Vol. 173, No. 16

Cofactor Requirements of BamHI Mutant EndonucleaseE77K and Its Suppressor Mutants

SHUANG-YONG XU AND IRA SCHILDKRAUT*

New England Biolabs Inc., 32 Tozer Road, Beverly, Massachusetts 01915

Received 16 January 1991/Accepted 13 June 1991

A mutant BamHI endonuclease, E77K, belongs to a class of catalytic mutants that bind DNA efficiently butcleave DNA at a rate more than 103-fold lower than that of the wild-type enzyme (S. Y. Xu and I. Schildkraut,J. Biol. Chem. 266:4425-4429, 1991). The preferred cofactor for the wild-type BamHI is Mg2+. BamHI is10-fold less active with Mn2+ as the cofactor. In contrast, the E77K variant displays an increased activity whenMn2+ is substituted for Mg2, in the reaction buffer. Mutations that partially suppress the E77K mutation wereisolated by using an Escherirhia coli indicator strain containing the dinD::lacZ fusion. These pseudorevertantendonucleases induce E. coli SOS response (as evidenced by blue colony formation) and thus presumably nickor cleave chromosomal DNA in vivo. Consistent with the in vivo result, the pseudorevertant endonucleases inthe crude cell extract display site-specific partial DNA cleavage activity. DNA sequencing revealed two uniquesuppressing mutations that were 'located within two amino acid residues of the original mutation. Bothpseudorevertant proteins were purified and shown to increase specific activity at least 50-fold. Like thewild-type enzyme, both pseudorevertant endonucleases prefer Mg2Y as the cofactor. Thus, the second-sitemutation not only restores partial cleavage activity but also suppresses the metai preference a$ well. Theseresults suggest that the Glu-77 residue may play a role in metal ion binding or in enzyme activation (allosterictransition) following sequence-specific recognition.

The BamHI restriction endonuclease purified from Bacil-lus amyloliquefaciens H cleaves the symmetric DNA se-quence 5'-GGATCC-3' between the two guanines on bothstrands (reviewed in reference 17). The coding sequence forBamHI has been cloned and sequenced, and the enzyme hasbeen overexpressed in Escherichia coli (2, 3, 10). Like alltype II restriction endonucleases, BamHI requires the diva-lent cation Mg2' as a cofactor for cleavage activity. Mn2+can be substituted for Mg2+ in the reaction, but the enzymeactivity is reduced (24). Mn2+ also decreases the enzymespecificity, allowing BamHI to cleave noncanonical se-quences (BamHI star sites; 10, 14). The DNA-binding activ-ity, however, is independent of metal cofactor (17, 28).The SOS regulon consisting of about 20 genes in E. coli is

negatively regulated by the LexA repressor (reviewed inreferences 19, 22, and 25). The repressor can be cleaved andinactivated by the RecA protein to derepress the genes in theregulon. The proteolytic activity of the RecA protein isactivated by a signal (e.g., single-stranded DNA or oligonu-cleotides) upon DNA damage or inhibition of DNA replica-tion. As the result of DNA damage or DNA replicationinterference, the genes in the SOS regulon are expressed atan increased level. E. coli indicator strain AP1-200 containsa dinD: :lacZ fusion in which the lactose operon is fused to aDNA damage-inducible promoter (12). --Galactosidaseexpression is increased upon treatment of the cells with UV,mitomycin C, or other DNA-damaging agents (5-8, 12, 21).It was shown that EcoRI temperature-sensitive mutants or

EcoRI mutants with partial cleavage activity induce the SOSresponse by causing damage to the E. coli chromosome (5, 7,8). Furthermore, EcoRI mutants with relaxed specificityhave been isolated by screening for SOS-inducible bluephenotype in the presence of EcoRI methylase (6).

