Monday, June 9, 2014

Insanity: genes 'versus' environment as causes

There are two statements about humans, or other organisms for that matter, that may seem inconsistent but that are both as solid as the Rock of Gibraltar, as fundamental and central as atoms to chemistry. Because of their deep importance, we feel that we must go beyond just stating these bedrock findings, and give rigorous proof of their truth.

Theorem 1. Genes cause every human trait!
Proof:  Without genes cells could not function, embryos could not develop, and there would be no humans to have any traits.

Theorem 2. Environment causes every human trait!
Proof:  Without their surroundings, differential gene expression would not occur, tissues would not differentiate, and there would be no humans to have any traits.

Given these incontrovertible proofs, it should seem rather trivially obvious that the search for the causation of any trait, should never be couched in one term or the other (genes or environment).  It is always genes and environment.  The legitimate interest in any search for biological causation should be clarified, first of all in the investigator's mind.  But this seems often, or even usually, not what happens when studies are designed, funding requested, or results reported.

Rock of Gibraltar; Wikimedia

As a result, there is a lot of needless contention and, sometimes, even abusive dismissal of points of view, in discussions of both the evolution and presence of human traits--be they normal or disease traits.

Unfortunately, it is not easy--even in principle--to separate these issues neatly.  For example, those who cling to genetic essentialism dismiss environmental causation as largely incidental or beside the point. Even 'epigenetics' is too close to environmental causation to be acceptable.  True enough, like many things in today's world of science (indeed, probably always typical of human affairs), we have fads and current epigenetic studies are in part just that--a way to get grants, get papers in the 'name' journals and get news attention.

However, epigenetics refers to the means by which gene expression is regulated.  Cells are always sensing their environment in various ways that lead to modification of their chromosomes that controls which genes are used and which are shut down.  In that very fundamental sense, every trait is an epigenetic trait.  Indeed, a given gene's 'environment' includes what's going on in the rest of the genome in that same cell.  The nontrivial scientific questions concern when, where, and how epigenetic mechanisms are at work in some context of interest--how the environment affects gene usage and its results.

Pure environmentalism is, however, also a fantasy.  Cells are genetically programmed to respond to their environments in various ways, ignoring some signals from the outside world, but responding to others, depending on its developmental stage and context.  Likewise intracellular conditions.  Genetic variation simply must be related to and circumscribes these responses.  It's this kind of sensing of environment-specific signals that causes one cell to become a stomach cell and another a brain cell, or for cells to divide or not, or produce particular proteins. Multicellular organisms wouldn't exist without differential cellular responses to environmental signals.

In this context, genomic causation and the evolution of genomic causation are similar.  Evolution only affects genotypes and environments that exist at any given time and place.  That variation changes in terms of its proliferation in its immediately local ecological context, which for convenience and perhaps having no better ideas, we usually consider as a phenomenon of relative frequency in some specified population. But these are abstractions of convenience whose relevance should be (but often rarely is) tested.  In any case, biomedical causation--like the genomic basis of autism or any trait you want to consider--similarly refers to some specific local context.

When incidence of a disease changes very rapidly, even if the trait has a genomic underpinning--and as we said above, every trait does--the change in incidence is very unlikely to be due to major changes in the local gene pool.  In that sense, it would be properly said to be 'environmental'.  Traits like autism, obesity, asthma, the success of Chinese ping-pong players, and many others are in this category.  Such attributions get under the collar of those whose careers and worldview are centered around genomes as the sole important determinants of life.  The response to rapid change may be acknowledged to be environmental, but in what seems to be a rather trivializing sense (yes, they might agree, China has manufactured millions of ping-pong tables which didn't before exist, but the champions are champions because of their Chinese genes).

A defense of the largesse of medical genomic research is that increased prevalence of disease is due to specific genotypes responding to the environmental change.  That is, environmental variation is leading to an epidemic of obesity because McFood triggers responses from some tractably few variants in the population that, before McFood, did not lead to disease.  Thus, the justification for expecting genomic 'causation' even when environment is the real 'cause'.

In fact, genomic variation changes just as fast as environments.  In a population (however you choose to specify it), alleles are coming and going.  A trait may in large part be the result of action of many different, individually rare, elements in the genome acting in their particular environment.  This coming and going means that the genes are as fluid in many ways as is the environment (however you may specify that).

