posted by Victor Hanson-Smith
In 2008, Shozu Yokoyama et al. published a compelling paper in which they reconstructed ancestral rhodopsin proteins in order to infer specific amino acid changes that explain phenotypic differences in vertebrate dim-light vision. In doing so, the authors shed light (pun intended) on the aquatic habitat of vertebrate ancestors. Sean Carroll commented (in his book “Making of the Fittest”) that Yokoyama’s work is “the deepest body of knowledge [to date] linking differences in specific genes to differences in ecology and to the evolution of species.” Indeed, this paper is remarkable because the overall story unites evidence from disparate scales of macro- and micro-analysis: the authors integrated molecular biology with paleontology and ecology. This paper also offers insight into limitations of dN/dS tests for positive Darwinian selection.
Before discussing this paper, let’s review the mechanisms of phototransduction.
Phototransduction, a review
Retinas (in eyes) contain rod cells. Within the exterior membrane of rod cells exists a transmembrane protein called rhodopsin, which is comprised of an opsin protein bound to chromophore ligand, called 11-cis-retinal. Incoming photons isomerize 11-cis-retinal into all-trans-retinal; the newly isomerized retinal cannot fit inside the opsin binding site, so the opsin conformationally changes. The bound combination of 11-trans-retinal and opsin is called metarhodopsin II.
Metarhodopsin II is unstable and quickly splits into its constituent parts: an opsin protein and 11-trans-retinal. The unbound opsin is now in active conformation and is free to bind with the alpha subunit of transducin (a G protein); in response to opsin activation, the alpha subunit of transducin exchanges its GDP molecule for a GTP molecule. By releasing GDP, the alpha subunit (with opsin attached) dissociates from the beta and gamma transducin subunits. Although the transducin beta-gamma complex is anchored to the cell wall, the opsin+alpha complex is now free to activate phosphodiesterase, which then proceeds to hydrolize cyclic guanosine monophosphate (cGMP) into 5’-GMP. cGMP mediates cellular calcium ion channels; the calcium channels close in the absence of cGMP (thus protons are not entering the cell). However, potassium ions (+) are still being pumped out of the rod cell, thus accumulating an intracellular negative electric charge. This electric charge is eventually discharged into the retinal ganglia, which sends an electrical signal to the brain.
Rhodopsin function is an ecological marker.
The authors explain that we can infer some properties of organism’s habitat (i.e. aquatic depth) by studying the functional characteristics of the organism’s rhodopsins. Across the vertebrate clade, rhodopsins phenotypically differ in their maximal absorptive wavelength. Most rhodopsins maximally react to photons in the wavelength range ~480 nm to ~500 nm (although, upper outliers of 700 nm are discussed in this paper). Surface-dwelling fish, amphibians, birds, and mammals use rhodopsins tuned for ~500 nm, which correlates to the wavelength of twilight. Fish living at depths below 200 meters use rhodopsins tuned for ~480 nm because the light distribution is narrower at those depths. Northern lampfish and Pacific viperfish vertically migrate at night, and their rhodopsins are tuned for the intermediate wavelengths of ~490 nm. This phenotypic gradient is—I presume—explained by subtle differences in how orthologous rhodopsins bind 11-cis-retinal.
What is the molecular basis of rhodopsin spectral tuning?
In order to determine the molecular mechanisms of rhodopsin variation, Yokoyama et al. reconstructed ancestral rhodopsin protein sequences for various vertebrate ancestral intermediates and then created these intermediates by mutagenizing extant rhodopsins. Experimentation in vivo reveals that most rhodopsin diversity can be explained by 15 amino acid mutations at only 12 sites. This observation underscores the larger evolutionary idea that phenotypic diversity emerges by orthologous accumulation of small mutations that are selectively advantageous. The reconstructed most-recent-common ancestor of vertebrate rhodopsins is tuned for ~501 nm, consistent with fossil evidence suggesting “the ancestors of bony fish lived in shallow, near-shore marine environments.” Furthermore, the reconstructed intermediate ancestors reveal a variety of evolutionary paths. For example, the ancestors of squirrel fish evolved from surface dwellers to intermediate-depth dwellers (with wavelength absorption ~497 nm); subsequent species evolved to live at deeper habitats (i.e., N. aurolineatus with rhodopsin wavelength ~481 nm), returned to surface habitats (i.e. N. argenteus with wavelength ~502 nm), or remained intermediate-depth dwellers (i.e., S. diadema with wavelength ~491 nm).
dN/dS tests have limitations.
The authors observed that several biologically significant amino acid changes occurred multiple times, implying that these sites are under positive selection. However, the parsimony method of dN/dS failed to detect any positively-selected sites. On the other hand, Bayesian methods detected eight positively-selected sites, but none of these sites coincide with those revealed by mutagenesis experiments! Furthermore, the Bayesian methods detected positive selection in closely-related genes, but not in distantly-related genes. The authors explain this apparent failure of dN/dS tests:
When nucleotide changes occur at random, the proportions of nonsynonymous and synonymous mutations are roughly 70% and 30%, repsectively. Hence, under neutral evolution, or even under some purifying selection, closely related molecules can initially accumulate more nonsynonymous changes than synonymous changes. However, as the evolutionary time increases, synonymous mutations will accumulate more often than nonsynonymous mutations.
We need more structural analysis!
The strength of Yokoyama’s paper is that it reveals specific amino acid mutational trajectories that lead to phenotypic rhodopsin diversity. However, the authors do not explain how specific amino acid change affects rhodopsin structure, binding affinity to 11-cis-retinal, or properties of photon absorption. I realize that a comprehensive structural analysis would require crystallizing multiple rhodopsins across the vertebrate clade, in addition to crystallizing one or more ancestral rhodopsins. As of writing this blog entry, the best structural work on rhodopsins comes from Krystof Palczewski’s structure of rhodopsin bound to 11-cis-retinal (Science, 2000), and Patrick Scheerer’s structure of opsin bound to transducin’s alpha subunit (Science, 2008). Given the monumental effort required to crystallize a single rhodopsin, I don’t expect to see this anlaysis anytime soon.
Yokoyama S, Tada T, Zhang H, & Britt L (2008). Elucidation of phenotypic adaptations: Molecular analyses of dim-light vision proteins in vertebrates. Proceedings of the National Academy of Sciences of the United States of America, 105 (36), 13480-5 PMID: 18768804