Previously, we isolated three mutant BamHI endonu-

* Corresponding author.

cleases (E77K, D94N, and E113K) that bind to DNA butcleave at less than 0.1% of the wild-type rate (28). In thisreport, we demonstrate that the purified E77K endonucleasediffers in cofactor preference from the wild-type enzyme. Byusing a dinD: :lacZ indicator strain, we isolated two suppres-sor mutations from the E77K mutant. We determined theeffects of Mg2+ and Mn2+ on the activities of the pseudo-revertant endonucleases.

MATERIALS AND METHODS

Bacterial strains, media, and reagents. E. coli AP1-200[endAl thi-J supE44 hsdRJ7 mcrB251 mrr-253 lacZ::TnJOdinD1::Mu d11734 (Kanr lacZ+)IF' lacIq lacZ::TnS] wasobtained from A. Piekarowicz (via E. Raleigh; 21). ADK21 isan E. coli K802 (26) derivative carrying a lambda prophage(A imm434 ind bamhIM+) that constitutively expresses theBamHI methylase (11). Plasmid pADE15 (bamHIM+ KanrCamr), a pACYC184 derivative, contains the gene coding forthe BamHI methylase (11). E. coli ER1755 (mutD5 kdgKSJxyl-5 mtl-i argE3 thi-J thr-J ara-14 leuB6 lacYl tsx-33supE44 galK2 hisG4 rfb-i mgl-51 rpsL3i/F' lacIq L8 pro)was from E. Raleigh (27). Luria-Bertani (LB) medium, LBagar, and SOB medium (20 g of tryptone, 5 g of yeast extract,0.6 g of NaCl, 0.2 g of KCl, 10 mM MgCl2, and 10 mMMgSO4 in 1 liter) were as described previously (15, 16).Media were supplemented with ampicillin at 100 ,ug/ml,kanamycin at 50 ,ugIml, and chloramphenicol at 30 ,ug/mlwhere needed. 5-Bromo-4-chloro-3-indolyl-3-D-galactopyra-noside (X-Gal) was added at the concentration of 40 ,ug/ml.[ot-35SdATP was from New England Nuclear. Restrictionenzymes, T4 DNA ligase, DNA size markers, Klenowfragment of E. coli DNA polymerase I, and the DNAsequencing kit were from New England Biolabs. The proteinsize marker was from Bethesda Research Laboratories.Sodium dodecyl sulfate-polyacrylamide gradient gels (10 to20%) were from Integrated Separation Systems. Homoge-

5030

JOURNAL OF BACTERIOLOGY, Aug. 1991, p. 5030-50350021-9193/91/165030-06$02.00/0Copyright ©) 1991, American Society for Microbiology

on May 6, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

COFACTOR REQUIREMENTS OF BamHI MUTANT E77K 5031

neous BamHI endonuclease was kindly provided by L.Dorner (New England Biolabs). The specific activity of theenzyme is 106 U/mg.

Plasmid DNA preparation, plasmid construction, DNA se-quencing, and transformation. Plasmid DNA was preparedby Qiagen midi columns according to manufacturer's proto-col and used as templates for dideoxy termination sequenc-ing (18, 23). Plasmid minipreparations were made by aboiling method (9). BamHI methylase-modified pBR322DNA (in vivo BamHI site methylation) was prepared froman overnight culture of ADK21 (A imm434 ind bamhIM+)(pBR322) by the Qiagen midi method described above. ApBR322 variant lacking BamHI site (pBR322-BamH-) wasconstructed by cleavage of pBR322 with BamHI endonucle-ase. The ends were filled in with the Klenow fragment of E.coli DNA polymerase I and ligated with T4 DNA ligase, andDNA was transformed into E. coli. E. coli cells were madecompetent by growth in SOB medium and CaCl2 treatment.The cells were transformed by a standard procedure (15).

