Thursday, February 14, 2008

Don't go marrying your cousin quite yet

I just read the Decode paper on reproductive success and relatedness (here. You can read other comments on it at Gene Expression here and here , and over at John Hawks. Overall I think it is a really interesting finding. One of the impressive things is that the number of grandchildren that a couple produced is still a decreasing function of relatedness out to 5 and 6 cousins. As the authors note this is pretty convincing evidence that this relationship is not due to a conscious choice by couples: (as ' Relationships at this genealogical distance are rarely known to the couples or their families and acquaintances in their social environment'), and so they conclude that this phenomenon is probably biological.

However I wonder whether there could be a non-conscious non-biological forces that could potentially give rise to this phenomenon. For example, in small communities there are fewer people to choose a partner from, so you are more likely to chose a relative (just by accident). In larger towns you have more choice and so are likely to pick some one less related to you (just by chance). If the number of children born per family is higher in smaller communities then partners with close relatedness will have more children will have more children purely because there are more of them in small communities. Does this make sense, or have I missed something? The authors adjust for spatial component in their model by putting the location of the couple (one of 21 counties) in as a fixed effect in the model. But this level of spatial structure might be crude compared to the information needed to detect an effect of town size. It is also a shame that they don't have marriage dates for the couples, as it would be interesting to know if the relationship between family size and relatedness is due to marrying earlier or a higher birthrate.

If this phenomenon has a biological basis that would be really interesting, and as noted elsewhere it could be due to subtly higher attractiveness between closer relatives (obviously not too close) or a high birthrate due to reduced genetic incompatibilities between mother and child. Even if it is not biological it is an interesting observation. It would also be great to see it replicated in different human populations. I wonder if this work could be replicated in macaque or some other model organism. I look forward to seeing more papers looking at this.

Thursday, February 7, 2008

KITLG and regulatory evolution

I thought I would post a few notes on the KITLG stickleback paper. Sticklebacks are a great system because many populations of marine stickleback have become isolated in flesh water lakes. A number of groups have done a lot of great work with them, and the availability of a good genetic map and draft genome is really accelerating them into a model system.

The paper is really nice and maps a major effect allele underlie losses in (gill and ventral) pigment in a fresh water population to the KITLG gene. This gene is known to effect skin pigmentation (along with spatial learning) in mice. The really great thing about the paper is they also find that the gene affects skin colour in humans. The authors use admixture mapping in African Americans to show that the gene has a sizable affect on the difference in skin pigmentation between Africans and Europeans (and presumably East Asians). The gene has a strong signal of a sweep in East Asians and Europeans. This is another great example of how a small set of genes seem time and time again to be the target of selection for changes in pigmentation.

I have a couple of comments on the paper, a number of these are acknowledged by the authors but i thought I would highlight them here.

The KITLG gene is expressed at different levels in the ocean and fresh water populations. This difference in KITLG expression level is controlled in cis (as allele specific the rt-PCR in F1 hybrids between the ocean and freshwater population reveals that the expression of the allele corresponding to the ocean population is lower than the allele corresponding to the fresh water population).

The authors suggest that the change at the KITLG gene in humans is also regulatory. But this seems somewhat speculative. The SNP they use for admixture mapping is in a conserved element. However, admixture mapping (unlike association mapping) can usually only narrow a signal down to a pretty broad region, as it relies upon recombination events in the 10's of generations since Africans arrived in America. Thus it is unlikely that the authors can be sure that changes within the gene are not responsible for the affect on pigmentation.

Also I felt that the authors were slightly quick to rule out coding changes in the stickleback KITLG underlying functional divergence. Obviously the expression level has changed, but that does not mean that change is automatically functional. The change in expression is not complete reduction of expression, so it is unclear how the protein levels are affected. The marine and freshwater fish are also different at 2 amino acid sites within the KITLG gene. The authors argue that these sites are not strong candidates for functional changes as they occur at non-conserved sites. But I find that this contrasts with the authors' views that regulatory evolution is a common path for genes with high pleiotropic effects because 'only mutations that may be compatible with viability and fitness may be regulatory mutations'. But the KITLG protein sequences of the freshwater and ocean stickleback do differ, and these changes (perhaps in combination with the regulatory changes) might underlie the changes in pigmentation. It seems slightly premature to suggest that this gene is a solid case for regulatory evolution (and the authors do acknowledge that).

Also the authors' own data suggests that this regulatory change in stickleback appears to have pleiotropic effects. The change in KITLG expression level in freshwater stickleback lowers expression in the gills (and the ventral skin) but also ups transcription in the brain. While the change in regulation in multiple tissues could be the result of multiple mutations, a parsimonious explanation is that a change to a single cis regulatory module causes these changes in different tissues. Thus while it is appealing that cis regulatory modules free up gene expression from the pleiotropy that coding changes might suffer, it is premature to conclude that such changes are easier to select upon than coding changes (especially as the gene in question has coding changes).

As I say I really like the paper, but felt that it slightly oversells the regulatory evolution aspect, which is interesting in light of the recent debate on the evidence for regulatory evolution (see here ).

Miller CT, Beleza S, Pollen AA, Schluter D, Kittles RA, Shriver MD, Kingsley DM.
cis-Regulatory changes in Kit ligand expression and parallel evolution of pigmentation in sticklebacks and humans.
Cell. 2007 Dec 14;131(6):1179-89