Proc. Natd. Acad. Sci. USA Vol. 89, pp. 5226-5230, June 1992 Biochemistry RecBCD enzyme is altered upon cutting DNA at a Chi recombination hotspot ANDREW F. TAYLOR AND GERALD R. SMITH Fred Hutchinson Cancer Research Center, 1124 Columbia Street, Seattle, WA 98104 Communicated by Hamilton 0. Smith, March 16, 1992

ABSTRACT During its unidirectional unwinding of DNA, to ensure even numbers of exchanges in conjugal and trans- RecBCD enzyme cuts one DNA strand near a properly oriented ductional crosses (13). At each end ofthe donor fragment one Chi site, a hotspot of homologous genetic recombination in RecBCD molecule is proposed to enter and promote just one Escherichia cohl. We report here that individual DNA molecules exchange. Such a mechanism would ensure exactly two containing two properly oriented Chi sites were cut with about exchanges and, hence, viability. 40% efficiency at one or the other Chi site but not detectably By what mechanism might RecBCD promote just one at both Chi sites. Furthermore, initial incubation of RecBCD exchange near each end ofthe donor fragment? This problem with Chi-containing DNA reduced its ability both to unwind is compounded by the high density of Chi sites in E. coli DNA and to cut at Chi sites on subsequently added DNA DNA: Chi occurs, on the average, once every 5 kilobases (kb) molecules much more than did initial incubation with Chi-free (18). A conjugational donor fragment one-quarter of the DNA; the activity was less severely affected. These chromosome long would contain about 250 Chi sites. Coor- results imply that RecBCD loses its Chi-cutting activity upon dination among the multitude of potential exchanges at these cutting at a single Chi site and provide a mechanism for sites seems difficult. Here we report that RecBCD, upon ensuring single genetic exchanges near the ends of DNA nicking DNA at a Chi site, loses the ability to nick DNA at molecules. a second Chi site. This observation provides a simple mech- anism for ensuring exactly two exchanges in conjugal and RecBCD (EC 3.1.11.5) is a multifunctional enzyme required transductional crosses. for by the major (RecBCD) path- way of (reviewed in refs. 1 and 2). The MATERIALS AND METHODS enzyme unwinds linear duplex DNA, from a flush or nearly Growth of Plasmids and Bacteriophages. Plasmids pBR322 flush duplex end (3), with the production of single-stranded x,0 pBR322 X+E224 (19), pBR322 y+F225 (19), pBR322 X+F (ss) DNA loops that enlarge at about 100 nucleotides (nt) per X+H (ref. 12; D. Dixon, personal communication), and sec as the enzyme travels along the DNA at about 300 nt per pBR322 y+E224 y+F225 (constructed by ligation of appro- sec (4, 5). The enzyme hydrolyses about two ATP molecules priate fragments from pBR322 y+E224 and pBR322 y+F225) per base pair unwound (6). When the enzyme encounters a were grown by chloramphenicol-induced amplification in E. Chi site, 5'-GCTGGTGG-3', it frequently nicks the Chi- coli N100 (galK recA thyA) and purified by alkaline lysis (20), containing strand about 5 nt to the 3' side of Chi (7, 8). followed by banding twice in cesium chloride/ethidium bro- Nicking at Chi occurs ifthe enzyme approaches Chi from the mide equilibrium density gradients. Bacteriophage A b2 cI857 right, as the sequence is written here, but not ifit approaches X+C151 susS7 (21) was grown by lytic infection of strain Chi from the left (8). DNA unwinding is postulated to JC8679 (F- A- thr-J leu-6 thi-) lacYI galK2 ara-14 xyl-5 continue after Chi cutting (as shown in Fig. 1), with 3'-ended proA2 his4 argE3 rpsL31 tsx-33 mtl-i recB21 recC22 sbcA23 ss DNA, bearing Chi near its end, being extruded as the supE44; ref. 22). Phage were purified (21) and DNA was enzyme continues to travel along DNA (9). This ss DNA isolated by phenol extraction and dialysis. "tail" has been proposed (9) to be a potent substrate for RecBCD Enzyme. RecBCD was purified and assayed as RecA protein, which, together with SSB (ss DNA-binding described (23, 24). The specific activity was 325,000 ds protein), forms joint molecules between ss DNA and homol- exonuclease units per mg ofprotein, which was measured by ogous double-stranded (ds) DNA (reviewed in ref. 10). The A280 and the molar