What we see is too often determined by what we expect to see

How we see is determined by a complex system of interacting eye parts, photoreceptors and brain functions, but what we see can be determined by what we expect to see.  Opsins are receptors bound to the membrane of photoreceptor cells of the retina of the eye that help organisms, including us, to see. They are involved in the conversion of light, photons, into an electrochemical signal that eventually gets translated into images.

Except when they aren't.  In a piece in last week's Science, "Opsins: Not just for eyes," Elizabeth Pennisi writes about opsins in sea urchins.  Sea urchins don't have eyes, so why do they have photoreceptors?

Adhesive tube feet: Janek Pfeifer, Wikimedia

Well, they may not have eyes, but they do have tube feet, and as it turns out, the photoreceptors, loaded with opsins, have been found on these tube feet, which are on the ends of their spines, and are used for feeding, moving, breathing. The photoreceptor cells in the tube feet aren't pigmented in the way that most opsin-containing photoreceptor cells are, which is why they had been overlooked.

But, opsins have been found in unexpected places before.  Pigment cells in amphibian skin, that react to light, dove brains, fish skin. So, photoreception is happening in all kinds of places.  And, sea urchins do tend to avoid light, which means they have to have a way of perceiving it so it does make sense that they have opsins.

Developmental biologist Maria Ina Arnone has been studying sea urchins for a long while.  As Pennisi writes,

Arnone proposed that the opsin photoreceptor cells in the sea urchin are positioned at the base of the tube feet such that they lie partially in the shadow of its calcite skeleton, allowing the skeleton to serve the same purpose as pigment in typical eyes—most opsins co-occur with pigment, which shields part of a photoreceptor cell so it can register the direction of incoming light. She has also shown that the photoreceptor cells connect to the five radial nerves in the brainless urchin, which may enable the input from the different photoreceptor cells to be compiled, much like an insect's compound eye does.

Arnone has found a second type of opsin, a "ciliary opsin," in different parts of the sea urchin, too, including the tube feet, the skin, perhaps muscle, and in the larvae.  She's not sure what this opsin does, and she can't say how one opsin became specialized for vision-related photoreception.

Arnone's work has led others to look for opsins in unexpected places, and they've now been found in the stinging cells of cnidaria, the phylum that includes hydra and jellyfish.  It turns out that the stinging cells are inhibited by light.  So, again, opsins are sensing light but for a purpose that has nothing to do with vision. 

And it's not just that opsins are doing their photoreception outside of eyes. It seems they also are sensing more than light -- e.g., in fruit flies they are involved in the mechanosensing required for hearing.  This is independent of light sensing.  The same group describing this role for opsins reports that they may also be involved in sensing temperature in these flies, and it's likely that opsins are involved in yet other kinds of sensory processes.

Pleiotropy, pleiotropy, pleiotropy everywhere!
And, opsins aren't the only molecules for which new roles were in the news last week. A paper in Nature Chemical Biology reports that tRNA synthetases, enzymes that have long been assumed to be specifically involved in translation of DNA into protein, have now been found to have nontranslational functions. "Although these new functions were thought to be 'moonlighting activities', many are as critical for cellular homeostasis as their activity in translation." And, apparently synthetases were recruited fairly early in evolution for their 'new' functions.

These new activities include but are not limited to (i) mediation of glucose and amino acid metabolism, (ii) regulation of the development of specific organs and tissues, (iii) control of the ying-yang balance of angiogenesis for the vasculature, (iv) triggering or silencing of inflammatory responses, (v) control of cell death and stress responses that may lead to tumorigenesis and (vi) amplification or inhibition of the immune response.

A new paper in PNAS reports that olfactory receptor genes, of which vertebrates have hundreds, were thought to be restricted to the nose and to function by detecting odorant molecules.  But some are now found to be expressed in other cells, including in kidney relating to blood pressure.  It was once suggested that they, so many and each so different, provided a kind of cell-type-specific 'address code' in the body.

There are many examples of newly understood functions for molecules that were once thought to do only one thing.  New functions for RNA itself are being discovered all the time, its role in gene expression only recently one of them.  We've seen RNA floating around in cells as long as it has been known about, and the extent of its function could have been described.  Except that no one thought of it.  Much of what used to be called "junk DNA" is now known to have function, and so on.

These are interesting cautionary tales about assumptions limiting what we see.  Opsins are for photoreception, tRNA synthetases are for translation of DNA to protein, olfactory receptors for smell.  Yes, scientists may stand on the shoulders of giants, but we also inherit the blinders of those who came before.

One reason why tunnel vision is a problem
If we understood what everything does, we wouldn't need science anymore.  Not knowing isn't the problem.  It's when assumptions block your view that it's a problem.  We've blogged before about the gene mapping study we've been involved in, looking at differences in head shape between inbred strains of large and small mice.  The mapping has, to date, identified 76 chromosomal regions that may contain a gene or genes that affect variation in head size and shape, a total of about 2500 possible candidate genes.

It turns out that function has been characterized for about 77% of the genes in these intervals, and information as to where they are expressed during at least one stage of embryonic development is available for about 60%.  So, not only are the data incomplete, but to add further difficulty, a not insignificant number of genes have been named for diseases they are associated with, or a single tissue in which they were first found, so that if they are in fact involved in the traits we are looking at, we won’t know it by name.  We might even be mislead.  

The gene called Brca1, Breast Cancer 1, for example, is associated with elevated risk of early onset breast cancer, but despite its name, it is not a gene for breast cancer.  It is a perfectly normal, functional DNA repair gene, expressed in a number of organs during different developmental stages and throughout life. The photo to the left, a figure from the expression database, GenePaint, shows BRCA1 expression at embryonic day 14.5 in a mouse.  The dark blue is where the gene is expressed: you can see that it's in the brain, the olfactory system, the tongue, the thymus, the liver, the lungs, and so forth doing, well, who knows what, really, at this stage of development.  It’s only when its function has been disrupted by mutation and the protein can no longer repair errors in DNA that it can be associated with cancer.

And, there are multiple genes identified as genes for a specific function – angiogenesis, or development of blood vessels, for example – but that are expressed in many different tissues.  So, gene names are often not reliable indicators of gene function, and all of a gene's functions generally are not known.  And in fact can't be known, since genes respond to environmental conditions, and gene function can't be assessed for every possible situation. 

We, and anyone doing GWAS or any other gene mapping studies, are faced with the task of poring over lists of genes for candidates for involvement in our trait of interest, be it facial shape and size or height or obesity or diabetes or schizophrenia, or anything else.  But, given the inconsistencies, vagaries, and incompleteness of the data, how are we supposed to choose?  We can’t rely on gene names or characterized function, and in our case we’ve found that about 95% of the genes in the map regions are expressed in the developing head, so we can’t pare down our list based on expression data either. And of course we can't do experiments with each of the 2500 genes in the map intervals.

We write all the time about the difficulties of predicting traits -- diseases, behaviors -- from genes.  There are many reasons why this is true.  Restricted vision is just one of them--whether in what opsins do or in what we 'see' them as doing!  But, few of us have the kind of vision that allows us to make sense of things in our data that are not what we expected to see.  Science progresses incrementally, in fits and starts, largely because that's true.  Thomas Kuhn wrote in his book "The Structure of Scientific Revolutions" about "paradigm shifts" which are leaps forward in our understanding.  They happen when someone finally understands the odd data that just didn't fit before.  Continental drift is probably the best example.  But, it doesn't always take, or make, a paradigm shift to really see what we're looking at rather than assume it's what we want it to be.