them with 3 natural molecules already in clinical phase for treatment of various cancers.

In conclusion, we developed an objective classifier for potential anticancer drugs based on their global effects on cancer cells. Applied to a larger scale, this method could constitute a first step in the screening method to select drugs with original mode of action.

Afin d’améliorer les thérapies contre le cancer, il devient actuellement cruciale de développer une méthode pour améliorer la sélection préclinique de nouvelles molécules, potentiellement anticancéreuses. Des publications précédentes ont mis en évidence que les spectres infrarouges de cellules cancéreuses exposées à différents agents thérapeutiques fournissent une empreinte globale de l’ensemble des changements métaboliques induit par ce médicament. Dans cette thèse, nous proposons d’utiliser la spectroscopie infrarouge pour mettre au point un critère de sélection basé sur le mode d’action des agents anticancéreux. Plusieurs aspects de la technique ont été investigués. Nous avons d’abord démontré la possibilité d’utiliser les spectres infrarouges de cellules cancéreuses de prostate PC-3 traitées avec 7 drogues pour classer ces molécules selon leur mode d’action. Nous avons ensuite reproduit les résultats obtenus sur PC-3 avec 6 autres lignées cellulaires et montré que la signature spectrale obtenue était largement indépendante de la lignée. Par la suite, nous avons étudié si l’effet sur le cycle cellulaire induit par de nombreuses molécules anticancéreuses, pouvait expliquer certains changements spectraux observés suite au traitement. Nous avons pu montrer que la majorité des variations spectrales n’étaient pas liées à une perturbation du cycle cellulaire. Nous nous sommes ensuite concentrés sur une classe de molécules en particulier: les lipides. Après avoir mis en évidence l’ensemble des informations contenues dans un spectre infrarouge de lipides, nous avons mis au point un modèle permettant de quantifier 5 classes de lipides dans des mélanges complexes tels que des extraits lipidiques provenant de cellules. Néanmoins, aucune variation du contenu en ces 5 classes de lipides n’a été observée pour les traitements utilisés dans cette étude. Enfin, nous avons appliqué la méthode mise au point dans cette thèse à une classe de molécules prometteuses : les polyphénols. Une approche globale s’avère particulièrement intéressante pour ces composés étant donné qu’ils présentent des mécanismes anticancéreux multiples et complexes. Nous avons comparé 3 molécules naturelles en phase clinique pour le traitement de certains cancers et 3 molécules synthétiques présentant une structure très proche. Par notre méthode, nous avons mis en évidence certaines similarités et différences de ces 6 molécules en termes d’effets globaux sur les cellules. En conclusions, nous avons développé un outil objectif de classification pour les molécules anticancéreuses potentielles, basée sur le mécanisme global des composés. Appliquée à plus large échelle, cette méthode pourrait constituer une première étape permettant de sélectionner les molécules avec un mode d’action original.