Mutagenesis. Plasmids were mutagenized by passagethrough a mutD E. coli strain (6). Plasmid pAEK14-E77K,carrying a G-to-A transition at codon 77 in the bamhIR gene,was transformed into ER1755 (mutD5). All Ampr transfor-mants on the plate were collected and pooled, and plasmidDNA was prepared by a boiling method (9). The mutagen-ized DNA was subsequently transformed into AP1-200 andplated on LB agar supplemented with ampicillin and X-Gal.Light to medium blue colonies containing a putative muta-tion(s) in the bamhIR gene were checked for plasmid-linkedblue phenotypes by isolating and reintroducing the plasmidsinto AP1-200. They were further screened for DNA cleavageactivity in crude cell extracts. The revertant plasmids werealso checked for gross DNA rearrangement by EcoRI,HaeII, HindlIl restriction digestion.

Deletion of the bamhlR gene. To map the suppressingmutations, a restriction fragment containing the entirebamhIR gene was deleted by a double NcoI and HindIllrestriction digest. The DNA ends of the vector were filled inwith Klenow fragment and religated with T4 DNA ligase at alow DNA concentration. The ligated DNA was transformedinto AP1-200 indicator cells and plated on X-Gal and ampi-cillin plates.

Preparation of crude cell extract and enzyme purification.Cells containing pAEK14-bamhIR+ or pAEK14-E77K orrevertant plasmids were grown at 37°C to 2 x 108 cells per mlin LB media. Isopropylthiogalactoside (IPTG) was added to0.4 mM to induce endonuclease production, and incubationwas continued for 2 h. Induced cells were chilled on ice,centrifuged, and resuspended in 0.5 ml of sonication buffer(10 mM Tris-HCl [pH 7.8], 2 mM EDTA, 10 mM P-mercap-toethanol). Cell lysis was completed by addition of lysozyme(12.5 F.g/ml) and sonication. Cell debris was removed bycentrifugation, and the supernatant was assayed for DNAcleavage activity. For enzyme purification, the procedurewas scaled up to 2 liters of cell culture and the protocol ofJack et al. (10) was used.

P-Galactosidase activity assay. Overnight cell cultureswere assayed for ,3-galactosidase activity as described pre-viously (16).

RESULTS

Enzyme purification. The wild-type BamHI purificationprocedure (10) was followed to purify the E77K endonucle-ase. It was purified to >95%, as judged by Coomassie bluestaining and densitometric scanning of the gel (Fig. 1, lane

r w- r- 0).- N- Nl- N-_ tLii:CCwx LoJ

Kds200

97

_ 68

m 43

29

_ 1814

1 2 3 4 5

FIG. 1. Purified wild-type BamHl and mutant endonucleases.The standard purification procedure included four chromatographicsteps (phosphocellulose, hydroxylapatite, heparin-Sepharose, andQ-Sepharose columns; 10). One microgram of each protein wasloaded on the gel. The molecular mass ofBamHI is 25,000 Da. Kds,kilodaltons.

2). The purified enzyme binds DNA in the gel retardationassay as described previously (28).

Cofactor requirement of the wild-type and E77K endonu-cleases in DNA cleavage reactions. Because the E77K endo-nuclease weakly induces the SOS response in the dinD: :lacZfusion strain (28), we tested the residual cleavage activity ofthe purified enzyme. E77K had nondetectable cleavageactivity in Mg2+ buffer (Fig. 2B, lane 1 on k DNA; Fig. 2C,lane 1 on pBR322 DNA). However, E77K displayed anincreased activity in Mn2+ reaction buffer. A weak partialdigest pattern was observed on k DNA (Fig. 2B, lane 4), andopen circular pBR322 DNA was accumulated (Fig. 2C, lane3). A faint linear pBR322 DNA was also observed in theoriginal gel picture (Fig. 2C, lane 3), but it did not reproducewell in the figure. Thus, it seems that Mn2+ has a stimulativeeffect on the E77K endonuclease activity. The specificactivity of E77K was estimated to be <102 U/mg of proteinin Mn2+ buffer.

