We use home-made retroviral vectors with GFPS65T, "Green Lantern"
(partially humanized S65T from Gibco) and EGFP. We also use EBFP, but did
less work with it. In most (but not all) constructs we get a clearly
separated positive population, that remains stable over an extended period
of time (3-15 weeks in continuos culture) without any selection or
resorting. Even the constructs that do not render very bright signal (7-15
times brighter than control) produce the same signal over an extended period
of culture. This was studied in HT1080, MCF7, NIH3T3, Rat1 and, somewhat
less thoroughly, - in K562 cells. The GFP-positive fraction remains stable
upon gamma-irradiation and treatment with at least several anti-cancer
drugs. GFP-positive cells do not lose tumorigenecity and can be easily
detected in tumors grown in vivo. When Southerns were done on individual
fluorescent clones of HT1080, many of them showed single-copy integration.
By the way, lack of coexpression of a marker gene and a gene of interest in
retroviral vectors happens in 5-10% of cases (in single-copy clones) and is
often caused by small deletions. This problem occurs much more often in
stable transfectants. So do not be surprised if selection for another marker
did not produce 100% GFP-positive population.
Although I generally would consider it a cheap self-advertisement, I
believe at this point I should post the references to two of our papers that
used GFP and show fluorescent profiles:
Schott ,B., Kandel, E.S., Roninson, I.B. (1997) Efficient recovery and
regeneration of integrated retroviruses. Nucleic Acids Res, 25(14) :
Kandel, E.S., Chang, B.-D., Schott, B. , Shtil, A.A., Gudkov, A.V.
Roninson, I.B. (1997) Applications of Green Fluorescent Protein as a marker
of retroviral vectors. Somatic Cell Mol. Genet., 23 (5) : 325-340.
Please note, that we are not the only ones that obtained and published such
I should say, that even with the strongest expressing constructs we see
continuous rather than discrete fluorescent profile if we use transient
transfection rather than infection. The likely explanation is that copy
number and expression levels are much more variable in this case. Moreover,
there is a continuos loss of transiently transfected DNA. Some cells
expressed it for only several hours while others got a stable integration.
To the best of my understanding, the events during adenoviral infection are
similar to those during transient transfection. By the way, we observed that
fluorescence of retroviraly infected (!) cells grows for the first 2 days
and if we analyze infected cells earlier we see less separation even with
our best constructs.
Although it has been discussed in this group before, I would like to
mention that "unhappy" (including dying) cells often have higher
autofluorescence. Thus, I would advise to do a very careful gating on side
and forward scatters. It may also help to visualize positive cells on 2D
plots, such as fluorescence vs. forward scatter, or green fluorescence vs
red fluorescence (I use PI to gate out dead cells). Higher autofluorescence
will inevitable decrease your resolution, especially on 1D plots.
Indeed we see some sort of normal distribution, rather than a straight
line, on the fluorescence histograms of "pure" clones. I suppose it reflects
random fluctuation during analysis, rather then a growing heterogeneity in
expression levels. For example the shape of a cells is not perfectly round
and the signal may depends on how the cell was positioned relatively to the
light source and detectors at the moment of analysis. There is also random
heterogeneity in cell size, especially in non-synchronized populations.
Of cause, some cells with very unstable karyotype, or some vectors with
frequently inactivated promoters will show you accumulation of negative
cells (we are having a hell of a time designing a reliable and strong
expression system for ES cells). The beauty of GFP is that you actually can
see it, while with other genes you suspect it only after your experiments
Hope these note are helpful,
U09577 at uic.edu