|
Author
|
Topic: Distinguishing Mechanisms of Co-option
|
John Bracht
Member
Member # 5
|
posted 26. March 2003 13:57
This thread is a continuation of ideas that I've been wondering about through the discussion on the Evolving Inventions thread, and I want to take these ideas in the direction of exploring the concept of cooption. In particular, I noticed that many of the counterexamples on the Evolving Inventions thread involved comparing different genes (such as Hox genes) that are similar in sequence, and then suggesting that this shows that evolution has coopted the gene from some initial function to its new function. While I think nobody can deny that genes performing different functions often show remarkable similariy between different organisms or even within a single organism's genome, the inference to a Darwinian mechanism underlying the change seems vacuous to me. However, it seems widely accepted among the mainstream scientific community to simply take these examples of similar genes doing different things as proof positive of the power of Darwinian mechanisms, working through the process of cooption, to create novelty.
To develop my thinking further, let me illustrate with some examples that I think will shed light on why I find such inferences vacuous and perhaps better explained by a design inference. I recently picked up John True and Sean Carroll's "Gene Co-Option in Physiological and Morphological Evolution" review, which summarizes some of the most compelling cases of cooption in the literature. I want to talk here about their example of the animal eye crystallins. True and Carroll state,
quote:
By far, the classic and best-studied cases of co-option in animal evolution are the lens crystallins of animal eyes. Crystallins are soluble proteins constituting 30-40% of the mass of the lens and have been characterized in many vertebrates and some invertebrates. These proteins occur in highly packed, transparent arrays in the lens, and function to refract light so that it may form a focused image on the retina.
I won't go into detail here, but the bottom line is that different lineages have different crystallin proteins, and these proteins are related to normal, cellular proteins and in many cases can still perform their normal, cellular functions (in addition to forming a regular array that is transparent to light in the lens). The various crystallin proteins are related to heat-shock proteins (chaperones), microbial stress-response proteins, arginosuccinate lyase, glycolytic enzymes, lactate dehydrogenase B, quinone oxidoreductase, glutathione S transferase, etc.
All these proteins are expressed in high concentrations in the lens of the eye, and True and Carroll ponder how this lens-specific expression originated:
quote:
It has been speculated that one possible evolutionary trajectory for the evolution of crystalliins is the acquisition of high, tissue-specific expression in the lens. It has been speculated that one possible evolutionary trajectory for the evolution this high-level expression of crystallins, which are ancestrally involved in stress responses, would be the ability of stress signals to induce high expression in specific organs. Subsequently, this inducibly high level of expression would then evolve into developmentally high expression in specific organs by regulatory changes. This could conceivably occur by alterations in cis-regulatory DNA sequences. For example, a stress-inducible promoter or enhancer could evolve lens-specific expression by acquiring a tissue-specific enhancer either by nucleotide substitutions or by a transposition or genome rearrangment that brought the transcription unit close to an extant enhancer.
Though it sounds reasonable initially, I think this scenario skips over severe difficulties with the model. The implication is that originally, the lens was formed with an opaque layer of ordinary tissue in place of the lens, in which the crystallins (heat-shock proteins, HSP's) would be induced by stress, providing the transparency necessary for vision. Later on, this stress-induced expression was encoded in the DNA. The problem, as I see it, is that simply inducing HSP's is not enough to cause an opaque tissue to turn clear (damaged tissue doesn't spontaneously become optically transparent). Thus, the genetic information for expressing some sort of tissue over the eye (which will evolve into a lens) is not enough to achieve selectable advantage. The mechanism for achieving optical clarity must be present at the same time the proto-lens tissue is added to the eye in order to gain selectable advantage over having a lens-less eye. What are the mechanisms that achieve optical transparency? Well, for one there must be the requisite machinery to lay down orderly arrays of HSP's or whatever proteins are being used as crystallins. Furthermore, the process of lens development is very complex and involves much more than just laying down lens crystallins. It turns out the lens cells are themselves laid down in highly regular arrays that are somehow encoded in the genetic program:
http://aussie-health.westga.edu/research/cataracts/development.html
Furthermore, the process of lens cell differentiation (required for optical clarity) is itself an amazing web of apoptotic responses:
quote:
Formation of the mammalian eye requires a complex series of tissue interactions that result in an organ of exquisite sensory capability. The early steps in eye development involve extensive cell death associated with morphogenesis. Later, suppression of programmed cell death is essential for tissue differentiation and in the adult, the immune privileged status of the eye is maintained in part through factors that induce inflammatory cell apoptosis. Experimental evidence suggests that suppression of apoptosis in cells of the lens lineage by fibroblast growth factors is one component of their action during lens morphogenesis. Fibroblast growth factors are also required for normal lens fiber-cell differentiation. This includes a degenerative step for organelles that is presumably an adaptation for the clearance of light scattering elements from the optic axis. The process of organelle degeneration may be related to apoptosis in a few of its features. Actively-induced apoptosis becomes important for eye development as the temporary ocular vasculatures regress. This too, is presumably an adaptation for the disposal of cells that would disturb the passage of light to the retina. Ocular macrophages appear to be essential for the induction of apoptosis in the endothelial cells comprising the ocular vasculatures. In the adult, inflammatory cells entering the eye are exposed to the pro-apoptotic agents transforming growth factor-2 and Fas ligand. The expression of these molecules in the eye, and their action in killing inflammatory cells, has evolved as a means of preventing inflammation and subsequent loss of vision. Thus, the eye offers a unique and versatile system for studying the role of programmed cell death in lens development, vascular regression and immune privilege.
