Copyright 0 1997 by the Genetics Socirty of America
Gene Replacement
With
Linear
DNA
Fragments in Wild-Type
Escherichia
coli:
Enhancement
by Chi
Sites
Patrick Dabert’ and Gerald
R.
Smith
Fred Hutchinson Cancer Research Center, Seattle, Washington 981 04
Manuscript received July 2 2 , 1996 Accepted for publication January 13, 1997
ABSTRACT
During conjugation and transduction of Escherichia coli even numbers of recombinational exchanges are required for replacement of a gene on the circular chromosome. We studied gene replacement using a related method of gene transfer (transformation with 6.5-kb linear DNA fragments) as an experimental model for conjugation and transduction. Two properly situated Chi sites, 5’ GCTGGTGG 3’, stimulated gene replacement -5@fold, more than the sum of the stimulation by the individual Chi sites. Gene replacement was dependent on R e d and RecB functions. Similar results were obtained with an alternative experimental model in which linear DNA fragments were generated from phage h by intracellular EcoRI restriction following infection. Dual Chi site-stimulation of these RecA-, RecBdependent recombination events thus did not depend upon the mode of delivery of the linear DNA into the cells. A single DNA
fragment with two Chi sites was sufficient for gene replacement. These results support a one Chi-one exchange hypothesis (“long chunk” gene replacement), stemming from studies with purified RecBCD enzyme, and argue against models in which Chi converts RecBCD enzyme to a state capable of promoting multiple exchanges on one DNA molecule. These results also provide a method for gene targeting in wild-type E. coli and suggest a method for gene targeting in other organisms.
I
N Escherichia coli conjugation and generalized trans- duction are efficient methods of replacing a chro- mosomal gene with an allelic copy from donor bacteria. Gene replacement in these situations occurs by the RecBCD pathway of homologous recombination, which requires RecBCD enzyme and RecA, single-stranded DNA binding, and other proteins (for a review, see SMITH 1989). Recombination by this pathway is not uni- form along the DNA, it occurs preferentially near a specific DNA sequence called Chi (5’ GCTGGTGG 3 ’ ) ,which interacts with RecBCD enzyme (for reviews, see
KOWALCZYKOWSKI et al. 1994; MYERS and STAHL 1994;
SMITH 1994). Chi occurs frequently in E. cola DNA, o n the average once per -5 kb or -1000 times per ge-
nome (FAULDS et al. 1979). By analogy with its enhance- ment of phage X vegetative recombination by the RecBCD pathway ( L A M et al. 1974), Chi would be ex- pected to enhance conjugational and transductional recombination. Although a Chi site in the donor DNA influences the distribution of exchanges in conjuga- tional and transductional crosses (DOWER a n d STAHL 1981), a requirement for Chi in such crosses has not been demonstrated because of the large number of Chi sites in E. coli DNA. Here we report strong stimulation by Chi of gene replacement in E. coli using relatively short DNA fragments with a n d without Chi sites.
CmrPsponding uuthor: Gerald R. Smith, Fred Hutchinson Cancer Research Center, 1124 Columbia St., Seattle, M7A 98104.
E-mail: [email protected]
’
Pruwnt addrrss: Institut National d e la Recherche Agronomique, Labmatoire de Biotechnologie de I’Environnement, Avenue desEtangs, I 1 100 Narbonne, France.
Genetics 145: 877-8851 (April, 1997)
Chi and its interaction with RecBCD enzyme have been studied both in intact cells a n d with purified com- ponents. In both situations, RecBCD enzyme interacts with Chi only if it approaches Chi, during its travel from a double-stranded DNA end, from the right of Chi (as written above) (KOBAYASHI et al. 1982; TAYLOR et al. 1985). Continued unwinding of DNA by purified RecBCD enzyme after its interaction with Chi produces single-stranded DNA substrates for RecA protein prefer- entially to the left (“downstream”) of Chi (PONTICFUI et al. 1985; DIXON a n d KOWALCZYKOWSKI 1991). In E.
coli cells, Chi stimulates recombination preferentially also to its left (STAHL et al. 1980; CHENG and SMITH
1989). There is, thus, concordance in these two aspects of Chi’s action (orientation dependence and direction- ality, respectively) both in intact cells a n d with purified components. In the experiments reported here, we used DNA substrates containing Chi properly placed and oriented to be active in accord with these estab- lished properties of Chi.
878 P. Dabert and G. R. Smith
4
Chi Chi 4-
+
n-El
FIGURE 1.-Model for “long chunk” gene replacementduring conjugation or transduction. Two Chi sites, oriented to enhance recombination in the interval between them, are hypothesized to be required for the replacement of a segment
of the recipient chromosome (circular thin line) with a seg- ment (“long chunk”) of the donor DNA fragment (linear thick line). The mosaic fragment in parentheses is expected only if the recombinational exchanges are reciprocal. See SMITH (1991) for further discussion.
a “long chunk” of donor DNA (Figure 1 ) . If each Chi site were able to stimulate exactly one exchange, then exactly two exchanges, and maintenance of chromo- some circularity, would ensue from recombination of a linear DNA fragment containing two properly situated Chi sites. Support for this idea came from the ability of purified RecBCD enzyme to act at one Chi site but not at two (directly oriented) Chi sites on the same DNA molecule (TAYLOR and SMITH 1992). In addition, in A
vegetative crosses, one Chi site reduces, but does not eliminate, the activity of another Chi site to its left (“downstream”) (STAHL et al. 1990).
As
noted above, a direct test of the requirement for two (oppositely oriented) Chi sites on the linear DNA fragment in con- jugational and transductional crosses is nearly impossi- ble because of the abundance of Chi sites in E. coli DNA. To test this requirement, we introduced into E. coli relatively short (6.5 kb) linear DNA fragments ma- nipulated to have 0, l , or 2 properly situated Chi sites. Our results show that gene replacement occurs most frequently with DNA containing two Chi sites and sup- port the one Chi-one exchange hypothesis.MATERIALS AND METHODS
Bacterial strains: The E. coli strains used in this study are listed in Table 1. Most strains were derived by PI transductions from strain V66, which contains the red7143 and Arac muta- tions. These mutations prevent genetic recombination via the RecF and RecE pathways, respectively, and recombination in this strain occurs primarily via the RecBCD pathway (for a review, see SMITH 1989). All rec genotypes were confirmed by phage spot tests (SCHULTZ et al. 1983) and by testing LJV sensitivity and antibiotic resistance. The designation rec+ indi- cates recAf, re&, r e c c , recD+ and, for strainV66 and its deriva- tives, includes recF143.
Media: LB broth and agar plates, tryptone broth (TB), BBL agar plates, minimal medium (OMBG) and phage suspension medium (SM) have been described (SMITH et al. 1986). Antibi- otics were used at the following concentrations: 100 pg/ml ampicillin (Ap), 12.5 pg/ml tetracycline (Tc), 45 yg/ml kana- mycin ( K m ) , and 35 pg/ml chloramphenicol (Cm). Minimal media were supplemented with the required amino acids at 20 yg/ml.
