114
4 Molecular Evolution
population" [122], it is clear that the findings of
molecular biology have significantly broadened
and enriched the neo-Darwinian (synthetic) theory rather than contradicted it. The neutral theory is a necessary supplement to neo-Darwinism
in the molecular context and is not its antithesis.
The attempts made in many publications to detect
clear cases of selection at the molecular level are
superfluous in as much as the neutral theory in no
way denies their existence. Gene transfer
between different species, and also between different eukaryotes, is now being discussed as a
possible mechanism of evolution. Should it be
shown that such a process is of frequent occurrence in evolution, this would make the reconstruction of phylogeny from molecular data more
difficult but, looked upon as a mechanism for the
emergence of new genetic variability, it fits easily
into the Darwinian theory. The same argument
would be valid for the recently suspected and
then refuted inheritance of an acquired immunotolerance in mice; if at all applicable, this would
be proof of gene transfer between immunocompetent somatic cells and germline cells but would
in no way be an argument in favour of Lamarckism against Darwinism [122].
The present great interest in questions of molecular evolution is indicated by the large number
of recent reviews and monographs on this theme
[68, 212,246, 369].
4.1 The Determination of Homology
Between Protein DNA Sequences
Proteins or genes are considered to be homologous when they have the same origin, i.e. when
they arise from the same gene. In many cases,
homology can be assumed unconditionally, e.g.
for proteins with the same function in closely
related species, for immunologically crossreactive proteins, and for proteins or genes with
extensive sequence coincidence. However, to be
certain about the homology between genes, it
must be shown that the apparent agreement is not
the result of either chance or convergence. The
existence of significant agreement between long
sequences due to convergent evolution can be
ruled out on probability grounds. There are, in
fact, many examples of proteins with similar function that show no detectable homology. For
example, there is no sequence similarity between
the ribonuc1eases of the bacteria and fungi, on
the one hand, and animals, on the other hand;
between phage and animal lysozymes; and
between the different protein super-families of
the peptide hydrolases or proteinase inhibitors.
Also, the Cu, Zn superoxide dismutases from the
cytoplasm of eukaryotes are not homologous to
the corresponding Mn and Fe enzymes from prokaryote and eukaryote mitochondria. It is true
that functionally similar amino acids are found in
similar spatial structures of different proteins, but
this does not produce statistically significant
agreement when long stretches of sequence are
considered.
The spatial structures of proteins are very conserved in evolution; as long as the basic function
of a protein does not change, then the responsible
spatial structure is maintained. It seems reasonable, therefore, to take into account spatial structure as evidence for distant relationships between
proteins, perhaps even "when all traces in the
sequence have been obliterated" [368]. For comparison of spatial structures, the distance
between the polypeptide chains that project into
each other can be calculated. A further possibility
is to describe polypeptides in terms of sequences
of rotational angles around the bonds N-C' and
c·_ccarboxyl and then compare them. The probability of chance in the agreement of spatial structures can be estimated by means of relevant comparisons of simulated folded chains [193, 368,
407, 420]. It must not be forgotten, however, that
convergent evolution of spatial structures to fulfil
similar functions can never be totally excluded.
Two proteins may be considered as homologues if their sequence agreement is higher than
would be possible by chance alone. In order to
test this, the sequences to be compared must first
be arranged opposite each other (aligned) so that
they show the highest possible agreement (S). A
series of appropriate alignments are then made
with random sequences of the identical amino
acids, and the average agreement, together with
the standard deviation thereof (Sr ± SDr), is
determined. If the quotient
A = (S-Sr)/SDr
(4.1)
exceeds a given value, which is arbitrarily set
between 3 and 5, then the agreement shown by
the original sequences is considered to be significant [87,93].