extinction coefficient calculated for concomitant action of purified RecBCD enzyme, RecA pro- RecBCD (5). A conversion factor of 5.6 x 109 enzyme tein, and SSB produces joint molecules from linear ds DNA molecules per ds exonuclease unit was used, calculated from and circular, supercoiled ds DNA (11). A Chi site in the linear the specific activity noted above, from the enzyme's subunit ds DNA determines the apparent size of these joint mole- molecular weights deduced from the DNA sequence (25-27), cules: thejoint molecules are apparently larger the farther the and on the assumption that the active form of the enzyme is Chi site is from one end of the linear ds DNA (the end a heterotrimer. "downstream" or 5' of the Chi site) (12). DNA Substrates. (i) Mid-labeled two-Chi substrates. Plas- The RecBCD pathway is the principal pathway of recom- mid pBR322 and its X+E and X+E X+F derivatives were bination during conjugation and transduction of E. coli. In linearized with Sty I and then 32P-labeled at the 5' termini by either case, a linear duplex chromosomal fragment from the treatment with calf intestinal phosphatase followed by incu- donor recombines with a complete circular chromosome in bation with [y_32P]ATP (New England Nuclear; 3000 Ci/ the recipient cell. Consequently, two (or any even number of) mmol; 1 Ci = 37 GBq) and polynucleotide kinase. Following exchanges are required to produce a viable (complete, cir- thermal inactivation of the polynucleotide kinase, the labeled cular) recombinant chromosome. The model in Fig. 1, based were cut with EcoRI. After thermal inactivation of the on one initially proposed for recombination of vegetative EcoRI, the DNA was mixed with a 10-fold molar excess of an phage A via the RecBCD pathway (9), provides a mechanism EcoRI "Taylomere" (23), a synthetic oligonucleotide that generates an EcoRI site on base-pairing with itself. Addition of The publication costs ofthis article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" Abbreviations: nt, nucleotide(s); ss, single-stranded; ds, double- in accordance with 18 U.S.C. §1734 solely to indicate this fact. stranded; SSB, single-stranded DNA-binding protein from E. coli. 5226 Downloaded by guest on September 30, 2021 Biochemistry: Taylor and Smith Proc. Natl. Acad. Sci. USA 89 (1992) 5227

A Loop-tail Twin-loop Chi nick I :.: - i Q - - ('hi

5 6 -:= =. - -

- _ mU --~~MM Mm t m m D-loop Holliday junlctionl

B __ ~ Z ....m. -C *

3mm** , - _

FIG. 1. A model of conjugal recombination promoted by RecBCD enzyme and Chi sites (13). Hfr DNA enters the recipient cell as a single strand and is rapidly converted to duplex DNA (14). (A) RecBCD-mediated interaction between one end of the linear duplex Hfr DNA (thin lines) and part of the circular F- chromosomal DNA (thick lines) is shown. (Step 1) The enzyme (stippled box) attaches to the ds DNA end and travels along the DNA, unwinding the 3'-terminthted strand and releasing it at a slower rate, to form a loop-tail structure. (Step 2) Base pairing between the extruded tails produces a twin-loop structure (4). (Step 3) The enzyme nicks one strand of a correctly oriented Chi site, generating a 3' ss DNA tail, with Chi at its end. (Step 4) Continued travel of RecBCD elongates the 3-terminated tail and releases the loop on the other strand to produce a gap. (Step 5) The ss tail, aided by RecA and SSB, invades the homologous region of the circular F- chromosome to form a D-loop. (Step 6) Nicking of the D-loop (possibly RecBCD-mediated; 15, 16), followed by strand annealing and ligation of the nicked strands, produces a Holliday junction. (B) Result of this scheme occurring at both ends of the duplex Hfr fragment: the Hfr DNA linked to the F- DNA by two Holliday junctions. Appropriate resolution of the junctions (possibly by the recG or ruvC gene products; ref. 17), as shown by the open arrowheads, produces a recombinant with the Hfr DNA substituted for part of the F- chromosome. In an alternative mode of resolution, the 3' ends of the D-loops (A, step 5) prime replication, which proceeds around the chromosome to generate a dimeric chromosome containing two Holliday junctions, resolution of which generates one recombinant and one recipient-type