Plasmids and phages: Plasmids and phages (Table 2 and below) were constructed according to standard techniques
(SAMBROOK et ul. 1989). All noncompatible DNA ends were made blunt either by filling in with the Klenow fragment of E. coli DNA polymerase I (for 3‘ recessed ends) or by exo- nucleolytic degradation with T4 DNA polymerase (for 3’ pro- truding ends). EcoRI restriction sites were inactivated by cut- ting with EcoRI endonuclease, filling in with the Klenow fragment of DNA polymerase I and ligating. Designations such as “ligation of a SpeI-Hind111 DNA fragment to a SmaI-
PstI DNA fragment” imply the formation of SrnaI/SpeI and HindIII/PstI junctions.
pDAl: a 5.25-kh Hind111 DNA fragment containing part of the his operon (CARLOMAGNO et al. 1988; JOVANOVIC et al. 1994) (Figure 2) from E. coli strain C600 was inserted into the unique Hind111 site of pBR322 (BOLIVAR et al. 1977) and selected by transformation of strain V66 (his@) to His+. The identity and orientation of this fragment was confirmed by restriction endonuclease analysis. The his operon is tran- scribed in the direction opposite to that of the ampicillin resistance (ApR) determinant of pBR322.
pDA2: the 3.0-kb SpeI-Hind111 his DNA fragment from pDAl was inserted into the SmaI and PstI polylinker sites of pUC19 (YANISCH-PERRON et al. 1985). The his operon is transcribed in the direction opposite to that of the ApR determinant of
pDA3: the 1.24-kb PstI-PstI kanamycin resistance (Km’)
DNA fragment from pUC4Km (VIEIRA and MESSING 1982) was inserted into the unique PstI site in the hisD gene of pDA2. The K m R gene is transcribed in the direction opposite to that of the his operon.
pDA4: the 4.3-kb his-KmR Kpnl-SphI DNA fragment from pDA3 was ligated to the replicative 2.1-kb PstI-PpuMI DNA fragment of pBR322. The his operon is transcribed in the direction opposite to that of the ApR gene.
pDA5: the 4.3-kb hzs-KmR KpnI-SphI DNA fragment from pDA3 was ligated to the replicative 2.1-kb XhoI-PpuMI DNA
fragment of pBR322 x+HF (DIXON and KOWAI.CZYKOWSKI
1993).
pDA6: the 4.3-kb h i s K m K KpnI-SphI DNA fragment from pDA3 was ligated to the replicative 3.8-kh EcoRI-SphI DNA fragment of pBR322.
pDA7: the 3.2-kb hisC HindIII-NdeI DNA fragment from pDA4 was ligated to the replicative 3.9-kb HindIII-NdeI DNA fragment of pDA6.
pDA8: the 5.1-kb hicKmR Xbal-NdeI DNA fragment from pDA5 was ligated to the replicative 2.1-kb EcoRI-NdeI DNA fragment of pBR322.
puc19.
pDA9: the EcoRI site of pDA7 was inactivated. pDAlO: the EcoRI site of pDA8 was inactivated. pDAl1: the EcoRI site of pBR322 was inactivated.
pDAl2: a blunt EcoRI linker “GGAATTCC” (New England Biolabs) was inserted into the AlwNI site of pDAl1.
pDA13: a blunt EcoRI linker “GGAATTCC” was inserted into the NdeI site of pDAl1.
pDA14: the 3.2-kb PstI-AJIIII tetracycline resistance DNA fragment from pDAl3 was ligated to the replicative 1.1-kb PstI-AflII DNA fragment of pDA12.
pDA15: the 5.9-kb hi>KmR BsaAI-BsaI DNA fragment from pDA9 was ligated to the replicative 1.2-kb BsuAI-BsaI DNA fragment of pDA14. This plasmid is
xo.
pDAI6: the 5.9-kh his-KmR BsuAI-BsuI DNA fragment from pDAl0 was ligated to the replicative 1.2-kb BsaAI-BsuI DNA fragment of pDA14. This plasmid is x+HE
pDAl7: the 2.8-kb x+H-hisG BsaI-Hind111 DNA fragment from pDA16 was ligated to the replicative 4.4kb BsaI-Hind111 DNA fragment of pDA15. This plasmid is x+H.
879 Chi Sites and Gene Replacement in E. coli
TABLE 1 E. coli strains
Strain
designation Genotype Source or reference"
AB1 157 thr-1 leuB6 A(gptproA)62 hisG4 thi-1 argE3 lacy1 galK2 ara-14 xyl-5 mtl-1 tsx-33 1
C600 thr-1 leuB6 thi-1 lacy1 tonA21 supE44 @Dl A- F- 1
DH5a thi-1 reoll AlacU169 (a80 lacZ AM15) hsdRl7 recAl endAl gyrA96 supE44 A- F- 2
QD5003 pro thi supF58 A- F- 1
V66 argA21 recFl43 his@ met rpsL31 galK2 xyl-5 A- F- r a C b 3
V67 As V66, plus recB21 3
supE44 rpsL31 kdgK5l $Dl mgl-51 A- F- rac-
V119 As V66, plus recA56 srl-300: : Tn 10 3
v220 As V66, plus recDl011 argA81: : Tn 10 4
V1904 As V66, but hisC C600 X V66
V1906 As V1904, plus recDlOll argA81::TnlO C600 X V220
V1908 As V1904, plus recB.21 C600 X V67
v1909 As V1904, plus recA56 srl-300: : TnlO V119 X V1904
V1910 As V66, but his+ ( A CZ Ind-) L of V1904
V1911 As V1910, plus recDlOll argA8l::TnlO ( A cIInd-) L of V1906
V2084 As V1910, plus recB2l ( A cZInd-) L of V1908
V2085 As V1910, plus recA56 srl-300::TnlO ( A d I n d - ) L of v1909
V2088 As V66, but re& zic::TnlO I. KOBAYASHI (strain BIK1288)
V2104 As V1904, plus re@ zic::TnlO V1904 X V2088
V2106 As V1910, plus re& zic::TnlO (A cZInd-) L of V2104
"A X B, P1-mediated transduction in which A is the donor and B is the recipient; L, lysogenization with A1624 (cZ Ind-);
Ir As expected from its genealogy, DNA from V66 fails to hybridize with a fragment of the cloned rac prophage (S. K. AMUNDSEN, references: 1, BACHMANN 1987; 2, HANAHAN 1983; 3, SCHULTZ et al. 1983; 4, CHAUDHURY and SMITH 1984.
personal communication).
was ligated to the replicative
x+F
4.3-kb BsaI-Hind111 DNA fragment of pDA16. This plasmid is x+F.pDA19: the 1455-bp BglII-Hind111 DNA fragment encoding the EcoRI modification methylase of pMB4 (NEWMAN et al.
