Statistical/probability issues in speciation

[Biochemist]

..Just wanted to repeat this snipet because I'm saying that the non-gene machinery does indeed control/direct the development of these new enzymes, and in almost every case the "new" stuff is indeed modification (maybe radical mods, but mods nonetheless) of an older "selected" sequence...

I think this is pure conjecture. We still end up with a gene that codes for proteins. You are suggesting that non-DNA activities feed back to the nucleus such that DNA is generated based on cytoplasmic requirements?

[Buffy]

...Actually, its important to note that self-modifying/evolutionary code gets quite ugly, with lots of "failed junk" lying around...just like you see in a cell...

Not really. Although there is indeed lots of non-coding DNA sitting around, it does not mean it is not critical. Unlike self-evolutionary code, we have incredibly efficient machinery.

Oh I overstate: its not "failed", its all potentially useful, just--as you say--unexpressed. The way that self-modifying code works is it has to use something for the "try this" fodder, so that stuff that's lying around gets used, cuz it prolly worked in the past, even if it gets modified... "Junk DNA" is a totally bogus (bogosity!), so I'll try to eschew its instantiation in my verbal elucidations.... :doh:

 

Cheers,

Buffy

[Biochemist]

And there's still no reason to *assume* that bits of each modification weren't expressed and possibly found "not detrimental" and recessed between environmental shocks, thus creating a "storehouse" of possible changes for when a nasty meteorite hits or CO2 levels go through the roof....

If they were expressed and "stored", they would show as trash proteins in the cytoplasm. There are very few of these. There would have to be (conservatively) billions of these for this model to work.

 

DNA can change, and be "stored" as a recessive allele. But the moment it expresses, it is time to fish or cut bait. If it expressed incrementally, we would expect to see gradualism in the fossil record. We don't. We see stasis, then new phyla.

 

The phyla burst onto the scene in just the environment when a recessive allele is likely to show itself.

 

The question is how did the recessive allele get there?

[Biochemist]

.The way that self-modifying code works is it has to use something for the "try this" fodder, so that stuff that's lying around gets used, cuz it prolly worked in the past, even if it gets modified... "Junk DNA" is a totally bogus (bogosity!)...

I think the non-codng part of the genome is a lot more than fodder to self modifying code. The size of the "junk" portion rises roughly conincident with complexity of embryological development. Some has to steer epithelial cells to be skin whiel they steer cardiac cells to be heart, all with the same genes.

 

The genome has to perform lots of functions in higher animals other than coding for protein (or RNA). Not so true on prokaryotes. They have almost no "junk" DNA.

[Biochemist]

Oh its cute when you're long winded! :doh:

Shucks. You'll turn my shy little head.

[Buffy]

..Just wanted to repeat this snipet because I'm saying that the non-gene machinery does indeed control/direct the development of these new enzymes, and in almost every case the "new" stuff is indeed modification (maybe radical mods, but mods nonetheless) of an older "selected" sequence...

I think this is pure conjecture. We still end up with a gene that codes for proteins. You are suggesting that non-DNA activities feed back to the nucleus such that DNA is generated based on cytoplasmic requirements?

:doh: Surely you jest! Of course non-DNA activities control changes to DNA! Now you can argue that "it all has to come from the DNA" but the *processes* that are effected by the non-DNA elements *dramatically* control how DNA is transformed, even if the "code in the DNA to create them" is relatively simple, and by the way stable (the same processes work in all cellular life, its the "kernel of the operating system"). Again, your math is based on the notion that the only way that changes occur is by purely random changes to sequences, and if something is "directing"--no matter how simply--the results are no longer Poisson distributed: not even close....

 

"We're taking a poll, are you voting for the saint or the sinner?"

Buffy

[Biochemist]

Of course non-DNA activities control changes to DNA! Now you can argue that "it all has to come from the DNA" but the *processes* that are effected by the non-DNA elements *dramatically* control how DNA is transformed, even if the "code in the DNA to create them" is relatively simple, and by the way stable (the same processes work in all cellular life, its the "kernel of the operating system"). Again, your math is based on the notion that the only way that changes occur is by purely random changes to sequences, and if something is "directing"--no matter how simply--the results are no longer Poisson distributed: not even close....

