In our previous post on the inheritance of disease, we referred to the ancient view that what we are is molded by what we experience in life, and is in turn transmitted to our children. To Darwin, the results would be screened for 'fitness'--survival and reproductive success--by natural selection as the core of the evolutionary process.
As we noted in our earlier post, there was great concern in Darwin's time about the nature of hereditary disease. It was only right to try to work out the principles of inheritance. Biobanking in those days would concern the offspring of consanguineous marriages, because if disease was hereditary, marrying relatives would increase or at least help reveal the nature of the risk.
In Darwin's 'Lamarckian' view of inheritance, while not simplistic and not entirely consistent, variation arose through life experiences and changed the physical elements (he called them gemmules) of our individual nature. The gemmules managed to get into the germline, and a child is a blend, as Darwin called it, of the gemmules it received from its two parents.
Now, Darwin clearly knew that offspring don't always bear their parents' traits. Some skip generations (e.g., what we know as recessive traits and incomplete 'penetrance'), some seem to come out of the blue as 'sports' (mutations), and males and females are clearly not simply blends of their parents. In The Variation of Animals and Plants under Domestication, where he expounded on his theory in the most detail, Darwin struggled to fit these apparent anomalies into his theory.
This is a highly deterministic view of life: you are made by your gemmules and you transmit them to your kids, and that will affect how well they do in life. While probability theory was certainly in existence in Darwin's time, to him ideas about chance were vague and informal relative to today's sampling and testing study designs. In his world it was natural to ask about consanguineous matings, because relatives would have similar gemmules that might reveal how inheritance worked, and this idea was only reinforced, after Darwin's time, in 1900 when Mendel's principles of inheritance were rediscovered. Dominant or recessive genes (or at least those examples that were clearly so were the ones studied and cataloged and used for research) were shown experimentally to be deterministic. In fact, Archibald Garrod's work published in 1902, on various recessive 'inborn errors of metabolism', as they appeared in offspring of cousin marriages, was the founding work in modern human genetics.
Given deterministic cause of human traits, the law of natural selection was to Darwin also essentially a deterministic screen of variation, in which every little advantage proliferated. It is no wonder that we have inherited, so to speak, a deterministic view of evolution, and hence of life, and hence of genetic causation. But this is essentially the long and subtle legacy of Darwinian's Lamarckian view of inheritance.
The validity of Darwin's theory of inheritance was suspected even at the time (and Darwin coyly said he was only tossing it out for testing), and we now know that it is fundamentally wrong--indeed, backwards in terms of causation. Genes, unlike gemmules, are causally related to traits, but not caused by them. They don't flow from all tissues into the gonads. They don't (by and large) change in response to circumstance as was thought.
And Darwin's essentially deterministic view of natural selection has a comparably long shadow into the present. We now know why the effects of genetic variants are typically statistical rather than deterministic. They do react to experience ('environment'), but the interaction is probabilistic. That means that prediction from gene to trait is usually not deterministic but only probabilistic as well.
This is why screening to find genes 'for' a trait like diabetes or heart disease or cancer or intelligence is a mistaken endeavor. When countless genes contribute to traits, they all vary in every population, the variants usually only have small statistical effects, and they interact with environments that also vary over time and among individuals, the gene-trait connection is typically weak.
The exceptions are the 'knockout' effects, when a defective gene basically just doesn't work or is wildly out of whack. There are many ways to break something, and hundreds of genes are known that, when broken, lead to disease in a basically deterministic way. But those diseases are typically congenital (present at birth or even earlier). They're not the target of genomewide association studies (GWAS), nor of the expensive gene-based biobanking initiatives now being established in many countries.
So why, given what we clearly know from GWAS, evolutionary theory, natural variation, and experimental studies, do we persist in thinking in an essentially Darwinian/Lamarckian way, that biobank kinds of resources are going to be so important to understanding the heritability of disease?
No comments:
Post a Comment