Why PCR reamp = smears.

Mon Aug 29 17:15:05 EST 1994

Maureen Bunger (MB) wrote asking for clarification about my (SCH) previous 
post trying to explain why reamping PCR products tends to produce smears.
SCH> The consensus response seems to be to dilute it more severely.  I
SCH> wonder if the explanation isn't something like this.  If you dilute
SCH> the original 1:1000, that's only 2^10.  So in ca. 10 cycles you should be
SCH> back where you started.  Now if you're going 20 or 30 cycles, you're
SCH> forcing the polymerase to work under conditions where reannealing of
SCH> the template is inhibiting further extension.  So you're prone to
SCH> aberrations like blocked partial extension and maybe even strand
SCH> switching.  This might make your smear.
SCH> This is pure conjecture.  Take it with a (diluted) grain of salt.

MB> Please clarify: Are you suggesting that a 1/1000 dilution is NOT enough.

SCH> Yes. I'd probably think more like 10exp-4 for starters and titrate
     down by factors of 10.  Different templates may differ in their 
     response. (see below).  Note: my recommendation is based on theory,
     not practice.

MB> are you thinking ... that the problem could be
MB> resolved by going say only 2 or 5 more cycles [beyound 10]?

SCH> Yes.  I like Rob Preston's suggestion of putting several identical tubes
     in the machine and pulling them at different cycle numbers (at least
     as a diagnostic experiment).

MB> I don't understand the logic behind the statement about "20 or 30
cycles ... inhibiting further extension. I don't see how this would
lead to abberrations. [Many related questions...]

SCH> The difference between my thinking and yours is that I believe that
     amplification stops before any reagents are exhausted.  I believe 
     that amplification stops because your buildup of product destroys
     the huge molar excess of primer:template needed to get the primer
     on the template instead of the template just reannealing to itself.
     I think this because:
      1) Our yield stops far short of exhausting the amount of dNTP and
         primer we supply to the reaction.
      2) Additional cycles will eventually bring out new nonspecific
         bands, so the enzyme is not dead.
      3) It explains the length effect.  That is, large fragments anneal
         kinetically faster, and hence will drop out of exponential 
         amplification earlier than small fragments.

MB> If I do a 1/1000 dilution after 30cycles, then I
MB> understand about being back where I started after 10 cycles, but
MB> then is there a problem with continuing another 10 or 20 cycles?

SCH> Possibly (this is conjecture).  The problem is that the other strand
     of the template may reanneal after the polymerase gets started.  Then
     the polymerase will plow into a replication fork.  The paused polymerase
     might then stop, or possibly even jump strands.  I have only the
     following suggestive evidence:
     1) In the olden days, E. coli PolI was shown to jump strands when
        in vitro replication was attempted w/o ss binding protein.
     2) When we tried extension times of 10 min. in an attempt to 
        amplify long fragments, we made some incredibly ugly smears.

MB> We are trying to use human exon primers to amplify
MB> possible homologous mouse exons. This has been done by someone else using
MB> 15ul straight from the first amp as template for the second. When I cut this
MB> down to 10ul I started getting significantly different results, (some
MB> work beautifully, some get multiple bright bands, others are smears).

SCH> The length of the fragment may be a factor that introduces variability.
SCH> Also the quality of the primers.  If priming is impeded by mismatches,
SCH> it may alter both the slope of the exponential phase, and the balance
SCH> between kinetics of primer annealing and template:template reannealing.
SCH> Note the issue is kinetic, not thermodynamic; computing Tm's won't help.
SCH> So you may just have to titrate to find a good working dilution. 

MB> Could you direct me to a reference possibly that talks
    about this stuff? 

SCH> Sorry; this is just my personal theory.

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

More information about the Methods mailing list