TENN 705 Biochemie Sclones Vol 6Nos Wy 001
Techniques: Recombinogenic
engineering - new options for cloning
and manipulating DNA
Joep PP Muyrers, Youming Zhang and A. Francis Stewart
Driven by the needs of functional genomics, DNA engineering by homologous
recombination in Escherichia colihas emerged as a major addition to existing
technologies. Two altemative approaches, Recfvdependent engineering and
ET recombination, allow a wide variety of DNA modifications, including some
Which are virtually impossible by conventional methods. These approaches do
not rely onthe presence of suitable restriction sites and can be used to insert,
delete or substitute DNA sequences at any desired position ona target
‘molecule. Furthermore, ET recombination can be used for direct subcloning
and cloning of DNA sequences trom complex mixtures, including bacterial
artificial chromosomes and genomic DNA preparations. The strategies
reviewed in this article are applicable to modification of DNA molecules of any
‘ze, Including very large ones, and present powerfull new avenues for DNA
‘manipulation in general
The completion of large DNA-sequencing projects,
including the Human Genome Project, has generated
an extraordinary amount of primary sequence data.
The noxt major challenge isto investigate the
‘components that make up a genome, and is often
called functional genomies. Escherichia coli vectors
‘that can contain large inserts, such as bacteria
artificial chromosomes (BACs)', PI vectors# and PL
artificial chromosomes (PACs)’, offer several
advantages for functional genomics. They can carry
sufficient DNA to encompass most eukaryotic genos,
‘including all cis-acting regulatory elements, as well
as many eukaryotic gene clusters, prokaryotic
rogulons and many complete viral genomes, ina
single molecule*. However, conventional cloning.
methods rely on the use of restriction enzymes and
in vitro purification steps, which preclude
engineering of large molecules. Consequently, the
usefulness of such molecules has been limited until
recently. Novel DNA engineering strategies that rely
on homologous recombination in vivo in E.coli
alleviate this imitation and allow a wide range of
‘modifications of DNA molecules at any chosen
position. These strategies are the scope of this review.
Program emis.
genearages com
Homologous recombination and DNA engineering:
recombinogenic engineering
Homologous recombination allows the exchange of
genetic infarmation between two DNA moleculesin a
precise, specific and faithful manner. These qualities are
optimal or engineering a DNA molecule regardless of
its size. Homologous recombination cecurs through
‘homology regions, which are stretches of DNA shared by
Ue two molecules that recombine, Because the sequence
ofthe homology regionscan be chosen reels, any
position on atarget molecule ean be specifically altered.
Because homologous recombination is a rare
‘event, some form of aelection is needed to identify the
cells that carry the recombinant. Hence DNA.
engineering by homologous recombination, or
rrecombinogenic engineering, makes use ofa selection
procedure that usually involv
‘genes, In certain cases, such as simply deleting a part
ofa BAC, Pl-vector or PAC (Refs 5-8), or generating.
mouse knockout constructs in one step*, the product
‘ofa single round of homologous recombination can be
the dosired product (Fig. 1a).In many other eases,
persistence ofthe selectable gene at the site of
recombination in the product is undesirable.