Wild-type BamHI requires Mg2+ as a cofator for cleavageactivity. Mn2+ could substitute for Mg2+ in the reaction, butactivity was reduced. Figure 2A shows that 2 U of BamHIwas required to completely cleave 1 ,ug of A DNA in Mg2+buffer (lane 1). However, 20 U ofBamHI was needed to givethe same result in Mn2+ buffer (lane 8). Thus, the activity ofBamHI was reduced 10-fold in the presence of Mn2+.Reduced BamHI activity in Mn2+ buffer has been reported(24).To determine whether E77K cleaves at the normal BamHI

recognition site, pBR322 DNA was first incubated withE77K overnight in a Mn2'-containing buffer to generate asmall amount of linear DNA, and a second cut was intro-duced by EcoRI digestion. Figure 2D shows that a fragmentthus released comigrated with the fragment (377 bp) gener-ated by wild-type BamHI and EcoRI cleavage, indicatingthat E77K introduced a double-strand break which maps tothe BamHI recognition site. When the BamHI site onpBR322 is modified by BamHI methylase or removed(pBR322-BamH1-), the DNA was no longer sensitive toE77K single strand nicking or double-strand cleavage (data

VOL. 173, 1991

on May 6, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

5032 XU AND SCHILDKRAUT

O 0C\ ;-0)c\oC\ %ir ac C\

~~~~IT

B.LJC0)

LLJ O-

N- N-N--±.L cruJ.L

C.

LLJtL2CL

_ CR(D -

N-N-N-L

D.a)

c\J

y Ir !S I: r E Iy- E - E cr v E rN-c-r 0 m r 0u mL mQJ InMD L

oc -)>

cc -->

bp6030

3104

1 2 34 5 6 78

Mg2+ Mn2+

1 2 3 4 5 6

Mg2+ Mn2+

1 2 3 4 51 2345

Mg2+ Mn2+

FIG. 2. DNA cleavage assays for wild-type BamHI, E77K mutant, and E77K revertants in Mg2+ or Mn24 buffer. One unit of BamHI isdefined as the amount of enzyme producing complete digestion of 1 pug of A DNA at 37°C for 1 h in a volume of 50 RI in a standard buffer (100mM NaCl, 10 mM Tris-HCl [pH 7.8], 6 mM MgCl2, 1 mM dithiothreitol). For comparison of DNA cleavage activities in Mg2' and Mn24buffers, the buffer pH of Tris-HCI was 7.0, because MnCl2 forms gelatinous Mn(OH)2 at pH 7.8, which rapidly darkens in the air as a resultof oxidation. (A) Cleavage of X DNA by BamHI in the presence of Mg24 or Mn24. The amount of enzyme is indicated above each lane. (B)Cleavage of A DNA by E77K and two pseudorevertants, R76K-E77K and E77K-P79T, in the presence of Mg24 or Mn24. A 100-ng sampleof purified endonuclease was used to cleave 1 ,ug of A DNA at 37°C for 60 min in each reaction. (C) Cleavage of pBR322 DNA by BamHI (10ng) and E77K (100 ng) in the presence of Mg24 or Mn24. CC, closed circular DNA; L, linear DNA; OC, open circular DNA. (D) Doubledigestion of pBR322 DNA by E77K and EcoRI (lane 3). The pBR322 DNA was first incubated with E77K (200 ng) in Mn24 buffer for 15 h,and the DNA was extracted once with phenol-CHCl3 and once with CHCl3 and then ethanol precipitated. A second cut was introduced byEcoRI. The double digestion of pBR322 DNA by BamHI and EcoRI (lane 2) generated two fragments, 377 and 3,986 bp. The size marker isHaeIII-cleaved 4X-174 DNA (New England Biolabs). (E) HindIII and BamHI digestion of A DNA. The HindIll-cleaved X DNA was cleavedwith BamHI (10 ng) or pseudorevertant endonucleases (100 ng) in a standard buffer (see above) at 37°C for 60 min. The largest HindIIIfragment (23,130 bp) was partially cleaved by the pseudorevertant endonucleases (lanes 2 and 3).

not shown). The low level of SOS induction by E77K in vivo(faint blue) was blocked by the introduction of a compatibleplasmid (pADE15-bamhIM+) expressing the BamHI meth-ylase. Therefore, cleavage in vitro and SOS induction byE77K require the presence of the 5'-GGATCC-3' sequence.

