In line with drestie's answer, the theory proposed by Margulis, et al is termed serial endosymbiosis theory (SET), which you probably already know. My understanding of the theory, (greatly simplified) is that possible evolution of eukaryotes from prokaryotes involved the symbiotic union of several previously independent ancestors. According to the theory, these ancestors included a host cell, an ancestor of mitochondria, an ancestor of chloroplasts, and, more controversially, a prokaryote that brought with it the structures that today provide cellular motion.(Source: The Serial Endosymbiosis Theory of Eukaryotic Evolution
by Jeremy Mohn)
An important work seeming to support Masrgulis definition of the SET theory occurred in the laboratory of Kwang W. Jeon, a biologist at the University of Tennessee. Jeon witnessed the establishment of an amoeba-bacteria symbiosis in which new bacterial symbionts became integrated in the host amoeba (Jeon 1991). In 1966, when the bacteria first infected the amoebas, they were lethal to their hosts. However, as time progressed, some of the infected amoebas survived and became dependent on their newly acquired endosymbionts within a few years. Jeon was able to prove this dependency by performing nuclei transplants between infected amoebas and amoebas lacking the bacteria. If left alone, the hybrid amoebas died in a matter of days. Yet if he reinfected these hybrids with the once-lethal bacteria, the amoebas recovered and once again began to grow (Margulis and Sagan 1987). So, nothing deffinintive but additional support for the theory relies on experiments in progress, in my opinion...