Update 12-4-2010: A discussion on the degree of non-functional vs functional phosphorylation started on Google Buzz. As I couldn’t find a way to embed the discussion, I’m linking to it here.
Some rebels have no cause. It turns out that some phosphorylations are pretty causeless too!
Phosphorylation: Because all the Hip Kids are doing it.
Phosphorylation is one of the ways by which cells can change what proteins do after they have been produced. The simple addition of a phosphate to proteins can change where a protein goes, what it is doing and who it interacts with. Phoshorylation is everywhere and plays important roles in the cell cycle, transmitting signals and stress responses, to name a few processes.
Kinases are the proteins that do all the phosphorylating, of which humans have about 500 different ones. They can put a phosphate group on serines, threonines and tyrosines. So, if we have 500 kinases, there must be about 500 proteins that are phosphorylated right?
Not quite. Kinases are highly promiscuous. The phosphorylation database PhosphoSitePlus now documents that 10,500 proteins are phosphorylated, on a total of 60,000 phosphorylation sites! That’s almost half of our proteins! Considering that phosphorylation is the cell’s favorite regulation mechanism, is each of these phopshorylations of functional importance? Or could some of them be non-functional?
A bacterial genome: organized and efficient.
Drift, Selection and Messy Bedrooms
Wait, did I just end that paragraph with the word non-functional? We’re a naturally selected species right? Wouldn’t this mean that nature would have selected against ATP-wasting, non-functional phosphorylation? Perhaps. But let’s not forget that it was Darwin himself who said:
I’m convinced that natural selection has been the main, but not exclusive means of modification.
And he was right, adaptation and natural selection are not enough to explain the evolutionary changes that we see. Perhaps surprisingly, natural selection as a model works better for bacteria than for us, multicellular species. The genomes of bacteria are like highly efficient distribution centers, where every inch of space is packed with genes that are ready to be processed. This what you expect from organisms that evolve to be as fit as possible. But our genomes? They’re more like your uncleaned bedroom. For one thing, it’s messy, with lots of useless junk lying around. And when your genome finally finds those notes it took from last month’s course, it first spends half an hour trying to find out what on earth that horrible handwriting is supposed to mean.
Your genome: unorganized and inefficient. And quite messy, too.
For phosphorylation it seems to be the same story. In bacteria phosphorylation is used in effective signalling circuits, sometimes involving no more than two components. But what about us? We’ve got complicated signalling pathways, cascades and feedbacks that no sane engineer would design. Nevertheless, it works.
Why would such complicated, wasteful systems evolve? Why aren’t our genomes and pathways as streamlined as in bacteria? Since all this complexity does not provide obvious fitness advantages, natural selection is an unlikely answer. We must turn our attention to genetic drift and neutral evolution, two underappreciated concepts in molecular and cell biology.
Biologists love these kind of diagrams, even though they only show a fraction of all the complicated interactions that take place. No engineer on earth would chose to design such a complex system, to reach a fixed set of goals. Soure:http://en.wikipedia.org/wiki/Signal_transduction.
The (criminally) short version of genetic drift is that in large populations, selection pressure is big, since there are a lot of competitors. Slight decreases in metabolic or reproductive processes are punished, hard! In smaller populations (like in us, animals), there’s more room for slight deviations to become fixed in the population. If these deviations introduce complexity, and future modifications become dependent on this introduced complexity, the entire system is slowly drifting towards ‘foolish designs’. Maybe the slogan for eukaryotes shouldn’t be survival of the fittest, but survivial of the “just good enough” or survival of the lucky. That’s nowhere near as dramatic though.
I’m doing the entire field great injustice by this short summary. If you want to read up on neutral evolution, I recommend you read this great overview by Psi Wavefunction (for the second week in a row!) or articles written by Michael Lynch.
Non-functional Phosphorylation: a Tale of Two Papers
So, where does this leave us in respect to non-functional phosphorylation? Is it harmless enough to allow eukaryotic cells to live happily after after, or is it too wasteful and harmful to even consider? In an a letter published in Cell (2008), Gustav Lienhard calculated that extensive non-functional phosphorylation would consume ~1% of a cell’s ATP production. Sure, it’s a waste of energy, but hardly one that couldn’t be overcome by our cells. He furthermore points out that current studies don’t account for the stoichiometry of phosphorylations. So if a given protein is only phosphorylated in 1% of the cases, its phosphorylation might perhaps hardly be called functional.
Proving that something is non-functional is hard. Really hard. Even if we not observe a direct function of a phosphorylation, there’s always the possibility we haven’t looked hard enough or the right tests haven’t been developed yet. But Christian Landry and Emmanuel Levy found a way to indirectly investigate the conservation of functional versus non-functional phosphosites.
Their idea was that phosphorylatable serines and threonines (pS/pT) are more conserved than ‘normal’ serines or threonines (S/T). But when they looked at phosphosites in two different yeast genomes, they found that pS/pT sites were polymorphic 0.62% of the time, whereas S/T residues varied 0.63% of the time, hardly a difference at all! Landdry and Levy thought that if a part of the pS/pT sites are non-functional, they would be subject to the same constraints as the S/T sites, explaining this surprising observation.
