(Normand et al. 2007). This genus of N 2 -fixing plant
symbionts belongs to the Actinobacteria. The genome size
of the three strains varies considerably: 5.43 Mb (Frankia sp.
HFPCcI3), 7.50 Mb (Frankia alni ACN14a), and 9.04 Mb
(Frankia sp. EAN1pec). Comparative genomic studies
revealed a core genome composed of 35–50 % of the
genes present in these strains, whereas 578, 1,563, and
1,289 genes were specific to each strain.
Most studies based on gene taxonomic distribution
provide similar results. Under the assumption that specific
genes result from recent HGT, estimated HGT rates are high.
Extrapolating these estimated rates to larger evolutionary
scales would lead to the conclusion that the amount of
genes acquired by a lineage during its evolution is several
orders of magnitude greater than the complete genome size,
meaning that all genes in a given genome have been
exchanged several times during evolution. Sequence similarity analyses (cf. Sect. 12.3.1) performed in the 1990s
suggest also high HGT rates among prokaryotes, especially
among organisms living in similar ecological niches, such as
hot ecosystems. Accordingly, HGTs are considered as very
abundant and even to represent the major evolutionary
driving force in prokaryotes (Lan and Reeves 2000).
Although more recent studies suggest that the HGT rates
estimated by these methods are likely overestimated, they
remain relatively high.
Phylogenetic methods (cf. Sect. 12.3.3) are also frequently used to estimate HGT rates. For instance, the phylogenetic analysis of 1,128 single copy genes present in at least
nine of eleven cyanobacterial genomes (Anabaena sp.
PCC7120,
Trichodesmium
erythraeum
IMS101,
Synechocystis sp. PCC6803, Prochlorococcus marinus
CCMP1375, P. marinus MED4, P. marinus MIT9313,
Synechococcus WH8102, Thermosynechococcus elongatus
BP-1, Gloeobacter violaceus PCC7421, Nostoc punctiforme
ATCC29133, and Crocosphaera watsonii WH8501)
identified 685 phylogenies inconsistent with the phylogeny
of the 16S rRNA gene (Zhaxybayeva et al. 2006). Assuming
that those inconsistencies result from HGT leads to the
conclusion that ~61 % of the studied genes have been
affected by HGT during the evolutionary history of these
nine cyanobacteria. This result is somewhat surprising
because the analyzed genes are conserved and well
distributed in these cyanobacteria (i.e., present in one copy
in at least 9 of the 11 genomes), which suggests that they
are ancient and may have been inherited from the last
common ancestor of the eleven cyanobacteria. Besides
multiple independent acquisitions from non-cyanobacterial
donors, these HGTs may result from HGT among
cyanobacteria and/or homologous gene replacement from
non-cyanobacterial donors (Figs. 12.9 and 12.10). To go
further, the authors analyzed the 700 genes for which
homologues are found in archaeal and bacterial lineages
(other than Cyanobacteria) in public databases. In 160
cases (23 %), the gene phylogenies did not recover the
monophyly of cyanobacteria, suggesting multiple HGTs
from cyanobacteria to other prokaryotes or vice versa. This
result illustrates one of the main weaknesses of HGT detection based on taxonomic distribution. Indeed, these
approaches allow detecting HGT resulting in the introduction of new genes in genomes, that is to say, genes with
no homologues in the recipient genome. In contrast, they
are inefficient to detect HGT leading to the replacement
of a preexisting gene. Nevertheless, the analysis of
cyanobacterial genomes reinforces the idea that HGTs are
frequent in prokaryotes.
Approaches based on atypical oligonucleotide and amino
acid composition and codon usage suggest also high rates of
HGT in prokaryotes. Among them, the most famous was
performed on E. coli K-12 MG1655 by Lawrence and
Ochman at the end of the 1990s. It shows that ~18 % of
the genome of E. coli K-12 (i.e., 755 genes on 4,288) results
from 234 HGT events since its divergence from the Salmonella lineage (Lawrence and Ochman 1998). Because of the
amelioration process (cf. Sect. 12.3.4.2), these estimates are
viewed as an underestimation of the real number of HGT
events. The average age of the detected HGTs was estimated
to be 14.4 Myr, leading to a rate of transfer of 16 kb/Myr/
lineage. Given that the divergence of the Salmonella and
Escherichia genera is thought to have occurred 100 million
years ago, the E. coli K-12 chromosome would have
acquired 1.6 Mb since this speciation event (Lawrence and
Ochman 1998). Extrapolating this rate over larger evolutionary scales implies that HGT have likely affected each gene
present in the genome of E. coli. This suggests that prokaryotic chromosomes are fluidic entities where the continual
introduction of new genes through HGT is balanced by gene
losses preventing the endless increase of prokaryotic
genome size.
HGT rate estimations vary considerably from one lineage
to another. For instance, the study of oligonucleotide composition of 116 genomes reveals HGT rates ranging from
0.6 % in the case of Buchnera (a gammaproteobacterial
intracellular symbiont of aphids) to 25.2 % in the case of
the archaeon Methanosarcina acetivorans C2A (Nakamura
et al. 2004), with an average of 14 % over all the studied
genomes. The low HGT rate in Buchnera can be explained
by a relative isolation due to their very particular lifestyle
(intracellular symbionts) compared to other bacteria. In the
case of M. acetivorans, a large fraction of the detected HGT
involved bacterial donors. However, other studies suggest
lower rates of HGT. For instance, a phylogenetic analysis of
205 single copy genes present in 13 gammaproteobacteria
(Buchnera aphidicola, E. coli, H. influenzae, Pseudomonas
aeruginosa, Pasteurella multocida, S. enterica subsp.