chromosome (see ref. 13 for details). Recombination during transduction and transformation is proposed to occur by the same mechanism. T4 DNA ligase and ATP to the mixture resulted both in Quantitation by Autoradiography. Several timed expo- religation of the (now 32P-labeled) Sty I site and in ligation of sures, using preexposed film, were analyzed with the "Whole the Taylomeres onto the EcoRI sites at the ends of the DNA. Band Analysis" feature ofthe Visage 2000 System (Millipore) Inactivation of the DNA ligase, followed by restriction with video densitometer. For all visible bands, the background- Nde I, produced two self-complementary DNA molecules, fitting algorithm was used to estimate the intensity of the one 2300 base pairs (bp) and 32P-labeled, the other 2000 bp and band above the visible lane background. The integrated unlabeled. The two DNA species were separated from each intensities for the multiple exposures of each band were other and from DNA products in which one or more of the plotted against the fraction of the radioactivity that had ligation reactions had failed by preparative electrophoresis in decayed during the exposure, and the greatest slope observed an alkaline agarose gel. The 32P-labeled 2300-bp fragment was in the middle portion ofthe plot was taken as the best estimate located by autoradiography and recovered by extraction with of the radioactivity present in each band. Geneclean (Bio 101, La Jolla, CA). (ii) Bgl II fragment of phage A. A DNA was digested with BgI II, the cohesive ends RESULTS were annealed by incubation at 370C, and the desired 651-bp A DNA Substrate Bearing Two Chi Sites. We wished to test fragment (nt 38104-38754) was purified in a 4% polyacrylam- whether RecBCD was able to nick DNA at a second Chi site ide gel. The fragment was labeled at its 3' ends by incubation after it had nicked at a first Chi site. We constructed variants with Sequenase (United States Biochemical), [a-32P]dCTP of the substrate shown in Fig. 2, bearing no X sites, the X+E (New England Nuclear; 800 Ci/mmol), and unlabeled dATP, site, or both the X+E and the X+F site. The substrate bore a dGTP, and dTTP and was purified by phenol extraction and synthetic hairpin oligonucleotide (23) at its "left" end, to ethanol precipitation. (iii) Nde I digests ofpBR322. pBR322 X0 ensure that any RecBCD traveling along the DNA entered at and pBR322 X+F X+H DNA were digested with Nde I and the right end and hence encountered the Chi sites in their purified by phenol extraction and ethanol precipitation. Con- active orientation. centrations were estimated both by A260 and by fluorimetry, The substrates contained a 32p label between the Chi sites, with T7 DNA as a standard. Relative concentrations deter- to allow detection of a fragment nicked at both of the Chi mined by these two methods differed by <4%. sequences. Other fragments indicated nicking at either Chi Denaturing Agarose Gel Electrophoresis. Alkaline agarose site or at the hairpin end. The presence ofinternal 32P resulted gels (1% agarose in 50 mM NaOH/1 mM EDTA) were run at in two types of unusual radiation-induced decay products. room temperature with buffer recirculation. Gels were dried First, decay of one 32P atom breaks the DNA substrate into onto Whatman DE51 paper for autoradiography. The sizes of an unlabeled 930-nt strand and a 3670-nt strand that may reaction products were measured by reference both to the contain a 32P atom. This decay product, which accumulates markers shown in Fig. 2 and to a BstEII digest of A DNA (23) on storage of the substrate (A.F.T., unpublished data), is run in the same gel. All the DNA species, even those with visible in lanes 1, 3, and 5 of Fig. 2. Second, the shortest extensive self-complementary regions, migrated strictly as a labeled fragment that can be produced by a single random function of their molecular weight. nick per DNA molecule is one that occurs immediately to the Downloaded by guest on September 30, 2021 5228 Biochemistry: Taylor and Smith Proc. Natl. Acad Sci. USA 89 (1992) +E xzh ground smear was somewhat increased by RecBCD enzyme's apparently random Chi-independent nuclease activity (8). Reaction of mid-labeled X+E DNA (lane 4) produced two I7() t ) (111 prominent bands of approximately equal intensity, of about /.1 RI ."1'. 