To calculate the similarity (S) of two sequences, values of 1 and 0 are given to each pair of
identical and non-identical amino acids, respectively, and the values for each position are added
together. Instead of this "unitary matrix", which
4 Molecular Evolution
population" [122], it is clear that the findings of
molecular biology have significantly broadened
and enriched the neo-Darwinian (synthetic) theory rather than contradicted it. The neutral theory is a necessary supplement to neo-Darwinism
in the molecular context and is not its antithesis.
The attempts made in many publications to detect
clear cases of selection at the molecular level are
superfluous in as much as the neutral theory in no
way denies their existence. Gene transfer
between different species, and also between different eukaryotes, is now being discussed as a
possible mechanism of evolution. Should it be
shown that such a process is of frequent occurrence in evolution, this would make the reconstruction of phylogeny from molecular data more
difficult but, looked upon as a mechanism for the
emergence of new genetic variability, it fits easily
into the Darwinian theory. The same argument
would be valid for the recently suspected and
then refuted inheritance of an acquired immunotolerance in mice; if at all applicable, this would
be proof of gene transfer between immunocompetent somatic cells and germline cells but would
in no way be an argument in favour of Lamarckism against Darwinism [122].
The present great interest in questions of molecular evolution is indicated by the large number
of recent reviews and monographs on this theme
[68, 212,246, 369].
4.1 The Determination of Homology
Between Protein DNA Sequences
Proteins or genes are considered to be homologous when they have the same origin, i.e. when
they arise from the same gene. In many cases,
homology can be assumed unconditionally, e.g.
for proteins with the same function in closely
related species, for immunologically crossreactive proteins, and for proteins or genes with
extensive sequence coincidence. However, to be
certain about the homology between genes, it
must be shown that the apparent agreement is not
the result of either chance or convergence. The
existence of significant agreement between long
sequences due to convergent evolution can be
ruled out on probability grounds. There are, in
fact, many examples of proteins with similar function that show no detectable homology. For
example, there is no sequence similarity between
the ribonuc1eases of the bacteria and fungi, on
the one hand, and animals, on the other hand;
between phage and animal lysozymes; and
between the different protein super-families of
the peptide hydrolases or proteinase inhibitors.
Also, the Cu, Zn superoxide dismutases from the
cytoplasm of eukaryotes are not homologous to
the corresponding Mn and Fe enzymes from prokaryote and eukaryote mitochondria. It is true
that functionally similar amino acids are found in
similar spatial structures of different proteins, but
this does not produce statistically significant
agreement when long stretches of sequence are
considered.
The spatial structures of proteins are very conserved in evolution; as long as the basic function
of a protein does not change, then the responsible
spatial structure is maintained. It seems reasonable, therefore, to take into account spatial structure as evidence for distant relationships between
proteins, perhaps even "when all traces in the
sequence have been obliterated" [368]. For comparison of spatial structures, the distance
between the polypeptide chains that project into
each other can be calculated. A further possibility
is to describe polypeptides in terms of sequences
of rotational angles around the bonds N-C' and
c·_ccarboxyl and then compare them. The probability of chance in the agreement of spatial structures can be estimated by means of relevant comparisons of simulated folded chains [193, 368,
407, 420]. It must not be forgotten, however, that
convergent evolution of spatial structures to fulfil
similar functions can never be totally excluded.
Two proteins may be considered as homologues if their sequence agreement is higher than
would be possible by chance alone. In order to
test this, the sequences to be compared must first
be arranged opposite each other (aligned) so that
they show the highest possible agreement (S). A
series of appropriate alignments are then made
with random sequences of the identical amino
acids, and the average agreement, together with
the standard deviation thereof (Sr ± SDr), is
determined. If the quotient
A = (S-Sr)/SDr
(4.1)
exceeds a given value, which is arbitrarily set
between 3 and 5, then the agreement shown by
the original sequences is considered to be significant [87,93].
To calculate the similarity (S) of two sequences, values of 1 and 0 are given to each pair of
identical and non-identical amino acids, respectively, and the values for each position are added
together. Instead of this "unitary matrix", which