An ID hypothesis: Front-loaded Evolution

Hello Genomicus,

Welcome to EvC

The front-loading hypothesis is an ID hypothesis that builds on Crick and Orgel’s directed panspermia hypothesis. In 1973, in a paper published in Icarus (“Directed Panspermia”), Crick and Orgel proposed that the earth was intentionally seeded with unicellular life forms.

I would note that Crick and Orgel have since rejected their views expressed in the 1973 paper as overly pessimistic - see Anticipating an RNA World.

1) The front-loading hypothesis predicts that the first genomes encoded genes that would be unnecessary (but beneficial) to early life forms, but necessary to the appearance of multicellular life forms and plants and animals. It predicts that the first organisms were not proto-cells, but highly advanced cells capable of terra-forming a hostile planet and able to shape future evolution in biased trajectories.

How do you intend to test this hypothesis?

2) The front-loading hypothesis predicts that prokaryotic homologs of important eukaryotic/metazoan proteins will be more highly conserved in sequence identity than the average prokaryotic protein.

The trouble with this prediction is that it does not, in fact, differ from the predictions of conventional theory. If a protein exists with homologes in Bacteria, Archaea and Eukarya* then it must have be strongly conserved, otherwise it wouldn't exist in all three domains. Such strongly conserved proteins will trivially be more highly conserved in sequence identity than other proteins.

This prediction makes sense from a rational design perspective because designing these prokaryotic homologs with functions that conserve their sequence identity will ensure that their 3D shapes will not be significantly changed by the blind watchmaker, preventing the appearance of eukaryotes (I realize that this prediction might sound a bit confusing – it’s past midnight where I am – so I’d be more than willing to elaborate on this).

Interestingly, the 3D shape of proteins is much more highly conserved than sequence identity. I'm not really sure how your justification follows, either.

* - It seems to me that Bacteria and Archaea are sufficiently distinct that they should be separately addressed in this kind of deep time discussion.

Edited by Mr Jack, : No reason given.

Other than within the confines of that particular incident? What incident are you speaking of? The work of Freeland et al. show that the canonical genetic code is highly optimized for error minimization. Why is there no phylogenetic tree consisting of sub-optimal codes in basal lineages, gradually leading to more optimal codes? What do you think is the best explanation for this phenomenon?

Because the selective advantage of a high quality genetic code is high, and the nature of horizontal gene transfer means that there is an advantage in a shared genetic code.

Living prokaryotes are not ancient throwbacks, they do not represent the vestiges of an ancient world - they are extremely highly evolved organisms.

(By the way we usually use [qs]whatever[/qs] rather than [quote]whatever[/quote], which is why our quotes appear in the more prominent light blue blocks and yours just have horizontal bars distinguishing them. Just a quirk of the software )

We could identify genes important to the development and function of multicellular organisms, and BLAST them against prokaryote genomes, and see if we get any significant hits.

Well, I can flat out tell you the answer to that one: we do. The problem for you is that this isn't, in any way, incompatible with conventional theory. Exaptation is common, and well understood. These genes perform important roles in single celled organisms, and are later exapted to perform different roles in multi-cellular organisms. You need to come up with something that will distinguish your front-loaded genes from simple exaptation.

Further, I'd suggest that your hypothesis should lead to us seeing many, probably most, such multi-cellular genes in prokaryotes - we don't. There are some, yes, but most are not found in bacteria; a few more in Archaea and a decent portion in single-celled Eukarya or in multi-cellular but simply differentiated Eukarya. It seems to me that if front-loading was true we would expect a majority of such genes to be present in all species, would you agree?

If a protein exists with homologs in all domains of life, then it does not follow from conventional theory that it must also be highly conserved in sequence identity - not any more so than the average protein.

Of course it does. Highly conserved proteins have higher sequence identities than average proteins whilst, conversely, proteins not under selective pressure rapidly diverge. Proteins shared between the three domains must be highly conserved and thus will tend to have higher sequence identities.