3300 and 1300 nt, the result of RecBCD nicking the DNA at X X+E (7). Cleavage both at X+E and at the hairpin produced no 1Rec X ;F unique labeled fragment (unlike the case with X+F discussed below), and so it is unclear what fraction of the cleavage at the hairpin (the 2300-nt band in lane 4) arose from enzymes '460()-- s | | | * e -sfl that had nicked at x+E. With the X+E X+F substrate the most prominent product after reaction of RecBCD was a band of -3800 nt (lane 6), the result of nicking at X+F. Note that both this reaction product nf *1)- _ and the unreacted substrate contain two 32P atoms, while all the other reaction products seen in Fig. 2 contain a single 32p atom. The other product of nicking at X+F was 800 nt but unlabeled. Faint bands of 3300 and 1300 nt, the result of nicking at X+E, are also present in lane 6, indicating that X+E was indeed present in the X+E X+F substrate. The full activity ofX+E in the X+E X+F DNA was confirmed by reaction with RecBCD after removal of the V+F site (see below). The 3300-nt fragment, which resulted from nicking at V+E, could arise with or without prior nicking by RecBCD at X+F. However, as the 3300-nt band in lane 6 is similar in intensity to the 1300-nt band, which can arise only from nicking at X+E without nicking at X+F, most of the 3300-nt fragment must have been generated by enzyme molecules that did not nick FIG. 2. Reaction of RecBCD with capped DNA molecules con- at X+F. taining two Chi sites. The diagram shows the structure of the capped The 2300-nt band in lane 6 results from RecBCD nicking at mid-labeled duplex DNA substrate. Each line represents a single the hairpin and again may arise with or without nicking at the DNA strand: the substrate is a self-complementary 2.3-kb duplex Chi sites. The band of 1500 nt seen in lane 6 is of a size molecule with a synthetic hairpin cap (23) on the left end. Asterisks consistent with that produced by an enzyme nicking at +F depict the positions of the single32P atoms, at the Sty I site, in each and at the hairpin. Experiments analogous to those in Fig. 2, strand. The substrate was prepared from a fragment of plasmid pBR322 (or its x+ derivatives), extending from the EcoRI site at nt but using substrates labeled at the EcoRI site at the left end, 0 to the Nde I site at nt 2300. RecBCD (1.5 units, 14 fmol) was confirmed the identity of the band (data not shown). The incubated with 4 fmol of capped DNA for 1 mn at 37cuin 20 mM presence of this product shows that RecBCD, after nicking MopsaKOH, pH 7.0/5 mM ATP/2 mM Mg(OAC)2/1 mM dithiothrei- DNA at X+F, can continue to travel through the DNA, hence tol. The products were separated in a denaturing agarose gel and passing X+E, and can still nick at the hairpin. analyzed by autoradiography. Bands are identified by their approx- No Detectable Nicking at X+E after Nicking at X+F. If imate sizes (nt) as follows: 4600, unreacted substrate; 3800, nick at RecBCD could nick the substrate at X+E after nicking it at x+F; 3670, nick at Sty I by 32P decay; 3300, nick at X+E; 2300, nick X+F, a 500-nt 32P-labeled fragment would be produced, but no at hairpin; 1500, nick at X+F and at hairpin; 1300,atenick XE. Size such band is visible in lane 6 ofFig. 2. The size markers in lane markers in lane 7, of the substrate with produced by cleavage a470-nt markerproduced by restriction enzymes that restriction enzymesPpuMi, EcoRI, Nrui, or PpuMtINru I, are 2880 7 include nt, 2300 nt (marker for nick at hairpin), 1300 nt (marker for nick at cleave very close to the X+E and X+F cut sites. It has icE), and 470 nt (marker for nicks at XfE and 20 m approximately the same size and sequence as the predicted RecBCD-generated fragment. The amount of this marker left of the 32p label (see Fig. 2): hence the background smear loaded corresponds approximately (based on previous exper- due to radiation-induced nicks (lanes 1, 3, and 5) ends iments) to the amount of X+E X+F fragment that would be abruptly at 1000 nt. produced if RecBCD could nick with high efficiency at both Nic3dng of the Substrate by RecBCD. Under the reaction Chi sites. No reaction product corresponding to