Consequently, recombinogenic engineering often
involves two rounds of processing. In the first round,
hhomelogous recombination is used to generate an
initial product by integration ofthe selectable gene,
‘together with additional fmetional elements, at the
intended site. In the second round, the extra
antibiotic-resistant
functional elements are used to remove the selectable
gene, thereby generating the final product. By
informed application of these principles, virtually any
DNA iteration ean be accomplished. Variations of
how additional functional elements can he used are
shown in Fig. 1 and inelude the following:
+ Acounterselectable gene (Fig. 1b; for recent examples
of the use of tho rpsl, ett and sacB counterselectable
senes, see Refs 810-18). Here, the desired second:
round product is identified by countersclection
Dbocause homologous recombination can eliminate the
counterselectable gene and flanking sequences,
replacing it with the DNA region of interest. This
approach can theret second:
round product that earries na sca'from the first
round of recombination, However, cunterselection is
typically less ficient than positive selection, as the
intended homologous recombination is only one of
several solutions for counterselection pressure. Any
other mutational event that alates expression of the
counterselectable gene will also grow under
counterselection pressure. Thus, candidates from
second-round-countersclection stratogies need tobe
serened find the intended recombinogenie
engineering product. In practice, the ratio of
anon rseed Pe sone-oon4oa8 4a TES 0S oho Scenes VrTe ie Mny
(2) One-stop positve selection _(b) Selection, countarsolaction
{€) Selection ste-specife
(€) Selection, restiction
recombination digestion
A B Aas 8 A= 8 At ate
near ONA Linesr NA Linear DNA Linesr DNA,
ne
Target
molecule
Target
molecule
Target
molecule
Fist
round
Festmund Frstround Frstround
rod rot prot
Second + Delton Delon
rom orem |
through» rowan
fcontioan Recomtnan\
ON finest conning ‘containing
n 8 5
recombinant
Escherichia colenromosome. targeting DNA hee stated as linear ONA i generoted BY
PCRampliieaton toned wo eglons, Asnd 8, omologousoegions inthe age ralecule
‘Alo includedisselertablemarer gene (im). The simplest exerese( ivolvesinsetion. by
between A ona (uncncanbe anywere beeen Dbp anda ast 0X8) ean Beuseatoaelee
remveit Three waysteremove he selecinble gene re state npanets bead. panel |
unter selectable gene fs iinsrted alongside the selecible gone nthe staund pode The
sireesmeassete lcd na second oundet hmelogous recombination courteseleton
‘anges or nserion of BNAregion ofirerest compel ee of any perationalsequences sed for
fengirering in ()sitespeciereeombinalion eget ses (S5RTs shown arevhesds} re
sprcteracombinsian|s useatoremove the smeatsteoleavea single SSRT inthe pros!
Tesvtion ates, Indcatedby an estes, re ncludedolnkineselectabie gene inte neat
targeting motel immediatly adjacent a one or oth ofthe homlogy regions nat show). The
poroaches ihusirated here are based on ET econbinogenicengineving bul te prices with
intended-to-unwanted produete varies wildly
leg. botwoen <1% (Ref. 14) and 15-85% (Refs 5,13)
forreasons that are still undefined.
+ Flanking site-specific recombination target sites
(SSRTe), such as the 34-bp recognition sites for
(Big. Le; Refs 5,6,16,17). Here, the first products
exposed tothe relevant site-specific recombinase,
which deletes the selectable cassette to leave a single
SSRT at the site of engineering, Thisis a high!
cffcient way to eliminate the selectable cassette and
is also useful method of placing an SSRT exactly
hero itis required. However, the remaining 34
(or 86) bp scar (the SSRT) can be a problem ifleftin
protein-coding, or regulatory, regions.
+ Flanking sites for restriction enzymo(s) (Fig Ld.
Here, the first product is purified, cut with the
relevant restriction enzyme and religated to give
second-round product free ofthe first-round
selectable gene, As with SSRTs, this is a useful
way of placing a restriction enzyme site (or sites,
including polylinkers) exactly where required. In
this strategy, the second-round product includes the
restriction site() and ean be used only for relatively
small DNA molecules because it relies on restriction
enzyme digestion
Recombinogenic engineering in coll
‘Although £. coliis the premier host for conventional
DNA engineering, Saccharomyces cerevisiae was,TENN 705 Biochemie Sclones Vol 6Nos Wy 001 ur
until recently, the preferred host for recombinogenic
engineering. This situation was due tothe
remarkable proficieney of &. cerevisiae at homologous
recombination, combined with certain complications
inherent in the endogenous E.coli homologous
recombination mechanism, Recombinogenie
engineering in S. cerevisiae is straightforward
because the linear DNA flanked by short homology
rogions can be inserted into endogenous DNA targets.
Linear DNA i particularly useful because DNA ends
significantly promote recombination'®. Furthermore,
‘the homology regions can be as short as 42
nucleotides. Consequently oligonucleotide synthesis
‘ean be used to make the homology regions'®. Success
in using yeast artificial chromosomes (YACs) and
‘yeast episomes in S. cerevisiae has produced several
recombinogenie engincering breakthroughs, the most
recent example being direct cloning by homologous
-recombination?®2!. Although YACs are better than
BACs, P1-veetors or PACs in one respect, that is they:
‘can carry larger inserts, recombinagenic engineering
in S. cerevisioeis hampered by several difficulti
as implicit genetic instability, low yields of purified
DNA products and the care and labour-intensive skills,
required to carry out the procedures.