Enterica serovar Typhimurium, Vibrio cholerae,
464
C. Brochier-Armanet and D. Moreira
symbionts belongs to the Actinobacteria. The genome size
of the three strains varies considerably: 5.43 Mb (Frankia sp.
HFPCcI3), 7.50 Mb (Frankia alni ACN14a), and 9.04 Mb
(Frankia sp. EAN1pec). Comparative genomic studies
revealed a core genome composed of 35–50 % of the
genes present in these strains, whereas 578, 1,563, and
1,289 genes were specific to each strain.
Most studies based on gene taxonomic distribution
provide similar results. Under the assumption that specific
genes result from recent HGT, estimated HGT rates are high.
Extrapolating these estimated rates to larger evolutionary
scales would lead to the conclusion that the amount of
genes acquired by a lineage during its evolution is several
orders of magnitude greater than the complete genome size,
meaning that all genes in a given genome have been
exchanged several times during evolution. Sequence similarity analyses (cf. Sect. 12.3.1) performed in the 1990s
suggest also high HGT rates among prokaryotes, especially
among organisms living in similar ecological niches, such as
hot ecosystems. Accordingly, HGTs are considered as very
abundant and even to represent the major evolutionary
driving force in prokaryotes (Lan and Reeves 2000).
Although more recent studies suggest that the HGT rates
estimated by these methods are likely overestimated, they
remain relatively high.
Phylogenetic methods (cf. Sect. 12.3.3) are also frequently used to estimate HGT rates. For instance, the phylogenetic analysis of 1,128 single copy genes present in at least
nine of eleven cyanobacterial genomes (Anabaena sp.
PCC7120,
Trichodesmium
erythraeum
IMS101,
Synechocystis sp. PCC6803, Prochlorococcus marinus
CCMP1375, P. marinus MED4, P. marinus MIT9313,
Synechococcus WH8102, Thermosynechococcus elongatus
BP-1, Gloeobacter violaceus PCC7421, Nostoc punctiforme
ATCC29133, and Crocosphaera watsonii WH8501)
identified 685 phylogenies inconsistent with the phylogeny
of the 16S rRNA gene (Zhaxybayeva et al. 2006). Assuming
that those inconsistencies result from HGT leads to the
conclusion that ~61 % of the studied genes have been
affected by HGT during the evolutionary history of these
nine cyanobacteria. This result is somewhat surprising
because the analyzed genes are conserved and well
distributed in these cyanobacteria (i.e., present in one copy
in at least 9 of the 11 genomes), which suggests that they
are ancient and may have been inherited from the last
common ancestor of the eleven cyanobacteria. Besides
multiple independent acquisitions from non-cyanobacterial
donors, these HGTs may result from HGT among
cyanobacteria and/or homologous gene replacement from
non-cyanobacterial donors (Figs. 12.9 and 12.10). To go
further, the authors analyzed the 700 genes for which
homologues are found in archaeal and bacterial lineages
(other than Cyanobacteria) in public databases. In 160
cases (23 %), the gene phylogenies did not recover the
monophyly of cyanobacteria, suggesting multiple HGTs
from cyanobacteria to other prokaryotes or vice versa. This
result illustrates one of the main weaknesses of HGT detection based on taxonomic distribution. Indeed, these
approaches allow detecting HGT resulting in the introduction of new genes in genomes, that is to say, genes with
no homologues in the recipient genome. In contrast, they
are inefficient to detect HGT leading to the replacement
of a preexisting gene. Nevertheless, the analysis of
cyanobacterial genomes reinforces the idea that HGTs are
frequent in prokaryotes.
Approaches based on atypical oligonucleotide and amino
acid composition and codon usage suggest also high rates of
HGT in prokaryotes. Among them, the most famous was
performed on E. coli K-12 MG1655 by Lawrence and
Ochman at the end of the 1990s. It shows that ~18 % of
the genome of E. coli K-12 (i.e., 755 genes on 4,288) results
from 234 HGT events since its divergence from the Salmonella lineage (Lawrence and Ochman 1998). Because of the
amelioration process (cf. Sect. 12.3.4.2), these estimates are
viewed as an underestimation of the real number of HGT
events. The average age of the detected HGTs was estimated
to be 14.4 Myr, leading to a rate of transfer of 16 kb/Myr/
lineage. Given that the divergence of the Salmonella and
Escherichia genera is thought to have occurred 100 million
years ago, the E. coli K-12 chromosome would have
acquired 1.6 Mb since this speciation event (Lawrence and
Ochman 1998). Extrapolating this rate over larger evolutionary scales implies that HGT have likely affected each gene
present in the genome of E. coli. This suggests that prokaryotic chromosomes are fluidic entities where the continual
introduction of new genes through HGT is balanced by gene
losses preventing the endless increase of prokaryotic
genome size.
HGT rate estimations vary considerably from one lineage
to another. For instance, the study of oligonucleotide composition of 116 genomes reveals HGT rates ranging from
0.6 % in the case of Buchnera (a gammaproteobacterial
intracellular symbiont of aphids) to 25.2 % in the case of
the archaeon Methanosarcina acetivorans C2A (Nakamura
et al. 2004), with an average of 14 % over all the studied
genomes. The low HGT rate in Buchnera can be explained
by a relative isolation due to their very particular lifestyle
(intracellular symbionts) compared to other bacteria. In the
case of M. acetivorans, a large fraction of the detected HGT
involved bacterial donors. However, other studies suggest
lower rates of HGT. For instance, a phylogenetic analysis of
205 single copy genes present in 13 gammaproteobacteria
(Buchnera aphidicola, E. coli, H. influenzae, Pseudomonas
aeruginosa, Pasteurella multocida, S. enterica subsp.
Enterica serovar Typhimurium, Vibrio cholerae,
464
C. Brochier-Armanet and D. Moreira