Gene, 97 (1991) 131-136
131
Elsevier GENE 03853
Ultraviolet light-induced plasmid-chromosome recF DNA; reciprocal;
(R~combin~t
nonreciprocal;
recombination in Edzerichia
plasmid
survival;
repair;
double-stand
cdi: the role of recB and
breaks)
John S. Mudgett *, Michael Buckholt and William D. Taylor Department
of Molecular and Cell Biology, The Pennsylvania State University, University Pa& PA 16802 (U.S.A.)
Received by R.E. Yasbin: 5 July 1990 Revised: 27 August 1990 Accepted: 31 August 1990
SUMMARY Bacterial host cells of different ret genotypes were used to investigate genetic requirements of ultraviolet light (UV)-induced homologous pIasmid-chromosome recombination. Plasmid DNAs which contained a wt or mutant ZacY gene were irradiated with UV prior to ~~sformation into E~c~e~~c~~acoli host strains which contained the complementary ZacY allele. Slung transformants were screened to determine the directions of UV-induced recombinational exchange between the bacterial and plasmid lacy genes, by assaying lactose utilization. Nonreciprocal chromosome-to-plasmid recombination was 100% dependent on the recA gene and > 80% dependent on the recF gene, but not dependent upon the recB gene of E. coli. In contrast, reciprocal plasmid-chromosome recombination was strictly dependent on the recA gene, greatly dependent (approx. 80%) on the veeF gene, and moderately dependent on the recB gene. Nonr~iprocal plasmid-to-chromosome recombination was only induced at very low frequencies, and appeared to be moderately dependent on the recB gene, but not dependent on the recF gene. UV-induced plasmid-chromosome recombination appeared to proceed by a two-step mechanism. In this model, the initial step is recF-dependent, recB-independent, and either resolves to become a nonreciprocal chromosome-to-
plasmid recombinant, or proceeds to the second step. The second step is moderately recB-dependent and results in the reciprocal exchange of plas~d-chromosome sequences.
INTRODUCTION Several groups have used plasmid DNAs, modified with various mutagenic agents, as molecular substrates for bacterial repair responses (Mizusawa et al., 1981; S&mid et al., 1982; Chattoraj et al., 1984; Luisi-DeLuca et al., 1984; Abbott 1985a,b; Mudgett and Taylor, 1986). Treat-
Correspondence to: Dr. W.D. Taylor, 108Althouse Laboratory, University Park, PA 16802(U.S.A.) Tel. (814)865-1968;Fax (814)X65-3634. * Present address: MS-M886, Los Alamos National Laboratory, Los Alamos, NM 87545 (U.S.A.) Tel. (505)667-0126. Abbreviations: bp, base pair(s); cfu, colony-forming unit(s); ds, doubfe strand(ed); kb, kilobase or 1000bp; UV, ultraviolet light; wt, wild type. 0378-l 119/91/$03.50
0
1991 Elsevier
Science Publishers
B.V. (Biomedical
Division)
ment of plasmid substrates prior to transformation into untreated bacterial host cells enabled defined substratelesion combinations to be used without the complications of host-cell modification. Chemical and UV-induced bulky lesions were found not to be highly mutagenic to the plasmid, but actively induced recombination between homologous plasmid and c~omosom~ sequences if the bacterial host had aviable ~ecAgene (Chattoraj et al., 1984; Luisi-DeLuca et al., 1984; Abbott 1985a,b; Mudgett and Taylor, 1986). We have recently demonstrated that y-irradiated plasmid DNA also recombined with homologous I?. co&chromosomal DNA in a recA + host strain at levels much higher than those observed for plasmid mutation (Mudgett et al., 1990). RecA protein is required for recombinational pairing and
132 strand exchange between complementary DNAs to repair postreplication gaps opposite UV-induced pyrimidine dimers (West et al., 1981; 1982; Radding, 1982; Wang and Smith, 1983). This initially asymmetric (nonreciprocal) transfer of homologous DNA can be resolved or converted to a symmetrical (reciprocal) exchange (Radding, 1978; 1982). This repair of daughter-strand gaps also requires the recF gene, but not the recB or reccgenes (Wang and Smith, 1983; 1984; Smith and Sharma, 1987), which have been associated with the repair of ds breaks (Wang and Smith, 1983; Smith and Sharma, 1987; Mahan and Roth, 1989). Similarly, it has been reported that the recF but not the recB or recC genes were required for interplasmidic recombination (Cohen and Laban, 1983) and UV-induced plasmid-chromosome recombination (Abbott, 1985b). The plasmid-chromosome recombination system of Abbott (1985a) appeared to be detecting only nonreciprocal chromosome-to-plasmid recombination. In studies designed to determine different directions of UV-induced plasmid-chromosome recombination, we found that nonreciprocal chromosome-to-plasmid and reciprocal plasmidchromosome recombination were induced at similar levels, while nonreciprocal plasmid-to-chromosome recombination occurred at much lower levels (Mudgett and Taylor, 1986). These results prompted us to investigate which ret genes are involved in the different directions of UV-induced
TABLE
plasmid-chromosome recombination. The experimental system employed consisted of bacterial hosts with different ZacY and ret alleles, and three plasmid substrates with complementing or nonexpressed ZucY genes. The aim of the present study was to determine the levels and directions of UV-induced plasmid-chromosome recombination between the ZacY alleles in wt, red -, recB -, and recF- E. coli host strains.