Again, the notion of improving on the poisson distribution is still conjecture. There is no basis for selection unless the phenotype is expressed incrementally. It is true that the preferential behavior of gene modifying processes (transposons, etc) is not random. But if you are suggesting that internal gene modifying processes steer the genome toward a functional improvement in the absence of phenotypical expression, you are agreeing that the next generation of genes is not selected: they are specified in the parent species code.

 

You aren't saying that are you?

[Pyrotex]

I am up to my neck in work right now. Sorry for not being able to spend more time. I'll catch up this weekend.

 

In the meantime, I would like to contribute a set of standards for determining the rates of evolution in chemical, bio-chemical, and macro-biological "soups".

 

To perform our statistical experiments, we must have a laboratory, and we must define an "experiment". We could go to extremes with this, but let's not. I suggest the liter (about a pint) as the volume of liquid within which our experiments are conducted. I also suggest th 24-hour day as the time boundary for our experiments. Whatever chemistry or small-scale bio-chemistry that can occur in one liter in one day is one experiment.

 

Now, how many experiments occur in our laboratory? We should not include all ocean water, for there is good reason to believe that ancient deep waters were as lacking in resources as today's. Let's include only those waters that have access to light and/or thermal energy -- and to minerals and mineral surfaces (which may act as catalysts). I suggest we assume coastlines and a quantity of thermal vents. Three hundred thousand kilometers of coastlines and vents is not an unreasonable number. Assume a kilometer of coastline/vent is on average 10 meters deep and 100 meters wide. This gives us 1000 X 10 X 100 cubic meters, or 10^6 cubic meters, or 10^9 liters per kilometer.

 

For 300,000 kilometers of coastlines and vents, we have 3 X 10^14 liters of liquid, or 3 X 10^14 "experiments" each and every day. Let's use the year as our standard unit of time, and assume only 333.3 days per year to make the numbers come out even. That gives us 10^17 experiments per year, where those bio-chemical experiments take on average one day.

 

We can arbitrarily coin three experimental rates based on our numbers above. If we are talking "fast" chemistry, we may assume 1,000 mini-experiments per day, giving 10^20 mini-experiments per year. Argument based on pure chemistry will usually require mini-experiments (10^20/yr); arguments based on life cycles of microbial life will usually depend on slower bio-chemistry (10^17/yr); and arguments based on life cycles of small animals with a 3-year (average) maturation will depend on even slower macro-biology (10^14/yr).

 

Is this acceptable?

[Biochemist]

In the meantime, I would like to contribute a set of standards for determining the rates of evolution in chemical, bio-chemical, and macro-biological "soups"....

 

To perform our statistical experiments, we must have a laboratory, and we must define an "experiment". We could go to extremes with this, but let's not. I suggest the liter (about a pint) as the volume of liquid...

I am not sure where you are going with this (although a liter is "about" a quart, not a pint.) It looks like you are setting up a format for a discusison of abiogenesis. I am OK with that, but it is not the topic of this thread. I set constraints on the problem such that there are far fewer variables (e.g., we already have DNA and the transcription machinery, we already have viable enzyme systems, we already have all of the fundamentals of a reproductive life form, we have a constrained time frame that is within bounds of the fossil record, etc).

 

Are you trying to go back to the biogenetic chemical soup as a starting point? If so, make it a different thread, and I would be happy to participate. If you are talking about my statistical model, are you trying to get a rough count of starting life forms?

[Pyrotex]

I am not sure where you are going with this (although a liter is "about" a quart, not a pint.) It looks like you are setting up a format for a discusison of abiogenesis. I am OK with that, but it is not the topic of this thread. ...

woops. my bad.

That's what happens when you jump in and make a post you've been thinking about for two days without reading the other posts.

I hate it when that happens!!! :)

[MortenS]

Time for me to enter the discussion...

 

I will start with Biochemist's initial points, and state which I agree with, which I disagree with, and which I would like to add some more information to.

 

1: Agreed

2: Agreed, although it is worth mentioning the redundancy in the genetic code, particulary in codon position 3.

3: Agreed, one theory is that natural selection acts (or acted) on the genetic code to reduce the effects of genetic errors (see point 2), and thus favored redundancy.