In B.coli, the endogenous homologous
recombination mechanism initiated by cooperation
between RecA (the most conserved strand invasion
protein in evolution)#*#* and RecBCD (a large,
functional enzyme composed of ReeB, RecC and
‘RecD subunits)##+" impedes the use of linear DNA
fragments, probably because RecBCD is, among other
functions, a very vigorous exonuclease. Therefore,
RecBCD rapidly digesta any exogenously introduced
linear DN
Avariety of RecA-dependent approaches for
‘recombinogenic engineering have been published,
For brevity only the approach that has found most,
"use will be described here; however, the interested
‘reader might wish to consider other variations!™2,
such
molecules
RecA dependent recombinogenic engineering
Consistent with the difficulties of using linearized
DNA in E.coli, early findings with RecA-dependent
‘recombinogenic engineering reported some
successes using intermolecular recombination
between two circular molecules", A significant
advance in RecA-dependent strategies was the
inclusion of a tomperature-sensitive (ts) plasmid
origin. This permitted the construction of a
targeting plasmid, grown at the permissive
temperature, followed by a first round of homologous
‘recombination to integrate the plasmid into its
target at a non-permissive temperature (Fig. 2).
After identification of correct homologous
‘recombinants and further culturing in the presence
of RecA, shifting back to the permissive temperature
identifies those recombinants that have lost the ts
origin (because its continued presence kills the
‘host). To improve the yield of intended product in the
DNA of interest
Target
Homologous recombination
through A
Fret round
DNA of imtorest
Cointegrant
LoS
Resolution through A Resolution through B
DNA of interest
Ae alia
‘Second round
7
Fig. 2 RecA dpender-nedatearecombinagenc engheetng The most
ley usedRecraepndenstegys shou Befoetecomansen,
‘ake mam). tempertre-ienaiv origina epson tr},
“recagene andtwonomology regions (a8 yea Te ntng
herelogy em Aar or srl any recombination rowan
Ceirnganisidetes by selection rhe Slate genet
pear hat sperrsivefor hese Reson ca the reve
second round, a counterselectable gene has been
included near the ts origin. This strategy was
developed for altering the B. coli chromosome",
applied to recombinogenic engineering in BACs
(Refs 11,15) and recently improvedTSIEN FRENDS io ochonct Scones Vol eNO Moy 2001
An outline ofthis RecA-dependent strategy is
shown in Fig. 2. Before the first round of
recombinogenie engineering, a dedicated targeting
plasmid containing several components must be built.
‘The components include: (1) ene( or selection and
counterselection; 2) homology region(s), typically
1 xb or longer; (3) ats origin; and (4) the recA gene,
Although the recA gene could be omitted ifa recA+
E.coli host is used, providing RecA in trans from the
targeting plasmid permits the use ofthe common recA
hosts and also limits the recombinogenie window.
After integration ofthe intact targeting plasmid into
the target to form a cointegrantin the first round
(Fig. 2), activating the ts origin by temperature shift
and exerting countersclection pressure drives
resolution in the second round, Because resolution.
climinates the backbone ofthe targeting plasmid,
including the recA gene, the host carrying the
second-round product reverts to rec; thus limiting
the recombinogenie window and securing the genetic
stability othe intended product, Resolution in the
‘second round can produce either the original target or
the intended recombinant.