For example, actin has a homolog in prokaryotes: MreB. What does conventional theory say about MreB's degree of sequence conservation, in contrast with other prokaryotic proteins (and without knowing the function of MreB)?

If I knew nothing about MreB I would predict that its sequence identity was higher than other proteins. As I mentioned, 3d structure is more highly conserved than sequence so it's possible that the sequence has entirely altered but it's a less likely possibility.

Hi Granny M,

I found an interesting comment from Ken Miller on this. He suggests that if front-loading were real, we would see the exact opposite of what Genomicus suggests; we would see enormous mutation rates on any front-loading sequences. They would be inactive and thus unchecked by natural selection. Those portion of the genome would be subject to runaway mutation.

If I'm understanding Genomicus correctly, this would not apply to his (her? I'm assuming, apologies if I'm wrong) ideas since he is suggesting that these front-loaded genes are active in their original form. His hypothesis is that the nature and selection of the genes present pre-prepares them for later adoption into functions vital for multi-cellular life and a particular form of multi-cellular life at that.

It's not that there is a silent library, it's that the active genes guide life's evolution down a desired path.

I've not been able to post in a while, and I see the thread has moved on a bit since I last posted. Rather than attempting to pick up dropped threads I think it's probably best if I make a more general reply to recent points. Genomicus: if there's any particular point you made in your responses to you that I've not responded to and you would like addressed, please feel free to raise it again and I'll answer it as best I can.

On the genomes of bacteria and the need for fidelity

I raise objection to your claim that bacteria should be more resistant to mutation and should benefit from more rapid mutation. In fact, both are likely untrue. The first reason to suspect this is straightforwardly empirical - error rates are lower in prokaryotic DNA replication mechanisms. Bacteria carry mutations at a lower rate than humans, and other other higher animals*.

The second reason is more theoretical: bacterial genomes are simply less redundant. Whereas animals tend to be diploid (they have two copies of each gene), prokaryotes are universally haploid (they have one copy of each gene): this means that any error is always expressed in prokaryotes whereas animals typically have another copy around to pick up the slack. Multi-cellularity also provides a further line of support because alterations expressed only in a certain cell type can drag down only that particular cell type and the rest of the organisms may be able to support that; single-celled organisms must function on their own. This also applies to environmentally relevant adaptations - a horse that can't suggest a certain plant can typically find a different plant to eat, whereas a lactose-metabolism deficient bacterium is stuck in its environment. Finally, the molecular pathways of multi-cellular organisms typically contain some redundancy, whereas bacterial molecular pathways are generally much more tightly regulated and the loss of single parts is more likely to destroy entire pathways.

On Pax6

In Message 64, you said:

Genomicus writes: Actually, the capacity for eye development is encoded at the root of the phylogenetic tree of life. Pax6 is a gene involved in eye development, for example. When you BLAST (blastp; default search parameters) the protein encoded by Pax6 (accession number: P63015) against the domain Prokaryota, you get significant hits (E-values < 1e-05). A PSI-BLAST search would almost certainly uncover hits with even greater significance. This suggests that eyes (and other major organs in Metazoa) were anticipated by the first genomes.

Did you carry out this BLAST yourself? Can you give more precise details on what you did? I have just attempted to repeat your BLAST to have a look at what is matching, but when I performed a blast with the European Bioinformatics Institute against their databases of Baceria and Archaea, I found no significant matches (link valid for 7 days, apparently). A similar search carried out using UniProt produced similar results - even the high E value results I did found matched simply in a high Pro/Ser/Thr-region which is unlikely to have profound biological significance.

Could you link to your results? Or, if you did not search yourself, perhaps you could identify your source?

As an aside I'd note that I would usually simply ignore any hits from BLAST with E-values as high as 1e-05 as simple noise; although I concede that deep-relationship searches must - by necessity - work with less-similar proteins.

* - Rapid-mutation mechanisms such as phase variation, and stress-response mechanisms - if they exist - aside.

Edited by Mr Jack, : No reason given.