Isolation of suppressing mutants from E77K variant. Be-cause second-site mutations can assist in defining criticalamino acids, we screened for mutations that could reversethe phenotype of an original mutation. We chose the E77Kvariant to screen because of its altered cofactor requirement.Our screening method to isolate suppressing mutations isbased on a dinD::lacZ fusion indicator strain (8, 12). TheSOS response and dinDl expression are increased by endo-nuclease activity (5-8, 20). A BamHI-overproducing plas-mid, pAEK14-bamhIR+ (in which the bamHIR gene isunder Ptac and lacI control) is lethal to AP1-200 cells(dinD: :lacZ indicator strain) in the absence of BamHI meth-ylase as a result of the significant level expression ofbamHIR gene from Ptac under noninducing conditions (10,28). However, plasmid pAEK14-E77K carries a G-to-Atransition at codon 77 in the bamHIR gene. This mutationallows the plasmid to transform AP1-200 cells in the absenceof its cognate methylase gene (28). The transformants appearfaint blue on an X-Gal plate, which is consistent with thelow-level cleavage activity of the E77K endonuclease invitro.

We mutagenized pAEK14-E77K plasmid DNA by passagethrough a mutD strain. To identify suppressing mutations,the plasmid DNA was prepared from 106 transformants ofthe mutD strain and transformed into AP1-200 cells in theabsence of BamHI methylase. About 105 transformants ofthe AP1-200 cells were screened on ampicillin and X-Galplates. This screening method would preclude wild-typerevertants or high-level-activity revertants. Plasmid DNAwas prepared from 14 medium blue transformants and re-transformed into AP1-200 cells. In all cases, the blue phe-notype was plasmid linked. Isolates that gave small darkblue colonies on indicator plates did not grow in liquidculture, presumably as a result of high-activity reversionwhich led to the lethal DNA damage. For stable mainte-nance, the plasmids were transformed into ADK21 cellsexpressing the BamHI methylase. Crude cell extracts wereprepared from all 14 clones and tested for DNA cleavageassays. Extracts from 5 of the 14 clones partially cleavedDNA in a site-specific manner (data not shown). Under thesame condition, E77K resulted in no detectable cleavage andwild-type BamHI gave complete cleavage (data not shown).No DNA rearrangements were found by HaeII restrictiondigestion of these five revertant plasmids (data not shown).The remaining nine clones contained nondetectable BamHIcleavage activity. These plasmids were shown by restrictionmapping and DNA sequencing to contain an IS5 insertion in

A.

c

I.-

E0co1

c} -oc c

-NIN-E. -F T-

+-UJ CLE0 N-~SNr- r-.-

mn rr i, -r

bp<-377

lanes

cof actor

1 2 32+ 2+Mg Mn

1 234

J. BACTERIOL.

on May 6, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

COFACTOR REQUIREMENTS OF BamHl MUTANT E77K 5033

76 77 78 79

Wild type A G A G A A AAA C C C (codon)R E K P (aa)

E77K A G A A A A A A A C C CR L K P

R76K-E77K A & A AAA AAA C C Cxi 2 K P

E77K-P79T A G A A A A AA C CR K Z

FIG. 3. Coding sequences (codons 76 to 79) and predicted aminoacids for the wild-type and mutant endonucleases. The mutatedbases and predicted amino acid (aa) substitutions are underlined.E77K indicates an E-to-K substitution at residue 77 in the BamHIprotein sequence. Abbreviations for amino acids: E, Glu; K, Lys; P,Pro; R, Arg; T, Thr.

the bamhlR gene (data not shown). It is known that overex-pression of the ISSa gene (coding for a putative transposase)in a multicopy plasmid downstream of a PlaC promotorinduces the SOS response (29).

Deleting the bamhlR gene for the five revertants. An NcoI-HindlIl restriction fragment (1,497 bp) containing the entirebamhIR gene was deleted from the five revertant plasmids(data not shown). The resulting plasmids no longer conferreda Lac' phenotype to AP1-200. Thus, the SOS response iscaused by the mutant bamhIR gene product.

