Hyg, Neo, and $$

[down-n-out]

I was quite interested in the recent discussion about cells mutating to 
become hyg resistant, and also about the relative sensitivities of 
various cell types to both hygromycin B and G418.  I have only just 
started to use hyg. for selection, as I have to select double 
transfectants and I've already used a neo plasmid.  I too have noticed a 
low frequency of hyg resistance occuring in my control cultures and was 
happy that this was explained in the discussion.  Also, working with T 
cells, I've experienced the annoyance of major base-line resistance to 
both G418 and hygromycin B.  This leads me to my question (I might add 
that this issue has come up a number of times in my discussions with 
other people, and I've heard people come down strongly on both sides).
Here it is:  After you have selected for stable transfectants, how 
important is it to keep the drug on the cells?
I know that some people maintain all their drug-resistant cell lines in 
drug-containing media.  As the person who was working with B cells said 
though, this is *extremely* expensive when working with cells that have a 
high baseline resistance.  I also know that many people remove the drug 
ASAP, even before all the control cells are obviously dead, with the 
argument that the drugs could be mutagenic and the less time the cells 
are exposed to it, the better.
I guess the reason that some people prefer to maintain the selective 
pressure is to ensure that their inserted DNA is not kicked out of the 
cell's genome, but how common is this really?  ES cells that are targeted 
with a neo insertion are removed from drug ASAP and the neo cassette 
remains stable.  When you think about the number of cell divisions 
required to generate a mouse, that's a lot of generations.  This may be a 
bad example, but I'm wondering if the frequency of foreign DNA 
"rejection" has anything to do with the type of gene product.  It seems 
reasonable to me that in the absence of drug selection, the introduced 
DNA would be likely to be eliminated or selected against if it encoded 
something that the cells were not entirely happy with.  If, however, the 
foreign DNA is relatively benign, would this be more or less stable?