1981) was ligated to the 3.9-kb BamHI-Hind111 DNA fragment of pACYC184 (CHANG and COHEN 1978).
pDA2O: the 940-bp PuuII-PstI DNA fragment encoding the
EcoRI restriction endonuclease of pMB4 was ligated to the replicative 3.8-kb HincII-PstI DNA fragment of pDA19.
The location, orientation, and integrity of the Chi sites on
TABLE 2
Plasmids and phages
Name Relevant properties"
pDA15
xo
his: : kanpDAl6
x+H
his:: kanx+F
pDAl7 x'H his: : kan
pDA18 his: : kan
x+F
A1766
xo
his: : kanX 1 768
x+H
his:: kan x+FA1771 x+H his: : kan
A1772 his: : kan x+F
Additional plasmids and phages used for constructions are described in MATERLALS AND METHODS.
Plasmids (pDA15 to pDA18) contain the 6.5-kb EcoRI frag- ment shown in Figure 2 and the 0.6-kb AlwNI-NdeI fragment of pBR322. The AlwNI and NdeI sites were converted to EcoRI
sites (see MATERIALS AND METHODS for details). Phages (A1766 to A1772) are derivatives of Agtll with the 6.5-kb EcoRI frag- ment shown in Figure 2 inserted into the EcoRI site in the p l a d segment of Agtll.
plasmids pDA16, pDA17 and pDA18 and the lack of Chi on pDA15 were confirmed by reaction with purified RecBCD enzyme (TAYLOR et al. 1985).
A phages 1766 (x"). 1768 (x+HF), 1771 ( x + H ) , and 1772
(x+F)
were constructed by insertion of the 6.5-kb nonreplica- tive EcoRI DNA fragment from plasmids pDAl5, pDAl6, pDA17 and pDA18, respectively, into the unique EcoRI site of Agtll (YOUNG and DAVIS 1983). Relative to the conventional A map, the ApR determinant is on the right side of the 6.5- kb insert. Packaging was with the Ready.To.Go Lambda Pack- aging Kit (Pharmacia).DNA preparation: Plasmid DNA was prepared from trans- formants of strain DH5a by the Qlagen procedure according to the supplier's instructions. Alternatively, the DNA was pre- pared by the alkaline lysis method followed by ultracentrifuga- tion in a CsCl gradient in the presence of ethidium bromide (SAMBROOK et al. 1989); similar results were obtained with the two methods. Plasmid DNA minipreparation and total DNA preparation were according to SAMBROOK et al. (1989) and GRUSS and EHRI.ICH (1988). Phage A stocks and DNA were prepared as published (SPRAGUE et al. 1978). Phage A DNA minipreparation was according to an unpublished procedure from D. EVANS (University of Guelph).
880 P. Dabert and G. R. Smith
with 70% ethanol, suspended in 100 pl of EcoRI buffer, and incubated again with 10 units of EcoRI for 1 hr at 37". The EcoRI enzyme was inactivated by a 10-min incubation at 65", and the DNA recovered by precipitation with ethanol. The DNA pellet was suspended in 100-150 p1 of TE [ l o mM Tris- HCI (pH 8), 1 mM EDTA], and kept at 4". The DNA purity and concentration were measured by determining the UV
absorption spectrum from 220 to 340 nm using a Shimadzu W-1201 spectrophotometer.
Transformation: Competent E. coli cells were prepared ac- cording to HANAHAN (1983) or by the following protocol de- rived from HANAHAN'S by W. WAHI~S (personal communica- tion); the two methods gave similar results. An overnight culture in LB, started from an isolated colony on an LRagar plate, was diluted 100-fold and grown at 37" until the OD(ioo reached 0.3. The cells were centrifuged at 5000 X g for 10 min at 4". The pellet was suspended in 0.4 volumes of ice- cold 100 mM CaCI,, and kept on ice for 20 min. The cells were collected as before and suspended in 0.01 volumes of ice-cold RbC1/CaCl2 buffer. After 10-30 min of incubation on ice, sterile glycerol was added to 20%, and the cells were frozen in a dry-ice/ethanol bath. The cells were stored at -80" for 5 6 months. RbCI/CaCl, buffer was composed by mixing equal volumes of 100 mM CaC12 and RbCl buffer (35 mM KOAc, 10 mM CaC12, 5 mM MgC12, 0.5 mM Lic1, 15% sucrose; adjust pH to 5.9 with 1:20 dilution of glacial acetic acid; then add solid RbCl and MnCI, to 100 mM and 45 mM, respectively). For transformation, linear DNA was heated for 10 min at 65", to melt cohesive EcoRI ends that annealed during storage, and chilled on ice for 2 1 0 min. An aliquot of the DNA was withdrawn and run later on an agarose gel to check its integ- rity. DNA (100 ng) was diluted to 50 pl with MCT buffer [IO
mM MgC12, 10 mM CaC12, 10 mM Tris-HC1 (pH 7.5)], and 120 pl of previously frozen cells (thawed on ice) were added. The mixture was incubated for 30 min on ice, heated for 5 min at 37", incubated at room temperature for 15 min, and diluted to 1 ml with LB. After 1 hr of shaking at 37", to allow expres- sion of antibiotic resistance, an appropriate amount of cells was plated on selective medium.
Transduction: E. coli cells were diluted 100-fold from an overnight culture in LB with Cm into 40 ml of TB supple- mented with 10 mM MgS04, 0.2% maltose and Cm and grown to an ODRio of 0.5 (-2 X 10' cells/ml). The viable cell concen- tration was determined at this time by plating dilutions on LB-agar plates. A 0.1-ml sample of the culture was infected with 0.1 ml of phage in SM at the appropriate multiplicity of infection (m.0.i.; generally 0.1) and incubated for 15 min at 37" without shaking. The infected cells were diluted to 1 ml with warm supplemented TB and incubated 30 min on a gyratory shaker at 37". Appropriate dilutions of the cells were plated on LB-Km agar plates. Unadsorbed phage, determined at the time of dilution of the infected cells, were < 10% of the input phage. Phage titers were determined on strain QD5003
(supF). Phages were EcoRI modified by growth on strain QD5003 (pDA20). The efficiency of EcoRI restriction (>98.5%) was determined by titration of A vir phage (EcoRI
modified and unmodified) on a sample of the cells immedi- ately before transduction.
