In two review articles in the 17 April issue Doolittle and Sapienza (p.

‘Selfish DN A

AY Orgel and Crick (p. 604) separately

suggested that much of the DNA in the genome of higher organisms could be described as ‘selfish’. They argued
that such DNA has no appreciable phenotypic effect and functions only to ensure its own self-preservation within
the genome. This view point stimulated a great deal of comment, some of which was published in the issue of 26
June (p. 617). Now the original authors have joined up with one of their critics and reassessed their ideas in the two
articles below. A further comment is added by H. K. Jain.

from L.E. Orel, F.H.C. Crick and C. Sapienza

DIFFICULTIES have: been caused by the
words ‘selfish’, ‘junk’, ‘specific’? and

‘phenotype’ that were used in the two.

reviews of selfish DNA!~.

Many people dislike the term ‘selfish
DNA’ and a more acceptable alternative
might be ‘parasitic DNA’. The word
‘parasitic’ does not imply that the DNA
can move between individuals, though
certain viral DNAs might do this. It does
imply that such DNA can usually move
between different chromosomes in ae
same cell.

The word ‘junk’ also seems to arouse
strong feelings. The idea behind it can be
clarified by considering what is meant by
‘specific’, We consider a sequence highly
specific if the change of any one of its bases
almost always has a considerable effect on
the organism. An example would be the
recognition site for a physiologically
relevant restriction enzyme. At the other
extreme are sequences whose deletion or
extensive alteration would produce a
negligible effect. Such sequences could
_ reasonably be called junk. However, there
is probably a continuum between these two
extremes, including fairly specific
sequences, where the alteration of most
bases will produce some effect (many
sequences coding for protein, and the
different signals for starting and stopping
transcription are likely to be of this type)
and sequences whose deletion or extensive
alteration will usually produce a smail
effect, such as a change in the local rate of
recombination. In some cases close
similarity of two sequences may be
important, while the base sequences
themselves may matter hardly at all — for
example, within the introns of two neigh-
bouring versions of a gene. The word ‘junk’
is perhaps too broad to cover ail those cases
for which ‘the effect of sequence on the
phenotype of an organism is small or zero.
We hope a more precise terminology will
evolve as the facts become better known.

The word ‘phenotype’ has also caused
difficulties in spite of Doolittle and
Sapienza’s careful use of ‘organismal

 

L.E. Orgel and F.H.C. Crick are at The Saik
Institute, San Diego, California and C.
Sapienza is in the Department of Biochemistry,
Dalhousie University, Halifax; Canada.

phenotype’ to make their meaning clear.
We obviously need two words: one to refer
to the phenotype of the organism and the
other to apply solely to the ‘phenotype’ of
the parasitic DNA, a distinction we would
certainly make in the case of a true parasite.
For the former we would suggest ‘organis-
mal phenotype’. and. for the latter,
following Cavalier-Smith}, ‘intragenomic
phenotype’, but we would allow the word
‘phenotype’ alone. to be used when. the
context makes the meaning clear. :

In our original definition we. said that
selfish DNA had two distinct properties:
(1) It arises when a DNA sequence spreads
by forming additional copies of itself
within the genome. (2) It makes no specific
contribution to the phenotype. By
‘phenotype’ we meant organismal
phenotype. We intended ‘specific’ to be
understood as ‘highly specific’ or: ‘fairly
specific’ in the discussion above. However
it has been pointed out to us by R.
Pritchard’ that ‘no... contribution’ is
unnecessarily strict. It. would have been
more useful to include also. DNA which
made a small contribution to the
organismal phenotype, either. positive or
negative. An example of the latter might be
a viral DNA which became ‘Part of the
genome.