Source: Nature Cell Death and Differentiation, December 1996, Volume 4, Number 1, Pages 12-20
To summarize this passage: Lens cells undergo a partial apoptotic (cell-death) response that results in clearing them of nucleii, mitochondria, golgi, endoplasmic reticulum, and other cellular organelles that would interfere with light transmission. However, they don't totally die, they become near-dead "fibers" that are still technically alive and able to dynamically respond to signals, such dopamine-induced adjustment of the refractive index. Furthermore, apoptotic pathways clear the developing lens of blood vessels (this from the Nature reference above). To preven any immune response in the eye, which could interfere with vision, the cells surrounding the lens secrete Fas ligand and tranforming growth factor beta-2, which induces apoptosis in those inflammatory cells which would quickly damage the delicate lens. It's all an amazing cascade of apoptotic responses, all of which are necessary for production and maintainence of optical clarity.
It is now obvious, then, that formation of a functional lense requires much more than just stress-induced expression of heat-shock proteins in a tissue. You need 1. orderly arrangment of these proteins into an array, 2. orderly arrangement of the lens cells into an array, 3. apoptotic responses in the lens cells and other tissues to produce optical clarity. So here's my question: did these multiple pathways get coopted at the same time as the crystallins? This seems a huge technical feat that the lens pulls off and the crystallins play only a small role in that feat. Somehow, all these different pathways must be coordinated into the end-goal, the lens. Furthermore, apoptotic responses are extremely dangerous things to mess with, since they have the capacity of doing enormous damage to an organism. Yet the eye is formed, in large part, by a delicately balanced, nested set of apoptotic responses. It's something like the blood clotting cascade, it seems balanced on a kinfe edge between just enough of a response and too much of a response (which would obliterate the lens entirely). How do all these processes get coordinated like this? Merely appealing to the similarities between crystallins and other proteins ignores the real engineering issues that have to be addressed here.
I also want to point out that the proteins that are utilized as crystallins don't seem to have been chosen at random; there seem to be some principles behind the "choices" cells have made. For example, regarding the the head-shock proteins as alpha-crystallins, True and Carroll note
quote:
The protein-stabiilizing chaperone function of alpha-crystallins may be particularly vital in the lens because degradation and extrusion of defective proteins cannot occur in this tissue and becasue the light-exposed environment of the lens causes large amounts of oxidative damage to proteins.
Or, about the cooption of gycolysis proteins:
quote:
Two taxon-specific crystallins, epsilon and tao, are active enzymes involved in glycolysis, This clycolytic function may be a clue to their mechanism of co-option as crystallins, becasuse differentiating lens fiber cells lose all of their organelles, including mitochondria. Hence, glycolysis in the cytoplasm is the sole source of energy for these cells.
Thus, the crystallin doubles as an energy source. Furthermore, there may be a purpose for the cooption of lactate dehydrogenase B (LDHB):
quote:
It has been speculated that the NAD/NADH-binding property of LDHB may have been important in its lens role, either for energy sequestration or for glare reduction, because the absorption maximum for NADH is at a wavelength similar to the range in which bird retinas are most sensitive.
Again, a good engineering reason to put those proteins in the lens of the eye.
The bottom line is that this "best example" of co-option in evolution is inferred based on protein similarities and an assumption that a Darwinian trajectory connects the two proteins. However, this inference actually ignores the real technical issues involved in forming a lens and ignores the fact that there are good design principles for many of the proteins being coopted.
Cooptation is common in human-designed artifacts. One obvious example: computer chips can be found in my HP laptop or in my car. However, it would be incorrect to assume that the car's computer was co-opted from the laptop (or any other computer) by a Darwinian process. Instead, the cooption process occurred by a design mechanism. This is a general feature, where designers can re-use designs for different purposes. Other examples include the re-use (cooption) of transistors, capacitors, resistors, etc., in different electronic devices like microwaves, radios, TV's, etc. The components are the same (or nearly the same) but are integrated in different design-driven ways.
So my question is this: what principles can help us distinguish between Darwinian co-option and designed co-option? We need to stop simply assigning co-option to a Darwinian process, since this begs the question.
I propose the following: if multiple components of a system are required for function and hence were coopted at the same time, then we ought to infer design. If only a single cooption event was sufficient to confer selectable advantage, we ought to infer Darwinian process (for example, the nylon-digesting bacteria are a good example of this single-component functional system). Since the lens of the eye requires multiple cooptation events at the same time to achieve selectable advantage (crystallins, crystallin matrix machinery, lens cell arrangment, apoptotic pathways), we can best ascribe this cooption scenario to design.
Rather that focus on the lens example in the discussion, I'd like to ask my fellow Brainstormers whether there are other ways to distinguish designed from non-designed cooption. How might we go about really testing the claim that a given pathway is the result of a Darwinian co-option, without simply begging the question?
In conclusion, I'd like to give a final quote from True and Carroll:
quote:
The acquisition of new roles by ancestral characters or new characters from old ones is known as co-option. Changes at the level of genes, either in regulation or function, underlie co-option of all the types of traits listed above. AS fundamental as this process has been in evolution, little is known about the mechanisms by which co-option of gene function takes place and whether particular modes of co-option are responsible for important eposodes of change in the evolutionary history of complex organisms.
[emphasis added.]
I suggest that the "acquisition of new roles by ancestral characters or new characters from old ones" is fundamental to any intelligent design process, in addition to a Darwinian process. If "little is known about the mecahnisms by which co-option of gene function takes place," there seems to be plenty of room for an intelligent or teleological mechanism to be involved in this transition. The question is, how do we distinguish the different mechanisms of co-option?