Ah, found it. Thank you, Wounded King.

I'm afraid I must disagree with your contention. At the deep level of relationship we're talking about here, protein similarity is much more appropriate to look at than DNA sequence. DNA changes much faster so expecting any kind of base-by-base similarity is unreasonable and would likely be meaningless anyway.

That the matches have >30% identity isn't that bad at all, when you factor in E-values. You can have a database match to your query sequence that has little over 20% sequence identity, but if the match is accompanied by a good E-value, it's still significant.

You don't have good E-values. 1e-5 is not a good E-value, it's bordering on noise.

I've been looking at the homology found by the NCBI database and I note this: what similarity exists is found only in the helix-turn-helix DNA binding motif of the better matching proteins. Do helix-turn-helix (HTH) motifs have deep evolutionary history? Yes, they likely do, but that really doesn't support your front-loading hypothesis because HTH motifs are simply a means to match specific DNA sequences, usually found in transcription factors.

In other words, an HTH motif forms part of a mechanism that switches genes on or off according to circumstance. That's not any part of anything specifically eye related about pax6 but rather a faint superfamily relationship. You'll get similar results from any HTH protein you choose to pick.

This freely accessible paper discusses the deep family relationships between HTH transcription factors in depth.

(edit)

To illustrate my point about E-values, try blasting P11388 against the prokaryotes (P11388 being human DNA topoisomerase ii-alpha), here where there is genuine, deep-rooted, homology you get E-values in the sub 1e-35 range across a stack of different bacterial species from the various branches of bacterial life.

Edited by Mr Jack, : Added DNA topoisomerase ii BLASTp details

In Message 70 I tackled your "front-loaded eye" statement by pointing out that the similarities found by BLASTP are found only in the regulatory part of the protein,he helix-turn-helix.

There's a small typo on your part here, which I shall correct for clarity's sake, the regulatory part of the protein is not the helix-turn-helix. You mean the DNA binding part. The regulatory part would be the part that actually interacts with other proteins to enhance or suppress transcription.

This "deletion event" happens to plants, but not animals. When they start off with the same genome. What causes this to happen? If you want to stick with common ancestry, then there must be some causal factor other than the genome that makes the "deletion event" occur. Well, what is it?

I don't see that this necessarily holds. Gene deletion happens, neither we nor Genomicus are going to dispute that, right? You don't need a special causal factor to account for the deletion. What you need is some kind of selective filter that will mean that your front-loaded genes end up in the right "buckets" but that isn't particularly difficult to explain either - it simply requires that the "animal" genes work well together and the "plant" genes work well together and straightforward selection will do the rest.

The problem, as I see it, is rather more fundamental. If there are to be pools of front-loaded animal and plant genes which are whittled down to the correct sets in animals and plants we should see there being more genes in prokaryotes than in plants and animals rather than less. And we should see distinct, non-overlapping, sets of genes present in plants and prokaryotes; and animals and prokaryotes - but there's no evidence of that either.

Finally, the idea that plants and animals have particularly distinct genetic sets is fairly untrue. In general we strong homologies between major families of plant and animal genes, but they occur with very different relative importances. For example, I'm currently doing my research project on a MYB transcription factor vital to pollen development. The name 'MYB' derives from 'myeloblastosis' since these genes were first identified in cancer research looking at chickens but, as it turns out, there are only a handful of MYB genes in animals whereas they're a hugely important group in plants, with almost 200 having been identified in Arabidopsis alone. A converse example would be myosin; which has diversified into a huge array of forms important in muscle tissue in animals but is largely confined to its original role, shared with other Eukaryotes, ferrying vesicles around cells in plants.

And, of course, plants and animals are hardly the be-all and end-all of Eukarya. Quite apart from the familiar seaweeds, algae and fungi, there are a host of unicellular eukaryotes.

I'm sure you're familiar with this image or others very like it. It's the biological tree of life showing the relationship between the major domains and kingdoms. You'll note that plants and animals are actually quite close together, and far removed from the two groups of prokaryotes. Were there separate seed populations of prokaryotes, one for plants, and another for animals shouldn't we expect to see these pools group with their descendent multi-cellular forms?