Analysis of gene replacement: Km' transformants or trans- ductants were patched onto fresh LB-Km plates and replica- plated onto minimal medium lacking histidine [or containing histidinol for strain AB1157 (hisC4)I and LB-Ap plates. His- Aps colonies were scored as gene replacement events. DNA from 66 independent His- ApS transformants and 35 trans- ductants was separately digested with Hind111 or BgllI, electro- phoresed, blotted, and hybridized with the 6.5-kb EcoRI frag- ment of pDA17 by the procedure of SOUTHERN (1975). Fourteen transformants with
xo
DNA (two from rec+ and 12from recD), 16 with x + H DNA (13 from ret' and three from recD), 14 with x+F DNA (seven from rec+ and seven from recD), and 22 with x+HF DNA (12 from rcc' and 10 from recD), and 35 transductants of ret' cells (four with
x'
phage in R- cells, three with ,yo in R+, seven with x + H in R+, five with x+P in R', five with x+HF in R-, and 11 with x+HF in R+) were analyzed. All showed the expected patterns (data not shown).After transformation we observed three other classes. Sev- eral independent
K
m
'
colonies from each class were purified, and their plasmid or total DNA content was analyzed by re- striction endonuclease analysis or Southern blot hybridization as above.The major nonreplacement class was His' Ap'. These cells contained free plasmid DNA identical to pDA15(xo) or pDAlG(x'HF) as judged by restriction analysis with EcoRI, PstI and XbaI endonucleases (data not shown). The frequency of this class among the
K
m
'
colonies varied from one linear DNA preparation to another and presumably reflected the extent of contamination of the linear DNA with intact or singly cut plasmid; it ranged from >99 to 97% without gel purification, and from 10 to < 1 % using the procedure de- scribed above.A second, minor class (<1%) was His' Ap'. These cells contained free plasmid lacking one EcoRI site and part of the Ap' gene. We suppose that they arose from exonucleolytic degradation and recircularization of plasmid DNA cut once by EcoRI (MCFARIANE and SAUNDERS 1996). Other types of degradation would delete the replication origin or the his and K m R genes of the plasmid and could not be scored in these experiments.
A third, minor class (<0.5%) was His- Ap'. These cells did not contain any detectable free plasmid DNA and generated
K m K His- Aps and K m s His+ Ap' cells during growth without selection, or His+ papillae when replica-plated onto minimal medium lacking histidine. Southern blot hybridization analy- sis indicated that they contained the entire nonreplicative 6.5- kb DNA fragment integrated into the his locus and sur- rounded by a duplication of the his DNA (data not shown). This structure is expected when the linear fragment circular- izes and integrates into the chromosome by a reciprocal ex- change within the his homology. Recombination between the duplicated region can excise the nonreplicative fragment and either restore a his+ gene or leave the K m K determinant i n
hisD.
Ectopic nonhomologous recombination of the linear DNA could generate
K
m
'
His' cells, either Aps or Ap', and would be scored as deleted (A$) or full-length (Ap') DNA integra- tions. They could be detected in our analyses only by Southernblot hybridization, but no such events were observed. After transduction we observed only two classes in addition to the expected His- Aps gene replacements: these were His+ ApR and His- Ap'. We inferred that they resulted from the integration of circular transducing phage DNA into the rcsi- dent A prophage or into the his locus, respectively. A5 with
transformants the His- Ap' clones generated His+ papillae on minimal medium lacking histidine. Southern blot hybrid- ization as above confirmed integration at his for the His- Ap' class (two with
xo
and four with x+HF, all in R+ cells), and integration at the A prophage for the His+ Ap' class (two withxo,
four with x'H, and four with x+HF, all in R+ cells) (data not shown).RESULTS
To test the hypothesis that gene replacement re-
Chi Sites and Gene Replacement in E coli 88 1
Donor DNA E ApR
I
GI
1 C . I E I+
-
1 kbGene
Replacement
I
KmR
Chromosome a * I . , . .
FIGL'RI.: 2.-Donor and recipient 1)NAs and the gene re- placement event. The top line shows the 6 5 k b KcoRI (E)
DNA fragment used as exogenous (donor) DNA. This DNA
contains Chi sites situated to stimrllate recombination in the Chi-free hi.K;I)C' DNA (thick line) homologous t o the chro-
mosomal (recipient) DNA (midtlle line). The Chi sites are 7
bp ( x + H ) and 13 bp
(x+!:)
from the his DNA. x Z H is a replacement of 18 bp (5' CT~~C;A<;(:AC~~GCCTCGA 3') for the 70 bp A s p I - P d fragment ofpBR322 ( D ~ s o s and K0\4',\1.(>ZM(OM:SKI 1993), and x'F is a C+T transition at bp 1497
(SvlTll e/ nl. 1981). The 1.2-kb katlam!.cin-resistance (KmK)
determinant (black box) is inserted a t the I?rd site of hisI1.
Parts of pRR322 DNA (thin lines), including the ampicillin- resistance (ApK) determinant, f l a n k the hi.$::kntt DNA. Single- headed arrows indicate the direction o f r e c o m b i ~ l a t i o n - s t i ~ ~ ~ ~ ~ - lation bv Chi sites, wl1ich are n o t t o scalc. Gene replacement
by homologous recombination between the ( G k b nonrepli- cative I h R I DNA fragment and the cl~romosomal I r i s locus inactivates the h i s oprron and protl~~ces KnmK His- ApS cells (bottom line). For transformations, the 6.5-kb k o R I fragment
was purified from plasmids pDA15
(x").
pDAl6 (x+HP'),pDA17 ( X ' I f ) , and pDA18 ( x ' F ) . For transductions, this frag- ment was inserted into the GoRl site of Agtll t o produce A
phages 1766, 1768, 1771 and 1772, respectively (Table 2);
infection of cells containing I h R I restriction endonuclease
generates linear DNA.
differ by the presence or the absence of Chi sites. Deriv- atives of plasmid pBR322, they contain a 3.0-kb DNA fragment of the L. coli his operon; this fragment, hisGDC', does not contain Chi (CARLOMAGNO et crl. 1988; data not shown). The hisD gene in the plasmids is interrupted by a 1.2-kb kanamycin-resistance de- terminant ( K m K o r Itan,) (see Figure 2 and Table 2 for plasmid structures, and MATERIALS A N D METHODS for constructions). When cut by EcoRI restriction endonu- clease, these plasmids produce a 6.5-kb nonreplicative linear DNA fragment. Homologous replacement of the chromosomal his+ locus with this fragment generates a His- K m K cell. T h e ampicillin-resistance (ApK) determi- nant of pBR322, present 0 1 1 one arm of the linear DNA,
is lost during homologous gene replacement. The Chi sites, when present, are situated to stimulate recombina- tion in the his::lrnn. segment according to the orienta-
tion-dependence and directionality of Chi's action in other recombinational events (see the Introduction). Plasmid pDA15
(x")
contains no Chi, and pDA16(x'HFj contains a Chi site -I0 bp from each side of the his::lzon segment. pDA17
( x ' H )
and pDA18 ( x + F )contain only x + H o r x + l < respectively to the left o r right of the Ais:: Iron segment. These Chi sites are expected to stimulate recombination in the his:: lznn segment be- cause in A vegetative crosses Chi stimulates recombina- tion over a several kilobase i n t e n d even when the Chi site is opposite a large region o f nonhomology (STAHI. and STAI-II. 197.5; MIERS Pt n I . 1995b).