John
[ 26. March 2003, 22:54: Message edited by: John Bracht ]
IP: Logged
|
|
yersinia
Member
Member # 324
|
posted 26. March 2003 22:35
Hi John,
Interesting topic, I'm sure it will get some discussion although I might only be able to contribute occasionally.
I noticed one immediate difficulty that needs to be cleared up at the beginning as it will cause much trouble otherwise:
You start your critique of the crystallin cooption scenario by saying,
quote:
Though it sounds reasonable initially, I think this scenario skips over severe difficulties with the model. The implication is that originally, the lens was formed with an opaque layer of ordinary tissue in place of the lens, in which the crystallins (heat-shock proteins, HSP's) would be induced by stress, providing the transparency necessary for vision.
On what do you base the assumption that the "layer of ordinary tissue" (the source tissues for the lense are very different between e.g. protostomes and deuterostome IIRC, but I digress) was originally opaque?
I can think of several arguments supporting the contention that the layer was *not* originally opaque, but I am wondering if you can provide arguments pro- and con- on opaqueness and then an argument for which you think is more probable.
yersinia
PS: To connect this to your main question, the way you analyze a question like the above will have larger implications for scenario construction and scenario testing, which then has implications for deciding between designed vs. natural cooption based on your proposed criteria.
IP: Logged
|
|
John Bracht
Member
Member # 5
|
posted 26. March 2003 23:02
Yersinia,
Maybe opaque was the wrong word; perhaps translucent would be better (though regular cells, say epidermis, can be either opaque or translucent depending on the thickness). The point is, normal tissue is not optically transparent and the implication in the review was that originally, lens cells expressed high levels of stress-induced proteins which somehow provided the optical transparency which was later encoded into the developmental system.
John
[ 26. March 2003, 23:04: Message edited by: John Bracht ]
IP: Logged
|
|
yersinia
Member
Member # 324
|
posted 26. March 2003 23:49
quote:
Yersinia,
Maybe opaque was the wrong word; perhaps translucent would be better (though regular cells, say epidermis, can be either opaque or translucent depending on the thickness).
??? From a random dictionary:
quote:
translucent
1. Transmitting rays of light without permitting objects to be distinctly seen; partially transparent.
2. Transparent; clear. [Poetic] ``Fountain or fresh current . . . translucent, pure.'' --Milton.
Syn: Translucent, Transparent.
Usage: A thing is translucent when it merely admits the passage of light, without enabling us to distinguish the color and outline of objects through it; it is transparent when we can clearly discern objects placed on the other side of it. Glass, water, etc., are transparent; ground glass is translucent; a translucent style.
...vs...
quote:
opaque
(a) Impenetrable by light; neither transparent nor translucent.
There are apparently some conflicts about whether or not transparent and translucent mean the same thing, but they are clearly closer to each other than "opaque".
But you're right, let's not harp on definitions:
quote:
The point is, normal tissue is not optically transparent and the implication in the review was that originally, lens cells expressed high levels of stress-induced proteins which somehow provided the optical transparency which was later encoded into the developmental system.
John
I think is the source of the problem. My impression of the model is that one starts with with your typical small wormlike invertebrate -- something like the flatworm planaria. As the entirety of the organism is less than 1 cm long and maybe a millimeter thick, it and its tissues are more or less light-transmitting, whether you want to call it "transparent" or "translucent" or what have you. Assuming that you've already got a pit with a dark photosensitive spot at the back (like e.g. planaria or lancelets) then all you need is for a surface layer of cells (light-transmitting based on the previous argument, as you recall) to express the protein a bit more (a typical regulational mutation), changing the index of refraction between the water and the protoeye a bit such that a little more light is focused on the photosensors, increasing sensitivity a bit (or, if you like, a little less translucent and more transparent -- there certainly is a gradual scale between these two). Microevolutionary processes eventually spread this mutation to fixation and you're off on the optimization trail.
So it's not like someone suddenly pulled back a curtain and the light suddenly floods in, perfectly focused. You can see how on the latter scenario, a natural origin appears very tough, while on the former scenario perfection is a very late-stage result.
My point in all this is that one's confidence in the design inference is critically dependent on one's understanding of the evolutionary model, and that an innaccurate understanding of the evolutionary model will lead to a great many false positive design inferences.
IP: Logged
|
|
John Bracht
Member
Member # 5
|
posted 27. March 2003 21:45
Yersinia,
I think you're making the same mistake that I was trying to critique in True and Carroll's essay; perhaps I wasn't clear enough. My point was (and is) that you (and they) are skipping over some technical difficulties. Let me try to clarify.
Your small planaria-like ancestor starts with a cup of light-sensitive cells. Initially, there is no lens, and we're trying to understand how the lens came to be added to the eye. Without a lens in place, there is nothing to obscure or block the transmission of light--in a sense, conditions of perfect optical transparency. These conditions allow for maximum information about light directionality, etc. Now, think about what happens when you add a surface layer of cells: suddenly, there is a light-scattering layer that will effectively cause the incoming light to be randomized and diffused.
This is where my initial point comes in. Even given the increased expression of heat-shock proteins (or other stress-induced proteins), the tissue won't become optically transparent, much less able to actively focus light on the retina (which would provide selective advantage). Furthermore, it seems intuitive to me that merely increasing expression of HSP's in the proto-lens tissue won't make any real difference. These molecules will just be added to the already crowded cytosol, and there will be too many mitochondria, golgi apparati, ER, nucleii, blood vessels, etc. in the light path. To get an increase in light transmission, the HSP's must be laid down in orderly crystalline arrays--something that doesn't happen spontaneously. The reason you need orderly arrays has to do with optics: you need the crystalline structure of the HSP proteins because otherwise light passing through the tissue will be randomized by the disordered molecules inside. Furthermore, you need to clear the light path of organelles and blood vessels by inducing the requisite apoptotic and pseudo-apoptotic pathways.
