cell signalling


Evelyn Fox Keller on genes, evolution, and epigenetics

This is another archival repost from the old blog, first published way back in March 2008.

I’ve been following CBC’s How To Think About Science series, and caught the Evelyn Fox Keller episode the other day. It was interesting, but there were a couple of issues that I just can’t let pass. Keller talks about the hype of genomics ten years ago — during the human (and other) genome projects, when huge amounts of a new kind of raw data were piling up, and everyone was speculating about the interesting things we could do with it. Leaving aside the fact that many of the claims about genomics have and are coming true (albeit, over a longer time-frame than mainstream media imagined it would), I have a problem with Keller’s own bit of hype.

It’s epigenetics, of course — reversible and heritable changes (both between generations of cells and generations of individuals) in gene expression patterns which do not alter the DNA sequence itself. Epigenetics is where all the hype is in biology at the moment. Don’t get me wrong: I think the field is interesting and exciting. But as the hottest newest branch of biology, everybody knows the name, and few know the details. It’s cited as the mechanism of faith healing, mind reading, and homeopathy. In Keller’s case, epigenetics is cited as a problem for theneo-Darwinian view of evolution. By “neo-Darwinism”, Keller particularly means the gene-centered view of evolution. The name “Dawkins” may have arisen once or twice. The problem that Keller thinks that epigenetics has for mainstream modern evolutionary biology is that organisms may be able to control their mutation rates in response to changes in environmental conditions, and thus alter the rate of evolution.

Keller is referring, I suppose, to the checkpoints and DNA repair mechanisms that spot and fix errors in DNA replication during gametogenesis (the production of sperm and eggs). It’s difficult to make a copy of three billion base pairs without making a few mistakes, and too many mistakes in too many important genes add up to a miscarriage. So there are some molecular machines which follow the copiers around, checking that they got it right. The machines do their best to fix the typos, and in extreme circumstances, will kill the cell if the mistakes are too big. Where epigenetics comes in is in the regulation of those molecular machines. Epigenetics hires and fires the copyeditors. Specifically, there are epigenetic mechanisms which pack away genes — wrap them around proteins called histones, to form structures called chromatin. Locked away in these packages, the genes can not be switched on, and no new copyediting machines can be produced.

The hypothesis that mutation rates may be under control by some mechanism which recognises changes in the environmental conditions and responds by altering the expression levels of the copyeditors is, I’m sure you’ll agree, a fascinating one. But a problem for the neo-Darwinian picture of evolution? I’m not sure I see the connection, there. Here is how I imagine such a mechanism working: in the cells producing sperm and eggs, a set of receptors monitor environmental conditions; when environmental conditions change, those receptors pass the message on to the nucleus, where a set of machines make the appropriate changes to gene expression. Why do I propose such a mechanism? Because just such mechanisms coordinate development, transmit the messages of hormones, detect pain smell light taste, determine the activity of drugs, and do a-hundred-and-one other things in the cell. They are the default way of getting a cellular response to an external stimulus. And it has already been empirically determined that such a mechanism exists in the case of DNAcopyeditors. The DNA copyeditors are not switched on 24/7 — after all, they are needed primarily during cell division. The mechanism which switches them on was discovered by researchers interested in cancers, who found that this mechanism is often damaged in tumours, leaving thecopyeditors in a permanent ‘off’ state.

Perhaps it doesn’t work this way. Whatever. My point is that it is very easy to imagine a mechanism by which environmental changes lead to heritable changes in mutation rates — a mechanism which can be created by the simple modification of another very similar mechanism. That modification? Orthodox neo-Darwinian evolution. The receptors and signals, the gene expression machinery and the chromatin re-modellers are all the product of orthodox neo-Darwinian evolution. And the system no doubt remains at the whim of natural selection. The idea that evolution itself evolves is fascinating, but it does not appear problematic or revolutionary to me.

