RT-PCR Question

Andrew J. Doherty a.doherty at bristol.ac.uk
Mon Oct 21 02:35:06 EST 1996


eric anderson wrote:
> 
> In article <543v1a$h4a at dfw-ixnews10.ix.netcom.com>,
> pjb4 at ix.netcom.com(Peter J Bartholomew) wrote:
> 
> > Hello,
> > As a minor part of my research project, I isolated total RNA from
> > estradiol-treated and control cells followed by RT-PCR.  This was done
> > to compare the mRNA levels of a certain gene.  Since the resultant
> > bands seemed of equal intensity following agarose gel electrophoresis
> > and ethidium bromide staining, I stated that the estradiol treatment
> > did not affect the mRNA levels.  In retrospect, I know I should have
> > done a northern, but was this conclusion still acceptable?
> 
> peter,
> 
> we're doing a very similar sort of study right now but with about 25-30
> different genes.  the key to believable semi-quantitative RT-PCR is that
> you MUST make sure that your number of amplification cycles is still in
> the linear range.  if you did 35 cycles or so, my guess is that you were
> out of the linear range and couldn't identify any differences that may
> have existed (this happens to us all the time).
> 
> the way around this is to repeat the PCR and take aliquots of your
> reactions at various points in the reaction (i.e. 20, 25 and 30 cycles).
> Unless you're using an integrating gel imaging system (e.g. Eagle Eye) you
> should also include tracer amounts of a labeled dNTP (we use 32-P a-dCTP)
> to make quantitation more accurate (via phosphorimager, densitometry,
> etc.).
> 
> finally, i would suggest trying to include primers for a housekeeping gene
> (we use GAPDH because we've seen variable actin expression in our system)
> in the same reaction as your gene of interest.  this will require some
> optimization but it makes quantitation much easier since you
> (theoretically) have an internal standard.  alternatively, you could make
> a construct of your gene of interest with a deletion between your RT-PCR
> primer sites and use a known amount of that as a competitive target.  this
> is probably a more accurate internal standard than what we do, but since
> we're working with 30+ genes, the construct making is a bit time
> consuming.  (see Biotechniques 21:268-279 and 280-285(Aug. 1996)) for more
> info. on this.
> 
> of course, the best way is to just do a northern and be done with it.
> 
> good luck,
> 
> eric
> 
> --
> Eric C. Anderson
> Memorial Sloan-Kettering Cancer Center
> Sloan-Kettering Institute
> 1275 York Ave. Box 470
> New York, NY  10028
> (212) 639-2977
> e-anderson at ski.mskcc.org

I couldn't agree more. If anyone out there wants to do semi-quantitative
RT-PCR, get a control gene to run alonside your genes of interest,
preferably in the same reaction. I used B-actin without too many
problems. You can also use this gene (or your gene of interest if you
know its intron/exon structure) to determine if your products are coming
from genomic contamination. Choose primers from different exons and you
can detect fragments generated which have an intron in between. NEVER
use standard PCR as a quantitative method - it's not quatitative!!!

Andy D
-- 
***********************************************
   Dr Andrew Doherty                          
   Department of Anatomy                   
   School of Medical Sciences             
   University Walk                               
   Bristol                                            
   UK
   BS8 1TD

   e-mail  A.Doherty at Bristol.ac.uk

************************************************



More information about the Methods mailing list