[Martin J Sallberg (OP)]

There is evidence for horizontal gene transfer from bacteria to eukaryotic algae (How to Thrive in Battery Acid and Among Toxic Metals), and even from vertebrate to vertebrate (How a quarter of the cow genome came from snakes). Logically, this implies that horizontal gene transfer should happen between different cells in the body as well. For the relevance of this to evolution, consider "Aneuploid neurons are functionally active and integrated into brain circuitry" and various studies of site-directed hypermutation in immune system cells. These non-Malthusian mechanisms of evolution may solve the problem I highlighted in an earlier post, "Why do not ecological instability wipe out lots of important species quickly?".

[CliffordK]

It is interesting to see genes jumping from single celled prokaryotes to eukaryotes.  Although the needs of a single celled organism may be quite different from a multi-celled organism. 

In mammals, the reproductive cells are relatively isolated, so diseases that cause mutations in some cells often don't affect the reproductive organs and cells.

The Cow/Snake similarities are apparently in a gene called BovB which is a Retrotransposon, where genes are copied from DNA to RNA, and back to DNA.  What is unclear to me is whether this activity is normal, or if it actually requires a virus infection. 

No doubt stable transfer of genes to eggs, sperm, and blastocysts will be a hot topic in the future.

[Martin J Sallberg (OP)]

How can you be so sure that "the reproductive cells in mammals are relatively isolated"? There is evidence for symbiotic retroviruses in the placentas of placental mammals. They can certainly modify the genome of the fetus. In males, new sperm forms throughout adult life. There is also evidence that different groups of animals form their proto-reproductive cells in different organs in their bodies during fetal development (Nieuwkoop and Sutasurya 1979, Davison 1984). This means that there must have been events in evolution where the old germ plasm died out and were replaced with somatic cells that assumed the role of reproductive cells- within an individual lifetime. Mammals are a distinct group in this regard, separated by a somatic germ plasm supplantation from their reptilian ancestors.

And why assume that transfer from soma to germline must be bad, "causing diseases"? Is it not possible that it can work the other way around? I mean, a mutation in some body cells curing a genetic disease, acquired by other cells in the body by selective horizontal gene transfer aka cellular observational learning, and then passed to the reproductive cells.

The vast majority of the inheritance documented in twin studies are not accountable for by the parts of the genome that neo-Darwinian theory accepts as functionally active, meaning that the ENCODE discovery of activity in almost the whole genome should be taken seriously and not be downplayed into insignificance. The argument used by the neo-Darwinists themselves as "evidence" that most DNA must be junk (that natural selection can only clean up so many harmful mutations per generation) then instead becomes evidence that there must be means of adaptive evolution vastly more efficient than purging by death or sterility of whole individuals can possibly be. Also consider that reading DNA costs much energy (opening and closing the DNA spiral, moving the reading enzymes, among others) which makes major amounts of junk activity very unlikely.