The same applies to evolution
The same applies to much of the selectionist arguments for traits we see today.  Environmental change is of course the 'cause' of natural selection, and in a new environment selection will favor those few genes that confer major advantage (and will remove the alternative variants that, in the new environment, don't function well any longer).  So, our traits today should reflect selection history in a tractable way and we can see what is 'good' and not so good based on Nature's past decisions.

You can see the tangle of interests and rhetoric in contending views that are prevalent these days and soak up a lot of resources that might be devoted to things that are overwhelmingly 'genetic' (like cystic fibrosis) or 'environmental' (like the effects of exposure to CT scans).  But even this is a bit of false reasoning in defense of vested interests, because even most such traits are not so genetic rather than environmental or vice versa.

"What?  You don't think even a broken gene, that doesn't even work at all, doesn't lead to consequences that are purely genetic?  You don't think even smoking is an environmental cause of disease?"  The answer is: that's right.  For example, most of us are walking around with many 'broken' genes that have been identified as causes of disease, yet we don't have the disease.  We do a gene knockout in lab mice and they may get the same disease as found with the same mutation in humans---but it may have little, different, or even no effect in other strains of mice.  And, of course, most smokers don't get lung cancer.

If the arguments for simple genomic selective or biomedical expectations were accurate, then mapping would identify those few genes that responded to selection pressures or that respond to environments and make us sick.  Instead, what we typically find is that traits like obesity and autism and many others (we fancy that Chinese ping-pong ability would be similar) are not arising in those with particular alleles in a few genes.  Instead, variation in tens or hundreds of genome regions contribute to the trait, in the current environment.

It is in this sense that one can make the argument that a trait like autism or obesity is mainly an environmental trait.  Even if the responses of many genes lead to the manifestation of the trait in those individuals, the genotype is different in each, and what is in common is some aspect(s) of the environment.   Similarly, environmental 'causes' may be as individual, and diverse.  So the chase for explanations makes no sense if one insists on denying one or the other category or type of causation.

Not so simple
How one relates Theorem 1 to Theorem 2 in a real-world problem is not at all easy, even to think about clearly.  It is not enough to say we'll do regression on all genomic and environmental variables (G1, G2,...Gn, E1, E2,...En), assuming such enumeration and identification could actually be done, and then add some product (non-linear) 'interaction' terms:  G1 x E1, G1 x E2,.....   This quickly leads to essentially infinitely many things to test, even with just the product assumption.  And then there is how to measure the factors.  If one gene is part of another gene's environment (in some types of cell, under some conditions) then genes can be both genes and environments.

Saying these factors 'interact' may mean there is some complex sort of network of factors that, acting in some ways as a unit, affects some other network in each individual, each uniquely, to generate some output--like a disease--and then you have to decide what the right measure is.  Are 'blood pressure', 'IQ', 'average ping-pong victory differential', or 'stature' reasonable measures of a 'trait' for which you wish to understand the causal basis? Maybe, say, the square of blood pressure or the 2/3 power, more accurately reflect what is going on.

So how genetic variants  and environmental exposures 'cause' outcomes deeply depends on definitions and what one chooses to study.  It is far from clear what to do even once you've made decisions about that.   In this light, scientists' confident assertions and grandiose study design claims are very far from actually even acknowledging much less reflecting the depth of the problem we want to solve.  So we can just say "it's genes and environment" or "only genes really matter", and try to buffalo our way through.

The question may be scientific, but the problem is sociocultural
The problem, as we often note, is the polarized view of contending parties in the face of what is so obvious.  And the explanation has to do with human society and psychology and resources.  The explanations are choices in that sense, and that makes them sociopolitical.  Each of us somehow formulates a view about what is important, what we mean by 'genes' and 'environment' and how they relate to each other.  We have to judge what is most important, what risk effects matter, where to put research resources.  Nature herself doesn't provide a guide.  Facing up to complex, uncertain causal situations is more difficult than carrying on with what one knows and has been doing....and an keep one's lab running and churning out papers.  But the ability to face up is also something a culture itself may affect.  Often, we do what we learned or what has fed us, avoiding the risk of acknowledging that there may be something better or very different to do.

Albert Einstein famously quipped that the definition of insanity is doing the same thing over and over again and expecting different results.  Maybe he was wrong.  Maybe that's the definition of human nature.

1 comment:

Patrick Clarkin said...

No comments? I guess this post made too much sense.