4: Agreed, although the functions of RNA is a bit glossed over here. RNA has a couple of important functions: m-RNA, r-RNA and t-RNA are the three best known, but RNA-molecules are also important components in processing of m-RNA via alternative splicing and polyadenylation, which is the main reason that only 25-30000 genes can code for the roughly 90-100000 proteins we find in mammals (not sure where you get the 300k number you have put forward earlier, I have not yet seen a single organism with an estimated 300k different proteins). Each protein coding gene make on average 3 alternatively spliced mRNA's.

 

5: Agreed

 

6: See comments in point 4.

 

7: Agreed

8: Not completely agreeing. Many proteins fold up pretty well in isolation from the rest of the cell. Chaperones are necessary to keep the folding protein from interacting with other proteins, since the cell is stuffed full of various proteins that may interact with the folding protein. There are examples of proteins that do not require chaperones at all, there are examples of proteins that use chaperones, but that can fold correctly in absence of chaperones when in a diluted solution, and there are proteins that cannot fold correctly without the presence of chaperones at all. Current estimates I have seen is that about 30% of human proteins require chaperones, but this is a relatively new field, so I guess this number can change.

 

9: Agreed

 

10: Agreed

 

11: Agreed

12: Wrong: Ribosomes do not exist inside the nucleus in the cell, mRNA has to be transported out of the nucleus, to the ribosomes that is freely in the cytoplasm, or in the endoplasmatic reticulum.

 

Now to the math section:

 

1. You are assuming a de novo formation of a specific protein. Evolution does not usually work like that. Usually, existing structures are modified. Only in successful frameshift mutations will you get de novo proteins that may look like something completely different. Also, you need to consider alternative splicing as a method of generating variations of proteins,

 

2. Not sure if this holds, but I am not going to debate this at this point.

 

3. Again you are calculating as if everything sprung into existence at once

 

4. Same as 3.

 

5. Same as 4.

 

6. same as 6

 

7. Autophagocytotis is stimulated by the lack of amino acids in the cell. Proteins destined for destructions get tagged with a tag (Ubiquitin) for a vesicle, the vesicle is then transported to the lysosome and the content of it digested. The lysosome itself does not have any means of recognising proteins as foreign or non-foreign.

 

The signals on proteins that determine them to be tagged are still largely unknown, although some clues have been gained: A. certain proteins with specific N-terminal amino acids degrade faster than others (Asp as N-terminal has a half-life of 3 minutes, while Ser as N-terminal has a half life of 20 hours),

 

B. proteins containing certain sequences of amino acids (e.g a socalled PEST sequence which is 8 amino acids long) have a

half-life of 5 minutes, but if these 8 amino acids gets removed, the half life increase to 50 minutes.

 

C. Signals may be hidden in the hydrophobic core of the protein, an abnormal protein with parts of the core exposed will therefore expose the signals, while a normal protein will hide the signals.

 

In addition, cells have mechanisms for removing Ub from the proteins, thus rescuing them from being destroyed.

 

 

Ok, so for the answer to your question:

 

I do have an explanation for how one ancestral globin protein became two gene families on two different chromosomes in mammals. Look at the figure in my attachment (shows human alfa globin and human beta globin phylogeny)

 

Start with one globin gene...then one chromosome duplication (or genome duplication). We know have two globin genes in two different chromosomes.

 

Let the two different globin genes collect mutations over time, as speciation etc happen in various lineages.

 

Unequal crossing over will cause two copies of the same gene on the same chromosome. If this chromosome then gets fixated in the population, all descendants will have the two copies of the gene on the same chromosome.

 

In one chromosome, the ancestral globin gene duplicated to (Epsilon + Gamma) and (Delta + Beta), later (Epsilon + Gamma) duplicated to Epsilon and Gamma, and (Delta + Beta) duplicated to Delta and Beta.

 

A similar story can be told for the alfa globin family.

 

 

We can also add myoglobin to this phylogeny, which would then be a sister group to both the alfa and the beta globin family, indicating that myoglobins split off from the alfa and beta globin lineage very early. If you add neuroglobins to the phylogeny, these split off even earlier than myoglobin.