Because there’s no agreed set of non-functional phosphorylation sites, Landry and Levy used proxies to investigate this hypothesis. They compared the rate of evolution in high and low abundantly phosphorylated sites, under the assumption that non-functional sites are phosphorylated less. Indeed, the highly phosphorylated sites evolved slower. They saw the same trend if they compared the evolution rate for phosphorylation sites for which a function was known, versus a dataset of phospho-sites without a known function.
In the end, they estimate that up to 65% of phospo-sites might be non-functional, a significant number! If it’s true that the majority of phosphorylations don’t affect proteins, this could save a lot of work for molecular biologists that are hunting for the functions these phosphorylations might have. What I find cool about this story is that Lienhard showed the possibility , and Landry and Levy the actuality, of non-functional phosphorylation. It’s a pity that stories like “Phosphorylation of Protein A on S192 does nothing at all!” don’t really make good papers.
To conclude: Two Door Cinema club sings that Something Good Can Work. Well, something good enough can work too! Even if a part of it is non-functional.
Lienhard GE (2008). Non-functional phosphorylations? Trends in biochemical sciences, 33 (8), 351-2 PMID: 18603430
Landry, C., Levy, E., & Michnick, S. (2009). Weak functional constraints on phosphoproteomes Trends in Genetics, 25 (5), 193-197 DOI: 10.1016/j.tig.2009.03.003
You might also like:
See our discussion of this http://www.landesbioscience.com/journals/cc/article/11066/
Clearly evaluating functionality of phosphosites by conservation alone is not going to cut it, as your paper illustrates. But I still think it’s entirely plausible that the phosphorylation system is robust enough to withstand a good deal of noise in the form of non-functional (less-functional?) phospho-sites, especially if the disadvantages these sites are not severe enough to be selected upon in small populations.
The problem with any functional/non-functional debate is, is that we humans are projecting function onto what we observe. A better question might be, whether all reported phosphorylations are significant enough, to be ‘worth reporting’. If the role of a phospo-site is to provide ‘background information’, do we really want to consider this as a significant feature of this protein? Or doesn’t it matter more or less than the specific position some amino acids might have?
(I’ve also posted this reply in the Google Buzz discussion linked to above)
The term ‘junk’ as in ‘Junk DNA’ should be composted. Although it isn’t functioning at a day to day level, it is vital to the process of mutation succession. Without it, only functional DNA could permutate, virtually guaranteeing 99% failure rate and continuously less fit progeny and eternal non-specialization. It ain’t junk, its storage, a treasure of past algorithms well worth their weight and volume in all except hyperspecialised species such as hummingbirds which trim all possible non-functional equipment to reduce starvation due to constant locomotion and liquid sugar diet. Most species can easily afford to retain their genetic gold, and would be far less fit if it was discarded.
I completely agree that ‘junk DNA’ is a horrible term, but I think it’s important to have a name for that part of a genome which came about by non-adaptive means. That this DNA can later be co-opted (structurally, fuctionally) is cool, but definitely not the main reason why it’s there.
The ‘junk DNA’ as protection against mutations is an interesting hypothesis, but one that has not been convincingly shown so far (and certain predictions it makes are certainly testable). Species seem to tolerate differing genome sizes remarkably well. Some newts have 40x as much DNA as humans do, while sea urchins get by on only 800 MB.
Continuing the analogy of junk DNA as a treasure trove of genetic diversity (which I really like!): there’s certainly some gems in there, but also a lot of brass nickels.
Thanks for the link. I definately agree with the ‘survival of the just good enough’ concept. Evolution is like peer-based exam marking, you don’t have to be ‘good’, you just have to be ‘better’!
“Why aren’t our genomes and pathways as streamlined as in bacteria?”
My answer will always be ‘cell nucleus’. Bacteria have a size restriction on their genome, so they *have* to make every base count. Eukaryotes can afford to hang onto random bits that might not be immediately useful as they have the space to do so.
I think your right that the nucleus played a huge role in raising the bars for how much noise a genome can take. I’m still a bit puzzled by the evolutionary origin of the nucleus though.. There’s enough interesting literature out there, I only need to find the time to read it!
At Kinexus Bioinformatics Corporation, we have spent a great deal of effort annotating phosphorylation sites in humans and other species. Our free PhosphoNET website (www.phosphonet.ca) presently features over 93,000 experimentally confirmed human phospho-sites and provides their evolutionary analysis in 20 other species. From our bioinformatics research, it appears that there is likely to be over 750,000 human phospho-sites in the 23,000 protein encoded by the human genome. Functional information is presently available for only a few thousand phospho-sites. Less than 10% of the human phospho-sites appear to be highly conserved, and these are mostly located in protein kinases. Interestingly, threonine phosphorylation is more conserved than serine or tyrosine phosphorylation, and highly conserved phospho-sites that have been shown to be functional are about 8-times more likely to be activatory than inhibitory. We have observed a negative correlation between the phosphorylation of most proteins and their expression levels. Based on this and other data, we believe that the major role of hyper-phosphorylation of proteins is to enhance their degradation. The addition of clusters of phosphates to proteins appears to permit their dissociation from subunits and unfolding, and this increases their accessibility to proteases. Because this appears to be a general phenomena, in most cases, phosphorylation at specific sites is not critical, only that many phospho-sites can be targeted.