EXPERIMENTAL
AND DISCUSSION
(a) The experimental system and procedures The experimental system used has been previously described (Mudgett and Taylor, 1986), and is summarized in Table I. In the present study, pBRM4Yl was substituted for previously described pBRPlY1 (Mudgett and Taylor, 1986). Briefly, plasmid pBRM4, which expresses the ZucY+ gene, was used to detect UV-induced nonreciprocal chromosome-to-plasmid recombination events in AB 1157, AB2463, AB2470, and JC9239 ZacYl bacterial host strains as Lac ~ colonies. Nonreciprocal plasmid-to-chromosome recombination events were detected as Lac - colonies using plasmid pBRM4Y 1 and the lacy’ bacterial hosts RDP147, WDTlOO, and WDTlOl. Reciprocal recombination events were detected as Lac + colonies using plasmid pBRM5 and
I
The experimental
system used to detect plasmid-chromosome
recombination Recombination
Plasmid
Bacterial
b
ABll57, AB2470,
pBRM4
AB2463 JC9239
RDP147,
pBRM4Y 1
WDTlOO
ABl157, AB2470,
pBRM5
AB2463 JC9239
a The bacterial
strains
c
genotype d
None
Lac +
10CY’
c to p
[Lac-] Lac +
lacy1
Lac +
lacy1
None
Lac +
lacy1
c
to p p to c
Lac +
lUCY+
[Lac-]
lacy1
Reciprocal
Lac +
lacy’
None c to p p to c Reciprocal
Lac -
lacy’ IacYl
p to c Reciprocal
WDTlOl
Plasmid
Colony phenotype
hosts =
lacy+
Lac _ [Lac+] [Lac + ]
lacy’ lacy1
genotypes are as follows: AB 1157 (recA + , lacYl), AB2463 (recA13, IucYl), AB2470 (recB21, lacYl), and WDTlOl (recF143,lacY+) were constructed (recA + , lacy+). The bacterial strains WDTlOO (recBZl,lacY+) AB2470 and JC9239, respectively, with UV-irradiated (300 J/mz) plasmid pBRM5 to induce recombinational were cured of plasmids by growth in the absence of lacy1 allele with the plasmid lacy’ allele. Lac + transformants agar. Lac + colonies were identified by replicate growth on selective and nonselective MacConkey
used and their pertinent
JC9239 (recF143, ZacYl), and RDP147 for use in this study by transforming
replacement of the chromosomal antibiotics and plasmid-deficient b The possible direction(s) of recombinational c The predicted
Lac phenotypes
d The predicted
lac genotypes
of recombinant of recombinant
exchange; colonies. plasmids.
c, chromosome;
p, plasmid. The recombination
The recombinants
scored
are indicated
direction(s)
by brackets.
scored are indicated
in bold letters.
P
V
VEV
HC
pBRM4 Y
P
V
Yl
P
V
HCV
V I
V I
CEP
V 1
V I
V I
CEP
A
VEVPV I I, 11
HC
pBRM4Yl
V I
I
A
V
V
VEV
CEP
1
pBRM5 A
Y
H 1 kb
Fig. 1. Restriction Plasmids pBRM4 Taylor,
maps of plasmids pBRM4, pBRM4Y1, and pBRM5. and pBRM5 were described previously (Mudgett and
1986). Both plasmids
pBRM4 was able to complement orientation
contained
the wt lacy’
the bacterial
with respect to the plasmid promoter;
was inverted
in orientation,
derived
from pBRM4
construct
a plasmid
and its associated
so lacy+ was not expressed. that contained
PstI and PvuII restriction
due to its
the pBRM5 lacy+ gene
by using plasmid-chromosome
(pBRM4Yl)
gene, but only
lacy1 mutation
pBRM4Yl recombination
the mutant
sites as previously
lacy1
was to allele
described
for plasmid pBRPlY1 (Mudgett and Taylor, 1986). Heavy line, pBR322 sequences; thin line, E%li sequences; open boxes, luc operon sequences. C, &I; E,EcoRI; H, HindIII; P,PstI; V, PvuII. The arrow indicates the Pl promoter
of pBR322
which is responsible
for lacy
expression.
