Endosymbiosis is important for three reasons. The first is that it provides organisms with biochemical capabilities that they otherwise would not have access to, such as eukaryotes and their display of cellular respiration as well as photosynthesis. Second, by carrying symbiotic organisms within cells there is a potential for much greater metabolic integration than would otherwise be the case with less physically linked organisms, potentially opening up otherwise unavailable niches, plus provides for the vertical co-transmission of host and symbiont, which is essential for such sophisticated coevolution. Lastly, and potentially of huge relevance, endosymbiosis provides a key mechanism of horizontal gene transfer, where genetic material flows from endosymbiont to host cell (extra-nuclear to nuclear genome). This latter phenomenon serves as the emphasis of this section.

The mechanism of this movement is equivalent to that described in the "You are what you eat" hypothesis. That is, genetic material that finds its way into a cell can, by accident, find itself in a cell's cytoplasm (i.e., as opposed to within an endomembrane member such as a phagolysosome) and, from there, can move into the nucleus where integration into the nuclear genome can then occur. The difference between the movement from endosymbionts, especially endosymbionts that are vertically transmitted within the germ line, and "You are what you eat" is that endosymbionts are present all the time with a cell and potentially in large numbers. Thus, if anything is likely to accidentally deliver DNA to a cell's cytoplasm, it is the endosymbiont. The consequence is that movement of endosymbiont DNA into the nuclear genome is surprisingly efficient. (Note also that we can also view this process as equivalent to the movement of DNA into cell chromosomes in general where any DNA that finds its way into the cytoplasm – such as by transformation, transduction, or conjugation – has a much greater potential of integrating into the host genome versus DNA that has failed to gain access to the cytoplasm.)

There exist four tendencies associated with endosymbiont evolution following their acquisition. These combine tendencies towards greater biochemical integration, on the one hand, with those of horizontal gene transfer on the other.

  1. The first is that endosymbionts are expected to become streamlined once they are acquired, losing functions that are redundant with those of their host's or otherwise unneeded within the intracellular environment. The pace of this loss of genetic material is probably aided by the lack of endosymbiont access to other bacteria, resulting in genomic erosion from genetic drift.

  2. The second mechanism is one of integration with the host metabolism, i.e., coevolution between the host and the endosymbiont genome. The result of this integration is further reduction in conflicts between host and endosymbiont as well as increases in the efficiency of metabolic integration.

  3. The third mechanism is the migration of endosymbiont genes to the nuclear genome, where these then can be expressed as nuclear genes and which as proteins migrate back to the endosymbiont to replace otherwise endosymbiont-expressed gene products, thereby rendering endosymbiont genes functionally redundant.

In combination these mechanisms give rise to a reduction in endosymbiont genome size and coding capacity, potentially to the point where in certain cases endosymbionts have ceased to even carry a genome. Lastly,

  1. Genes that migrate from the endosymbiont to the nuclear genome may take on functions that effectively are independent of endosymbiont functioning or endosymbiont-associated metabolism.

In short, the host cell's nuclear genome becomes a hybrid between that of the original host cell and that of the endosymbiont, with resultant net increase in the sophistication of the nuclear genome in combination with a net decrease in the sophistication of the endosymbiont genome. In fact, the endosymbiont's impact on the nuclear genome can be so great that it constitutes a complication in sequencing nuclear genomes of organisms containing relatively recently acquired endosymbionts, since it can be difficult to distinguish between DNA that has migrated into the nuclear genome versus bacterial and/or endosymbiont DNA that instead is simply contamination acquired during prepping for sequencing (Hotopp et al., 2007) .

Ultimately, not all genes from all endosymbionts migrate to the nucleus but instead are retained by the endosymbiont genome, which begs the question, why don't all endosymbiont genes migrate to the nucleus? This question is especially pertinent given that the endosymbiont must retain a complete translation apparatus – RNA polymerase, tRNAs, ribosomes – if it is going to express even just one endosymbiont protein-coding gene, so why then bother? Two obvious general explanations are that either the retained genes are crucial to endosymbiont survival or functioning as an organelle, but cannot otherwise be imported into the endosymbiont, or that the endosymbiont survival or functioning is enhanced by the potential to regulate genes from within. These can be described more formally as a hydrophobicity hypothesis and, for mitochondria in particular, a redox-regulation hypothesis (van der Giezen and Tovar, 2005) . There "hydrophobicity" refers to the barriers to importation of the protein following hypothetical nuclear expression while "redox" refers to the oxidation-reduction character of the electron transport system, and regulation of the synthesis of its constituents. Notwithstanding the reasons why, there appear to be limits to gene loss by most, though perhaps not all endosymbionts, resulting in the retention of cytoplasmic inheritance associated with organelles that once were themselves free-living cells.