Why PCR reamp = smears. (fwd)

Klaus.Matthaei at ANU.EDU.AU Klaus.Matthaei at ANU.EDU.AU
Mon Aug 29 18:37:51 EST 1994

>This replies to the following:
>Forwarded message:
>> From: HARDIES at thorin.uthscsa.edu
>> Subject: Re: Why PCR reamp = smears.
>> I've been reading this thread about PCR products producing smears on
>> reamp, and trying to think of a reason for it.
>> For example, Mruiel Liberto writes: "I have diluted the original
>> 1:1000.  Could purity be causing a problem, even after a thousand fold
>> dilution (into water) ???"   Good point.
>> The consensus response seems to be to dilute it more severely.  I
>> wonder if the explanation isn't something like this.  If you dilute
>> the original 1:1000, that's only 2^10.  So in ca. 10 cycles you should be
>> back where you started.  Now if you're going 20 or 30 cycles, you're
>> forcing the polymerase to work under conditions where reannealing of
>> the template is inhibiting further extension.  So you're prone to
>> aberrations like blocked partial extension and maybe even strand
>> switching.  This might make your smear.
>> This is pure conjecture.  Take it with a (diluted) grain of salt.
>I think Steve Hardies and others who say to go for the high dilutions are
>on the mark, due to the excessive taarget input without high dilution.
>It seems to be difficult for people to relate to the incredible capacity
>of the PCR exponential process, so amounts of target can be way too
>excessive for the "standard" number of cycles used for PCR (20-30).
>It helps to think of copy number in the input DNA, NOT nanograms.  In a
>routine human genomic 30cycle rxn, we use 50 ng DNA. That's 15,000 copies
>of (single-copy/haploid genome) target, and it gives good gel-visible yield
>(i.e., 100's of ng of product) after 30 cycles, and 30 cycles is probably
>overkill.  BUT, if the target is, say, 0.01 ng of some 200 bp PCR product
>that you want to hyperamplify, that's already 50,000,000 copies in the input,
>and 30 cycles is way way way way way way way way tooooo many.  You'd get
>all sorts of smear-mischief going on in the late cycles.  If I INSISTED
>on using 0.01 ng of 200 bp input target (way more than enough for 30 cycles)
>then the reasonable thing to do would be to reduce the number of cycles
>to give 50,000,000/15,000 = 3,300-fold less amplification.  That would be
>12 cycles less, since 2(exp 12)=3300, so I'd run 18 cycles and be smear-free.
>I could also go home earlier.  You can test this yourself by setting up
>multiple duplicate PCRs in one of your smear-experiments, pulling out
>one tube after 10 cycles, one after 12, one after 14, etc., and I'll bet
>you'll see that a real nice band appears in the lower cycles and then
>degenerates into smear after too many cycles.
>Rob Preston
>rapr at med.pitt.edu

I agree with Rob.  ALWAYS think in terms of copy number.

MOST PCR failures are because of either too much template, too much enzyme
or too low an annealing temperature.

With template remember that 1 cell contains about 7pg DNA, therefore a
standard  PCR for a single copy mammalian gene can be performed with about
10-100ng DNA in 35 cycles or less, so you can work out how many copies that
is.  With nesting you can detect 10 copies in 20 cycles followed by 33
cycles using ~1% of the first round in the second round.

For controls: 1fg of a 10kb plasmid is about 100 copies of target.  For
standard 35 cycle PCR 1pg of plasmid is usually enough to detect easily
(depending on your cycler, primer and enzyme combination).

I hope this helps.

Cheers, Klaus

Dr Klaus Matthaei                               |           |
The John Curtin School of Medical Research      |  _--_|\   |
The Australian National University              | /      \  |
Snail mail: Canberra, ACT 0200, Australia       | \_.--._/<<|
E-mail: Klaus.Matthaei at anu.edu.au               |       v   |
Landline: 61 6 249 3782
                        "Sometimes I'm the Louisville slugger"
                        "Sometimes I'm the Ball"
                                Dire Straits

More information about the Methods mailing list