 

This is a phylogeny done on human proteins only. It gets more interesting once you put the globins of many organisms into the same phylogeny. Then you get much better support for when the various duplications in the globin gene families occurred. As more and more genomes gets sequenced, I am sure we will get a more detailed history on the evolution of all kinds of proteins.

 

In closing:

 

if you want to calculate the probability of new enzyme system arising, you need to take into account:

- gene duplication

- genome duplication

- that it is more important what class the amino acid belongs to rather than what amino acid it is.

- cumulative selection, as provided by natural selection, and even by genetic drift, not single step creation ex nihilo as your calculations attempt to.

 

The power of cumulative selection over single step selection has been shown many times.

 

A short program that displays the difference between single step selection and cumulative selection, is a little program called MONKEY (http://members.aol.com/dwise1/cre_ev/monkey.html) (named after the infinite monkeys typing randomly and eventually producing Shakespeare's collected works).

 

I used the following sentence as the target sentence:

"EVOLUTION OF ENZYMES"

 

which is a 20 letter large sentence.

 

Now, consider the follwing model:

Start with a random 20 letter string.

 

Create two offsprings (copies) of the string.

Let one (picked at random) of the strings receive ONE random mutation

The offspring that is closest to the target sequence get to replicate, if none are closest, one is randomly picked out.

 

Let the selected one again have two offspring, and select randomly one of these offsprings to receive a random mutation, and so on.

 

Stop the iterations when one of the descendants is similar to the target sequence.

 

When I ran this several times, I arrived at the target sequence within 1000-3000 iterations.

 

Try to calculate the probability of picking the letters out at random in the correct sequence.

 

The probability should be (1/28)^20 or so, much different from 1/2000.

 

 

I made a sentence with 80 letters, and arrived at the correct sequence in around 9-11000 iterations.

 

If I increase the number of offspring, it will go even faster. If I increase the model from 2 offspring to 10 offspring, the 80 letter sentence is found in about 800-2000 iterations.

 

The program is not a model of evolution (since evolution does not have a goal, or rather, the goal is allways in the previous generation), but it does demonstrate the power of cumulative selection over single step random assembly of whole sentences.

 

 

 

Puh...this post is getting so long that I loose sight of what I have written, so I guess I will take a break here...(maybe focusing on one topic at a time would be an idea for me :))

[Biochemist]

Mort-

 

Always a delight to have you in the discussion-

12: Wrong: Ribosomes do not exist inside the nucleus in the cell, mRNA has to be transported out of the nucleus, to the ribosomes that is freely in the cytoplasm, or in the endoplasmatic reticulum.

You are (of course) absolutely correct. My brevity clouded my text.

1. You are assuming a de novo formation of a specific protein. Evolution does not usually work like that. Usually, existing structures are modified. Only in successful frameshift mutations will you get de novo proteins that may look like something completely different. Also, you need to consider alternative splicing as a method of generating variations of proteins,

I suppose we could debate "usual," but my only point is that there appears to be numerous circumstances where whole enzyme systems arrive de novo. Whole body plans arrive de novo as well, but that is another issue.

7. Autophagocytotis is stimulated by the lack of amino acids in the cell. Proteins destined for destructions get tagged with a tag (Ubiquitin) for a vesicle, the vesicle is then transported to the lysosome and the content of it digested. The lysosome itself does not have any means of recognising proteins as foreign or non-foreign.

Correct again, but I am not sure this is relevant. It is true that lysosomes have no (known) recognition scheme other that the ubiquitin tags, but the reliability of the ubiquitin centric tagging seems very strong. That is, both foreign proteins and naturally occurring proteins are succesfully tagged for destruction. The issue is that the system is highy effective at eradicating foreign, non functional proteins. If something gets transcribed and does not get used, it gets trashed.

I do have an explanation for how one ancestral globin protein became two gene families on two different chromosomes in mammals. Look at the figure in my attachment (shows human alfa globin and human beta globin phylogeny)

I do understand that there are good examples of expansion of protein families. I just don't think this case is the general model. There are hundreds of cases where the family starts de novo. The first globin is the issue (and the basket of related enzyme systems that function with it).

