'semiquantitative' RT-PCR

T.D. Laing tdlaing at nospam.dres.dnd.ca
Tue Jun 30 10:14:59 EST 1998

In article <NJ_l1.189$Gy3.774481 at fozzy.nit.gwu.edu>, Steven Sullivan
<sullivan at gwis2.circ.gwu.edu> wrote:

> I see lots of papers with 'semi' or comparative RT-PCR data in the
> developmental biology journals -- e.g. one amplicon compared in different
> treatments or stages, with amplification of a 'housekeeper' gene as a
> normalizing control.  I'm still a bit unclear as to how teh authors
> determine 'linearity' in these experiments though.  For a given primer
> set, must one test 1) a range of cDNA input amounts, or 2) a range of
> cycle numbers?

Yes to both.  Different primer sets will amplify with different
efficiencies in the same reaction tube, as the reactions will compete for
the same pool of enzyme and nucleotides, and the target and housekeeping
genes are likely present in different amounts.  You test a range of cDNA
input amounts to determine the best amount of cDNA to add to get the best
result.  For the cDNA, I would choose some representatives from your
samples, because that is the system you are trying to quantitate.

For any quantitative or comparative approach, testing a range of cycle
numbers for each primer set is necessary.  Remember that PCR is
exponential, but plateaus after a certain number of cycles for each primer
set and cDNA amount as enzyme, nucleotides and primers are exhausted.  The
exponential portion of the curve is linear.  For the best direct
comparison the amounts of both your target and your control gene products
should fall at about the midpoint on their amplification curves for any
given cDNA amount.  This requires titrating your primers so that when both
are added together to the reaction, both products end up within the linear
portions of their curves.

Also remember that your target is likely not expressed to the extent that
your housekeeping gene is.  So it usually takes fewer cycles to see a
given amount of housekeeping gene (because there are more cDNA copies)
than the target for a given amount of cDNA added.  The method I'm familiar
with, primer dropping, starts off by adding the primers for the gene with
the longest cycle number first, then adding the 2nd primer set. 
(reference below)

(Your predetermined cycle number will depend on whether you use EtBr
staining or radioactive detection of products.  I've seen it done both

>  And is this done for *each* different RNA sample (e.g. 
> injected versus uninjected Xenopus animal caps) or can one jsut use any
> tissue in which the target RNA is expressed (e.g. a whole Xenopus embryo) 
> *once* and never have to do it again for that set?

You should test your primer sets on representative RNA samples.  For
example, I wouldn't use whole embryo RNA if looking at animal caps.  But
after finding your appropriate cycle range and cDNA amount with your
representative sample(s) for your target and control genes, I don't see a
problem with using those conditions with the remainder of your samples. 
However, to work well your control gene (e.g. GAPDH) should be expressed
at similar levels in all your samples.

  People I've spoken
> with seem to have different ideas over which combination of these
> approaches constitutes 'rigor'.  I've already encountered the phenomenon
> of being able to RT-PCR up a supposedly 'unexpressed' mRNA simply by doing
> more than the published number of cycles. 

Sometimes the "unexpressed" mRNA is an artifact.  Other times there may
not be enough copies of the target for it to be visible for the set of
conditions you're using (i.e. you're still in the lag phase, not the
exponential phase, of your amplification curve at a given number of

> None of the PCR protocol books I've seen discuss this in detail, but if
> anyone can point me at a good reference please do. 


Wong H, Anderson WD, Cheng T, Riabowol KT Monitoring mRNA expression by
polymerase chain reaction: the "primer-dropping" method.  Anal Biochem
1994 Dec;223(2):251-258 

There are probably other methods out there, but this is the one I'm most
familiar with.


T.D. Laing
tdlaing at dres.dnd.ca

Remove "nospam" from my e-mail address in the header to reply.

All the standard disclaimers apply.

More information about the Methods mailing list