Selected applications of RecA-dependent
recombinogenie engineering are summarized in
‘Table 1. All RecA-dependent strategies appear to
roquire relatively long homology region(s) and hence
the construction of dedicated targeting plasmids,
Furthermore, efficiencies are often low. [The term
‘efficiency’ refers throughout to both the total number
of candidate recombination products achieved and
the ratio of intended (romologous) to unwanted (non
‘homologous or rearranged) recombination products
‘among those candidates,]
Recombinogenic engineering using ET recombination
An alternative recombinogenic engineering strategy
was developed whereby recombination is not
dependent on RecA, but instead mediated by
phage-derived protein pairs, either RecE/Rec from
the Rac phage or Reda/Redf, from 2 phage". To coin
simple term, this recombinogenie engineering
strategy was initially termed ET recombination
(or ET cloning)® and has also been called L-mediated
recombination® and GET recombination". As
established from both fundamental studios!®238
and applied studies with recombinogenic
engineoring® "2199 or genotypic manipulation of
E.coli hosts “8%, the RecB/RecT and Reda/RedB
protein pairs are functionally and operationally
equivalent, RecE and Reda are 5’-3" exonucleases,
and RecT and Red are DNA annealing proteins,
Alunetional interaction between Ree and Ree, or
Redotand Redf, is also required for homologous
recombination™
‘As with recombinogenie engineering in
S. cerevisiae, ET recombination encompasses the
advantages inherent in the use of linear targeting
DNAs and thus the increased efficiencies promoted by
DNA ends. To secure this advantage, the endogenous
RecBCD exonuclease activity must be absent or
inactivated. I'the E.coli host in which BT
recombination takes place is not recBCD, a recBCD
phenocopy can be induced by the expression of Redy,
the phage protein that inactivates ReeBCD
(Ref. 88).A further similarity to S. cerevisiae is found
in the fact that the RecE/ReeT and Reda/RedB protein
pairs need only short homology regions for useful ET
recombination efficioncies™#116173936,
With BT recombination, as illustrated in Fig. 1a,
a linear targeting DNA carrying short homology
regions flanking a selectable gene (sm) is integrated
into. circular target DNA. Although long homology
rogions can bo used”, the length of homology
required for efficient recombination is
;-60 nucleotides, and thus short enough tobe made
by oligonucleotide synthesis. Ina very convenient
application, the synthesized oligonucleotides also
include polymerase chain reaction (PCR) primers for
amplification ofthe selectable gene. Thus the linear
targeting DNA can be a PCR product and no
dedicated targeting plasmid needs to be constructed,
For example, PCR products have been used to'shave’
target molecules, such as various BACs, by deletion of
‘Table 1. Summary of recombinogenic engineering applications+®
Modification Target motecaTe
E-collehromosome PIs/EACS/PACS
RecA ET RecA, ET RecA, et
Precise sequence deletion $2782) 4617.85-85) + (10-1215,1629) +8) 1252631) 159.58,36)
Precise sequence insertion
Small (6009, - +6. +110) +69 - “0
0g SSRThesriction sites
Large, eg. rotein tags + QN82 46788-3855 101215.29 + 1.7) + (25.2628.30) +669)
GFR.IRES. promoters, genes, ete
Point mutations - « - sea = “6
modieatonsin varius target mokectes ere neate Accomp shnets of engineering exercises by RecA-deenden recombination RecA
subaivised ini sll (20 bp sarge 40 bp} to cstinguish between those barea onthe convenience fered by algonurleoiseTEIN 78055 ochericrScoces VraeNoS My 20 7°
‘unwanted DNA precisely in one step, POR products
have also been used to generate targeting constructs
for knockout experiments in mouse ES cellsin one
stop?. Here, the selectable gene was chosen beeause it
conveys resistance to antibiotics in both E. coli and
cukaryoticcolls. Throe such genes are usoful, namely
‘those that convey resistance to kanamyein/neomycin,
hygromycin and blasticidin. ET recombination was
used to place one of these selectable genes exactly
where desired for later sein knockout experiments.
Strategies involving ET recombination and two
rounds of recombinogenic engineering are illustrated
in Fig. 1b-a. In particular, POR produets carrying
selectable and counterselectable gene(s) (Fig. 1b)
‘have heen used to point mutate a plasmid? or BACs
(Refs 13,14).As described above, Figs eand Id
illustrate simple ways to delete the selectable gene
‘cassette and to introduce short sequences encoding
SSRTS or restriction site(s) exactly where desired. In
‘these cases the SSRTs/restriction site(s) can also be
‘programmed into the synthesized oligonucleotides.
Similarly other short ‘DNA-of interest’ sequences,
such as a polyadenylation signal ora short protein
‘ag, can be programmed into a synthetic
‘oligonucleotide for convenient introduction. In eases
in which the added DNA-of interest is beyond the
limits of oligonucleotide synthesis, itean be
introduced either in the second round of
selection/eounterselection (Fig. 1b) orby use of
plasmid template constructed tocarry the DNA-of-
interest beside a selectable gene. (The construction of
these plasmids is easily accomplished using one-step
positive selection; Fig. 1.) note of caution
‘concerning the fidelity of PCR is appropriate here. In
‘many cases, the risk that a PCR amplification step
introduces mutations in the DNA-of-nterest is
“unacceptable, Fear of PCR mutagenesis does not
apply, however, toamplification of selectable genes,
‘counterselectable genes, SSRTs, restriction sites or
short DNA-of-interest regions because there are easy
ways to test function or screen out the mutants,
Ifwanted, the DNA-of.interest can be constructed
in adedicated plasmid that is linearized by restriction
enzyme digestion. These linear DNAs necessarily
‘have some non-homologous sequence at their very
ends, However, erminal stretches of non-homology,
‘which are either the few nucleotides left by restriction
enzyme cleavage, or up to several kilobases between,
the end and the homology region, are eliminated by
recombination through the homology regions toleave
the intended product.
