Tue Sep 27 14:24:36 EST 1994

Nancy Templeton writes:

> 	We are trying LA (long and accurate) PCR of genomic DNA.  We have been 
> successful at some genomic loci and not others.  What should we try on loci
> that are difficult?  We are using accurate primers that should amplify the
> region of interest. 

First a disclaimer: I'm speaking from theory, not experience.

For loci that produce no product (after excluding a flaw in the primer design)
I'd suspect a base composition problem.  I notice that Wayne Barnes' procudure
calls for 3.5 mM Mg and that he's going over a 57% G+C template.  For ordinary
PCR buffer with 50 mM K+ or Na+, I'd expect this template to not denature at
95 C.  I notice that his buffer has no K+ or Na+.  So maybe that saves it; or
maybe I'm being misled by the vaguries of trying to extrapolate the ion effects
to mixed ion solutions.  In any case, he must be real close to not denaturing
the template.  Other templates with higher G+C (even in a local region) may
not denature.  I'd look first at dropping the Mg by increments down to 1.3 mM.
Then I'd try adding formamide by increments up to 10%.  Unfortunately, these
things will also mess with primer performance, so you may have to adjust
annealing temperature downward at the same time.

>	Also we have noticed possible amplification product when run on a 
> low percentage agarose gel (0.4%); however, this product will not migrate 
> out of the wells.  We have tried various methods (i.e. SDS, 
> heating, etc.) to resolve this problem to no avail!!

This is the usual failure mode, to quote Barnes in his PNAS91, 2216 (94) paper.
He doesn't know what it is either.  I can add that this stuff appears when you
increase extension times even in control reactions where the template is a pure
plasmid clone and the intended PCR product is only a few kb.  The only thing
I can figure it might be is a network: ie. a branched annealed structure made
among heterogeneous truncated products and possibly products involving strand
jumping by the polymerase.  Since you get some much material, I have to believe
that at earlier cycles you must have gotten correct amplification and then at
later cycles the product to product reannealing competed with primer extension
and promoted the formation of this stuff.  So the first thing I'd do is just
cut back on cycle number.

I don't think you can recover your product out of the smuck.  I think you have
to adjust your PCR conditions to avoid making smuck.  Barnes says he got rid of
it by making the primers longer.  However, it's real unclear why this worked.
Increasing the stability of the primer itself doesn't make sense, because 
you're just lowering the stringency of the annealing and actually promoting
mispriming.  When he lengthened them he also moved them, opening the door to
all kinds of alternative interpretations.  For example, he might have moved
them to a region of more or less secondary structure thus achieving either
greater or lesser priming efficiency.  You could imagine that greater priming
efficiency solves the problem by making sure most templates get started in
time to finish each cycle.  Alternatively, you could imagine that lesser
priming efficiency solves the problem by causing you not to accumulate as
high a concentration of product by the end of your experiment.  Also note
that solving one problem at the expense of promoting mispriming may cause the
procedure to work great on a noncomplex control template and to not work on a
complex genomic template.

So playing with the primers at this time is pure trial and error.  I would
guess that the problem progresses in the order: band -> smear -> everything in
the well.  So if you change a condition and go from everything in the well to a
smear, you're going in the right direction.

Hope this helps

Steve Hardies, Dept. of Biochem., Univ. of Texas HSC at San Antonio
Hardies at

More information about the Methods mailing list