Long PCR smear

Anne Buchanan ABW1 at psuvm.psu.edu
Sun Jun 12 08:37:14 EST 1994

We've been trying, in our lab, to optimize use of DNA from irreplaceable
samples, so we've been working on a method which will, in theory, produce
essentially unlimited sample.  We digest genomic DNA with a blunt 6 base
cutter, ligate double stranded adapters of differing lengths to the cut pieces,
and amplify up the whole genome using one of the adapters to prime the rxn.
This product is used as template for subsequent gene-specific PCR.  This worked
fine when we used a 4 base cutter, except that with such a frequent cutter, the
probability that the locus of interest would be cut was unacceptably high, so
we decided to go to the less frequent cutter and long PCR.  Now the average
piece length of the cut DNA is around 5000 bp, ranging from something like 12
kb on down, and we've been using Klentaq for the first round of PCR, with 10
minute extensions for 12 cycles, then adding 15 seconds each cycle for 13 - 18
cycles.  We're also testing the system with lambda phage because we can count
the pieces we expect to see.  So far, we seem to be reproducing the 15 pieces
of the lambda phage, and getting the expected smear around 5 kb with the
genomic DNA, and can do gene-specific PCR with the resulting PCR product.

The problem is this, and someone else posted a similar problem a while back -
we sometimes get long smears starting at the well, even in the negative
control, and we don't know why.  It's a problem, because we expect a smear even
when all goes well, but not this long.  The fact that someone else gets this
result with long PCR makes me think it's something about the reaction, not our
template.  There is usually background smear when we amplify the lambda phage,
as well.  It's been suggested that maybe the primers are forming long chains,
somehow (this seems so unlikely to me, because these would be *really* long
chains).  Anyway, we thought we might understand better what was going on if we
reamplified DNA from a small slice of the gel of the amplified genomic DNA, so
we did this.  We thought we'd get a band of the length we'd cut out, but
instead we got the familiar smear, and no discrete band - we removed a tube
from the thermal cycler every 5 cycles to check the effects of number of
cycles, and the only difference more cycles made was longer and longer smear.
This could be a problem with reamplification, of course, but it looked
suspiciously like the familiar smear to us.  My question is, does anyone have
any idea what is going on, or how to test it?  Thanks in advance.

Anne Buchanan
Dept. of Anthropology
Penn State University

More information about the Methods mailing list