The evolution of “irreducibly complex” antifreeze

proteins in a polar fish

A new paper in Proceedings of the National Academy of Sciences shows how a functional
protein (an antifreeze protein in the blood of an Arctic fish) can be assembled out of
scraps of genome that have no function at all. Moreover, the protein doesn’t become
functional—e.g., being secreted into the fish blood to keep it from freezing—until the
very last step of gene assembly, so the sequence looks “irreducibly complex”. But we
can construct a perfectly naturalistic evolutionary sequence by looking at DNA in
relatives and putative ancestors. This shows that the appearance of “irreducible
complexity”—the existence of an adaptation that doesn’t seem to function until all its
parts are in place—does not require a Designer, but can arise from natural processes.

The molecular mechanism and reconstruction of the evolutionary path is complicated,

but I’ll try to present it in a stepwise fashion. Zhuang et al. used known phylogenies of
fish related to the two “antifreeze” fish in the family Gadidae, a group of codfish. These
codfish have functional antifreeze glycoproteins (AFGPs) that act to keep their blood
from freezing when the cod swim in super-cold polar waters. The proteins do this by
keeping ice crystals from forming and acting as sites of nucleation that could turn the
fish into popsicles.

The fish’s AFGPs consist of three bits: the antifreeze protein itself, which consists of
repeats of the amino acid sequence threonine-alanine-alanine (Thr-Ala-Ala), a second
secretory protein that gives a signal to the genome to enable the antifreeze protein to be
secreted into the blood, and a promoter region that is necessary to allow the DNA
sequence to be transcribed into RNA (which then makes the antifreeze protein).

What’s remarkable about this configuration is that every bit of it, including the two
proteins themselves and the promoter sequence, was cobbled together via translocations
and duplications of DNA (this happens passively in the genome) until all the elements
were in place. And the “functional” gene couldn’t function right up to the very end, when
the promoter sequence moved to the right place to allow the protein-coding region to
produce an RNA transcript. This entire series of steps was reconstructed by sequencing
the DNA of relatives that don’t have functional AFGPs, so we could see the evolutionary
order in which things were assembled, and where the functional bits originally came
from.

Here’s how it occurred; this is Figure 4 from the paper, and I give its caption:
Evolutionary mechanism of the gadid AFGP gene from noncoding DNA. The color codes of the sequence
components follow Fig. 1. (A) The ancestral noncoding DNA contained latent signal peptide-coding exons
with a 5′ Kozak motif, adjacent to a duplication-prone 27-nt GCA-rich sequence. (B) The 27-nt
GCA(Ala)-rich sequence duplicated forming four tandem copies. (C) A 9-nt in the midst of the four 27-nt
duplicates became the three codons for one AFGP Thr-Ala-Ala unit and underwent microsatellitelike
duplication forming a proto-ORF. (D) A proximal upstream regulatory region acquired through a putative
translocation event. (E) A 1-nt frameshift led to a contiguous SP, a propeptide, and a Thr-Ala-Ala-like
cds in a read-through ORF. (F) Intragenic (Thr-Ala-Ala)n cds amplification, fulfilling the antifreeze
function under natural selection.

The evolutionary stages posited (and supported by sequence and phylogenetic analysis)
are A-F. First, there is a sequence of GCAGCAGCA in an ancestor, a sequence that would
normally code for repeated alanines, but wasn’t functional (this is found in a relative). It
expanded through duplication: A —>B.

Then, a mutation from a guanine to a cytosine base in another ancestor converted one
Ala-Ala-Ala sequence to a Threonine-Alanine-Alanine amino acid triplet, which itself
expanded through successive duplications (B —>C). The gene now had four Thr-Ala-Ala
units, but was still nonfunctional. But this was to be the core of the functional protein in
the future; it’s the dark blue bit seen in C through F above.

Another part of the genome had a nonfunctional sequence that could serve as the
secretory protein to get the dark blue protein secreted into the blood. A deletion of a
single nucleotide (C—>E) rendered it capable of producing a signal protein (the purple
bit in D-F). But the entire system was still nonfunctional because it lacked a promoter
region.

Finally, the system became functional when the protogene moved to a location near a
nonfunctional DNA region that could serve as a promoter for the nascent gene. Now a
gene producing a repeated Thr-Ala-Ala protein could function and secrete it into the
blood.

Further, natural selection could now act on the functioning gene to make it more
effective, simply by selecting for those genes that had even more duplications of the
Thr-Ala-Ala segment, so we had a big protein of repeated units that could act as an
antifreeze in fish blood (E—>F). (More repeats = better antifreeze protection.)

It’s a bit more complicated than this, but this is the essence of how the final protein
came to be. And it’s not speculation, because all the bits can be found in other species or
posited in ancestors, and so this reconstruction is fairly sound. Moreover, it involves
processes known to operate in the DNA: the moving of bits around by translocation,
duplication of sequences, etc. No divine intervention is required to do this, even though
the protein isn’t functional until it’s put together with the secretory protein and the
promoter.

One might ask this reasonable question: “Well, if the nascent antifreeze protein is just
sitting there and not doing anything before it becomes active, why isn’t it inactivated by
mutations?” That’s a good question, and one answer is that the process took place
reasonably quickly so that mutations (which are, after all, rare) didn’t have time to turn
the dark blue protein core into gibberish. And once that core formed, duplication of the
Thr-Ala-Ala would be rapid, promoted by natural selection because more repeats confer
greater antifreeze activity.

So here we have an “irreducibly” complex system, functional as an antifreeze system

only at the very end, but one that formed purely through natural and well-known
genomic processes.