So my point is that you need all these components (machinery to lay down HSP arrays, nested apoptotic pathways, orderly arrangement of lens cells) to achieve as much acuity as a lens-less eye. Without them, there will be no selectable advantage to putting a proto-lens tissue over the eye. To go beyond functional equivalence and get a focusable, dynamic, optically transparent lens certainly would require all the adaptations I've described, and more (like muscles to adjust the focus).
[ 27. March 2003, 21:52: Message edited by: John Bracht ]
IP: Logged
|
|
charlie d.
Member
Member # 159
|
posted 27. March 2003 23:00
quote:
Your small planaria-like ancestor starts with a cup of light-sensitive cells. Initially, there is no lens, and we're trying to understand how the lens came to be added to the eye. Without a lens in place, there is nothing to obscure or block the transmission of light--in a sense, conditions of perfect optical transparency. These conditions allow for maximum information about light directionality, etc. Now, think about what happens when you add a surface layer of cells: suddenly, there is a light-scattering layer that will effectively cause the incoming light to be randomized and diffused.
Perhaps. But, on the other hand, a layer of translucent skin may have protected the photoreceptors without significantly impairing vision - a significant enough improvement, perhaps. In fact, as far as I know, the original eye of all crystallin-bearing organisms may have started out under a layer of skin (I am not sure the "exposed" planarian eye is necessarily the ancestral state). Basically, it seems to me that unless we look at a complete compendium of existing eye structures and crystallin compositions, there isn't much to talk about, except what we project should or should not be there.
John, since you have already done some research, is there some recent comprehensive review on this?
IP: Logged
|
|
yersinia
Member
Member # 324
|
posted 27. March 2003 23:28
Well, we can see how quickly this gets into hard-core biology which must be understood before we can even begin to consider critiques.
For instance:
quote:
Your small planaria-like ancestor starts with a cup of light-sensitive cells. Initially, there is no lens, and we're trying to understand how the lens came to be added to the eye. Without a lens in place, there is nothing to obscure or block the transmission of light--in a sense, conditions of perfect optical transparency. Now, think about what happens when you add a surface layer of cells: suddenly, there is a light-scattering layer that will effectively cause the incoming light to be randomized and diffused.
But we have to stop right there, because although I know next-to-nothing about eye development, I know that in vertebrates the lens is derived from the ectoderm and the rest of the basic eye from brain tissue. Ectoderm predates vertebrate eyes by a long shot.
As for the supposed light-scattering qualities of a layer of cells in front of the proto-retina, you have to keep in mind that life on the micro-scale doesn't really work the same as at our scale. There are a great many average cells that you basically can't see with a light microscope without the addition of staining techniques. Even there, many of the sub-cellular structures are not distinguishable, which is why electron microscopy was such a revolution. The basic problem is that the wavelength of visual light is quite long (400 - 700 nm) relative to subcellular structures.
And anyhow, it's quite clear that image-formation had nothing to do with the first eyes, which were at best directional light/dark detectors. E.g. the planaria "eye" (treat it as an analog, the phylogenetic position of planaria is controversial IIRC):
 (Hmm, if this doesn't work go here)
...look at all of those nasty bits of nerve cell in front of the retina (the retina is mostly a pigment shield here, ensuring that light only reaches the photocells from one direction). It seems fairly clear that a few cells in front of the eyecup don't make much difference for a small, light/dark detection structure.
As the organism evolves to be larger, however, the situation might change, at which point a surface layer could be lost, or could be capitalized on to become a proto-lens. Both have evidently occurred (e.g. lost: nautilus).
quote:
This is where my initial point comes in. Even given the increased expression of heat-shock proteins (or other stress-induced proteins), the tissue won't become optically transparent, much less able to actively focus light on the retina (which would provide selective advantage).
As for rudimentary lense capability, basically all you need is a fluid more dense than the outside fluid and the light will bend inward:
Increasing the concentration of soluble proteins is one thing that would fit the bill.
quote:
These molecules will just be added to the already crowded cytosol, and there will be too many mitochondria, golgi apparati, ER, nucleii, blood vessels, etc. in the light path.
Blood vessels? Who said anything about blood vessels? I doubt that cephalochordates even have much resembling blood (they may have hemolymph or something), let alone vessels. When you are small enough the oxygen, CO2, and wastes can pretty much diffuse wherever they need to go.
As for the rest, see the above discussion about the wavelength of light. If, as the organism (and protolens) evolved to be larger, the various organelles began to be a hinderance, they could be gradually subtracted the end of development as the levels of protein were gradually raised. No cliffs here.
quote:
The reason you need orderly arrays has to do with optics: you need the crystalline structure of the HSP proteins because otherwise light passing through the tissue will be randomized by the disordered molecules inside.
"Randomized" light is an obscure notion. Light changes direction when the density of the medium changes (see figure above). Individual molecules have virtually no impact on visable light as they are so much smaller than a single wavelength of light.
Crystallization can come later (although I suspect that a great many proteins begin to form self-similar, repeating patterns when at high concentration).
With "randomization", you may be referring to the light-scattering by particles e.g. in milk -- but you need a substantial thickness of material in order for this to occur. A cell a few micrometers thick ain't going to do it even it contained materials able to scatter light, which hasn't been established.
quote:
Furthermore, you need to clear the light path of organelles
...tell that to the planaria eye...
quote:
and blood vessels
...see above...
quote:
by inducing the requisite apoptotic and pseudo-apoptotic pathways.