I didn't say that the causal factor needed to be "special". I just said that there has to be one. Two lineages evolve in different directions. This cannot be solely caused by the same gene passed down from the same common ancestor, or they'd both evolve in the same direction. There must be another causal factor.

I don't see why you feel chance is an insufficient factor?

Thus, under the front-loading hypothesis, we would predict that important proteins in eukaryotes, animals, and plants will share deep homology with unnecessary but functional proteins in prokaryotes.

This seems a contradiction in terms; what do you mean by unnecessary but functional? Because I assure there are no known unnecessary but functional genes in prokaryotes (excepting, I suppose, the few molecular parasites known) under the usual meaning of these terms.

Kinesins don't seem to have any representation in Prokaryota. Using the rat kinesin Kif18B mRNA sequence (Accession Number NM_001039019.1) I carried out a BLASTn search - that's a search looking for "somewhat" similar sequences - so that the search was as broad as possible and searched in "prokaryotes (taxid:2157)".

As I suggested earlier, I believe BLASTp is more appropriate for these searches than BLASTn. BLASTp produces an identical result, however. Nor do our sloppy searches produce results different from those in the wider literature, those interested can read two open access papers on Kinesin evolution here and here.

Perhaps, but I think you'd be willing to agree that loading the first genomes with rhodopsins, globins, actins, kinesins, - or their sequence/structural homologs - that this would increase the chances of Metazoan-like life forms appearing on the scene.

I wouldn't, no. Metazoa use proteins derived from globins and actins, sure, but other proteins would have served just as well; and the broader set of hurdles to metazoan-life are sufficient to control the chances of such life.

I described two predictions made by the FLE hypothesis, one regarding levels of sequence conservation in prokaryotic homologs of cilia, and the other regarding deep homology and proteins in eukaryotes. These predictions are not made by conventional theory.

I'm afraid I missed the details of that first prediction, I'm assuming the second is this, from your opening post:

quote:

---

2) The front-loading hypothesis predicts that prokaryotic homologs of important eukaryotic/metazoan proteins will be more highly conserved in sequence identity than the average prokaryotic protein. This prediction makes sense from a rational design perspective because designing these prokaryotic homologs with functions that conserve their sequence identity will ensure that their 3D shapes will not be significantly changed by the blind watchmaker, preventing the appearance of eukaryotes (I realize that this prediction might sound a bit confusing – it’s past midnight where I am – so I’d be more than willing to elaborate on this).

---

We discussed this earlier in the thread, but it's perhaps time to pick it up again, in message 47, you said:

Mr Jack writes: Of course it does. Highly conserved proteins have higher sequence identities than average proteins whilst, conversely, proteins not under selective pressure rapidly diverge. Proteins shared between the three domains must be highly conserved and thus will tend to have higher sequence identities. I don't see how this would be the case. If a protein has homologs in all domains of life, this does not mean that it is necessary for life per se; it means that it is beneficial. And most proteins are, of course, beneficial. This doesn't constrain their sequence identity above average.

When we're talking about sequence identity, we're talking about levels of conservation across different kinds of organisms, different species, families, phyla, domains - whatever level you're choosing - not just selection within a single group. Of course, most proteins in a given organism will be selected for in that organisms offspring, and have been selected in its ancestor, but high sequence identity between larger groups speaks to something more fundamental - it means that protein has been under substantial selection in all those organisms and all their respective ancestors right back to their joint common ancestor.

It follows, therefore, that a protein found across multiple domains will show higher sequence identity within those domains than average proteins of those domains. Otherwise it would be insufficiently conserved to have made it between domains.

A much better comparison would be to look for proteins conserved with a domain but not between domains (e.g. a protein found in all prokaryotes but in no eukaryotes) and compare the level of sequence identity of that to the sequence identity of proteins shared between domains. But, even then, you're not really comparing like to like.