Optimization of DNA shuffling for high fidelity recombination
Optimization of DNA shuffling for high fidelity recombination
Huimin
Zhao
and
Frances H.
Arnold*
Division of Chemistry and Chemical Engineering 210-41, California Institute of Technology,
Pasadena
, CA 91125,
USA
Received December 19, 1996;
Accepted January 27, 1997
ABSTRACT
A convenient `DNA shuffling' protocol for random recombination of homologous
genes
in vitro
with a very low rate of associated point mutagenesis (0.05%) is described. In
addition, the mutagenesis rate can be controlled over a wide range by the
inclusion of Mn
2+
or Mg
2+
during DNase I digestion, by choice of DNA polymerase used during gene
reassembly as well as how the genes are prepared for shuffling (PCR
amplification versus restriction enzyme digestion of plasmid DNA). These
protocols should be useful for
in vitro
protein evolution, for DNA based computing and for structure-function studies of evolutionarily related genes.
The method of DNA shuffling, or `sexual PCR', is used to recombine homologous DNA sequences during
in vitro
molecular evolution (
1
,
2
). While randomly recombining the DNA sequences, the technique also introduces new point mutations at a relatively high rate
(0.7%;
3
). Though these point mutations may provide useful diversity for some
in vitro
evolution applications, they are problematic for others, especially when the
mutation rate is this high. Much lower mutagenesis rates are desired, for
example, during the
in vitro
evolution of long genes or whole operons (
4
), during recombination of beneficial mutations already identified previously (
5
), for DNA-based computing, or when the method is used to differentiate adaptive from
neutral or deleterious mutations in evolutionarily-related sequences (
6
).
In order to optimize the DNA shuffling technique with regard to fidelity, we
have attempted to minimize the number of point mutations introduced in each
step. Here we describe a convenient DNA shuffling protocol which randomly
recombines genes with a very low rate of associated point mutagenesis. In
addition, we show how the mutagenesis rate associated with DNA shuffling can be
controlled over a practically useful range by appropriate changes in the
protocol.
DNA shuffling consists of four steps: (i) preparation of genes to be shuffled,
(ii) fragmentation with DNase I, (iii) reassembly by thermocycling in the
presence of a DNA polymerase, and (iv) amplification of reassembled products by
a conventional PCR. Point mutations may be generated during each of these
steps. Lorimer and Pastan reported that use of Mn
2+
instead of Mg
2+
during the DNase I fragmentation step improves the fidelity of DNA shuffling ~3-fold (
7
). Our protocols include this improvement. Conventional PCR with
Taq
polymerase is usually used to prepare the genes to be shuffled (step 1) as well
as to amplify the reassembled products (step 4) (
1
,
7
). However, the fidelity of
Taq
polymerase is the lowest among commercially available thermostable DNA polymerases. In fact, between 33 and 98% of the amplification products will contain mutation(s) when a 1 kb fragment is amplified for 20 effective cycles (one million-fold amplification) using
Taq
polymerase. Extensive studies have revealed that fidelity during PCR depends on the specific conditions and DNA polymerase
used (
8
-
10
). Avoiding PCR where possible and using higher fidelity DNA polymerase during
amplification and reassembly should further reduce the point mutagenesis rate
associated with DNA shuffling.
Wild-type subtilisin E and its thermostable mutant 1E2A genes were randomly recombined by DNA shuffling using the conditions summarized in Table
2
. Gene 1E2A, obtained by directed evolution of wild-type subtilisin E, differs by 10 base changes (
6
; Fig.
1
). The enzyme encoded by this gene retains wild-type activity. The ~1 kb fragments encoding mature subtilisin E from residue -15 (from the prosequence) to the C-terminus (including 113 nt after the stop codon) were
obtained by restriction digestion of plasmid DNA and purified from a 0.8%
agarose gel using the QIAEX II gel extraction kit (QIAGEN, Chatsworth, CA).
After DNA shuffling, the gene library was amplified in
E.coli
HB101 and transferred into
B.subtilis
DB428 competent cells for expression and screening, as described elsewhere (
6
). Screening for protease activity was carried out at 37 oC in 96-well plates using suc-Ala-Ala-Pro-Phe-
p
-nitroanilide (0.2 mM) as substrate, as described previously (
11
). All PCR reactions were done on a MJ Research (Watertown, MA) PTC200 thermocycler. Sequencing was done on an ABI 373 DNA Sequencing System using the Dye Terminator Cycle Sequencing kit (Perkin-Elmer, Branchburg, NJ).
