How to Avoid PCR Errors

Brian Foley brianf at
Sat Feb 19 07:54:05 EST 1994

Jim Owens (jow at wrote:

: Could you clarify something?  How did you know a mutation when you saw
: one?  Were you looking at a known sequence?  Were you sequencing several
: different clones of the same PCR fragment?

	Yes, I am sequencing mutant alleles of the elongation factor 2 
gene (actually the cDNA derived from those alleles) from Chinese hamster
ovary cells.  The CHO-K1 EF-2 gene and cDNA have already been published 
and put into GenBank.  I can tell which mutations were RT-PCR artifact 
and which really exist in the genomic DNA (from which the mRNA was derived) 
because I do four separate PCRs on each sample and clone each product 
in a separate ligation/transformation.  If a particular mutation really 
exists in the DNA, I find clones with this mutation about 50% of the time 
from each tube (CHO-K1 has two functioning EF-2 genes, it is diploid).  
If the mutation is an RT-PCR artifact I usually find it in only one clone 
or only in other sister clones from the same PCR reaction, but not in the 
other three.

: I am in the habit of sequencing RT-PCR fragments directly and am
: concerned about how much I should worry about sequencing errors?  The
: fragments are from sequences involving transcripts that cross chromosomal
: translocations that involve genomic sequences that are already known.

	Whenever you see a change from the published sequence, you will 
not know if it is an alleleic variation, or a PCR artifact until you do 
some follow-up work.  I have been able to follow up some of my mutations 
by noticing that if real, they would create an RFLP.  I can then do a 
Southern blot to find out if that mutation exists in the genome.  Others, 
I have to do more PCR and sequencing as descried above.
	When I started this project I assumed that RT-PCR errors would 
not be a big problem.  They are.  So now I always just do the four PCRs 
and 4 ligations for each mutant cell strain.  It is much easier to just 
do them all at once and get it over with, than to have to go back and do 
another RT-PCR at a later date.  Especially considering that RNA is not 
all that stable.

*  Brian Foley               *     If we knew what we were doing   *
*  Molecular Genetics Dept.  *     it wouldn't be called research  *
*  University of Vermont     *                                     *

More information about the Methods mailing list