ZucYl host strains, as pBRM5 contained a wt but nonexpressed ZucY gene. Nonreciprocal plasmid-to-chromosome recombination events could also give rise to Lac + colonies using pBRM5 (Table I) (Mudgett and Taylor, 1986), but these frequencies were determined independently using pBRM4Y 1, and plasmids from recombinant Lac + colonies were examined to confirm the reciprocity of pBRM5 exchange, as described below. (b) Inactivation of UV-irradiated plasmid DNA Plasmid survival was calculated as the ratio of transformants (in cfu/ml) at a given UV dose to the number of transformants (in cfujml) at zero dose. Exponential survival curves for (transformants from) UV-irradiated plasmid DNAs are shown in Fig. 2. Plasmid survival, for the three plasmids used, was greatest in AB 1157 (recA + ), followed by the recB -, recF- , and recA - bacterial strains. This is in agreement with observations that recA + -dependent repair pathways may slightly enhance UV-irradiated plasmid survival (Mudgett and Taylor, 1986; Strike and Roberts, 1982; Roberts and Strike, 1981). In our study, however, the recF- mutation slightly decreased UV-irradiated plasmid survival for pBRM4 and pBRM5, and moderately decreased pBRM4Yl survival (Fig. 2), while the same recF mutation (bacterial host) did not affect plasmid survival in a previous study (Roberts and Strike, 198 1). Our results also contrast reports that the same recB mutation (bacterial host) reduced UV-irradiated plasmid survival to levels much less than observed in recA - cells
I
I
I
100
200
300
.Ol +
0
UV Dose (J/m 2, Fig. 2. Survival of UV-irradiated plasmids pBRM4, pBRM4Y1, and pBRM5 in unirradiated E. coli host cells. Plasmid DNA was irradiated with
UV
and
previously
transformed
described
into
(Mudgett
cells were plated on MacConkey ampicillin/ml
competent
and Taylor,
for determination
bacterial
host
cells
1986). Transformed
agar (Difco) supplemented of plasmid
survival
with 50 pg
and Lac phenotype;
Lac + colonies were red while Lac - colonies were white. Plasmid was isolated viously
from recombinant
described
contains
(Mudgett
characteristic
this enabled
colonies and
for restriction
Taylor,
analysis
1986). As the lacy1
PstI and PvuII sites (Luisi-deLuca
us to physically
as
bacterial
confirm the recombinational
DNA as preallele
et al., 1984) exchanges
by
restriction endonuclease analysis of recombinant plasmids (Table (Top) Survival of UV-irradiated pBRM4 in ABl157 (recA + , lacY1;
n ),
I).
AB2470 (recB_, lacY1; A), JC9239 (recF_, lacY1; 0) and AB2463 (recA_, lacY1; +). (Middle) Survival of UV-irradiated pBRM4Yl in RDP147
(recA +, lacy+
(recF_, lacy’; 0). AB1157 (recA’,lacYl; lacY1; 0)
; n ), WDTlOO (ye&-,
lacy’
; A) and WDTlOl
(Bottom) Survival of UV-irradiated pBRM5 in n ), AB2470 (recB_, lacY1; A), JC9239 (recF_,
and AB2463
(recA_,
lacY1; a).