Successful applications of ET recombination are
‘summarized in Table 1, In contrast to RecA strategies,
the efficiencies achieved with ET recombination, using
either RecE/RecT or RedaRed protein pairs, are
consistently high, and typically more than 80% of
‘candidates are the correct recombinant, Hence, itis not
necessary to analyse many clones to obtain the desired
‘recombinant (fora more detailed deseription of the
involved experimental procedures, se Ref. 39).
As evidence ofthe very high efficiencies achievable
‘with ET recombination, a recent publication has shown
‘that correet recombinants ean be identified without the
need to employ antibiotically selectable gones®, Using
oligonucleotides and short PCR products, absolute ET
recombination efficiencies approaching 1 in 100 total
colonies have been shawn. Hence, selective PCR
screening methods on pools and individual colonies ean
bboused to identify correct recombinants.
Limiting the recombinogenie window
[Recombinogenic engineering, either Rec dependent
orby BT recombination, oocurs through homology
regions. Any two regions homologous to each other
‘within # recombinogenic host can recombine. In
practical terms, this means that unintended
‘additional homology rogions in the targeting DNA
should he avoided, Probable unintended homologies
include all or parts of plasmid origins, selectable
markers or other commonly used DNA sequences such
‘as fragments carried over from common cloning
vectors. Whereas unintended homologies in the
targeting DNA can be avoided simply by good design,
repeats occur randomly in cloned sections of
cukaryotic DNA and cannot be avoided.
Intramolecular recombination between such repeats
is clearly undesirable. Conveniently, the homology
regions that direct the intended ETT recombination
product are stimulated by nearby DNA ends, whereas
potential intramolecular recombination in the target
DNA doos not have this advantage. Thus, intended
EP eventa should be favoured over unintended
background instabilities. Inour experience, thishas
proven tobe the case, but moreexperienceusinga
broader range of target molecules is needed to settle
the matter, Regardless of the outcome, the issue of
‘genetic instability isan important concern for
rrecombinogenie engincering and highlights the
importance of strategies that limit the recombinogenic
‘window; that i, the time-span in which recombination
cean take place, tothe smallest interval required for
generation of the intended product.
In addition to the RecA-dependent suicide
strategy outlined in Fig. 2, ats RecA host has also
been used as a way tolimit the recombinogenic
‘window to a period of expression aterculturo at the
permissive temperature!®®®, For ET strategies, the
rrecombinogenie window i limited by regulated
‘expression of at least one of the components ofan ET
protein pair (RecE/RecT or Reda/Red) from an
inducible promoter7IS7:S3, Most plasmids from.
