Methods and reagents: Long and accurate PCR

by Paul N. Hengen, Ph.D. *

Methods and reagents is a unique monthly column that highlights current discussions in the newsgroup bionet.molbio.methds-reagnts, available on the internet. This month's column discusses some problems encountered during the amplification of large DNA segments by the polymerase chain reaction (PCR), and gives tips for doing long and accurate PCR. For details on how to partake in the newsgroup, see the accompanying box.

A recent discussion came about of a technique for amplifying very large fragments of DNA with high fidelity, developed by Wayne M. Barnes (Barnes@biolgy.wustl.edu) at the Washington University School of Medicine, St. Louis [1].

Barnes proposes that the limiting factor in long-range PCR is premature termination of the extension product owing to misincorporation of nucleotides. The long and accurate (LA) PCR, or `Barnesian' method, employs a mixture of two thermostable DNA polymerases, one that is highly processive and one with 3' to 5' exonuclease activity, allowing `proofreading' of the product.

After studying different combinations of enzymes, including Amplitaq/Pfu, KlenTaq1/Vent, KlenTaq5/Pfu, Stoffel/Pfu, KlenTaq1/Deep Vent and Pfu exo-/Pfu exo+, it was found that the best combination for obtaining PCR products as large as 35 kb was a 16:1 ratio of KlenTaq1, an exonuclease-free mutant of Taq polymerase, and Pfu polymerase, which has 3'-exonuclease actvity. The mixture was termed KlenTaqLA-16.

Unfortunately, despite the more recent publication of conditions for reliable `long PCR' [2], some people unaccustomed to Wayne's world of LA PCR are trying to use enzyme combinations without much success.

Instead of distinct bands of expected PCR products, a smear is commonly seen extending the length of the gel when a portion (25-50 ul of a 50 ul reaction mixture) of the amplified material, either from complex genomic DNA or from very large fragments cloned within cosmid vectors, is electrophoresed through an agarose gel, stained with ethidium bromide and visualized under UV illumination.

Smears were seen by netters regardless of the enzyme combination used. One reported seeing the proverbial smear every time a combination of Taq and Vent polymerases was tried, but that a faint band of DNA the correct size for the expected amplified product was also visible behind the background of smeared DNA. Another person could not see any bands when attempting to amplify a fragment of 12 kb from a cosmid clone using the KlenTaqLA-16 combination, even though this is much smaller than the largest product obtained by Barnes.

Interestingly, the smears appear even in control samples lacking DNA template, suggesting that the material within the smear may be partially composed of amplified contaminating DNA, complexes of primer and enzymes, primer-dimer concatamers or coagulates of buffer components.

One person wrote that concatamers of primers alone are an unlikely source since they would have to have grown extremely large during the course of the experiment to account for the breadth of the smear within the gel. Others are convinced that the culprit is bovine serum albumin (BSA), added by companies to enyme stocks to protect the enzyme activity, or supplied in concentrated PCR buffers. It was also mentioned that the use of gelatin in place of the BSA might circumvent the problem.

One way of lessening the amount of smearing is to heat the reaction mixture to 95 degrees C for five minutes before adding the polymerase. This may destroy the BSA or any other contaminating proteins, or may allow more specific annealing of primers. On the other hand, it might cause more trouble, because the extended denaturation step may damage the DNA and thwart amplification of large strands. Some people have also tried heating their amplified products to 65 degrees C for five minutes before loading onto an agarose gel, in the hope of disrupting the smear-causing complex.

Others suggested that a higher number of PCR cycles (25-30) with genomic DNA as template may lead to incompletely extended DNA strands, causing the diversity of DNA lengths that would appear as a smear. Lowering the number of cycles should then relieve the smearing. One person reported that samples taken every few cycles result first in the appearance of the expected DNA fragment, followed by larger-sized bands, and finally smears. However, another netter who removed samples every five cycles, noticed that the only difference seen over the course of many cycles was a longer smear.

After all the tweaking of PCR conditions, netters keen on amplifying large DNA fragments are frustrated from not being able to get clean results.

References

[1] Barnes, W. M. (1994) Proc. Natl. Acad. Sci. USA 91,2216-2220

[2] Cheng, S. et al. (1994) Proc. Natl. Acad. Sci. USA 91,5695-5699

Paul N. Hengen
National Cancer Institute
Frederick Cancer Research and Development Center
Frederick, Maryland 21702-1201 USA
e-mail: pnh@ncifcrf.gov



REPLY


Tips and tricks for long and accurate PCR

Although I will admit to experiencing some problems recently (and describe a promising cure; Table I c), I would first like to defend my method against those who do not follow it, yet complain that it doesn't work.

The conditions for LA PCR are very narrow at this stage of development. I am always bumping into the edges of the useful window of conditions, and I often thank my lucky stars that I hit the window in my first few experiments, or else I would have concluded that there was nothing there, and would not have found the `long' conditions.

Anyone attempting to do research on this method should, at least for now, follow every detail strictly. One person (an NIH virologist from Quebec) has called me saying that he did this, and he is very happy, having immediately reproduced my 35 kb amplicon. He says that his lab always uses filtered tips, and has separate areas for reaction setup and gels, so he is all set to make new progress. Netters who do not follow my recipe exactly and get poor results are just repeating experiments that I have done and (mostly) reported.

If you wish to join me in doing research on the method, you should choose conditions listed in the first column of Table I. This method may not yet be ready for routine application, but fully-licensed formulae should be available commercially within three months. If you are using something from the second column, even any single item above the last three new ones, expect less success. By less success I mean that amplifying 6-8 kb will work, but amplifying 20 kb and above will result in decreased yields or worse.

