Hi, everyone - long time reader, first time poster:

I'm also among the "strongly opposed" camp regarding mixing colonies for
expression.  Namely, when expressing proteins which might be slightly toxic to
their host cells - either as a direct toxin, or by somehow altering the host's
gene expression - we've observed a *strong* colony-to-colony variability in
expression level.  Our sense here (and we've done no direct investigations to
the matter, but rather are going on what collaborators have told us), is that
if the gene is toxic, the small amount of leaky synthesis that occurs prior to
induction (perhaps even too little to make out on a gel) might serve as a
strong enough selective pressure to turn off its expression.  This could be
through mutation of the gene itself, the lysogen, or any number of
uncharacterized and otherwise unlikely epigenetic phenomena.

We also see that those cells which *do* express the target gene tend to grow
more slowly, which is circumstantial evidence supporting our model.  
Namely, if
the cell hasn't somehow altered its expression to turn off the transgene, then
the leaky expression during lag- and early log- phases causes a mild toxicity
which slows growth.  Of course, once you hit mid log and induce, you expect
growth to slow down substantially, so there's really no difference post
induction.  However, I'd guess - as is consistent with my own experiments -
that mixing a bunch of random colonies would allow the anti-expression 
selected
individuals to outgrow their slower, protein-expressing counterparts, and take
over the medium.

One thing you can do is - as others have suggested - is restreak from your
glycerol stocks, pick ~10 colonies, and in parallel test their expression on a
5-10mL scale.  Save a bit from each prior to induction and propagate that in
uninduced culture so that they'll grow as a starter stock while you check
induction on a gel.  Then, only play with the guys that induce.  In 
general, we
also see that letting the starter culture overgrow *too* much can be bad.  For
genes which express robustly, it's not a problem, but for our problem 
proteins,
we never let the overnight starter go for more than 12 hours before 
inoculating
growth cultures.

Another thing to consider is just cloning the gene - as is (if the codon usage
isn't a *serious* problem - i.e., requiring a rare codon tRNA or what 
have you)
- into a vector which allows tighter expression control.  If it's lag-phase
toxicity that's selecting against expression and killing your protein growth,
then a vector which shows less 'leakiness' in early growth might 
circumvent the
problem altogether.  Your choice of growth medium might help in this 
regard, too
- though I've never checked directly.

Also, try inducing overnight at 15C to room temp, rather than at 37.  
Who knows?

Best of luck,
Dave Shechner

Quoting Artem Evdokimov <[log in to unmask]>:

> In my humble opinion starting with a mixed culture is a bad idea, because:
>
> Successful expressors have heightened metabolic load even in the absence of
> inducing agents. Therefore, during cultivation their 'expression-less'
> companions will progress more rapidly and undergo more cell divisions -
> leading to an increased fraction of 'empty' cells in the resulting culture.
> Furthermore, even after induction the 'expression-less' cells will continue
> to grow, thus depriving the expression-competent ones from nutrients
> necessary to build your protein of interest.
>
> There is a good deal of precedent regarding lack of expression from
> previously expressing cells (specifically E. coli, even more specifically -
> the DE3 cryptic lysogens) after glycerol stocks are used to start cultures
> rather than single colonies. There is a recent article in Microbial Cell
> Factories journal that analyzes this phenomenon - and the authors insist
> that the cells do not lose the plasmid, but rather mutate the lambda
> polymerase to become inefficient. This obviously does not happen to ALL the
> cells in the glycerol stock - the proportion of mutant must be relatively
> low, but repeated re-culturing brings them to the forefront so to speak, by
> allowing the mutants to win over their competitors.
>
> Therefore, in practical terms - if you just transform a bunch of cells with
> a pure plasmid - it's perfectly OK to use the entire transformation for
> growth. On the other hand, aged glycerol stocks made of such transformations
> (i.e. without purifying single colonies) may pose a problem later.
>
> Artem