I think Marc missed my main concept. In humans, and other furry mammals, X-inactivation occurs relatively early in embryogenesis. The examples I gave were to illustrate how tissues that might be sampled in DNA modification tests (or even behavioral tests) might derive from a single blastomere. In other words, the population being sampled is clonally derived and not random.

Let's imagine that in Mrs. Smith's embryos, X-inactivation occured at the 64-cell stage. Let's say that 1 of those 64 blastomeres gives rise to all (or most) of the lymphocytes in the resulting person. Now, let's look at DNA from blood samples from her twin daughters (naturally, named Gina and Tina).

Gina randomly inactivated the maternal X in her lymphoctype-progenitor-blastomere, whereas Tina inactivated the paternal X. And let's say there is a gene anywhere in the pathway for DNA methylation on the X chromosome. All (or a majority) of Gina's blood-derived DNA will be methtlated by the paternal allele and all of Tina's by the maternal allele. This would occur through her life, not just during embryogeneis. It could explain the observed difference in DNA modification in Gina vs. Tina without invoking lifestyle and other environmental factors.

Clearly, all genes required for cellular or organ function are not located on the X-chromosome, and that is not relevant. To get the results being discussed, only a single gene relating in any way to DNA or histone modification needs to be x-linked to get the observed effect. And my point is that these results would be cleaner in male twins, since there is no X-inactivation in them.

To answer Marc's other question, the paper in question was flagged as "hot" because it has been cited in 91 papers since it was published. I hope some of those papers properly controlled for X-inactivation-related effects.

"Female twins/clones are not truly identical because random inactivation of either the maternal or paternal X-chromosome, early in embryogensis, give rise to large patches of tissue derived from one or the other. One twin may have a hypothalamus, ovary, lymph node, or skin patch derived from her mother's X, while the other twin may have that same tissue derived from her father's X."

This assumes that all the genes necessary for the development of a particular organ/cell-type reside on the X, which I reckon is never the case.

As for the claim that different states of X-inactivation might have differing effects on the expression of other genes, sure, that I agree that could well be the case, but if there is a random unskewed distribution of each inactivated X allele in the population, it shouldn't significantly effect the statistical correlation they're observing here. Besides, considering they're correlating with age, I would have thought that X-inactivation, a mechanism that occurs at one point during embryogenesis, would not be responsible for any perceived changes in DNA/histone modifications as we the twins get older?

Unrelated issue: why has an article from 2005 been flagged as 'hot' on the front page of The Scientist?

This study will open the doors for the understanding of environmental agents (which we consume through food, drinks, medications, pesticides and other air borne factors) imprinting on genes, modifying the genes thereby leading to pathological outcomes such as cancer. True, long term observations are needed and results will not be available for a few years. However, crucial trands could be identified rapidly and studies expanded in a short time. Great!

Data from female sets (the majority of the study) will be confusing at best, or possibly not relevant. Female twins/clones are not truly identical because random inactivation of either the maternal or paternal X-chromosome, early in embryogensis, give rise to large patches of tissue derived from one or the other. One twin may have a hypothalamus, ovary, lymph node, or skin patch derived from her mother's X, while the other twin may have that same tissue derived from her father's X. This alone could possibly affect the expression of other genes being measured in this study.

I have an identical twin sister (we are 67 now) and we have very different health histories. My twin has had 2 autoimmune diseases- ulcerative colitis and RA for over 30 years, while I have had neither, although my daughter (39) has had Crohn's disease for 14 years. We have both had cancer, however- she had breast cancer (estrogen +) over 10 yrs ago, while I had endometrial cancer 6 yrs ago- we are both cancer free now. My sister lived in Miami for most of her adult life, while I spent most of that same time in the Burlington, VT area. We are both divorced and I did the majority of child-rearing of my 2 children, while she divorced later after her 4 children were grown. I have often wondered if location (urban vs rural), or life-style differences contributed to our varied health histories.