...selective cell death is unnecessary until you have a reasonably thick lens trying to acheive a reasonably precise focus. The basic slightly-increased-light-gathering directional light/dark detector does not appear to need it.
quote:
So my point is that you need all these components (machinery to lay down HSP arrays, nested apoptotic pathways, orderly arrangement of lens cells) to achieve as much acuity as a lens-less eye. Without them, there will be no selectable advantage to putting a proto-lens tissue over the eye. To go beyond functional equivalence and get a focusable, dynamic, optically transparent lens certainly would require all the adaptations I've described, and more (like muscles to adjust the focus).
You're not starting with a lense-less camera eye like a nautilus eye, you're starting with a directional light-dark detector formed from endoderm, and where acuity isn't even an issue. All we need is a bit higher concentration of solute in the overlying ectoderm cells to get the ball rolling (which probably wouldn't occur until the organism was a bit larger than a planaria).
Muscles to adjust the focus of the lens are optional, e.g. lamprey eyes don't have them lens muscles (see here).
Perhaps to make my point more succinctly, the probability of evolutionary scenarios cannot be assessed apart from extensive calibration against extant organisms (analogy/homology) and known phylogenetic relationships. Think like a biologist and you will have something with which to check whether or not your intuitive notions about likelihood and difficulties are accurate or not (and if you do no such calibration then you have a high likelihood that they will not be accurate).
yersinia
PS: Keep in mind that we've been discussing all of this in a cellular, deuterostome context, but there are evidently other ways to get eyes with lenses, e.g., this single-celled dinoflagellate apparently has one:
PPS: The tremendous range of complexity in extant eyes is the primary stumbling block for the attempt to discover barriers for evolution therein. This is, I think, why Behe has pretty much conceded the post-eyespot evolution of the eye to the evolutionists.
(The evolution of the photoreceptor cascade is another matter entirely and will require delving into the photosensitive systems of single-celled eukaryotes and prokaryotes, which is only at an early stage at the moment although it is worth typing "Chlamydomonas" into pubmed if you want to see some of what I'm talking about.)
[ 28. March 2003, 00:19: Message edited by: yersinia ]
IP: Logged
|
|
yersinia
Member
Member # 324
|
posted 27. March 2003 23:50
charlie,
I believe that the definitive reference is by Salvini-Plaw and Mayr (1977) although I can't find the full reference and haven't read it myself (so there!).
Anyhow it is discussed every time a new article on eye evolution comes out. They concluded that complex eyes had originated something like 15 times independently. Some took the discovery of common hox genes for eye development to challenge this conclusion but my sense of it is that the consensus is that all the eyeless homologs indicate is that the common ancestor of bilaterans had a primitive eyespot that used the hox gene.
Here is a page discussing it:
Evolution and Complexity
http://www.stephenjaygould.org/ctrl/futuyma_complex.html
...with a nice pic of snail eyes ranging from cups to full-on lens eyes.
IP: Logged
|
|
yersinia
Member
Member # 324
|
posted 28. March 2003 00:12
Ah, here is a more detailed discussion of the above reference in context:
Curr Opin Genet Dev 2002 Aug;12(4):430-4. Pax genes and eye organogenesis. Pichaud F, Desplan C.
I'll quote a chunk as it summaries Mayr's views and has some interesting tidbits on planaria and other eyes:
quote:
Pax6 and the evolution of the eye
Until the realization that Pax6 was involved in the development of very divergent types of eyes, it was widely accepted that various eyes represented examples of convergent evolution. Such convergence reflects the critical advantage of being able to perceive the environment as soon as the sophistication of the nervous system allowed the interpretation of images (i.e. after the split between major branches). Therefore, eyes are thought to have arisen independently ~40–60 times through evolution [27]. This view is also supported by the wildly divergent design of the contemporaneous eyes as well as by the fact that two types of photoreceptors, ciliary (e.g in vertebrates) and rhabdomeric (invertebrates) exist, sometimes in the same species. In addition, phototransduction uses significantly different pathways in the two types of photoreceptors (phosphodiesterase versus phospholipase C).
However, this view has recently been challenged by several intriguing resemblances between the genetic and patterning processes that control the development of eyes from very distant species, suggesting a common origin to these eyes [1]. One of these resemblances is the utilization of the Pax6 protein, which is essential for eye development in all species where it has been studied and was thus termed the `master regulatory gene' for eye organogenesis. Strikingly, ectopic expression of many Pax6 genes is also sufficient to redirect the developmental program of Drosophila (and in some cases of Xenopus) tissues toward an eye fate.
Other protagonists are shared for the development of many eyes. For instance, So/Six is a gene that encodes a transcription factor that has the capacity to induce ectopic eye structures in flies and is expressed in the eye in many species (see below [28]). Further similarities between the morphological routes that lead to retinal patterning have also been noted (e.g. waves of hedgehog in the morphogenetic furrow in the fly imaginal disc and of SHH in the developing fish retina, conservation of the atonal gene as the determinant of the first retinal cells; for review, see [29]). Therefore, in spite of obviously distinct evolutionary origins, the fly and vertebrate eyes do share many molecular and morphogenetic components.
These `functional homologies' are very surprising and suggest that, despite ~700million years of independent evolution, and structures that are based on widely different designs, all contemporary eyes might share a common ancestral `eye' whose development was controlled by Pax6.
