30
evasion of the host immune system. In addition, many antibiotics
are known to increase mutations rates in bacteria, which can
increase the evolvability of bacterial pathogens.
DNA mutations have been studied for more than a century
but never directly observed. Initially, the occurrence of mutations
in genomes of organisms was inferred from the detection of muta-
tional events in genetic markers using selection based on gain or
loss of a function. Fluctuation test, one of the first quantitative
tests for measuring mutations is based on this property [ 1 , 2 ]. The
fraction of the new phenotypic variants in a growing population is
measured and the number of cycles of cell division estimated. This
allows for estimating mutation rates. However, these estimates are
biased due to the fitness effects of mutations. For example, variants
carrying deleterious mutations are often affected for growth and
have longer generation times compared to the nonmutated cells.
As a consequence, they are outgrown in the final population by the
nonmutated cells that divide more rapidly. The rates of variants
with deleterious mutations obtained by fluctuation analysis are
therefore underestimated. In addition, fluctuation analysis is appli-
cable to minority of DNA mutations, which produce an identifi-
able phenotype, i.e., lethal, synonymous, and some deleterious
mutations cannot be detected. Extrapolating to whole genomes
from data obtained from few loci is likely to be inaccurate because
mutation rates vary between different chromosomal sites due to
the differences in base composition, transcriptional activity and
variations in DNA repair efficiency [ 3 ].
More straightforward strategy for identifying and quantifying
mutations is DNA sequencing. DNA sequencing gives precise
information on the mutation site and the mutation type of the vast
majority of DNA mutations. It became very rapid and inexpensive.
For example, next-generation DNA sequencing generates several
billions of nucleotides of DNA sequence in less than 2 weeks [ 4 ].
However, 1% of sequenced bases are identified incorrectly due to
errors introduced during sample preparation, DNA amplification,
and image analysis [ 4 ]. This is problematic for detecting the rare
mutations and those with strong negative effect on the fitness
when sequenced bacterial populations are heterogeneous, which is
generally the case. Single-cell genome sequencing can solve this
problem. However, sufficient amount of DNA for this approach is
obtained by massive whole-genome amplification, which produces
many artifacts. Finally, because cells must be killed to extract DNA,
it is impossible to perform single-cell time-course analysis using
DNA sequencing methods.
The vast majority of spontaneous mutations are due to the
DNA polymerase errors occurring during DNA replication. While
the error rates of replicative DNA polymerases are of the order ofMarina Elez et al.