If you want to calculate the probability of new enzyme system arising, you need to take into account:

- gene duplication

- genome duplication

- that it is more important what class the amino acid belongs to rather than what amino acid it is.

- cumulative selection, as provided by natural selection, and even by genetic drift, not single step creation ex nihilo as your calculations attempt to.

I do understand the genetic drift mathematics, but they are not applicable to my initial question. All of these specific mechanisms for genome expansion only decrease the end-game probablity if they express themselves as they do in examples of genetic drift (like the ensatina salamander- I like them because they live in my back yard).

 

The exercise I am trying to address is the sudden arrival of new body plans during the Cambrian Explosion. That is why I picked 300 million years as the time frame. We essentially seem to have jumped from phyla like proto-coelenterates to mammals in 300 million years. There do not appear to be significant interim fossil expressions. That is, there appears to be no interim states that resulted in selection fodder (like my dear ensatina salamander). Unless you contend that the fossil record is really incomplete (as some do) then you have to accept one of two scenarios:

 

1) some sort of randomized activity resulted in new enzymes systems, or

2) a relatively small change in the existing genome of the parent species resulted in significant morphological change in the daughter species.

 

If (2) above is true (as it seems to be in a number of known cases) it suggests strongly that the precursur code for the daughter species was already extant in the parent species.

 

I can't think of a third option.

A short program that displays the difference between single step selection and cumulative selection, is a little program called MONKEY (http://members.aol.com/dwise1/cre_ev/monkey.html) (named after the infinite monkeys typing randomly and eventually producing Shakespeare's collected works).

 

I used the following sentence as the target sentence:

"EVOLUTION OF ENZYMES"....

When I ran this several times, I arrived at the target sequence within 1000-3000 iterations.

Again, this model assume that there was an extant influence that erased the model farthest from the target sequence. This would only by applicable if the interim sequences were expressed, and favorably selected. We do not have any evidence of such. We do know that this occasionally happens (like my salamander) but not that it created new body plans.

 

Outstanding post, Mort. Thanks for jumping in!

[Biochemist]

4: Agreed, although the functions of RNA is a bit glossed over here. RNA has a couple of important functions: m-RNA, r-RNA and t-RNA are the three best known, but RNA-molecules are also important components in processing of m-RNA via alternative splicing and polyadenylation, which is the main reason that only 25-30000 genes can code for the roughly 90-100000 proteins we find in mammals (not sure where you get the 300k number you have put forward earlier, I have not yet seen a single organism with an estimated 300k different proteins). Each protein coding gene make on average 3 alternatively spliced mRNA's.

Incidentally, you are correct on the protein count in the proteome, too. I don't know where I got that number either.

[MortenS]

Mort-

 

Always a delight to have you in the discussion-

Thanks...

 

 

 

I suppose we could debate "usual," but my only point is that there appears to be numerous circumstances where whole enzyme systems arrive de novo.

 

I'd like to know one or two examples of such an enzyme systems...so that I can research them.

 

Whole body plans arrive de novo as well, but that is another issue.

 

Only in the fossil record. Not in the genomic record. Hint: check out Hox genes and genetic switches, and how they are turned on and off in the various phyla that have been investigates so far:

 

It is probably a topic worthy of a separate discussion, so I will let it pass for now.

 

Correct again, but I am not sure this is relevant. It is true that lysosomes have no (known) recognition scheme other that the ubiquitin tags, but the reliability of the ubiquitin centric tagging seems very strong. That is, both foreign proteins and naturally occurring proteins are succesfully tagged for destruction. The issue is that the system is highy effective at eradicating foreign, non functional proteins. If something gets transcribed and does not get used, it gets trashed.

 

Probably not relevant, I agree. I just got the notion that lysosomes mysteriously managed to separate wanted from unwanted proteins, and spared the wanted ones. I hold that there is a tagging that tags both wanted and unwanted proteins, but that the tagging efficiency is increased when proteins show certain "signals" such as misfoldings, or specific sequences. Untagging mechanisms may work to "rescue" proteins that are tagged, but not misfolded.

 

 

I do understand that there are good examples of expansion of protein families. I just don't think this case is the general model.