Table I. The way to successful LA PCR

______________________________________________________________________________

LA PCR                              Classical PCR
______________________________________________________________________________

Klentaq1 (a) plus                   Full-length wild-type Taq
1/16 Pfu or 1/50 Deep Vent          Only one enzyme
(by volume; 1/160 or 
1/500 by units)

Thin-walled tubes                   Original-thickness tubes

RoboCycler                          P.E. original or 480 cycler

pH 9.2                              pH 8.3

16 mM (NH4)2SO4, no KCl             50 mM KCl

50 mM Tris                          20 mM Tris

3.5 mM MgCl2                        1-2 mM MgCl2

5 sec 95 C melt                     60 sec 95 C melt
(set Robo block to 
99 C for 30-40 sec)

2 ng lambda DNA                     20 ng lambda DNA

33 ul reaction volume               100 ul reaction volume

20 cycles                           30 cycles

68 C extension temperature          72 C extension temperature

33-nucleotide primers               20-22-nucleotide primers

11-24 min extension                 3-10 min extensions
(longer at later cycles)

Hot start or Taq Antibody           Cold start, no Antibody
start (for genomics) (b)

Filter tips (c)                     Non-filter tips

UVA + 8MOP before template (d)      No treatment
______________________________________________________________________________

(a) The level of Klentaq1 should be titred in 0.2 ul increments from 0.6 to 1.4
ul per 100 ul reaction, or at least from 0.8 to 1.2 ul.  When I stated in my
paper that Taq/Pfu shows the LA effect, I meant that Taq benefits from added
Pfu, but, as stated, it does not, in my hands so far, come up to the
performance of Klentaq1/Pfu.

(b) The high pH, auto-extend cycles and hot start are from the adaptation by
S.  Cheng [1,2] at Roche.  As single substitutions from her system into my
system, I have not yet found the following to work above 20 kb:  Taq, Tth,
DMSO, glycerol, 9600 cycler, 22-nucleotide primers.  Apparently, it is
essential to go all the way in the direction of either system.  I am not yet
claiming any success with genomic samples above a few kilobases, but Suzanne
Cheng can amplify up to 22 kb from human DNA.

(c) The problem I have had (and cured, I believe) with my own method is that
primers and targets, from 11 kb up to 35 kb, work fine only for a few weeks to
months.  Then they begin to fail more-or-less regularly, depending on the
experience of the person doing the experiments.  Since I have switched to
filtered tips (which cost 8 cents each), and since I began to chlorox my
pipettor barrel weekly, this problem has apparently gone away.  I remade all of
my stocks using the new tips, sometimes working at a sterile bench.  The reason
for the `bad' reactions may be carry-over of a `bad seed'.  This bad seed is
made in small quantities even in a good reaction, and if it gets into the air
of a pipet barrel, thence into the dNTP and/or primer stocks, the bad seed
recruits good product into the unidentified stuff, described in my paper, at
the top of some of the agarose-gel wells. Originally, I had been using one
pipet for everything, from making or dividing reactions before the PCR to
loading the gel.  The bad-seed reaction takes over above cycle 16, so it can
usually also be avoided by doing no more than 16 cycles.  I am now unsure of
the proper conclusion for the 27-nucleotide versus 33-nucleotide experiment in
my paper.  Perhaps the proper conclusion is not `27-nucleotide work better than
33-nucleotide primers', but `New primers work better than old primers (that
have been used for a while and whose bad seed is everywhere)'. I am still
recommending 33-nucleotide primers until I know for sure.

(d) Treatment of the enzyme(s) with 8-methoxy-psoralen (8MOPS) and UVA
radiation [3] seems to be a good idea for inactivating endogenous contaminating
bacterial DNA, but I am just trying it now.

References

[1] Cheng, S., Fockler, C., Barnes, W. M. and Higuchi, R. (1994) Proc. Natl. Acad. Sci. USA 91,5695-5699

[2] Sharkey, D. J. et al (1994) Bio/Technology 12,506-510

[3] Jinno, Y., Yoshiura, K., and Niikawa, N. (1990) Nucleic Acids Res. 18:6739.

Wayne M. Barnes
Department of Biochemistry 8231
Washington University Medical School
St. Louis, MO 63110, USA
email: Barnes@biolgy.wustl.edu


Any statements made by the author are not meant to advocate the use of a particular commercial product or endorse any company. All opinions are those of the author and do not reflect the opinion of the National Cancer Institute or the National Institutes of Health.

Copyright: This manuscript is not copyrighted by Elsevier Publishing Company. However, you may not reproduce any portion for resale or edit the text for redistribution, sale, or otherwise without written permission from the author.


You found this at the World Wide Web (WWW) Uniform Resource Locator (URL):

ftp://ftp.ncifcrf.gov/pub/methods/TIBS/aug94.txt

Any reference to this column must be cited as the following published article:

@article{Hengen1994Augtibs,
author = "P. N. Hengen",
title = "Methods and reagents - Long and accurate {PCR}",
journal = "Trends in Biochemical Sciences",
volume = "19",
number = "8",
pages = "341-342",
month = "August",
year = "1994"}



*******************************************************************************
* Paul N. Hengen, Ph.D.                           /--------------------------/*
* National Cancer Institute                       |Internet: pnh@ncifcrf.gov |*
* Laboratory of Mathematical Biology              |   Phone: (301) 846-5581  |*
* Frederick Cancer Research and Development Center|     FAX: (301) 846-5598  |*
* Frederick, Maryland 21702-1201 USA              /--------------------------/*
* - - -  Methods FAQ list -> ftp://ftp.ncifcrf.gov/pub/methods/FAQlist - - -  *
* -  TIBS column archive -> http://www-lmmb.ncifcrf.gov/~pnh/readme.html - -  *
* - The BEST Molecular Biology HomePage -> http://www-lmmb.ncifcrf.gov/~pnh/  *
*******************************************************************************