In the following experiments, we introduced the puri- fied 6.5-kb nonreplicative linear DNA fragment from these four plasmids into his' I;. coli cells and selected K m K colonies. Homologous recombination events (KmK His- Ap') were scored by replica-plating the KinK
colonies onto minimal medium lacking histidine and o n t o LB agar supplemented with ampicillin. We used two methods to introduce linear DNA into the cells, transformation of Ca"-treated cells with purified linear DNA and phage A-mediated transduction of cells con- taining EcoRI restriction endonuclease to generate lin- ear DNA within the cells.
Transformation experiments: Chi sites enhance gene replacement in ret' cells: Using the linear, nonreplica- tive (5.5-kb fkoRl fragment of plasmids pDAl5
(x")
and pDAl6(x'HI.),
we transformed Ca'+-treated cells of strain V1904 ( w + ) to K m K . [Strain V1904 and related strains are r~cF243 and have an active RecBCD pathway but an inactive RecF pathway (SCHVLTZ rt d . 1983: SMITH 1989). We shall refer to these strains as I w I toindicate that they are rrrA+, rd", rc.cC*, and wrIl+. Re-
low we show that r~cF143 and r p c p strains behave simi- larly.] T h e K m K transformants were analyzed for gene replacement as described above and detailed in MATERI- ALS AND METHODS. From 63 independent transforma- tions, each with 100 ng of DNA, nine replacements were obtained with
xo
DNA and 402 with x+HFDNA (Table3). The frequency of gene replacement was 1.4 per microgram of
x"
DNA and 64 per microgram ofx'HF
DNA. Two Chi sites on the exogenous DNA thus stimu- lated gene replacement -45-fold.
To test the integrity of the poorly transforming
xo
DNA, we transformed in parallel strain V1906( r e c D l O I l ) . rPcD mutant cells lack RecBCD exonuclease activity and Chi stimulation of recombination but are recombination-proficient and support gene replace- ment (CHAUDNURY and SS41TH 1984; AMLNXES PI d .
1986; SHFXTLI. pt nl. 1988; R u s s ~ r . ~ . pt nl. 1989). In this strain, we observed the same frequency of gene replace- ments with the two DNAs (260 per microgram) (Table
S ) , indicating that the
xo
DNA was competent for gene replacement. Moreover, supercoiled pDAl.5(x")
and pDA16(x'HF)
plasmids transformed the rpT+ and rwD882 P. Dabert and G. R. Smith
TABLE 3
Chi sites enhance gene replacement by transformation in ret' E. coli
Supercoiled DNA"
Transformants per
microgram of DNA No. of gene replacements
Linear DNA
x lo-"
ReciDient cells and from
i
transfbrmations Chiexperiment number X0 X+ HF n X0 X + HF stimulation"
~ ."
V1904 (rec")
TI
T:!
T3 T4 T5 T6 T7 T8 T9 TI0 T11 Sum
Mean (per pg of DNA)
V1906 (re&)
Mean (per pg of DNA)
AB1 157 (ret+)
Sum
T1 T2 Sum
Mean (per pg of DNA)
C600 (rec") T1 T2
Sum
Mean (per pg of DNA)
ND' 7.9 2.4 3.0 1.4 0.7 0.6 3.9 4.5 10.7 1.3 ND 4.2 2.3 3.0 1.3 0.7 0.5 5.0 3.9 12.3 1
.o
6 6 3 5 3 3 4 6 5 15 7 63 0 0 0 1 0 0 2 2 0 4 0 9 1.4 48 92 15 23 12 26 4 143 4 26 9 402 64 >16 >31 >5 23 >4 >9 2 71 > 1.3
6.5
> 3
3.6 3.4 45
41/38d 1055
260
976 260
1.5 1.9 1 .0
0.4 6.0 0.3 6.0 5 5 10 0 1 1 1 .o 1 13 14 14 - 13
3.2 3.2 14
1.7 1
.o
1.3 1
.o
5 5 10 0 1 1 1.0 11 10 21 21 >3.6 10
1.4 1.2 21
~~ ~
' I Supercoiled or linear DNA (100 ng) and 1-8 X lo8 viable cells were used per transformation.
"The frequency of gene replacements (number per pg of DNA) obtained with linear x+HF DNA divided by the frequency with linear
x0
DNA. When no gene replacements were obtained withxo
DNA, three were assumed, to give the 95% confidence limit according to the Poisson distribution.~~
' Not determined. " X' DNA/x+HF DNA.
mid; we assume that the same was true for the linear X'HFand
xo
DNA and that differences in transformant frequencies reflect differences in recombination, notDNA uptake.
We verified that the K m R His- ApS transformants had
the expected homologous replacement. These trans- formants were genetically stable during nonselective growth, and analysis of their chromosomal DNA by Southern blot hybridization revealed the pattern ex- pected for homologous replacement (see MATERIALS AND METHODS). Other types of K m R transformants re- sulted from other events; for example, K m R His+ ApR
cells arose from contaminating circular plasmid DNA in some preparations (see MATERIALS AND METHODS). These types were not considered in the following analy- sis of gene replacement.
To test the generality of gene replacement stimulated by Chi in other strains, we transformed two distantly related E. coli rec" strains, AB1157 and C600 (BACHMANN
1987). The results (Table 3) were similar to those with
strain V1904 and suggest that the genetic factors, in- cluding recF, by which these strains differ did not greatly influence the frequency of gene replacement and its stimulation by two Chi sites.
Chi stimulated gene replacement depends upon the
RecBCD pathway: Since homologous recombination
Chi Sites and Gene Replacement in E. coli
TABLE
4Chi-stimulated gene replacement in transformation depends upon R e d and RecB activities
Supercoiled x+HF DNA Linear x+HF DNA
Transformants per pg Gene
Strain Genotype of DNA X n" replacements6
V1904 rec+ 2.9 10 28
V1909 recA56 2.6 18 0
V1904 rec+ 12.3 12 13
V1908 recB 2 1 2.5 22 0
Number of indeDendent transformations with 100 ng of X'HFlinear DNA per transformation.