nicking at conditions used, linear (uncapped) duplex substrates of the both Chi sites was visible, even after considerable overexpo- lengths used here were completely denatured by RecBCD sure of the gel (data not shown). As the enzyme must have (A.F.T., unpublished data; ref. 7); the enzyme evidently traveled past+Eafter nicking atx+F (as shown by the 1500-nt unwinds DNA in the x+F-to-hairpin fragment), we infer that nicking at one Chi "loop-tail" mode rather than the "twin- prevents nicking at a subsequent Chi on the same molecule. loop" mode (ref. 4; Fig. 1), leaving ss DNA behind it. RecBCD Quantitation of Chi Nicking. Measurement of the optical can neither initiate on DNA with ss tails (3) nor unwinding long densities ofthe bands in the autoradiograph of Fig. 2 allowed nick Chi on ss DNA Hence the examined at (8). products confirmation and extension of the above qualitative conclu- below must have resulted from interaction of a single RecBCD sions. As seen in Table 1, either Chi was nicked in about 40% molecule with each substrate DNA molecule, as the substrate of the molecules. Similar frequencies of Chi nicking were has only one -end available for RecBCD to enter. observed atV+Ein the X+EorX+EX+F substrates from which We first examined the action of RecBCD on capped sub- the X+F regions had been removed by digestion with Ava I, strates with no Chi sites. Reaction of mid-labeled x0 DNA showing that X+E in both the substrates used in Fig. 2 was produced a prominent 2300-nt fragment, resulting from cleav- fully active (data not shown). age of the substrate at the "hairpin" (Fig. 2, lane 2). As Ifefficient nicking at both Chi sites occurred, the predicted RecBCD enzyme cannot cut such structures from the outside intensity of the 500-nt fragment so produced would be the (ref. 23; A.F.T., unpublished data), it must have entered the product of the fractional nicking at the individual sites. right end of the molecule, traveled the length of the duplex Allowances must be made for the presence of only one 32p DNA, and then cleaved the hairpin from within. The back- atom in the 500-nt fragment and for the lower recovery of Downloaded by guest on September 30, 2021 Biochemistry: Taylor and Smith Proc. Natl. Acad. Sci. USA 89 (1992) 5229

Table 1. Quantitation of Chi nicking Chi Sites on Separate Molecules. The above experiment Relative band intensity clearly showed that nicking at a Chi site prevents interaction of RecBCD with a subsequent Chi site on the same DNA ick at N ick t +E Nick at Nick at molecule but could not address the question of whether the hairpin c X+F X+EandV+F effect is mediated via a change in the enzyme or a change in Substrate (2300 nt) 3300 nt 1300 nt (3800 nt) (500 nt) the topology of the substrate. To address this question, and X0 22 ND ND ND 0.4 to investigate whether other activities of RecBCD were X+E 19 23 19 ND 0.6 altered by Chi, experiments were conducted in which the two X+E X+F 11 ND 5 43 0.5 Chi sites were on separate DNA molecules. Relative intensities of gel bands (Fig. 2) are expressed as percent- RecBCD was first incubated with linearized pBR322 X+F ages ofthe intensity ofthe equivalent unreacted substrates. Separate X+H DNA (12) with actively oriented Chi sites near each end values are reported for the two fragments that result from a nick at of the DNA, or with linearized pBR322 DNA devoid of Chi X+E. For the hypothesized 500-nt fragments, the intensity of a sites (19). After these initial reactions, a sample was incu- rectangle of the size of the 470-nt marker band was measured at the corresponding location in each lane. Values for the corresponding bated with 3H-labeled T7 DNA to assay the ds exonuclease position in the unreacted control lanes (0.1-0.2%) were subtracted activity of RecBCD (24). The remainder of each reaction from the observed values. ND, not determined. mixture was incubated with a 32P-labeled Bgl II fragment of bacteriophage A DNA bearing X+C (21), and the products small fragments (-50%, based on recovery of the size were analyzed by nondenaturing polyacrylamide gel electro- markers). The predicted intensity of the 500-nt fragment is phoresis. After staining with ethidium bromide, fragments 5% ofthe input