I said I had a second issue with the programme, didn’t I? Ah, yes, still on the topic of Dawkins and the idea of the selfish gene. Keller suggests that the ideas expressed by Dawkins have been surpassed and overturned by the modern developments of molecular biologists. Developments such as the fact that genes have complicated networks of interactions with each other. Gosh. It’s almost as though Keller hasn’t read The Selfish Gene. In TSG (Or was it The Extended Phenotype?), Dawkins is very careful to point out the fact that the “genes” of population geneticists — Mendelian particles of inheritance, and the “genes” for which the word “gene” was coined a century ago — are not quite the same thing as the “genes” of molecular and developmental biologists. Dawkins’ selfish genes need not be defined by start and stop codons, upstream promoters, or discreet messenger RNA products. Which makes Keller’s criticism largely irrelevant.

Whatever. Who cares? Somebody slightly mischaracterised an obscure academic problem, buried in an obscure podcast. Well, the main reason I care is that Keller is herself telling us that we should be more precise when talking about genes. When first used, the term “gene” was just a placeholder for a phenomenon we understood little about, she reminds us. Over time, we’ve filled in the details. The problem is, the population geneticists and evolutionary biologists have filled in different details to the molecular geneticists and developmental biologists. They’ve all continued to use the term “gene”, but they’re now using it to mean different things to each other. Oh, wait, haven’t I heard this somewhere before?

I guess it’s just that Richard Dawkins is so shrill and screechy that it’s impossible to read him carefully.

Filipe V. Jacinto and Manel Esteller, 2007. Mutator pathways unleashed by epigenetic silencing in human cancer. Mutagenesis 22(4):247-253; Free full text


Robustness: a new battlefield in the evolution wars?

This is another archival repost from the old blog, originally written in oct 2007.

Evolution, from the point of view of the geneticist, is the change in allele frequencies that occurs in populations over time. New alleles are created by random mutation, and others slowly disappear through natural selection or pure chance (“genetic drift”). When making random changes to a complex and finely tuned system, there are far more ways to break it than to improve it. But in fact, the vast majority of mutations have no apparent effect on fitness at all. There are many layers of redundancy in biology: changing the letters of genes does not always alter the building blocks of the proteins they encode; changing the building blocks of proteins more often than not has little or no effect on the parts of the protein that matter; many genes have more than one copy in the genome; and so on. Much of this has been known since the early twentieth century, and is part of what is called “neutral evolution”.

A similar, but not identical concept is “robustness”. Robustness is essentially a measure of how much change a system can tolerate without loosing its function. Robustness is related to neutral evolution: in more robust systems, a larger proportion of mutations could be described as neutral. Robustness is particularly used for describing networks of components, such as networks of genes activated during development, or signalling networks. Loss or change of function of an individual signalling component is often tolerated because there are several other routes by which the message is transmitted, and so the end result may be unaltered. Robustness would appear to be advantageous: the mutation rate can remain at a level that generates beneficial mutations, and mutations that may prove useful in future situations, without too many deleterious mutations arising; mutations in the somatic cell line do not instantly render a cell useless (though one can imagine this being a bad thing, e.g. in cancer); populations are more varied, which may be beneficial during hard times and environmental change; and so on. So the big questions are: is robustness itself an adaptation (and if so, for the obvious reason, or for a completely different reason?), an accident that has since become advantageous, or just one of many equally good ways that life could work? Is robustness an inevitable consequence of the evolutionary process, either as a product of natural selection or random drift, or both? Is robustness a prerequisite for complex multi-cellular life as we know it? And to what extent do the answers to these questions apply also to the other examples of redundancy described above?

There are a lot of assumptions made about what the answers to these questions are, and they reflect a bigger argument in evolutionary biology. In both the specific and the general arguments, the difference in terminology tends to be bigger than the difference in theory: both sides accept each other’s arguments, they just believe their own to be more important or interesting, and therefore talk about evolution in different terms. The emphasis of the “adaptationist” (pejoratively “Darwinian fundamentalist” and “Panglossian”) camp is on structures created by natural selection; the caricature is that everything in biology has a selective advantage. The emphasis of the opposing (“neutralist” or self-styled “pluralist”) camp is on the role of processes other than selection, i.e. chance, in the form of genetic drift and random long-term trends. In reality, the adaptationists admit a role for non-selection processes, and the pluralists admit the power of natural selection as the explanatory force in biology. The principle difference between the two camps is that, in the absence of evidence either way,adaptationists presume a feature is an adaptation, while neutralists presume it isn’t (and chide the adaptationists for making presumptions). A symptom of this divide is that the pluralists will speak in terms of neutral evolution, and the adaptationists will speak in terms of robustness; neither will admit that both ways of approaching the subject might be useful.