There is obviously, a. ‘continuum of
possible selective advantages (positive or
negative) to the organism,. We had
excluded from our. definition of selfish
DNA. those cases. where. the: selective
advantage is very high. To decide whether a
repeated sequence is parasitic or not, one
must determine whether the presence of the
repeated sequence in the population. is
mainly due to the efficiency with which the
sequence spreads intragenomically or
mainly due to the. reproductive success of
those individuals in the. population who
possess repeated copies of the sequence.
Only in the former case do we consider.it
useful to use the.term selfish or parasitic
DNA, as opposed to-useful or. symbiotic
DNA — the borderline. between the two
maynotbesharp. __

In considering the spread of parasitic
DNA one should not: underestimate -the
power of natural selection. -For example, if

a particular transposon: was.-inserted:.at_

random, it would run the risk of
inactivating: many genes and thus be
selected against. A transposon which
usually inserted at sites between genes
would be at a selective advantage. Sites
very near essential genes (as pointed out by
Bruce Grant*) may be harder to delete than
thosein the middle of long stretches of junk
and so parasitic DNA in the former
positions is likely to survive longer. Effects
of this type would lead to the selection of
selfish DNA sequences that inserted
preferentially at special sites in the genome.

Competing theories differ in their
analysis. of the factors determining the
amount of non-specific DNA and of the
way in: which it comes into existence.
Although we:cannot at present decide on
the: quantitative contribution of the
different types of non-specific DNA to the
genome, it is still helpful to classify the
various theories. ..

. We proposed? that the amount of non-
specifi ic DNA present in a given genome is
often. determined by the balance between
the intragenomic spreading of selfish
sequences and phenotypic selection against
excess DNA -— the weaker the phenotypic
selection against non-specific DNA the
larger the DNA-content of the genome. In
another group of theories it is proposed
that there is an optimal: DNA content for
each organism, which may be substantially
greater than the amount of specific DNA
that is needed to define the phenotype. The
amount: of. non-specific DNA is then
principally determined by the difference
between the optimal DNA content and the
essential content. of specific DNA. The
theories are not mutually exclusive, but
differ substantially in emphasis in their
explanation of C-values. |. .,..

, Cavalier-Smith’s proposal?:® is an
interesting example. of an. ‘optimal DNA
content’ theory. One of his ideas, which we
misinterpreted in our previous paper?, is
that in large cells, particularly in oocytes,
the transport of messenger. RNA across the
nuclear membrane may become.a limiting
factor and that the only way to increase the

The “selfish DNA’” design at the top of the page
was. created. by: Linda. Angeloff-Sapienza of
Hatifax, Canada and originally appeared on the
cover of the issue of 17 April, 1980...”

 

 

0028-0836/80/510645-02$01 .00

21980 Macmillan Journals Lid
646

Nature Vol. 288 18/25 December 1980

 

rate of transport is by increasing the
number of nuclear pores by extending the
surface of the membrane. If the area of
nuclear membrane is determined by the
DNA content of the nucleus, it follows that
selection for a larger cell must lead to an
increase in the DNA content of the
genome. Thus, rather surprisingly, extra
non-specific DNA is selected for because it
allows such a cell to grow faster. While we
do not question the logic of the argument,
given the various assumptions, we do not
find all the assumptions particularly
plausible. It may be that there is sometimes
selection for increased cell volume and
increased nuclear volume. In cells so
selected, non-specific DNA _ can
accumulate. Whether it does so because
large cells with large nuclei require such
accumulation, or because they simply
permit it remains to be seen. We feel that
more experimental work is needed to
unravel the complexities of the situation.
In particular, we should like to know in
which stages and in which organisms the
surface of the nuclear membrane is
saturated with nuclear pores.