 

Oh, take a look at the phylogenies of protein families at:

 

http://www.treefam.org/cgi-bin/misc_page.pl?home

 

Just enter a name of a a couple of protein families,and check out the most similar sequences. The Treefam traces genes back to metazoa.

 

There are hundreds of cases where the family starts de novo.

 

I would like to know the name of one of these hundred gene families that appeared to have arrived de novo around the cambrian explosion. My claim is that there most likely will be a precursor found in archaea, bacteria or one- celled eukaryotes, or a distant relative gene in plants.

 

The first globin is the issue (and the basket of related enzyme systems that function with it).

 

But then we are not discussing vertebrates anymore. Since globin genes are found in plants, fungi, protists and bacteria as well as animals, it has probably been present since the dawn of life (or at least back to about 3-2.5 bya).

Read: http://www.findarticles.com/p/articles/mi_m1200/is_15_161/ai_85175717

 

 

 

The exercise I am trying to address is the sudden arrival of new body plans during the Cambrian Explosion. That is why I picked 300 million years as the time frame. We essentially seem to have jumped from phyla like proto-coelenterates to mammals in 300 million years.

 

But then we mainly need to discuss morfogenesis, not the evolution of all sorts of functional proteins. Basically, regulatory genes and genetic switches can explain differences in body plans pretty well. There is still much to understand about regulatory genes and genetic switches for a lot of organisms, but the big picture is starting to get clear.

 

There do not appear to be significant interim fossil expressions.

That is, there appears to be no interim states that resulted in selection fodder (like my dear ensatina salamander).

Yeah, that is an unfortunate fact of the fossil record, I am afraid.

 

Unless you contend that the fossil record is really incomplete (as some do) then you have to accept one of two scenarios:

 

I do contend that the fossil record is not only incomplete, but that it is biased towards easily fossilized specimens (bones, teeth, shells), occuring abundantly in certain environments, for a long enough time period, so that fossilization was favorable. Organisms that did not have hard parts, lived in environments where fossilization was not favorable, or had a short geological life time did not make it to the fossil record. That this is so, is evidence by the many softbodied phyla we have today, that have not left a single fossil at all. Yet phylogenetic studies based on DNA place their divergence back several 100 million years.

 

The fossil record is very incomplete, even if we find all fossils that have been formed (which is the reason why I am much more interested in phylogenetic studies utilizing DNA and proteins than doing phylogeny of fossils). In reconstructing the history of life, comparative morphology, genomics and biochemistry is needed in addition to the fossil history.

 

Fossils are snapshots of what life looked like, and they are useful for establishing minimum age estimates for when organisms appeared on earth for the first time. They are, unfortunately, not very useful for establish maximum times, since the finding of one fossil earlier will push the boundary further back.

 

 

1) some sort of randomized activity resulted in new enzymes systems, or

2) a relatively small change in the existing genome of the parent species resulted in significant morphological change in the daughter species.

 

First of all, I am not sure if new enzymes are required for the evolution of new body plans (unless you consider DNA-binding proteins attaching to genetic switches as enzymes) The evolution of new genetic switches, or turning off genetic switches, during development of an organism should be enough for morphological change.

 

 

If (2) above is true (as it seems to be in a number of known cases) it suggests strongly that the precursur code for the daughter species was already extant in the parent species.

 

I agree. Evolution builds on code that is already present. Descent with modification. The modifications are due to recombination and mutation.

 

I can't think of a third option.

It's late, and I cannot either...

 

 

Again, this model assume that there was an extant influence that erased the model farthest from the target sequence. This would only by applicable if the interim sequences were expressed, and favorably selected. We do not have any evidence of such. We do know that this occasionally happens (like my salamander) but not that it created new body plans.

 

We have no fossil evidence for it, but we do have some genetic evidence for how body plans are built, during embryogenesis. Trace the phylogeny of the regulatory genes involved in morfogenesis and maybe you find some clues :)

[Biochemist]

Another outstanding post, Mort-

 

You raised a lot of good content here, and I think I would like to follow up on one point for the moment. I don't mean to ignore the rest of your (highly articulate) comments, but I am trying to keep the question focused on my view of the probabilistic problem.