Introduction to cell signalling (part 1)

This is an archive of a piece originally published on the old blog in 2007.

A particular interest of mine is cell signalling. This is the field of molecular biology that looks at how biochemical events in the cell go on to cause responses elsewhere in the same cell, and in other cells. Signals may be triggered by some aspect of the environment – the senses, or chemicals produced by invading pathogens, for example – and are propagated by a variety of mechanisms – hormones and neuronal action potentials over long distances; reversible chemical modifications activating enzymes in cascades, or releasing stores of small molecules within the cell. These details of the mechanisms are not important. With signalling I had barely got my feet wet before I was pushed into the depths working as a research assistant on a particular aspect of signalling in a cancer. I therefore quickly picked up a lot about these mechanisms.

But I was very confused. In my cells, response X to drug A is dependent on MAPK – one of a series of postal workers who handle the signal (just the first of today’s bad analogies). But so is response Y. And response Z to drug B. And my colleague is working on a different cancer in a different cell type in a different species, which doesn’t even respond to drugs A or B, but which is still doing things with MAPK! This would be fine if MAPK really were like a postal worker: it could pass on as many signals as it liked. But MAPK is not just the medium, it is the signal: it has an off and an on state, a zero and one. This confusion was not remedied when I was taught signalling formally in a final year undergraduate module. We were taught the pathways that propagate the signal from its cause to its effect (the cause and effect themselves were largely glossed over). There was the MAPK family of pathways, membrane derived messengers, cyclic AMP, and calcium ion pathways. But this was just rote learning of facts, not an introduction to themes and concepts. The concept of “cross-talk” was briefly introduced: the different pathways are able to interact and influence each other. But this concept is itself misleading.

It is often difficult to distinguish between cause, intermediate signals and response. A convenient “unit of history” may be defined as the period starting shortly after the assassination of Franz Ferdinand and ending with the Treaty of Versailles. During this unit, a large number of interacting intermediate events led from the assassination to the treaty. But this unit of history is not self-contained: World War I had many different causes converging, and many different effects diverging afterwards. History is not just a series of causes and effects, but a network of interacting causes and effects.

Lets look at the “responses” to signals. A typical “response” may be to alter the rate of expression of a set of genes. But gene expression itself resembles signalling in many ways: a cascade of recruitment of protein-based machinery to the gene and intermediates, and chemical modifications to other bits of machinery to modify their activity. In many cases, the gene product may itself be classified as a signalling molecule, going on to activate further pathways. Another response may be to release (or re-stock) stores of sugars (and fat), and again, some aspects of this process are hard to distinguish from the signalling itself, and the result – altering the concentration of sugars in the cell – may have effects on some of the other ongoing signalling events. Other “responses” may be to proceed with (or halt) cell division (itself a series of signalling and expression events); or to secrete some proteins out of the cell, where they may be picked up by neighbouring cells, or end up in the blood, where they are picked up by distant cells (where, of course, they go on to activate further signalling).

When we think about signalling (and when it is taught), we tend to think about it in terms of distinct packets, with a cause and an effect, and a series of intermediates. But like history, this can be a misleading way of thinking, especially when the events occurring before and after this packet are themselves complicated and important. To talk of “cross-talk” between historical events would also be misleading: it would suggest a number of distinct moments when the otherwise discreet parallel cascades of historical events were able to influence each other. Signalling, like history, is a network of interacting events. Dividing signalling into units is useful for studying it, but only if you understand that what you are looking at is a simplified model, and not the complete picture.