‘which the ET'mediating proteins are expressed rely on
the ParaB promoter and its properties of L-arabinose
induction and glucose repression‘. Frequently, such ET
plasmidsalso allow expression of Red so that ET
recombination in any B. coli strain (including those
‘expressing RecBCD) can occur. Thus, these plasmids
offer a simple way toimport ET recombination intonew
hhosts, particularly ones that already contain a BAC of
interest, and to limit the rocombinogenie window.me TREN 0S oho Scenes Vrae es My
DNA of DNAof
A interest a _ierest
Purlied DNA preparation | Intact ereutar target
as
‘a [ Recombinant } 5
DNA of interest a
User tagetingmotcles ae generale polymersecaineation PER ocentan aepleaen
Puried DNA proprton loin oan actu trget rola uber). The partied ONA
ani a genomic DNA preprtton ands co-ecropraed th ne nes een mole aan
molecule creategplsriarecombinanthatistertiiedy seectonr the selectable
Recent developments
Following the recombinogenic engineering initiatives
pioneered in S. cerevisiae®2%, the potential of BT
recombination for direct cloning and subcloning was
explored. The principle ofthis strategy isillustrated
in Fig. 3. In this variation, the linear targeting
‘molecule is a PCR-amplified plasmid backbone that
contains a selectable gene and an origin of replication,
‘The oligonucleotides used for PCR also contain
‘homology regions that are chosen to define the exact
boundaries ofthe DNA region ta he cloned or
‘subeloned. The chosen DNA region is copied into the
plasmid backbone by homologous recombination. By a
single-step process, chosen regions from plasmids and
BACs, including a23-kb region of mouse genomic
DNA, were subcloned at the high efficiencies that are
typical of BT exercises, Similarly chosen regione
from purified E.coli, east and mouse genomic DNA
preparations were directly cloned'?, This presents a
simple and efficient way to subelone that does not
involve the tedium of purifying DNA, the required
eogereering ne co's
inomtegedrereby
Irhuenmuyrerscoen,
Supported by nts fom
Feundsion orogrmmon
prosence of convenient restriction sites or the
‘mutational risk of PCR. It also opena up a new way to
clone without the need to make and serven libraries,
‘and to.amplify a chosen DNA region by a means other
than PCR, Furthermore, unlike PCR, DNA molecules
ofup to atleast 28 kb have been subcloned in a single
step. These subcloned molecules (unlike those
generated by PCR) are unlikely to contain any
‘unwanted mutations because the subeloned region is
amplified in vivo by the endogenous E. coli replication
‘machinery and hence has been fully proaf reviewed.
Concluding remarks and perspectives
‘Tho strategies presented outline how precise
‘modifications, including single-base changes, and
variably sized insertions and deletions, can be made
to DNA molecules in £ coli of any size
Recombinogenic engineering strategies will
undoubtedly accelerate progress in functional
genomics, as they allow straightforward engineering
oflarge DNA molecules, which are important for
functional studies of genos in all aspects such as
development, homeostasis and disease, Importantly,
the strategies also provide alternatives to
conventional, small-molecule engineering using
restriction enzymes, POR and DNA ligase. Moreover,
ET cloning and subcloning provide an alternative to
PCR amplification, Because any recombination-
based strategy is carried out in vivo, any
recombinant molecule is generated and inspected by
the endogenous F. coli replication and repair
ayatems, Therefore, the recombinant product is
virtually ensured tobe free of any undesired
‘mutations, Thus, with a little forethought, the
‘mutagenic ambiguities involved in the use of PCR
or EtBr/UV exposure during DNA preparation can
be avoided by use of a recombinogenic engineering
alternative,
Although recombinogenic engineering strategios
were previously reliant on the use ofS. cerevisiae, the
convenience of using B. coli as the host is now
available, Consequently, commonly used cloning
‘molecules, which replicate in E coli,can be engincered
directly: Moreover, whereas recombination potential is
constitutively activein yeast, the E.coli-based
recombinogenie engineoring systems have been
developed to limit the recombinogenic window.
The possibilities offered by homologous
recombination prompt new ways of thinking about
DNA engineering. Engineering strategies that were
previously fat all, only applicable to small DNA.
‘molecules auch as plasmids, can now be applied to
DNA molecules of any size and composition. Ata
‘minimum, the experimenter nevds tobe aware that,
when conventional methods fail, are flawed, or are
difficult to apply, alternative solutions exist. Adding
recombinogenic engineering tothe repertoire of
frequently used molecular biological techniques is
likely to pay off, as it will provide effective solutions
for many DNA engineering objectives.1 Shinya, H eta. (1992) Cloningand stable
‘maintenance of 900-bane pale fragments
human DNA in Baherchiacal wing an
Pistorbased van, Pr, Natl, vad. Sel USA.
0, 974-8707
2 Sternberg, Ni (1982 Cloning igh molecular
‘ripht DNA Supmontab the biter P
sytem, Trad Get, =18
$ Touro, PA. etal (1099 Anew bacteriophage
Priderived eta forthe propagation a arge
‘human DNA fragmenta, Not Gone 6, 4-89
4 Brase, W eof (2000) Fareand wich BACS now
toalaforhorpsvius genomic, Tends Genet
16, 254-258,
5 Thana, ¥ eta (1996) new ogiefor DNA
gineringwsingrenmbination in
Eosherichia col Net Gene, 20, 129-128
6 Miyrery PR.