...I would like to know the name of one of these hundred gene families that appeared to have arrived de novo around the cambrian explosion. My claim is that there most likely will be a precursor found in archaea, bacteria or one- celled eukaryotes, or a distant relative gene in plants....

This is a good postulate to evaluate.

 

I think I will research some of the enzyme systems in mammals that relate to higher-level organ function (eyes, endocrine systems, lungs, kidneys) and get back to you on this.

 

Thanks for the thoughtful reply.

[Biochemist]

Mort-

 

I think that we could have a good discussion (both pro and con) by discussing the phylogenetic development of insulin.

 

Good summary here:

 

http://www.findarticles.com/p/articles/mi_qa3746/is_200004/ai_n8899929

 

My hypothesis is that the modulation of glucose metabolism in mammals would have required multiple complex adaptive steps. That precursors to insulin exist in lower phyla is undeniable. But the modulation of insulin (in mammals via serum glucose in the pancreas) and the counterbalancing activity of glucagon (both of which are required) suggests that the existence of a particular protein precursor is immaterial without the concomitant parallel development of a) insulin release modulation, including feedback to keep serum glucose at homeostatic levels, :) availability of the counterbalancing hormone (glucagon) with roughly similar homeostatic controls to keep glucose up, and c) development of receptors on target end-organs (pancreas and most metabolic tissues for insulin, liver and muscle for glucagon) for both hormones.

 

It would he hard to count the number of separate enzyme systems that would have had to deploy in parallel, but there is a lot of information on this, so I think it would be a worthwhile discussion.

[MortenS]

Interesting...I would have to study this in some detail, its been a while since I looked at insulin. Is the following a good representation of how insulin works together with other enzymes? I have kept glucagon out for now, haven't started on researching that yet.

 

 

When glucose level is high, insulin (INS) attaches to the alpha subunits on the insulin receptor (INSR) in the cell membrane of some cells. (e.g.muscle cells and fat cells)

As insulin attaches, the beta subunits of the insulin receptor phosporylates (via ATP), thus activating the catalytic activity of INSR. This in turn, phosphorylates proteins such as insulin receptor substrate 1 (IRS-1), which lets phosphoinositide 3-kinase (PI3-kinase) attach to IRS-1. PI3-kinase then phosphorylates the cell membrane lipids. The phosphorylated cell membrane lipids acts as docking sites for the Ser/Thr protein kinase B(=PKB-AKT-1) and 3-phosphoinositide-dependent protein kinase (PDK-1). When these two are close together, PDK-1, phosphorylates PKB-ATK-1. Activated PKB-ATK-1 then moves into the cell, and activates other enzymes via phosphorylisation.

Eventually, an enzyme (not sure which one) attaches to vesicles in the cytoplasm that has glucose transporters (GLUT4) in its membrane, and stimulates these to move to the cell membrane and fuse with it (fusion involving VAMP, syntaxin, Munc 18C and NSF). GLUT4 can now move glucose into the cell. I have not found a description of the mechanism where GLUT4 is removed from the cellmembrane, but some form of endocytosis must happen.

 

I hope this is a reasonably detailed and accurate representation of what is known about the insulin pathway.

 

Heh, and now the job starts...it is going to take me some time to look into the phylogenetic history of all these enzymes.

 

Here is the list I am going to try and look at

 

INS

INSR

IRS-1

PI3-kinase

PKB-ATK-1

PDK-1

GLUT4

VAMP

Syntaxin

Munc 18c

NSF

 

as well as

glucagon and the glucagon-like peptides.

 

We'll see where this all will take us. It would be interesting to find an organism that LACK either insulin or glucagon, but not both, as that would put a nail in your hypothesis, as far as I can see.

 

I do think that the maintenance of homeostasis of blood glucose is important in mammals, at least, I just do not hold to the idea that the homeostasis of glucose levels have allways been important.

 

Anyway, I will be collecting some studies done on both glucagon, and the various enzymes in the insulin signal pathway, to see where this leads us. I do suspect an independent origin of glucagon and insulin, and that they have gained new functions during evolution, and that they are now working as antagonistic pairs in vertebrates. NCBI is down at the moment for me, so I cannot search for literature or download sequences to my phylogeny programs.