54 4 Rational Efforts to Streamline the Escherichia coli Genome
irrelevant under defined laboratory or industrial conditions. It was estimated
that even under poor nutritional conditions, only 75–80% of the genes have
detectable activity [35, 36]. Moreover, the genome is loaded with prophages and
transposable elements (mostly residing on the accessory genome) (Figure 4.1),
which, although occasionally contribute to fitness under certain conditions,
could be viewed as dispensable genomic parasites. Finally, the fact that a large
proportion of the genes lack a known function, despite of decades of E. coli
research, suggests that they may be unimportant.4.4 Random versus Targeted Streamlining
There are natural organisms possessing a nearly minimal number of genes, often
in the range of 400–600. These organisms are typically obligate host-associated
bacteria, and phylogenetic studies indicate that the small gene sets evolved from
much larger genomes through massive loss of genes no longer required in the
intracellular environment. This suggests that nutrient-rich, constant environ-
ment and low population size favor genome reduction. It was estimated that the
free-living ancestor of Buchnera has lost 75% of its genome since it switched to
an endosymbiotic lifestyle approximately 200 million years ago [37]. As an anal-
ogy, culturing a population of cells by serial passage under conditions favoring
loss of genetic material (limiting nutrients for DNA synthesis, periodic popula-
tion bottlenecks, defects in mismatch repair) could lead to smaller genomes
[38–40]. Such an undirected procedure would have several advantages. First, no
a priori knowledge of the genome is required. Second, high-fitness, rapidly grow-
ing cells are automatically selected. Third, this approach allows the exploration
of different orders and combinations of deletion events. Unfortunately, since
DNA synthesis requires little energy [22], there is no strong selection for smaller
genome per se. Experimental work along this line so far has not resulted in major
genome reduction. The 0.05–2.5 bp per genome per division deletion rate,
obtained in an experimental evolution test with Salmonella enterica [41], is too
low for practical application. Similarly, a long-term laboratory evolution experi-
ment applying serial passage of E. coli cells in a single medium yielded only a few
deletions totaling 38 kb in 20 000 generations [42]. Clearly, an experimental
approach based on selectable deletion formation is needed for satisfactory results
on a realistic time scale. An interesting approach partially fulfilled this require-
ment. Using an engineered, composite transposon, serial random deletions were
created in E. coli [43]. Transposon-inserted cells were selected in each cycle
by their antibiotic resistance. Subsequent induction of an “inner” transposon
resulted in deletion (or inversion) of a neighboring genomic segment along with
the loss of the resistance cassette, and a new cycle could be initiated. Unfortunately,
there are some drawbacks: only one-fourth of the transposon-inserted cells
undergo the proper rearrangement, replica plating is needed to find the proper
clones, small deletions are favored, and the construct leaves a 64-bp exogenous
sequence in the genome in each cycle. In conclusion, due to lack of an adequate
deletion selection scheme, random deletion methods are currently not applied to
genome streamlining. Instead, targeted genome reduction schemes are favored.