Cavalier-Smith? also cites the widely
different DNA contents of germ cells and
somatic cells in some invertebrates as
evidence against the selfish DNA hypo-
thesis. However these observations can
also be explained in terms of the selfish
DNA theory. Such DNA ‘needs’ only to
remain in the germ line to function para-
sitically. On the other hand, organismal
selection might sometimes be stronger
against surplus DNA in the soma than in
the germ line. Thus representation in the
germ line but not in the soma may some-
times be an optimal strategy for parasitic
DNA. As for B chromosomes, in many
cases the evidence appears to us to give
some support to the idea (originally pro-
posed by Ostergren’ in 1945) that they are
largely parasitic, but there is certainly
evidence that they sometimes have
phenotypic effects which may possibly be
useful®.?,

Smith!® has pointed out that the DNA of
vertebrates usually has about 42 per cent
GC whereas the GC content of invertebrate
and prokaryotic DNA varies over a much
wider range. The theory of parasitic DNA
has rather little to say on this point. There
are many factors which might affect the
GC content of an organism’s DNA. If
much of the parasitic DNA has descended
rather recently from insertion elements
which themselves originally coded for
proteins, then it would not be surprising if
their present GC content were similar to

 

. Doolittle, W. F. & Sapienza, C. Nature 284, 601 (1980).
Orgel, L. E. & Crick, F. H. C. Nature 284, 604 (1980).
. Cavalier-Smith, T. Nature 285, 617 (1980).

Pritchard, R. (personal communication).

Grant, B. (personal communication).

. Cavalier-Smith, T. J. Cell. Sci. 34, 247 (1978).
Ostergren, G. Bot, Notiser 2, 1$7 (1945).

. Jones. R. N. dat. Rev. Cytol. 40, 1 (1975).

. Ames, A. & Dover, G. Chromosoma (in the press).

10. Smith, T. F. Nature 285, 620.(1980).

11. Zuckerkandel, E. ( personal communication).

12. Dover, G. Nature 285, 618 (1980).

13. Loomis, W. Devi Biel. 30, F3-F4 (1973).

t4. See. for example, Proudfoot, N. Nature 286, 840 (1980).

O90 OA te

that of genes which still code for protein.
This may, perhaps, explain the € constancy
of GC in vertebrates.

As for our own ideas, we now feel that
there may perhaps be reasons why too little
DNA can in some cases produce a selective
disadvantage. For example, Zucker-
kandel!' has suggested that there may bea
minimum size for a ‘domain’ necessary for
stability of the chromatin in the folded
state. Thus a domain containing only a few
genes might benefit from having some non-
specific DNA as ‘padding’. This would
mean that there is indeed an optimal
amount fortotalDNA.  .

In our original paper? we feel that we did
not put enough emphasis on the distinction
between sequences which are repeated,
exactly or nearly exactly, in many tandem
repetitions and sequences which are more
widely dispersed over the chromosomes
and which occur in only one or a few copies

in any one place. It seems plausible that’

these two types of sequence evolved
different mechanisms. It is possible that the
mechanisms generating the tandemly
repeated type are usually more ‘ignorant’
(in Dover’s sense!2) than the more
dispersed type. If the latter have any
specific function it is likely to be that of the
control, at one level or another, of gene

expression, whereas the tandemly repeated

_type seem more likely to influence

chromosome mechanics.

One possibility to which we feel we
should have given more weight is that of
‘dead genes’, also called ‘pseudo-
genes’ !3.!4; that is, sequences which can no
longer code for a protein (or a structural
RNA) but which appear to have descended
from a sequence that did. Whether these
conform to our definition of parasitic
DNA remains to be seen, but we suspect
this is unlikely, since they usually exist in
only a single copy, or as multiple tandem
copies in only one place.

In our recent experience most people will
agree, after discussion, that ignorant
DNA, parasitic DNA, symbiotic DNA
(that is, parasitic DNA which has become
useful to the organism) and ‘dead’ DNA of
one sort or another are all likely to be
present in the chromosomes of higher
organisms. Where people differ is in their
estimates of the relative amounts. We feel
that this can only be decided by
experiment. We expect that due to the
recent advances in genetic engineering and
related techniques much sequence
information will accrue in the near future.
This should help to decide between the
different alternatives. , Q

 

1028-0836) 80+ £10646-02$01 00

¢1980 Macmifian Journals Ltd