substrate. Measurements ofthat region ofthe corresponding to nicking at X+F or X+H on the X+F X+H gel (Table 1) revealed no Chi-dependent increase in intensity: pBR322 DNA were seen (data not shown), demonstrating the the intensity of the band must thus be -0.3% of the input efficient nicking at Chi sites under these reaction conditions. substrate. Thus, nicking at X+F reduced nicking at X+E to No specific bands were seen after incubation with X° pBR322 <6% of the predicted value. DNA, confirming the absence of Chi sites in that DNA. The The predicted frequency of DNA molecules nicked only at fraction of the 32p molecules that had been unwound, or that X+E, by enzyme molecules that passed X+F without nicking had been nicked at Chi, was measured by Phosphorimager it, is 24%, calculated as (fraction nicked at X+E alone) x (1 analysis of the separated 32P-labeled fragments (Fig. 3). - fraction nicked at V+F). The observed frequency, 10%o, Prior incubation with Chi-containing DNA caused a par- was calculated as twice the value of the 1300-nt band in the allel reduction in the unwinding and Chi-cutting activities of X+EX+F substrate, as that fragment contains only one 32p. It RecBCD on the second DNA, and did so at a much lower thus appears that the Chi-mediated alteration to RecBCD is concentration than did prior incubation with the X° DNA. not totally dependent on the enzyme cutting at the Chi site. Fifty percent reduction occurred with about 40 fmol of X+ Finally, the predicted frequency at which RecBCD would DNA (lane 3), compared with 160-200 fmol ofX° DNA (lanes nick at a hairpin, after nicking at X+F, is (fraction nicked at 11 and 12). This estimate of 40 fmol is consistent with the hairpin in X°) X (fraction nicked at X+F). After allowance for amount of RecBCD added (19 fmol) and the observation, the single 32p in the 1500-nt fragment, this was estimated to discussed above, that the maximum frequency of Chi cutting be 5% of the input DNA. The observed frequency of the in brief(1-min) reactions is about 40%. Complete inactivation fragment, 2%, shows that the ability of the enzyme to travel of RecBCD presumably requires multiple passes through the along DNA and to nick at the hairpin is not abolished by its DNA. Three passes of each enzyme molecule through DNA having already nicked at a Chi site. would require 60 fmol of DNA and would predict (1-0.4)3 = Unlabelled X+F XCi ZX %+F X+ I

-= 0.8 . 086 Unwoundx mi - 'r Substrate f

-

c 0.4 .

Substrate W_S^ m awI " 0.2 vxX F X+I

Chi A.ss Fragment 40 80 120 160 200 fmol unlabelled DNA 2 4 6 8 10 12 14 16 18 20 Chi cuttiney:A Unwinding:o Exonuclease:m

FIG. 3. Inhibition of RecBCD by incubation with Chi-containing DNA. (Left) RecBCD (2 units, 19 fmol) was incubated for 15 min at 20°C with the indicated amounts of unlabeled pBR322 DNA (either X° or X+F X+H) linearized with Nde I in 20 Al of 20 mM Mops-KOH, pH 7.0/5 mM ATP/0.2 mM EDTA/3 mM Mg(OAc)2/1 mM dithiothreitol/0.01% acetylated bovine serum albumin/6.5 AM SSB. One microliter was then removed for ds DNA exonuclease assay (in 50 A.l of nuclease reaction mix without ATP to produce 0.1 mM ATP for the assay; 24). Twenty fmol of 3' end-labeled Bgl II V+C fragment of A DNA was then added, and incubation was continued for a further minute (except for lane 18, in which the second incubation was for 20 min). Reactions were stopped by addition of EDTA (16 mM) and SDS (0.8%) and run in a nondenaturing polyacrylamide gel. The samples in lanes 13-15 and 20 were denatured at 90°C before loading. Lanes 19 and 20 contain unreacted ds and ss [32P]DNA substrate. In lane 16 the reaction with unlabeled DNA was terminated by heating to 65°C for 5 min. The sample was then returned to 20°C, 2 units of RecBCD and the [32P]DNA were added, and the mixture was allowed to react for 1 min. In lane 17, additional RecBCD (2 units) was added immediately before the [32P]DNA. Samples were analyzed in a nondenaturing, 4% polyacrylamide/0.5% agarose gel in E buffer (20). The gel was dried onto Whatman DE51 paper and analyzed on a Molecular Dynamics (Sunnyvale, CA) Phosphorimager system. Backgrounds for the measured bands were the means of similar-sized rectangles immediately above and below the bands