Best cell line for expression cloning but not COS-7 c

David de Graaf degraaf at
Wed Jan 14 14:52:00 EST 1998

In article <34BBEFF4.FDB at>, dspinella at, wrote:

> David de Graaf writes:
> > If you really want to do this by transfection, I could recommend 293T
> > cells, since they also allow for episomal vectors. If you want a non-human
> > cell line, you might want to give LLC-PK1 procine kidney cells a try. You
> > can routinely electroporate 50% of these cells. Look up a Laurent Journot
> > paper in Nature on the expression cloning of PACAP receptors.
> > 
> > My personal bias: don't use transfection. You will have multiple
> > integrants in every cell, forcing you to recover your DNA and retransfect
> > and select multiple times. Consider doing this the right way using a
> > retroviral expression library. It just means you have to clone the library
> > into a different plasmid. A number of mouse cell lines now become
> > excellent targets. For example, mouse 3T3s are 100% infectable. Check Gary
> > Nolan's web site at Stanford for more details.
> > 
> > David
> I absolutely agree with you here.  I should have mentioned this in my
> response to the original posting. We are just beginning to play with
> retroviral expression cloning, but it certainly makes a world of sense
> to use a system in which each "transfected" (or infected) cell expresses
> only one exogenous gene.  Thanks for the insight! -- Dom Spinella

With a good dialectical background I immediately want to contradict your
last statement based on what I wrote earlier. Yes, I still think
retroviral vectors are the way to go, especially if you are not tied to a
badly infectable cell line, *but* the prevention of superinfection by
retroviral vectors is a myth. 

If you want a good reference, look up Brigitte Schott's and Igor
Roninson's paper on the subject. At 100% infection you get as many as 6
integrations per cell on average. That is still better than 100s of
integrations with some transfection methods. You can also be fairly sure
that all 6 are expressed, whereas with transfections, a large portion of
the integrations can be transcriptionally silent.

Brigitte also has an amazingly simpe wat to recover integrated virus by
long PCR and retransfecting this PCR product into a packaging cell line
without recloning. That saves a few steps right there. If you want to be
fairly sure that you do not get more than 1 insertion per genome, you have
to go to an infection rate of about 5%. 

David de Graaf

More information about the Methods mailing list