How does/would sperm-based germline modification work?

[Nathan McCorkle]

I've got an idea that you could make a condom-catheter type
microfluidic. I've read about male/female sorting of sperm, but how do
you edit the DNA? I know about CRISPR, transposons, and probably some
other similar things I can't remember, but any other ways that have
been thought of or experimented with?
Is adding a new chromosome a better idea? Would this prevent the 'new'
chromosome from being passed during future breeding with a
non-modified person? (I think that would make the neo-human less
offensive to anti-GMO folks)
Previously I've asked what political/social climate would be most
conducive to this work, and I think the answer was 'south korea', but
that was a few years ago, and I wonder if there's a better place these
days. Not necessarily a country with active research, but one that
would be supportive of active evolution, rather than being fearful?

--
-Nathan

[Koeng]

New YACs are known to be highly unstable, which are actually why they started using BACs instead during the human genome project. I'd assume it'd be the same for a new human one.

Possibly use a viral system to excise the DNA in the presence of, lets say, lactose and tetracycline, but only when in germ line cells. 

[Michael Tellier]

I remember a paper one/two years ago from William C. Earnshaw group which succeed in making a HAC (Human Artificial Chromosome) to pass through a few mouse generations. I don't think it has ever been used in human though but yeah stability is one of the major problem (also to be sure the HAC doesn't interact with other chromosomes).

Currently it seems CRISPR is what is used for germline editing both in different species. In human, George Church lab has made a try but the paper is not yet published and there are rumours of success by Chinese groups but until the publications came out we don't really know...

The idea for using CRISPR is to edit the hESC and then to develop them in sperm or gametes which is far more easier to perform than modifying the embryo directly (hESC is easy to obtain and grow whereas embryo is far more difficult to obtain and there is not a lot available).

The best country for doing this kind of experiments right now must be China since they don't seem to have any problem in performing it. In USA some scientists have already called for a moratorium but we still don't know if it will be followed. 

[Mega [Andreas Stuermer]]

http://www.pnas.org/content/103/47/17672.full

Here is a protocol how to do it. Basically, sperm has a layer that protects it from foreign DNA (alledgedly) so you purify it, mix it with DNA and do artificial/instrumental insemination of a female specimen. This episome is said to be totally stable, and is replicated during mitosis (almost as stable as a chromosome). It has an element that attaches to the chromosomes during mitosis and is co-replicated. It does not integrate so no chance of getting cancer from random integration and hetereochromatin unwinding/activation of downstream genes.
There also is an improved version of the plasmid with a native human promoter (gets less silenced).

[Michael Tellier]

There is only one episome per cell or the number is variable? Depending on what you want to do, gene dosage can be really important. 

For genome engineering, if you want to modify a native gene (editing, knock-out, insertion of a tag) and CRISPR is the best one right now. If you want to add one or more genes, there are several possibilities (replicable episome, retrovirus, transposons, plasmid integration, artificial chromosomes) but I think it is better to avoid anything that integrate in the genome due to the possible deleterious effects (even more for germ lines modification). For replicative episomes/artificial chromosomes, one problem might be the possibility of cross-over with other chromosomes which could be problematic (not sure if it is known). 

[Mega [Andreas Stuermer]]

Copy number was determined 5 to 10 per cell IIRC

[Mega [Andreas Stuermer]]

For some reason the episome does not seem to integrate. At least it was described. Probably it is recognized as a chromosome which also won't integrate into another chromosome (hopefully).

Cross-over needs some homology right? Else it would happen between native chromosomes too extremely frequently?

[Michael Tellier]

The fact that it doesn't seems to integrate is clearly interesting. I did some work on plasmid integration and it is dependent on DNA repair (NHEJ). For integration to happens, the plasmid needs to get linearized and be close to a genomic double strand break for the NHEJ to integrate the plasmid. This also explains why plasmid integration is not an efficient process (< 1%) since linearized DNA is prone to recircularization by the NHEJ or is degraded. 

http://www.nature.com/gt/journal/v10/n21/full/3302074a.html

For the cross-over, I was thinking about the human promoter region which might be similar to some endogenous promoter. I must say after that I don't really know the DNA size and similarity which is required for a cross-over to happen. But since we are talking about organisms which contains billions of cells,  even an unlikely event could happen (after that, an integration/cross-over is not obligatorily deleterious).

[Mega [Andreas Stuermer]]

How about using a mouse promoter or yeast promoter? It will be different...

[Michael Tellier]

A mouse promoter should work without problem. For the yeast one, it is a good question. I don't know how efficient an endogenous yeast promoter is in mammalian cells. I suppose there should be some activity but I am not sure if it will be at the same level as the one you see in yeast. 

Looking at this paper, a human promoter should be ok since the length of homology required for an efficient crossover is in kb whereas a promoter is below 500 bp.

http://escholarship.umassmed.edu/cgi/viewcontent.cgi?article=1018&context=rivera