How is Pax6 involved in eye development in various species? Was Pax6 already present very early on in a primitive species where it was involved in the development of a photosensitive organ (`proto-eye'), before functional image-forming eyes first appeared (a form of pre-adaptation)? Or instead, are we missing a critical early link when Pax6 first appeared, a real ancestor `eye' from which all eyes evolved––that is, an organ with photoreceptors, pigment cells and the ability to focus light to form images or to detect motion? An important task to answer this question is to define the level of sophistication of such a `primitive eye', or to find an alternative explanation to the re-utilization of similar processes. If PaxC in the very primitive coral Acrospora, which does not appear to have an `eye', were a bona fide Pax6 gene [13], this would suggest that Pax6 did precede the formation of eyes. It should be noted, however, that some cnidarians do have organized eyes, made up of two cell types, nerve cells and pigment cells with Tripedalia cystophora having genuine anatomical eyes. In addition, crystallin genes have been characterized in jellyfish [30]. Planaria, another group of less `primitive' organisms, might also hold a key to this mystery [31]. These animals present various light-detection systems that range from simple eyespots to more complex eyes. These complex eyes are composed of photoreceptors and pigment cells that form a cup capped by a lens, and are thus good candidates for an ancestral visual system. Strikingly, these eyes are characterized by the expression of a Pax6 homologue in both the pigment cells and in the photoreceptors and their organogenesis seems to rely on a conserved genetic module with Pax6 and So [31]. This suggests that the different types of eyes––the camera-like eyes of vertebrates and compound eyes of invertebrates––might share a common ancestor `proto-eye' (eyespot) and have diverged mostly due to modification of an ancient module, for instance by modification of the regulatory regions of these genes or of their targets. The apparition of a more elaborated `image-forming eye' would then be considered as a convergent evolutionary process that occurred independently in many species.
[...]
Conclusions
It is clear that Pax genes play an important role in eye development and that, at least in vertebrates, different Pax genes play antagonistic roles. The `functional homologies' found in contemporaneous eyes might reflect the early recruitment of a module in a primitive light-gathering organ or `proto-eye', presumably a transcriptional complex controlling photoreceptor differentiation. Gehring and Ikeo [1] have proposed an attractive model of `intercalary evolution' whereby genes used for downstream functions are recruited for more upstream functions when the structures become more sophisticated.
[...]
27. L.V. Salvini-Plawen and E. Mayr , On the evolution of photoreceptors and eyes. Evol Biol 10 (1977), pp. 207–263.
IP: Logged
|
|
Argon
Member
Member # 276
|
posted 28. March 2003 11:14
Yersinia writes:
quote:
As for the supposed light-scattering qualities of a layer of cells in front of the proto-retina, you have to keep in mind that life on the micro-scale doesn't really work the same as at our scale. There are a great many average cells that you basically can't see with a light microscope without the addition of staining techniques. Even there, many of the sub-cellular structures are not distinguishable, which is why electron microscopy was such a revolution. The basic problem is that the wavelength of visual light is quite long (400 - 700 nm) relative to subcellular structures.
Yes. I'm reminded of "glass" fish that I used to see in aquariums and pet stores. One could see all their organs through their muscles and layers of tissue. This is very pronounced in fish fry and in many aquatic insects. For example, see here. Basically, a few layers of cells doesn't obscure much, unless those cells contain absorbing materials and pigments.
I don't think having organelles in the light paths disrupts vision in humans and so I'm not too sure why they'd be a problem the development of the vertebrate eye.
[ 28. March 2003, 11:15: Message edited by: Argon ]
IP: Logged
|
|
Frances
Member
Member # 169
|
posted 28. March 2003 12:21
John Bracht argues for the gap of ID when he states.
quote:
If "little is known about the mecahnisms by which co-option of gene function takes place," there seems to be plenty of room for an intelligent or teleological mechanism to be involved in this transition. The question is, how do we distinguish the different mechanisms of co-option?
Bracht seems to argue that since little is known about co-option of gene functions there may be plenty of room for intelligent or teleological mechanisms. The question of course is two fold: 1. Would our increased understanding of co-option lead to less and less room for intelligent design 2. Will intelligent design approach help us fill in our knowledge about co-option?
What is the ID mechanism and pathway that is being proposed here? That seems to be the crux of the matter. But I thought that according to Dembski ID does not deal in mechanisms and pathways hence its focus on rejecting scientific pathways. John seems to be suggesting that there may be ID mechanisms of co-option.
Perhaps one way of distinguishing between ID and regularity/chance processes may be the observation that much of life shows signs of degeneracy rather than redundancy. The latter seems to be more prevalent in technology while the former seems to be related to the mechanisms of evolution.
Also I would like to point out that John's examples of cooptation in technology (microchips in his car) do not seem to be relevant to co-option. Unless the microchips somehow perform a different function than 'micro-processing instructions'.
quote:
This is a general feature, where designers can re-use designs for different purposes. Other examples include the re-use (cooption) of transistors, capacitors, resistors, etc
Transistors, capacitors etc are not used for different purposes, they still perform the original function. John objects that co-option is being assigned to Darwinian processes because it 'begs the question' but Darwinian explanations have been given and can be tested. What about ID explanations?
Hence the somewhat ad hoc definition quote:
I propose the following: if multiple components of a system are required for function and hence were coopted at the same time, then we ought to infer design. If only a single cooption event was sufficient to confer selectable advantage, we ought to infer Darwinian process (for example, the nylon-digesting bacteria are a good example of this single-component functional system). Since the lens of the eye requires multiple cooptation events at the same time to achieve selectable advantage (crystallins, crystallin matrix machinery, lens cell arrangment, apoptotic pathways), we can best ascribe this cooption scenario to design.