" Total gene replacements from n transformations
(recA56), and V1908 (recB2I) with x+HF linear DNA. The results (Table 4) show that no gene replacements were observed in either the recA or the recB mutant. Taking into account the number of transformations performed in each background and the transformation frequency of these strains with supercoiled DNA (which includes a measure of cell viability), we would expect 45 and five gene replacements in the recA and recB mu- tants, respectively, if the frequencies of gene replace- ments were as high as that in the ret' strain. The differ- ences observed between these strains and the rec' strain are highly significant ( P = 3 X and
7
X lo-',respectively, according to the Poisson distribution). The same results would be obtained if the linear DNA con- tained a substance that prevented transformation of the recA and recB strains but not the
reef
strain. This possibil- ity was excluded by transforming the recA and recB mu- tants with supercoiled pACYC184 DNA (25 ng) either alone or mixed with the x+HF linear DNA (100 ng). The same frequency of pACYC184 (Cm") transformants was obtained in both cases (data not shown). Together, these results show that Chi-stimulated gene replace- ment depends upon the recA+ and recB+ genes and, thus, upon the RecBCD pathway.One
Chi
site enhances gene replacement at an inter- mediate level: The model presented in the Introduc- tion proposes that recombination of linear exogenous DNA with the circular endogenous chromosome is max- imal with one Chi site activated from each end of the linear DNA. To test this hypothesis, we transformed strain V1904 (reef) with linear DNA that contained only one Chi site, X+Horx+F,
oriented to stimulate recom- bination in the his::kan region (Figure 2). These DNAs produced gene replacements at frequencies between those of thexo
and x'HFDNAs (Table 5). The relative frequencies of gene replacements were 1, 9, 17, and77
with the
xo,
x'H,x'F,
and x+HF DNAs, respectively. The second Chi stimulated about ninefold (x+HF us. x + H ) or about fivefold [x+HFus.x+F;
these values are similar to the degree of enhancement by addition of a single Chi site to X in vegetative crosses (MYERS andSTAHL 1994; SMITH 1994)]. Our observations demon-
883
strate that two Chi sites are needed for the maximal frequency of gene replacement.
Transduction experiments: The low number of transformants obtained with linear DNA (usually <IO
per transformation; Tables 3-5) is likely limited by inef- ficient entry of DNA into Ca'+-treated E. coli. We there- fore tried another method of delivering linear DNA into the cells. This phage h delivery system, described
by NUSSBAUM et al. (1992), contains two components:
X phages bearing the same 6.5-kb nonreplicative EcoRI
DNA fragments used above and bracketed by EcoRI sites and EcoRI restriction endonuclease encoded by a plas- mid in the recipient cells to release this 6.5-kb linear DNA fragment. This system allows efficient and nearly synchronous delivery of linear DNA into the cells. To prevent the replication of transducing phages that might escape the EcoRI restriction, the recipient cells carried a X cl(Ind-) noninducible prophage.
Chi
sites enhance restriction-stimulated gene replace- ment: We first tested the effect of Chi sites in the in- fecting phage DNA in the presence and absence ofEcoRI restriction. Chi sites increased the frequency of gene replacements -70-fold, but only when the recipi- ent cells carried the EcoRI restriction determinant (Ta- ble 6). With
x'
phage, the frequency of gene replace- ments (His- Ap') was 0.13 X per phage, and withx +
HF phage, it was 9.3 X per phage, or 72-fold higher. Not only was the frequency of gene replace- ments increased by Chi, but also the fraction of that class among the selected K m R colonies was increased,from 5 to
78%,
by Chi. Thus, in the presence of two Chi sites and EcoRI restriction the majority of the selected transductants had the desired gene replacement.Two other phenotypic classes were found among the selected
Km"
transductants (Table 6). We deduce that they were formed by integration of the nonrestricted circular X DNA into the X prophage locus to form His+ ApR cells or at the his locus to form His- Ap" cells (see884 P. Dabert and G. R. Smith
TABLE 5
Gene replacement by transformation is enhanced by two Chi sites more than by one Chi site
Supercoiled DNA" Linear DNA"
Transformants per No. of No. of gene Gene replacements Chi
Chi sites in DNA pg of DNA X transformations replacements per 100 ng DNA stimulationb
None 6.8, 39 3, 6 0, 2 <1, 0.3 1
X+H 6.8, 46 8 , 8 9, 23 1.1, 2.9 9
X + F 8.3, 38 8, 8 13, 46 1.6, 5.8 17
x + H x + F 6.5, 50 3, 7 26, 143 8.7, 20 77
Strain VI904 (ret+) was transformed with 100 ng of supercoiled or linear DNA for each transformation. Data are from two
The frequency of gene replacements (combined number per pg of DNA) obtained with the indicated DNA divided by the experiments, each with all four DNAs, done on separate days.
frequency with
x0
DNA.the greater length of homology at the X prophage. The frequency of these integrations per infecting phage was about the same in the presence or absence of Chi: re- spectively,
2.2
or 2.1x
for integration at the X prophage and 0.5 or 0.4 X for integration at his. Therefore, Chi specifically enhanced the frequency of gene replacements, as expected if linear DNA (suscepti- ble to RecBCD enzyme and Chi activation) produces gene replacements and if circular DNA (not susceptible to RecBCD enzyme or Chi activation) produces integra- tions.Chi sites in the infecting phage DNA had no signifi- cant effect in the absence of the EcoRI restriction deter- minant. The frequency of the K m R transductants
(2
Xlop3 per phage) was about the same with or without Chi, and the distribution of the three phenotypic classes was about the same (Table 6). Most of the K m R trans-
ductants had apparently integrated the circular X DNA,
as expected in the absence of restriction. Only -1%
had gene replacements.
Further evidence that gene replacement is most effi- cient with linear DNA came from experiments with phage previously modified by EcoRI modification en- zyme and hence resistant to restriction. The frequency of J S m R transductants with X'HFEcoRI-modified phage was 1.8 X per phage in cells with the EcoRI restric- tion determinant and 2.0 X
lo-'
in cells without it(Table 6 ) . Fewer than 1% of these transductants had gene replacements (His- Ap')
.
These results parallel those with unmodifiedxcHF
(or x") phage in the non- restricting host and imply that gene replacement re- quires cutting of the infecting DNA.The next experiments showed that, as in the transfor- mation experiments, two Chi sites enhanced gene re- placement more than one Chi. In the restricting cells, the frequencies of gene replacement per IO5 phage
TABLE 6
Chi sites on restricted (linear) DNA enhance gene replacement by transduction in rec' E. coli Percentage in each class
K m R Gene
Chi sites transductants No. His+ ApR His- ApR His- Aps replacements Chi
in DNA Restriction per lo5 phage" analyzed (Int. at A)" (Int. at his)b (G.R.)b per
lo5
phage' stimulationdNone
None
+
2.7 (1.2-5.3) 371 79 16 5 0.13 1x'H x+F - 200 (100-330) 1260 88 11 1 2.0 -
x + H x'F
+
12 (5-22) 1187 18 4 78 9.3 72- 180 (75-320) 689 86 13 1 1.8
-
x'Hx'F (-y 180 (150-200) 183 85 15 0 <3 -
"Derivatives of strain V1910 (rec') containing plasmid pDA19 (R-M+) or pDA2O (R'M') were infected with A phages 1766
(x")
or 1768 (x'HF) at an m.0.i. of 0.1. K m R transductants were analyzed for gene replacement (His- ApR) as described inMATERIALS AND METHODS. Data are the means (with ranges in parentheses) from nine experiments with two stocks of
x'
phage, 13 experiments with two stocks of A x'HF phage, and two experiments with modified x+HF phage; the experiments, in most cases each with X' and x+HF phage, were done on 13 separate days. Standard deviations of the percent in each class ranged from 1 to 5%."
'Integration of the A transducing phage at the A prophage (Int. at A) or at the his locus (Int. at his) or gene replacement
(G.R.).
