Gene loss is the most important event in the process of genome reduction that appears associated with bacterial endosymbionts of insects. These small genomes were derived features evolved from ancestral prokaryotes with larger genome sizes, consequence of a massive process of genome reduction due to drastic changes in the ecological conditions and evolutionary pressures acting on these prokaryotic lineages during their ecological transition to host-dependent lifestyle. In the present thesis, the process of genome reduction is studied from different perspectives. In the first chapter, genome rearrangements have been studied in a set of 31 complete γ-proteobacterial genomes that includes five genomes of bacterial endosymbionts of insects. This is carried out by comparing the order of a subset of 244 single-copy orthologous genes presents in all the genomes and calculating the number of inversions and breakpoints between each genome pair. This reveals that inversions were the main rearrangement event in γ-proteobacteria evolution, with a progressive increase in the number of rearrangements with increased evolutionary distance. However, significant heterogeneity in different γ-proteobacterial lineages was also detected, with a significant acceleration in the rates of genome rearrangements in bacterial endosymbionts of insects at initial stages of the association. In the second chapter, the structure and functional capabilities of Sodalis glossinidius has been studied. S. glossinidius is the secondary endosymbiont of tsetse flies, and it´s at very initial stages of genome reduction process. It´s genome is experiencing a massive process of gene inactivation, with 972 pseudogenes (inactivated genes) that were described but not annotated in the original annotation of the genome. In this chapter, a complete functional re-annotation of this genome was carried out, that includes the characterization of 1501 pseudogenes though analysis of S. glossinidius intergenic regions. A massive presence of CDSs related with mobile genetic elements and surface proteins were detected, being also the functional classes most affected by pseudogenization. The reconstruction of the metabolic map of S. glossinidius revealed a functional profile very similar to that of free-living enterics, with inactivation of L-arginine biosynthesis pathway, whereas the comparison with Wigglesworthia glossinidia (tsetse primary endosymbiont) reveals possible cases of metabolic complementation between both tsetse endosymbionts at thiamine, coenzyme A and tetrahydrofolate biosynthesis level. Finally, in the third chapter of the thesis, the complete reductive evolution process associated with S. glossinidius was studied from a systems biology perspective through the reconstruction of their genome-scale metabolic networks at different stages of this process and the prediction of their internal reaction fluxes under different external conditions through Flux Balance Analysis. This revealed the decisive role of the pseudogenization of genes involved in L-arginine and glycogen biosynthesis and specially the pseudogenization of the key anaplerotic enzyme phosphoenolpyruvate carboxylase in the ecological transition to a host-dependent lifestyle experienced by S. glossinidius. A progressive decrease in network robustness to gene deletion events and to changes in particular reaction fluxes were detected. Finally, reductive evolution simulations over the functional network of S. glossinidius under different external conditions revealed a higher plasticity in minimal networks evolved in a nutrient-rich environment, and allow defining different sets of essential and disposable genes based on their presence or absence in minimal metabolic networks. These essential genes had more optimized patterns of codon usage and more restricted patterns of sequence evolution than disposable genes that could be lost without affecting the functionality of the network. However, lineage-specific estimates of dN and dS in S. glossinidius and Escherichia coli revealed that common features of ancient bacterial endosymbionts like acceleration in the rates of sequence evolution and the loss of adaptative codon usage were starting to affect S. glossinidius evolution.En esta tesis doctoral, el proceso de reducción genómica característico de bacterias endosimbiontes de insectos ha sido estudiado utilizando diferentes aproximaciones computacionales basadas en la genómica comparada y la biología de sistemas. Por un lado, las dinámicas de reordenaciones genómicas han sido estudiadas en un subconjunto de 31 genomas completos de γ-proteobacterias que incluyen 5 genomas completos de endosimbiontes bacterianos de insectos, revelando una aceleración significativa de las tasas de reordenaciones en estos genomas en etapas iniciales del proceso de reducción. Posteriormente, el genoma de Sodalis glossinidius, el endosimbionte secundario de la mosca tsétsé, fue re-anotado con el objetivo de evaluar el impacto de los procesos de inactivación génica y proliferación de elementos genéticos móviles en etapas tempranas del proceso de reducción, asi como su impacto sobre las capacidades funcionales de la bacteria en el contexto ecológico de su coexistencia con el endosimbionte primario ancestral Wigglesworthia glossinidia. Finalmente, el proceso completo de reducción genómica en S. glossinidius ha sido estudiado a través de la reconstrucción de su red metabólica a diferentes etapas de este proceso y su análisis funcional mediante Análisis de Balance de Flujos, evaluando la robustez de las redes frente a sucesos de deleción asi como las dinámicas evolutivas de genes esenciales y no esenciales en base a su presencia en redes mínimas evolucionadas a partir de la red funcional. Este análisis permitió identificar sucesos de inactivación génica con efectos drásticos sobre las capacidades funcionales del sistema como los genes implicados en la biosíntesis de arginina y glicógeno, y especialmente la inactivación de la enzima fosfoenolpiruvato carboxilasa, asi como una disminución progresiva de la robustez de las redes frente a diferentes sucesos mutacionales asociada al proceso de pérdida génica. Finalmente, simulaciones de evolución reductiva sobre la red funcional bajo diferentes condiciones de entorno ha permitido definir conjuntos de genes esenciales y delecionables en base a su presencia o ausencia en las redes mínimas producto de las simulaciones, revelando una mayor conservación a nivel de secuencia y un uso de codones más optimizado en genes esenciales frente a genes cuya pérdida no afecta a la funcionalidad del sistema.