Ignores imho the historical pathways. The suggestion that multiple cooption events were needed seem to be 'begging the question' in the example of the eye. Especially since John seems to be arguing that science may not understand the mechanisms involved (allowing space for ID). So how can John on the one hand argue that "little is known about the mecahnisms by which co-option of gene function takes place" and on the other hand argue that we should infer design because of the need for multiple co-option events. Is this not an argument that appeals to what we do not know?
[ 28. March 2003, 12:31: Message edited by: Frances ]
IP: Logged
|
|
John Bracht
Member
Member # 5
|
posted 28. March 2003 16:21
Frances brings up some good points and helps get the discussion back toward the direction I want to take (I'll address some issues related to lens evolution a bit later).
Frances asked:
quote:
Bracht seems to argue that since little is known about co-option of gene functions there may be plenty of room for intelligent or teleological mechanisms. The question of course is two fold: 1. Would our increased understanding of co-option lead to less and less room for intelligent design 2. Will intelligent design approach help us fill in our knowledge about co-option?
There is an embedded assumption here that needs to be addressed. Namely, that Darwinian (or non-intelligent) explanations are the default explanation, and will eventually crowd out design explanations. Implicit to Frances's thinking is the idea that all co-option occurs by a Darwinian pathway and design inferences are always based on gaps reasoning, where science is simply ignorant of the "true" Darwinian mechanism. I don't want to give Frances a soapbox to promulgate this worn-out argument again (and will appeal to moderation if he tries to side-track the discussion that way). In his view, then, increases in knowledge will always provide non-intelligent mechanisms to explain all co-option events.
My point, though, is that this is an ASSUMPTION and not based on evidence. We ought to question this assumption and ask why we put non-intelligent mechanisms on a pedistal as the "default" explanation. How do we really know that non-intelligent mechanisms are responsible for the co-option events we infer in nature? Why should I accept this philosophical position?
This is the question I am trying to get at in this thread: how do we know that Darwinian co-option events really occurred by a non-intelligent mechanism? My experience is that there is no "test" that Darwinian thinkers apply to co-option events; rather they simply look at protein similarities and use that as "evidence" for their view. My point is that a design-driven co-option event would look exactly the same from our vantage point and hence the Darwinian comparison-of-similarity approach doesn't really test different mechanisms that might have been responsible for a given system. Obviously, those concerned about truth will wonder if there is a non-question-begging way to distinguish intelligent from non-intelligent co-option. So it's ironic that Francis asks this question:
quote:
John objects that co-option is being assigned to Darwinian processes because it 'begs the question' but Darwinian explanations have been given and can be tested. What about ID explanations?
Francis, show us how to test the Darwinian explanations. My point is that you're concept of "testing" is nothing more than begging the question. I want to see a rigorous test proposed that can distinguish co-option mechanisms and prove the Darwinian explanation for (say) lens crystallins; if you can provide one, please do so. I'd like to see it. But don't go assuming a Darwinian mechanism and then concluding a Darwinian mechanism.
So, I'd like this thread to focus on this question:
quote:
How do we independently test Darwinian co-option scenarios? And, as a corollary, how might we independently test design-based co-option scenarios?
Finally, Frances argues that
quote:
Transistors, capacitors etc are not used for different purposes, they still perform the original function.
I beg to differ. Transistors, capacitors, etc., are used for very different purposes--generating radio waves, sending electrons down a vacuum tube, tuning an electronic circuit, etc. They perform radically different functions. When we look at the computer example, I'm sure there are specialized functions that the car's chip performs that my laptop wouldn't be able to perform. They are each adapted to the functions they will carry out (different functions) even though they are both clearly similarly-designed chips.
Besides, I can easily turn this argument to biological examples of co-option. The TypeIII secretory system performs the function of secreting proteins, while the flagellum (which was supposedly co-opted from the type III system) also uses the type III homologous proteins to carry out the SAME function of secreting proteins (as part of the assembly process for the flagellum). Since, according to Francis, the function hasn't changed, this is not co-option.
It would be interesting to discuss different forms of co-option. Perhaps Francis would want to develop this as a means of testing different co-option mechanisms; namely, that intelligent co-option is of one form and non-intelligent is of a different form. I challenge him to provide rigorous and useful definitions of terms and criteria if he chooses to do this--it would be most interesting (and again would make a positive contribution to this thread).
Yersinia,
You make some good points about the transparency of thin layers of cells. I have been thinking about thicker layers of tissue, the sorts of layers that might form a lens. My critique applies more to the transition from thin layer to thick layer of tissue, and the inevitable blurring that will occur unless the other adaptations I mentioned are implemented.
You mention that a fluid more dense inside than the outside fluid will tend to bend light, which is correct. But a flat sheet will tend to bend all light rays evenly, producing a "shifted" pattern, not focusing anything into a proto-image. That requires a thicker layer of cells and a curved surface that can alter light rays differently depending on their spatial position, producing an image. This is something you're not going to get until you thicken the lens tissue and address the engineering problems of optical clarity, lens cell arrangement, apoptotic pathways, etc.
Furthermore, I still don't see how increasing the relative expression of HSP's will increase the density of the cells. My understanding is that cells are pretty much completely "packed" with proteins (around 300mg/ml) and so adding some more protein won't really change the density all that much--you'll just change the relative amounts of different proteins in the cell (or perhaps bloat the cell with protein).
At any rate, I don't want to bog the thread down too much with the details of lens evolution. Instead, I'd like to focus on the main question of the thread: How do we independently test Darwinian vs. intelligent co-option mechanisms?
Any ideas?
John
IP: Logged
|
|
charlie d.