Chi Sites and Gene Replacement in E. coli 885
TABLE
7Gene replacement by transduction is enhanced by two Chi sites more than by one Chi site
Percent in each class Gene
K m R replacements
transductants No. His+ ApK His- Ap' His- Aps per 1oi Chi
Chi sites in DNA per lo5 phage" analyzed (Int. at (Int. at his)" (G.R.)6 phage stimulation'
None 2.6 (1.5-4.2) 140 83 13 4 0.10 1
X+H 7.7 (1.7-13) 259 38 6 55 4.2 42
X + F 6.5 (0.8-11) 226 40 6 53 3.4 34
x + H x + F 8.9 (4.7-13) 300 21 4 75 6.7 67
" A derivative of strain V1910 (reff) containing plasmid pDA20 (R' M+) was infected with h phages 1766 (x"), 1771 ( x + H ) ,
1772 ( x + F ) , or 1768 (x'HF) at an m.0.i. of 0.1. K m R transductants were analyzed for gene replacement (His- Ap') as described
in MATERIALS AND METHODS. Data are the means (with ranges in parentheses) from three experiments, each with all four phages, done on three separate days. Standard deviations of the percent in each class ranged from 3 to 8%. Concurrent infections of strain V1910 (pDA19) (R-M') gave 190 (range 110-270) K m K transductants per lo" phage. Among 300 analyzed for each phage
type 86 2 5% were His' Ap', 13 ? 4% were His- ApK, and <3% were His- Ap'. No significant differences were seen for the four phage types or for the three experiments.
"Integration of the X transducing phage at the h prophage (Int. at A) or at the his locus (Int. at his), or gene replacement
(G.R.) .
The frequency of gene replacements (per loi phage) obtained with the indicated phage divided by the frequency with
x'
phage.were 0.1, 4.2, 3.4, and 6.7 with the
xo,
x f H , x+F a n dxfHF phage, respectively (Table
7).
T h e fraction of gene replacements (His- Aps) among the selected K m Ktransductants was 4, 54, a n d
75%
for 0 , 1 , and 2 Chi sites, respectively. Thus, by both measures, two Chi sites enhanced gene replacement more than one Chi. Single Chi sites enhanced gene replacement relatively more in transduction than in transformation (Tables 5 and7).
We believe this reflects incomplete cutting of the infecting DNA in transduction and the effect of Chi- like sites in the A DNA when it remains attached to one e n d of the his::kan DNA (see DISCUSSION).Transduction-mediated gene replacement depends upon the RecBCD pathway: T o test the rec gene, and hence recombination pathway, requirement for Chi-en- hanced gene replacement, we infected recmutant EroRI- restricting hosts with unmodified X'HFphage. No gene replacements were obtained in recA o r recB mutant hosts; the frequencies were reduced by factors of >60 a n d >150, respectively, from that in the ref+ host (Table 8). These hosts were recF143, to inactivate the RecF pathway. T o test directly for an involvement of re@ we infected rpclk+ cells with
x"
and X+HFphage. The results were not greatly different from those with the recF143 host (compare lines 1 and 2 with lines7
and 8 in Table 8): two Chi sites stimulated gene replacement -100- fold in both hosts. These results were similar to those obtained by transformation (Table 3; strains V1904, AB1 157 and C600). Infection of a recD mutant host gave gene replacements at the same frequency with bothx"
and x+HF (12 X per phage), about the same frequency as X+HFin recD+ cells (9.3 X per phage). These results show that thex0
phage were competent to produce gene replacements but did s o at low fre- quency in ref+ cells. Thus, as in transformation, Chi-stimulated gene replacement in these A transductions of
ret+
cells proceeds via the RecBCD pathway and is maximal with DNA containing two Chi sites.One h e a r DNA fragment is sufficient to produce one gene replacement: The model presented in the Introduction and Figure 1 supposes that a single linear DNA fragment is required to produce one gene replace- ment. To test this supposition, we determined the de- pendence of the number of gene replacements ob- tained on the number of phage used. In two separate experiments the m.0.i. was varied from 0.01 to 18 x'HF
phage per cell. Throughout this range the number of gene replacements was a linear function of the number of phage used (Figure 3 ) ; the frequency of gene re- placements was constant at 4 x per phage. Since at low m.0.i. most of the infected cells had only o n e injected and hence linearized phage DNA, a single DNA fragment is sufficient to produce a gene replace- ment.
DISCUSSION
The data reported here show that gene replacement in wild-type (rec') E. colican be achieved with short (6.5 kb) linear DNA that has two properly situated Chi sites. These gene replacements proceeded by the RecBCD pathway and were enhanced -50-fold by two Chi sites. In transformation experiments, the Chi sites aug- mented an essentially undetectable event into a repro- ducibly detectable one; with purified linear DNA 90-
100% of the selected K m K transformants had the de-
sired gene replacement. In transduction experiments in EcoRI-restricting cells, thousands of K m R transductants
886 P. Dabert and G. R. Smith
TABLE 8
Chi-stimulated gene replacement in transduction depends upon R e d , RecB, and RecD activities
Percent in each class
Gene No. of Chi sites K m R transductants No. His' ApR His- ApR His- Aps replacements
Strain Genotype" experiments in DNA per 10' phage" analyzed (Int. at A)' (Int. at his)' (G.R.)' per lo5 phage
V1910 reef
V1910 rec+ V2085 recA56
V2084 recBZ1
V1911 recDlOll V1911 recDlO11
V2106 recp
V2106 re@
13
9
2 3 2
2
1 1
x+HF None x+HF
x+
HF x+HF Nonex+
HF None12 (5-22) 1187
2.7 (1.2-5.3) 371
<0.15 0
0.3 (<0.15-0.5) 15
29 (28, 30) 180
27 (27, 27) 179
24 120
4.5 89
18 79
93
56
53 15 88 -
4
16
7 2 2 3 9 -
78
5
0
42 45 82 3
-
9.3 0.13 <0.15
<0.06
12 12 20
0.13
"All strains except V2106 also carry the recF143 mutation (see Table 1).
'Derivatives of the indicated strains containing plasmid pDA20 (R+M+) were infected with A phage 1768 x+HF at an m.0.i. of 0.1. hR transductants were analyzed for gene replacement (His- Ap') as described in MATERIALS AND METHODS. Data are
the means (with ranges or individual values in parentheses) from the indicated number of experiments. Data for rec+ infections are from Table 6.
'Integration of the A transducing phage at the A prophage (Int. at A) or at the his locus (Int. at his), or gene replacement (G.R.).
molecule with two Chi sites was sufficient to produce one gene replacement. We discuss the implications of these findings for the mechanism of homologous re- combination and for methods for gene replacement in wild-type E. coli and other organisms.