(Roberts and Strike, 1981). The authors of this previous study suggested that the RecBCD protein, exonuclease V, was acting as an essential component of excision repair (Roberts and Strike, 1981). The major difference between
134 our system and the system of Roberts and Strike is that our plasmids were able to recombine with the host chromo-
I
I recB
pBRM4 0.10
some. Plasmid-chromosome (vecF-dependent) recombination may therefore have been involved in the repair of plasmid damage that could not be removed by other repair pathways (e.g., excision repair) in the recB- hosts. (c) Plasmid-chromosome recombination plasmid DNA Nonreciprocal chromosome-to-plasmid
I
0.12
wt
0.08
-
0.06
-
of UV-irradiated L recA
recombination
was assayed using plasmid pBRM4 as described in section b. UV irradiation induced high levels (10% of the pBRM4 transformants at 300 J/m’) of Lac - nonreciprocal chromo-
0.12
some-to-plasmid recomb&rants in AB 1157 (recA + ) host cells, but no recombinants were detected when AB2463 (recA - ) cells were used (Fig. 3). These results were similar to those observed previously for plasmid pBRM4 (Mudgett and Taylor, 1986). AB2470 (recB-) bacterial host cells had normal levels of induced chromosome-to-plasmid recombination, while JC9239 (reck_) cells were about 80% deficient in chromosome-to-plasmid recombination (at 300 J/m”) (Fig. 3). This is in agreement with relative plasmid-chromosome recombination frequencies determined by Abbott (1985b). Plasmid pBRM4 DNAs isolated from Lac - recombinant colonies were digested with PstI and analyzed on agarose gels to confirm that recombination transferred the ZucYl mutant gene from the chromosome to the plasmid substrate as previously described (Mudgett and Taylor, 1986). All (24/24) plasmid DNAs isolated from AB1157 (recA ‘) and almost all (23/24) plasmid DNAs from AB2470 (recB-) host cells were found to contain the ZucYl allele (data not shown). Less than half (9/24) of the JC9239 (vecF_) Laccolonies were found to contain recombinant plasmids, suggesting that chromosome-toplasmid recombination was even more deficient in the recF- host. The nonrecombinant lacy- plasmids were presumably the result of plasmid mutation.
0.08
Nonreciprocal plasmid-to-chromosome recombination frequencies were determined using pBRM4Yl and lacy’ host strains (Table I). Only very low levels of these exchanges (Laccolonies) were induced (Fig. 3), as previously reported (Mudgett and Taylor, 1986). The frequency of these events was apparently reduced when recB- cells were used, while recF_ cells had almost normal levels of plasmid-to-chromosome recombination (Fig. 3). Plasmid isolations confirmed that pBRM4Yl retained the lacy1 allele in all (24/24) of the recombinants examined (data not shown). pBRM5 was used to assay UV-induced reciprocal plasmid-chromosome recombination in ZacYl host cells (Table I). Nonreciprocal plasmid-to-chromosome exchanges could also give rise to Lac + colonies in this system (Table I), but the low levels of these events (observed with
I .
0.10
I
I
pBRM4Yl
-
0.06 0.04 0.02 wt recF recB
0.00 --
I
0.12
0.10
-
0.08
-
0.06
-
I
I
pBRM5
0
T-
100
200
300
UV Dose (J/m 2 ) Fig. 3. Plasmid-chromosome pBRM4,
pBRM4Y1,
recombination
and
pBRM5
of UV-irradiated
in unirradiated
plasmids
E. coli host
cells.
Plasmid pBRM4 was used to assay nonreciprocal chromosome-to-plasmid recombination, pBRM4Yl to assay nonreciprocal plasmid-to-chromosome
recombination,
mosome
recombination
and pBRM5 (Table
to assay reciprocal
I). Experiments
were
described in Fig. 2 legend. (Top) Ratios of UV-induced recombinants to total surviving transformants in ABl157
plasmid-chroperformed
as
Lac- pBRM4 (recA + , lucY1;
n ), AB2470 (recB_, lucY1; A), JC9239 (recF_, lacY1; 0) and AB2463 (recA ~, lacy 1; +). (Middle) Ratios of UV-induced Lac - pBRM4Y 1 recombinants to total surviving transformants in RDP147 (recA + , lacy’ ;
n ), WDTlOO (recB_, lacy+ ; A) and WDTlOl (reck_ ; lacY+; 0). The inset is an expanded representation of the same data. (Bottom) Ratios of to total surviving transforUV-induced Lac + pBRM5 recombinants mants in AB 1157 (recA + , lacy1 ; n ), AB2470 (recB -, lacy1 ; A), JC9239 (recF_, lacY1; 0) and AB2463 (rec_4 _, lacY1; +). Error bars represent standard deviations experiments.
ofthe data from between two and seven independent
pBRM4Y 1) would not have contributed greatly to the frequencies of Lac + colonies detected. These Lac + colonies were subsequently confirmed by plasmid analysis to be mostly (about 80%) the result of reciprocal pBRM5 recombination in all cell strains.
