Considering gene duplication, Steve Hardies says that:
duplicate -> drift awhile -> stumble into new function
seems simplistic and unrealistic.
Here's another possibility:
New function -> adaptive conflict -> gene duplication.
This is what seems to happen in the gene recruitment of
enzymes as crystallins, the structural proteins of the eye lens.
There have been numerous independent events in which existing
enzymes have become major components of the lens (e.g. LDH-B is 50%
of total protein in hummingbird lens). This occurs without prior
duplication. The gene just acquires high expression in lens while
maintaining its normal expression elsewhere. The protein now has
two distinct roles as enzyme and crystallin. Being subject to two
sets of selective pressure the gene may find itself in an adaptive
conflict. Indeed, the new role in lens may require unusual sequence
changes in the enzyme of no obvious benefit to its catalytic
function. Under these circumstances there may be selective
advantage in duplication and specialization. This seems to have
happened in the case of delta-crystallin/ argininosuccinate lyase
in birds and we have recently found evidence of another example in
This scheme could conceivably apply elsewhere. For example, an
enzyme could acquire an additional useful function through sequence
drift. (For example it has been proposed that `-enolase is also a
cell surface plasminogen receptor.) If this function was useful in
some cell types but not in others there could be pressure to
duplicate and specialize, leading to the appearance of
enzymatically indistinguishable isozymes with different tissue
What is attractive about this is that the both the gain of function
and the subsequent duplication can have some selective component.
In contrast, the classic scheme requires both unselected
duplication and chance acquisition of new function.
graeme at mge2.nei.nih.gov