That's been one of my main points!
Enjoy.
Jim
****
>Cascades
>A running theme in much of molecular biology is genetic cascades, in
>which one gene triggers another gene, which triggers another. Two
>examples of cascades cited by Behe are the formation of blood clots
>and the complement system. The complement system is a set of
>proteins that are activated by antibodies; these proteins then create
>holes in the cell membranes of the invading bacteria and thereby
>disrupt the specific balance of solutes and ions required for the
>bacterium to live.
>A major claim of Behe's is that biochemical cascades are "irreducibly
>complex". The claim is that without all the parts of the cascade, the
>cascade cannot function, and that therefore known evolutionary
>processes could not produce such a cascade by sequential addition of
>steps. p.87:
> Because of the nature of a cascade, a new protein would immediately
> have to be regulated. From the beginning, a new step in the cacade
> would require both a proenzyme and also an activating enzyme to
> switch on the proenzyme at the correct time and place. Since each
> step necessarily requires several parts, not only is the entire
> blood-clotting system irreducibly complex, but so is each step in
> the pathway.
>We can easily see that this broad statement is false; it is possible
>to posit such an evolutionary process. Furthermore, we can go from
>such a process to the expected results of such a process, and thereby
>make predictions as to what might be found in the living world.
Before I begin, I should point out that I only have access to the text
version of Keith's article and the version of his figures from this
format
is a little confusing. But I think I have succeeded in recreating his
model: a gene duplication occurs so that two X's can be activated to
generate target. Then a mutation occurs in one X so that the activated
X
cannot generate target. Then a mutation occurs in the other activated X
so that it cannot autocatalytically generate itself. Thus, the
duplicate
becomes dependent on the stimulus and original to become activated.
Finally, the duplicated gene mutates so that it no longer responds to
the
stimulus. Thus, the original no longer produces product, but can
activate the duplicate, while the duplicate can no longer respond to the
stimulus, but can generate the target.
Okay, so what's the problem with Keith's model?
First, Behe made the specific claim that the clotting-cascade could not
have evolved step-by-step. Keith claims this is false and offers a
abstract model to explain a speculative series-of-events. But given
that
we have a fairly good handle on the clotting and complement cascades,
why didn't Keith speculate about how *they* evolved? Instead of
talking about X's and Y's, why not talk about the things in question -
the
proteins involved in the cascade? Imaginary speculations about things
that could happen is not the same as providing evidence for things that
*did* happen. Keith thinks the clotting-cascade did indeed evolve step-
by-step. But he offered no evidence. Nor did he speculate about how
*it* happened.
And this is not a nit-pick. Chalkboard evolution is no substitute for
the
real thing. In chalkboard evolution, the only constraint is one's
imagination. But in the real world, there are often more constraints
than one's imagination. So let's consider some of the relevant factors
that have been neglected in Keith's chalkboard evolution.
1. Keith begins with the following situation:
>A stimulus triggers the conversion of X to X*, and X* autocatalytically
>drives this conversion also. X* then acts
>on a target.
This *seems* like a simple system, but it may actually be a candidate
for irreducible complexity (IC). And that would mean Keith was
assuming the existence of IC to show that IC does not exist! Thus, the
problem with chalkboard evolution. Keith needs to pick a clotting or
complement protein as his starting place and show how that protein
came to be. This is important because Keith's protein must recognize
the "stimulus" and then undergo conformational changes so that a target
can now be acted upon. And all of this probably involves several
specific
amino acids in very specific positions working as a "whole." In other
words, this protein's activity itself may not be capable of being
constructed "one tiny Darwinian step at a time."
2. Keith relies heavily on gene duplications to create his cascades.
For
example, in part of his explanation, he said:
>Post gene duplication.
>This is a completely symmetric system. (The diagram is not, but could
>be rearranged into one.) This arrangement is neutral; the species has
>gained no advantage. On the other hand, duplications such as this are
>a common event.
Well, how common *are* gene duplications? That is, what is the
determined frequency of gene duplication?
Keith then says that his post-gene duplication arrangement is
"neutral."
That's easy to say in the world of chalkboard evolution, but what about
the real world? As I see it, Keith has created a system where the
stimulus is more readily detected. If there is more X (due to
duplication),
there is more likelihood that X would react with the stimulus and thus
produce
target. So while Keith claims the system is neutral, I do indeed see
the
advantage of sensitivity to stimulus.
So let's imagine that this post-gene duplication state is selected for
because cells are now more sensitive to the stimulus. For thousands,
maybe millions of years, the organism would adapt with this new
sensitivity. That, in turn, spells trouble for the rest of Keith's
model.
For now, mutations that would eliminate the ability for one gene to
generate target will be disadvantageous. An organism that is not as
sensitive to the stimulus would be at a disadvantage in a population
where such sensitivity was selected for.
In his next step, Keith envisions a mutation that eliminates the ability
of the duplicate to be autocatalytic. He claims this would be neutral,
but again that is not clear. For autocatalytic activity is associated
with
amplification. He'll likely lose some of that amplification with this
loss.