135 The frequencies of Lac” colonies (mostly reciprocal recombinants) induced by UV irradiation of pBRM5 are
(d) Conclusions (1) The recA mutation
shown in Fig. 3. About 9% of the AB1157 transfo~~ts were Lac+ pBRM5 recombinants at 300 J/m”. pBRM5
UV-~~adiated plasmid survival, as has been previously observed. The reel3 and recF mutations also had minor effects on plasmid survival, in contrast to previous studies in which the same recF- host had no effect, and the recB- host gave very low plasmid survival (Roberts and Strike, 1981). The obvious difference between the two plasmid systems is that our plasmids were able to recombine with the bacterial chromosome. This suggests that in a recB- host, recFdependent plasmid-chromosome recombination was able to compensate for the loss of recB-mediated repair pathways previously described (Roberts and Strike, 1981). (2) The recF gene, but not the red gene, was required for the majority of the homologous nonreciprocal chromosome-to-plasmid recombination observed. Reciprocal recombination appeared to utilize both the recB and the recF genes. These results are consistent with UV-induced plasmid-chromosome recombination proceeding by a two-step mechanism. The initial step may be a recF-dependent, redindependent filling of UV-induced postreplication gaps with chromosomal DNA. This structure can either resolve to become a nonreciprocal chromosome-to-plasmid recombinant or it can proceed to the second step. The second step is moderately recB-dependent and results in the reciprocal exchange of plasmid-chromosome sequences, with the possible involvement of ds breaks.
(reciprocal) recombination was moderately dependent upon the recB gene and almost completely dependent upon the recF gene of E. coli. As before, recA was absolutely required for pBRM5 recomb~ation (Mudgett and Taylor, 1986). Plasmid isolation from Lac’ recombinants, and digestion with PstI showed that 128/162 of the Lac+ colonies contained plasmids which had acquired the ZacYl allele, indicative of reciprocal exchange (Table I). About half (85/162) of the pBRMS-derived recombinants contained mixtures of r~ombinant and nonre~omb~~t plasmid molecules (data not shown), as has been previously reported (Mudgett and Taylor, 1986). These results were independent of bacterial strain used. The existence of nonrecombinant molecules in the reciprocal recombinants was previously shown not to be due to Lac phenotype, the plasmid iucY gene, or segregative mechanisms (Mudgett and Taylor, 1986). Investigations are currently underway to determine the mechanistic origin of these nonrecombinant molecules. The results obtained in this study appear to be consistent with a two-step model proposed for c~omosom~ recombination (Mahan and Roth, 1989). This model predicts that homologous recombination begins as a recBCD-independent nonreciprocal exchange which does not involve ds breaks (Mahan and Roth, 1989). A late rec~CD-dependent step then utilizes ds breaks that are generated as intermediates in recombination and results in a reciprocal exchange (Mahan and Roth, 1989). This model is consistent with reports that ds break repair requires the action of RecBCD enzyme (Wang and Smith, 1983; Smith and Sharma, 1987; Mahan and Roth, 1989). As gap repair is reported to be recF-dependent (Wang and Smith, 1983; 1984; Smith and Sharma, 1987), UV-induced plasmid-chromosome recombination may therefore begin primarily as a recF-dependent, recB-independent nonreciprocal exchange which acts to ftil plasmid gaps with chromosomal DNA, but does not lead to the fo~ation of ds breaks. This exchange may be resolved as a nonreciprocal chromosome-to-plasmid recombination event, or it may be converted to a reciprocal chromosome-to-plasmid recombination event, or it may be converted to a reciprocal exchange during which ds breaks are generated as intermediates in recomb~ation, thus (moderately) requiring the RecBCD enzyme. Alternatively, nonreciprocal recombination (and its resolution) may not require RecBCD, but RecBCD may promote the resolution of reciprocal events (possibly through a Holliday structure) witbout ds break induction. The reason that so few nonreciprocal plasmidto-chromosome events arose from resolution of these proposed recombination intermediates is unknown at this time.
had only a moderate
effect on
We thank Ronald D. Porter for cell strains, helpful discussions and critical reading of this manuscript. This work was supported in part by NIH-NC1 grant CA44658 to W.D.T.
REFERENCES Abbott,
P.J.: Mutational
and recombinational
events in carcinogen-mo-
dified plasmid DNA. Mutation Res. 145 (1985a) 25-34. Abbott, P.J.: Stimulation of recombination between homologous
se-
quences on carcinogen-treated plasmid DNA and chromosomaf DNA by induction of the SOS response in Escherichia coli K- 12. Mol. Gen. Genet. 201 (1985b) 129-132. Chattoraj,
D.K., Cordes,
K., Berman,
M.L. and Das, A.: Mutagenesis
and mutation transfer induced by ultraviolet DNA. Gene 27 (1984) 213-222. Cohen,
A. and Laban
A.: Plasmidic
light in plasmid-cloned
recombination in Esche~~ch~ co& Moi. Gen. Genet. 189 (1983)