of interest. (Right) Data are plotted as the fraction of the maximum values observed for the reactions in lanes 1-12. The maximum value observed (as percent of the sum of the ds, ss, and Chi-cut bands) was in lane 6: 78% unwound, 5% Chi-cut band. This sample gave 15% degradation of the T7 [3H]DNA in the exonuclease assay. Solid lines indicate first incubation with unlabeled X° DNA; dashed lines indicate first incubation with X+FX+H DNA. Downloaded by guest on September 30, 2021 5230 Biochemistry: Taylor and Smith Proc. Natl. Acad Sci. USA 89 (1992) 20% remaining Chi-cutting or unwinding activity: the ob- cellular ds exonuclease activity (32, 33), while purified served value is about 10%. The inhibition of RecBCD activ- "RecBC" enzyme, apparently devoid ofRecD, lacks both ds ities by higher concentrations ofX° DNA may well be due to and ss exonuclease activity (34). The dissociation ofRecD at the presence of Chi-like sequences on that DNA (28, 29). Chi, hypothesized to explain the action of Chi (35), is thus The ds exonuclease activity was inhibited by prior reaction unlikely to be responsible for the changes, reported here, in with Chi-containing DNA, but not as severely as were the RecBCD activity upon Chi cleavage. Alternative possibilities unwinding and Chi-cutting activities. Sixty femtomoles ofX+ include a covalent modification, such as phosphorylation, or DNA reduced Chi-cutting to 6% and unwinding to 12% of a conformational change of the enzyme. their maximal values, but only reduced the ds exonuclease to 40% of its maximal value. In similar reactions (data not We thank Dan Dixon and Steve Kowalczykowski for gifts of SSB shown) the enzyme retained 31% of its ds exonuclease and the pBR322 plasmids containing %+H. We thank our colleagues activity and 35% of its ss exonuclease activity, even after for helpful comments on the manuscript. This work was supported reaction with 200 fmol of x+ DNA. This retention of exonu- by research grants from the National Institutes of Health. clease activity allows us to infer that the enzyme has suffered a qualitative change, rather than total inactivation. 1. Taylor, A. F. (1988) in Genetic Recombination, eds. Kucherlapati, R. & Smith, G. R. (Am. Soc. for Microbiol., Washington), pp. Inhibition by reaction products was also ruled out by the 231-263. results in lanes 16 and 17: freshly added RecBCD was fully 2. Smith, G. R. (1990) in Nucleic Acids and Molecular Biology, eds. able to unwind and cut at Chi on the 32P-labeled DNA in the Eckstein, F. & Lilley, D. M. J. (Springer, Berlin), Vol. 4, pp. 78-98. presence of the products of the first incubation. The reaction 3. Taylor, A. F. & Smith, G. R. (1985) J. Mol. Biol. 185, 431-443. products in lanes 13-15, in which duplicates oflanes 1, 3, and 4. Taylor, A. & Smith, G. R. (1980) Cell 22, 447-457. 5 were heat-denatured before electrophoresis, ruled out the 5. Roman, L. J. & Kowalczykowski, S. C. (1989) Biochemistry 28, 2863-2873. possibility that the failure to observe the Chi-cut fragment 6. Roman, L. J. & Kowalczykowski, S. C. (1989) Biochemistry 28, was due to its being produced but not released. 2873-2881. We conclude that prior incubation with Chi-containing 7. Ponticelli, A. S., Schultz, D. W., Taylor, A. F. & Smith, G. R. DNA resulted in loss ofthe enzyme's ability to either unwind (1985) Cell 41, 145-151. or cut at Chi on a separate DNA molecule, as well as a 8. Taylor, A. F., Schultz, D. W., Ponticelli, A. S. & Smith, G. R. reduction in its ds exonuclease (1985) Cell 41, 153-163. activity. 9. Smith, G. R., Schultz, D. W., Taylor, A. F. & Triman, K. (1981) Stadler Genet. Symp. 13, 25-37. DISCUSSION 10. Radding, C. M. (1988) in Genetic Recombination, eds. Kucherla- pati, R. & Smith, G. R. (Am. Soc. for Microbiol., Washington), pp. We report here that RecBCD does not detectably cut at two 193-229. Chi sites on one DNA molecule. We also report that RecBCD 11. Roman, L. J., Dixon, D. A. & Kowalczykowski, S. C. (1991) Proc. to at Natl. Acad. Sci. USA 88, 3367-3371. appears be qualitatively changed upon cutting Chi. 12. Dixon, D. A. & Kowalczykowski, S. C. (1991) Cell 66, 361-371. The first observation provides a basis for ensuring that 13. Smith, G. R. (1991) Cell 64, 