Mechanism of homologous recombination: A model for conjugational and transductional recombination predicts that two Chi-stimulated, RecBCD-dependent
exchanges are required, one near each end of the linear
*-
.
rc
. I /a
c
103
1
02
1
0'
exogenous DNA (see the Introduction, Figure 1)
(SMITH 1991). The data reported here support that pre- diction: the frequency of gene replacement was highest with two Chi sites, each situated to promote recombina- tion in the homologous region between them (Tables
3 and 5-7). These events were dependent upon the
recA+ a n d recBf genes ( i e . , the RecBCD pathway; Tables 4 and 8). O u r results therefore support a key feature of the "long chunk" model of E. coli recombination.
Although two Chi sites were required for the maximal frequency of gene replacement, one Chi site on either one side of the homology or the other gave significant stimulation relative to the Chi-free DNA (Tables 5 and
7).
Chi-like sites ( i e . , sequences differing from Chi at1 bp) may be responsible for this residual activity and for gene replacement in the absence of Chi. Some Chi- like sites have low but detectable activity (SCHULTZ e1
al. 1981; CHENC a n d SMITH 1984, 1987). In the 5.3 kb of pBR322 a n d his DNA flanking the selected K m R
determinant there are six Chi-like sites properly situ- ated to stimulate the selected event; two are on the left arm, and four are on the right arm (oriented as in Figure 2 ; data not shown). Although not all of these sequences have, to o u r knowledge, been directly tested for Chi-like activity, they, or perhaps other more dis- tantly related sequences, may be responsible for the low frequency of gene replacements in the absence of Chi.
in MATERIALS AND METHODS. Data are from experiments done
Chi Sites and Gene Replacement in E. coli 887
on either the right or the left side of the expected 6.5- kb fragment (Figure
2).
The many Chi-like sites in thisA arm, in conjunction with a single Chi site in the 6.5- kb fragment, could promote gene replacement at a re- duced frequency relative to two fully active Chi sites. Chi-like sites may also play a role in conjugational and transductional crosses.
In the transformation experiments the stimulation by two Chi sites (77-fold) was greater than the sum of the stimulation by the individual Chi sites (9- and
17-
fold; Table 5). This result indicates that the two Chi sites act cooperatively, as expected from the one Chi- one exchange hypothesis and the requirement for two
exchanges in gene replacement. This result argues against models in which Chi converts RecBCD enzyme into a state capable of promoting multiple exchanges on the Chi-containing DNA molecule. This result does not, however, bear on the question of Chi’s effect, via RecBCD enzyme, on recombination of another DNA molecule. In the transduction experiments, the stimula- tion by two Chi sites (67-fold) was approximately equal to the sum of the stimulation by the individual Chi sites (42- and 34fold; Table
7).
As noted above, however, Chi-like sites in A may have augmented the stimulation by the single Chi sites.The observation that a single DNA molecule can pro- duce a gene replacement in a Chi-stimulated manner (Figure 3; Tables 3-8) implies that Chi can stimulate recombination in cis (i.e., of the molecule on which it resides as predicted by the model discussed in the Introduction). The finding that Chi changes RecBCD enzyme has raised the possibility that Chi acts in trans, and in some situations, it does so (TAYLOR and SMITH 1992; KOPPEN et al. 1995; MYERS et al. 1995a). The data reported here show that trawaction of Chi is not neces- sary for stimulation of recombination. Chi activity in P1- mediated transduction and Hfr-mediated conjugation also suggests that Chi can act in cis (DOWER and STAHL 1981). The switch from cis to trans action of Chi may depend on the ratio of the number of broken DNA ends to the number of RecBCD enzyme molecules avail- able in the cell. In conjugational and transductional crosses, and perhaps in cells with a small amount of DNA damage, Chi may act primarily or exclusively in cis. Both types of Chi action may occur in cells with heavily damaged DNA (i.e., when there are many bro- ken DNA molecules to be repaired).
Gene replacement in E. coli and other organisms:
Two methods for gene replacement with short ( i e . , i n vitro manipulated) DNA have been described previously for E. coli and other bacteria. One method involves inte- gration of circular DNA at a homologous locus to form a direct duplication; excision by homologous recombi- nation in a different interval produces the desired re- placement (e.g., GUTTERSON and KOSHLAND 1983; JOYCE and GRINDLEY 1984; PARKER and MARINUS 1988; HAMILTON et al. 1989;
KULAKAUSKAS
et al. 1991). Thismultistep method often requires proper cellular [e.g.,
pol4 (Ts)] or plasmid [e.g., ori (Ts)] mutations. Another method employs rec mutant cells in which the RecE pathway (recB recC sbcA mutants), the RecF pathway
( r e d recC sbcB sbcC or D mutants), or the RecBC (D-) pathway (recD mutants) is active; in such mutants Chi- independent gene replacement by linear DNA occurs at frequencies similar to that of the Chi-dependent events in rec+ cells studied here (e.g., JMIN and SCHIM- MEL 1984; WINANS et al. 1985; SHEVELL et al. 1988; RuS
SELL et al. 1989) (Tables 3 and 8). Such mutants are not generally available for bacteria other than E. coli. With either method it is often necessary to transfer the gene replacement ( . g . , by P1-mediated transduction) to another suitable strain for further work. The method used here allows gene replacement in one step to any transformable rec+ host. In some cases, the higher effi- ciency A-mediated transduction method may be more suitable.
The Chi-stimulated method described here should be applicable to other bacteria. Since Chi stimulates recombination promoted by RecBCD-like enzymes from numerous enteric bacteria (SCHULTZ and SMITH 1986; SMITH et al. 1986; MCKITTRICK and SMITH 1989; WEICHENEUN and WACKERNAGEL 1989; RINKEN et al.
1991), the procedure used here is likely to work with many or all enteric bacteria. For other bacteria, identi- fying sites equivalent to Chi would allow construction of appropriate vectors containing dual recombination- enhancing sites. BISWAS et al. (1995) identified a site (5’ GCGCGTG 3’) that stimulates accumulation of high molecular weight plasmid DNA in Lactococcus lactis, much as Chi does on plasmids in E. coli (DABERT et al. 1992). This method may successfully identify RecBCD- enzyme-interacting sites in other bacteria. In eukaryotes with mitotic recombination hotspots, such as Saccharo-
myces cueuisiae and humans (see review by SMITH 1994), the responsible sites may aid gene targeting in somatic cells in a manner analogous to that shown here for Chi. Although Chi sites stimulate gene replacement -50- fold in the experiments reported here, the frequency of replacements per x+HFDNA molecule is still lower by a factor of
-lo3
than the frequency of replacements in conjugation and P1-mediated transduction. In our X-mediated transduction experiments the frequencywas
-7
X lop5 per infecting phage (Table 6; Figure 3).On the assumption that linear DNA and circular DNA enter @+-treated E. coli cells equally efficiently, we calculate a similar frequency in the transformation ex- periments (64 replacements per microgram of linear DNA + 3.4 X lo6 transformants per microgram of circu- lar DNA = 2 X lop5; Table 3). Finding conditions or factors that increase the efficiency of gene replacement with short linear DNA will aid strain constructions and will permit a physical analysis of DNA recombining in