Member
Member # 159
|
posted 28. March 2003 17:10
I don't have much time these days, but I'd like to jump in and drop a quick idea regarding John's question (and then run away from the consequences, as fast as i can :-) )
quote:
This is the question I am trying to get at in this thread: how do we know that Darwinian co-option events really occurred by a non-intelligent mechanism? My experience is that there is no "test" that Darwinian thinkers apply to co-option events; rather they simply look at protein similarities and use that as "evidence" for their view. My point is that a design-driven co-option event would look exactly the same from our vantage point and hence the Darwinian comparison-of-similarity approach doesn't really test different mechanisms that might have been responsible for a given system.
I think the answer lies in the prevailing operational metaphors for evolution and design. That is, evolution as bricolage, and design as engineering. In undirected evolution, we expect to find the unexpected: indirect routes, obvious "chances" missed, repeated evolution of the same structure through different means, etc. In other words, contingency. In design, we expect to find logic and consistency to dominate.
Now, let's look back at the crystallins: different lineages use entirely different proteins for the same purpose. Why? Sure, as John mentions in the beginning post, some of these proteins actually have apparent good engineering reasons to be utilized, but that raises the question of why didn't all crystallins use the same solution. For instance, if the epsilon and tao crystallins are particulary well engineered, as John argues, because of their double duty as metabolic enzymes, then why didn't the designer utilize them in most, maybe all crystallins? What good engineer would reinvent the wheel every time? Of course, the standard "we can't argue the goals and modus operandi of the designer" objection doesn't cut it here, because that's exactly what the "co-option is good engineering" argument is all about. (Incidentally, I guess RBH would claim that the extreme variation in crystallins is either strong evidence for undirected evolution or for MDT, which of course would be indistinguishable in this respect.)
Note that the same is true for the other main example of co-option cited in True and Carroll's review, i.e. anti-freeze proteins (and for those we even have evolutionary intermediates, but that's another story).
So, I'd say if co-option is the result of foresight, and not contingency, that the design proponents have to come out with an explanation for its obvious heterogeneity.
IP: Logged
|
|
John Bracht
Member
Member # 5
|
posted 28. March 2003 18:06
Thanks, Charlie, this is precisely the kind of input I'm seeking.
If I understand correctly, you seem to be saying that heterogeneity without apparent reason for that heterogeneity should be viewed as evidence for a Darwinian mechanism. Design mechanisms should produce heterogeneity only if there is a good reason for it. Is this a fair summary of your views?
(notice I'm not endorsing this view; it's not at all clear to me that the assumption is correct. However, I just want to be clear that I understand the general idea.)
Assuming this is a correct understanding of your view, I would like to probe the underlying assumption that a designer would only use heterogeneity if there are good reasons for that heterogeneity. Is it intrinsically better design to avoid heterogeneity unless there are good reasons for it? Do human designers always do this, or do human engineers sometimes use very different solutions to very similar problems?
Furthermore, might the crystallin example actually be a "reasonable heterogeneity" example; maybe the lactate dehydrogenase in bird's eyes is glare-reducing (by binding NADH) in this case because birds are often in a high-glare environment and there is a premium on quick, visual acuity while in flight?
Furthermore, it's my understanding that some of the crystallins discussed were actually mixed in various lenses. Thus, there might be a glycolysis crystallin plus a glare-reduction crystallin in the same lens. I don't have enough expertise to know, and it wasn't extremely clear from the True and Carroll review, but it seems that there might actually be overlapping subsets of these crystallins. Might it be possible that these overlapping subsets might make sense from an engineering standpoint? Has anyone thought of looking at this before (motivated, no doubt, by a Darwinian perspective)? Perhaps when I get some time I'll look into it. This seems like some design-motivated research that could prove extremely valuable.
Thanks, Charlie, for your contribution. I think this discussion is going in some positive directions.
John
IP: Logged
|
|
RBH
Member
Member # 380
|
posted 28. March 2003 19:34
In reponse to Francis's remark, John wrote quote:
I beg to differ. Transistors, capacitors, etc., are used for very different purposes--generating radio waves, sending electrons down a vacuum tube, tuning an electronic circuit, etc. They perform radically different functions. When we look at the computer example, I'm sure there are specialized functions that the car's chip performs that my laptop wouldn't be able to perform. They are each adapted to the functions they will carry out (different functions) even though they are both clearly similarly-designed chips.
John is confusing levels of analysis here. Actually, at one level, all capacitors perform exactly the same function (storage) as do all transistors of a particular type, all inductors, resistors, and so on. And your laptop could actually (and fairly easily) perform the functions an automobile's chip performs. It probably wouldn't fit under the hood, but wiring could be kluged up to allow it to replace a car's chipset.
At one level all those components are interchangeable with their peers. It is the combination of elements that yield the higher-level functions like those seen in oscillators and bistable flip-flops and so on. I spent some years as an electronics technician back in the days of discrete components (even tubes!), and built many a bread-boarded circuit out of whatever components happened to be in my workbench drawers.
Further, John's paragraph contains the same conflation I recently noted in a posting on ARN, the conflation of "purpose" and "function." In the several "purposes" John describes, the capacitors are performing exactly the same function. In one circuit that function may provide nonlinearity for an oscillator, in another a storage medium for generating current spikes, in yet another for filtering spikes from a circuit. But in every case, the capacitor itself functions in exactly the same way.
To say that a capacitor functions to store electrons is very different from saying its purpose is to store electrons. The latter implies intentionality, goal-seeking, and similar connotations. It contains an implicit default much like that John criticizes Francis for assuming. "Function" and "purpose" are not synonyms!
RBH
[ 28. March 2003, 22:28: Message edited by: RBH ]
IP: Logged
|
|
|
Printer-friendly view of this topic