19-27. exactly two exchanges occur between the recipient chromo- 14. Ippen-Ihler, K. A. & Minkley, E. G., Jr. (1986) Annu. Rev. Genet. some and a linear donor fragment contributed during conju- 20, 593-624. gation or transduction of E. coli (Fig. 1). During conjugation 15. Wiegand, R. C., Beattie, K. L., Holloman, W. K. & Radding, a of the donor chromosome is injected into the C. M. (1977) J. Mol. Biol. 116, 805-824. portion (Hfr) 16. Williams, J. G., Shibata, T. & Radding, C. M. (1981) J. Biol. Chem. recipient (F-) cell; this DNA is initially ss but is converted to 256, 7573-7582. ds DNA by complementary strand synthesis (reviewed in ref. 17. Lloyd, R. G. (1991) J. Bacteriol. 173, 5414-5418. 14). During transduction mediated by phage P1, a linear ds 18. Faulds, D., Dower, N., Stahl, M. & Stahl, F. (1979) J. Mol. Biol. DNA fragment of the donor chromosome, encapsidated in a 131, 681-695. P1 coat, is injected into the recipient cell. Conjugational 19. Smith, G. R., Kunes, S. M., Schultz, D. W., Taylor, A. & Triman, have an distribution of with a K. L. (1981) Cell 24, 429-436. fragments exponential lengths, 20. Maniatis, T., Fritsch, E. F. & Sambrook, J. (1982) Molecular number average of about 500kb (1/10th ofthe chromosome), Cloning: A Laboratory Manual (Cold Spring Harbor Lab., Cold whereas transductional fragments are uniformly about 100 kb Spring Harbor, NY). long. These fragments would contain, on the average, about 21. Sprague, K. U., Faulds, D. H. & Smith, G. R. (1978) Proc. Natl. 100 or 20 Chi sites, respectively. Acad. Sci. USA 75, 6182-6186. The pattern of inheritance of genetic markers from a 22. Gillen, J. R., Willis, D. K. & Clark, A. J. (1981) J. Bacteriol. 145, or transductional in 521-532. conjugal donor suggests that, the major- 23. Taylor, A. F. & Smith, G. R. (1990) J. Mol. Biol. 211, 117-134. ity of the cases, the linear donor DNA is integrated by two 24. Eichler, D. C. & Lehman, I. R. (1977) J. Biol. Chem. 252, 499-503. exchanges, one near each end of the donor fragment (13). We 25. Finch, P. W., Wilson, R. E., Brown, K., Hickson, I. D., Thomp- propose that these exchanges occur at the first (or with kinson, A. E. & Emmerson, P. T. (1986) Nucleic Acids Res. 14, decreasing probability the second, or third, etc.) properly 4437-4451. oriented Chi sites encountered RecBCD its travel 26. Finch, P. W., Wilson, R. E., Brown, K., Hickson, I. D. & Em- by during merson, P. T. (1986) Nucleic Acids Res. 14, 8573-8582. from each end of the donor fragment. Chi-cutting activity, 27. Finch, P. W., Storey, A., Brown, K., Hickson, I. D. & Emmerson, which is essential for Chi's stimulation of recombination (2), P. T. (1986) Nucleic Acids Res. 14, 8583-8594. is lost upon cutting at one Chi site; this loss would ensure that 28. Cheng, K. C. & Smith, G. R. (1984) J. Mol. Biol. 180, 371-377. exactly one exchange occurs near each end of the donor 29. Cheng, K. C. & Smith, G. R. (1987) J. Mol. Biol. 194, 747-750. fragment. These exchanges would occur within the terminal 30. Yagil, E. & Shtromas, I. (1985) Genet. Res. 45, 1-8. 1% to 10% of the donor Previous 31. Stahl, F. W., Thomason, L. C., Siddiqi, I. & Stahl, M. M. (1990) fragments. reports (30, 31) Genetics 126, 519-553. indicate that a Chi site reduces the hotspot activity ofanother 32. Chaudhury, A. M. & Smith, G. R. (1984) Proc. Natl. Acad. Sci. Chi site "downstream" (to its "left") of the first Chi site in USA 81, 7850-7854. phage A. Our results provide an enzymatic basis for this 33. Amundsen, S. K., Taylor, A. F., Chaudhury, A. M. & Smith, G. R. effect. (1986) Proc. Natl. Acad. Sci. USA 83, 5558-5562. Upon cutting at Chi, RecBCD loses the ability to either 34. Palas, K. M. & Kushner, S. R. (1990) J. Biol. Chem. 265, 3447- unwind or cut at Chi upon subsequently added DNA. The 3454. 35. Thaler, D. S., Sampson, E., Siddiqi, I., Rosenberg, S. M., Stahl, enzyme is changed but not totally inactivated upon cutting at F. W. & Stahl, M. (1988) in Mechanisms and Consequences ofDNA Chi: it retains much ofits ds and ss exonuclease activity. recD Damage Processing, eds. Friedberg, E. & Hanawalt, P. (Liss, New nonsense mutations eliminate the RecD subunit and intra- York), pp. 413-422. Downloaded by guest on September 30, 2021