Of course, even if increased sensitivity to stimulus was not
advantageous, there's another route that also spells trouble for Keith's
model. His post-gene duplication state builds redundancy into the
system. In the chalkboard world, he calls it "neutral," but it could
just
as easily be called advantageous or disadvantageous. It could be
advantageous because a population with a redundant system is
buffered against harmful mutations. If an organism mutates one path
so that only one is capable of generating product, it's descendents are
more likely to suffer disadvantageous mutations in the pathway than
are the others who have both pathways generating target. On the other
hand, to get to this advantageous state, it seems to me that such
redundancy would be initially disadvantageous, as it costs energy to
build X.
So, when Keith says:
>So, duplication plus a loss of function, plus one of two
different
>loss-of-function mutations can convert a single step pathway into
a
>two step cascade. The initial steps are neutral, neither
advantageous
>nor disadvantageous. Such neutral mutations occur regularly. The
>final step locks in the cascade. It is potentially advantageous,
because
>multiple levels of cascade give opportunities for both
control
>(probably a long-term advantage) and for amplification. (One X*
can
>activate many Y's, increasing an initial
signal.)
I'm not so sure that the initial steps are neutral. But what if they
were?
Then we're back to evolving a complicated systems by *pure chance*.
That is, selection doesn't enter the picture until the "final steps."
And I
don't understand why chance is a better explanation than a designer.
Finally, there is one more serious problem with Keith's model in that it
relies on gene duplication as if it were a magic wand in a just-so
story.
Y'see, when gene duplication occurs a multi-gene family is created. But
those who invoke gene duplication in their evolutionary speculations
almost always fail to account for an attribute of multi-gene families -
the process of gene conversion. Gene conversion occurs when the
sequence of one gene is converted to the same sequence of another
*very similar gene* (this is the reason why hundreds/thousands rRNA
genes are identical). Let's consider a hypothetical example. Gene A is
duplicated. So now a chromosome contains gene A and gene A' (the
duplicated gene) in tandem. Then, a mutation occurs in gene A' so now
we have gene A and gene A* (the duplicate is mutated). Keith envisions
this mutant form to be part of his model, but what if the mutant form
was converted back to the original sequence?
Well, gene conversion does exactly that. In fact, the frequency of gene
conversion is much higher than the mutation rate. Measurements of
artificially created two-gene families shows gene conversion to occur
1-10 thousand times more often than the mutation rate. Thus, while
Keith's chalkboard analysis allows him to keep his mutant products for
the duration of his model, in the real world, these slightly different
duplicates would probably be converted back to their original sequence
before the next step could take place (if the original is converted to
the
mutant state, this would be a selective disadvantage).
Keith claims:
>This is, of course, a model. A model should make predictions, and this
>model does. If a pathway evolved by such a manner, then consecutive
>steps in the pathway should be catalyzed by homologous proteins (a
>characteristic of both the blood coagulation and complement cascades,
>two examples given by Behe). If the proteins are encoded by adjacent
>genes (tandem duplication) or one gene shows the hallmarks of reverse
>transcription (no introns and a poly-A tail in the genomic sequence),
>all the better. We would also expect to find systems in nature with
>fewer or greater levels of cascading.
Fine. But we would also expect to find many examples of evolving
cascades, would we not? Or have cascades stopped evolving? That is,
we should find many examples of cascades containing the various
components
of the "intermediate" steps in your model. Are there any?
Keith adds:
>Is this the case with either the complement system or the clotting
>cascade? I'll confess I do not know for sure. Many of the proteins of
>both these systems fulfill the similar sequence criterion, though this
>is a weak test. The much stronger test, and major violation of the
>"irreducible complexity" postulate, would be to find varying cascade
>systems in the living world. If, for example, I found a system with
>one less level than human, that would already show that the human
>system is not "irreducibly complex", violating Behe's claim that "not
>only is the entire blood-clotting system irreducibly complex, but so
>is each step in the pathway." Again, it's not my speciality, and I am
>unaware. Indeed, it is quite possible that it remains an unexplored
>area. Does the clotting system function in the same manner in fishes?
>In Agnatha (jawless fishes such as hagfish and lampreys)? In lancets
>(invertebrates thought to be representative of the ancestors of
>vertebrates)?
First of all, jawless fish and lancets are just as "evolved" as humans.
They've been evolving alongside us for millions or years, thus the
systems they possess may have little or nothing to do with the one's we
possess.
Secondly, I think Keith may be wrong on this point. Looking for
"simpler" systems in other nonhuman organisms would *not* mean the
*human* system is not irreducibly complex. It would simply mean that
what is irreducibly complex in one system need not be in another
system (if I want to kill mice with a hammer, I don't need a spring).
Systems do not exist on chalkboards. They exist among the
*context* of the specified complexity of living cells. And while you
can
tinker with things in the realm of imagination, cells have to continue
existing in a nondisadvantageous state during this tinkering. What
Keith
would have to do is to *connect* the systems one-tiny-step-at-a-time in
a nondisadvantageous manner.
>Behe's silence on these points, the fact he did not even
>consider such an opposing model, or worse, that he never considered
>the predictions of his own model, does not bode well. Indeed, it is
>the critical examination of one's own model which separates cargo
>cults from science
People in glass houses ought not throw stones. ;)
I detected no such "critical examination" coming from Keith when it came
to his own model. For example, he did not even consider gene conversion
when he invoked gene duplications as a crucial part in his model. Oh
well.