Determination of the codon change for the five revertants.The entire bamHIR gene (642 bp) was sequenced for fiveclones that displayed partial cleavage activity. There areonly two unique mutations among the five clones (twooccurred in codon 76; three occurred in codon 79). Both ofthem are proximal to the original mutation (Fig. 3). The twopseudorevertants are named R76K-E77K and E77K-P79Taccording to the position of the predicted amino acid substi-tutions. Under noninduced conditions, the pseudorevertantendonucleases reduced AP1-200 colony size and resulted ina medium blue phenotype on indicator plates. In IPTG-induced liquid culture, the cells lysed, presumably becauseof the increased DNA damage caused by the pseudorever-tant endonucleases.

I-Galactosidase activity assays. The pseudorevertantswere found to form medium blue colonies on X-Gal plates.To quantitatively measure the level of SOS induction, weassayed the 3-galactosidase activity induced by the mutantendonucleases in liquid cultures. The results are summarizedin Table 1. The pseudorevertant endonucleases increasedthe 3-galactosidase level approximately fourfold. Another

TABLE 1. ,B-Galactosidase levels induced by mutantBamHI endonuclease

Strain 3-Galactosidase activity(Miller units)"

AP1-200 ................................... 1.4AP1-200(pBR322) .................................... 1.5AP1-200(pAEK-E77K) .............................. 2.0AP1-200(pAEK-R76K-E77K) ...................... 7.8AP1-200(pAEK-E77K-P79T) ....................... 7.2AP1-200(pAEK-T153I)b ............................. 11.7

a Average of three independent assays.b BamHI mutant T1531 displays approximately 1% of the wild-type cleav-

age activity in vitro (28).

BamHI mutant, T1531, which displays a partial cleavageactivity in vitro (28), resulted in an eightfold increase in,B-galactosidase activity. The SOS response is induced by thelow-level constitutive expression of the mutant endonucle-ase under PtaC control since the Ptac and lacI system is nottightly repressed in vivo (10).

Purification of the pseudorevertant endonucleases and theircofactor preference. The wild-type BamHI purification pro-cedure (10) was followed to purify the R76K-E77K andE77K-P79T endonucleases. They were purified to >95%, asjudged by Coomassie blue staining and densitometric scan-ning of the gel (Fig. 1, lanes,3 and 4). The purified pseudo-revertant endonucleases, R76K-E77K and E77K-P79T, havea specific activity of 5 x 103 U/mg of protein in Mg2" buffer(Fig. 2B, lanes 2 and 3). In comparison with the E77Kendonuclease, both pseudorevertants increased activity atleast 50-fold. The pseu,dorevertants are still 200-fold lessactive than the wild-type enzyme (106 U/mg of protein in 100mM NaCl, 10 mM Tris-HCl [pH 7.0], 6 mM MgCl2, 1 mMdithiothreitol). Both pseudorevertants demonstrate reducedactivity in Mn2+ buffer (Fig. 2B, lanes 5 and 6), as does thewild-type enzyme. The magnitude of reduction, however, isless dramatic than for the wild-type enzyme (approximately2-fold versus 10-fold). Hence, the suppressing mutations thatpartially restored activity also partially restored the cofactorpreference.

Substrate specificity and stability of R76K-E77K and E77K-P79T mutants. Since R76K-E77K and E77K-P79T mutantendonucleases were isolated by indication of increased SOSresponse, we tested whether the SOS induction could beblocked by the expression of BamHI methylase. AP1-200cells carrying the revertant endonuclease plasmids weretransformed with a compatible plasmid, pADE15-bamhIM+,and plated on X-Gal indicator plates. These transformantsdid not demonstrate a blue phenotype. Expression of theBamHI methylase blocked R76K-E77K and E77K-P79Tendonuclease damage in vivo, indicating that the lesion isBamHI site dependent.The pseudorevertant endonucleases appear to be less

stable than wild-type BamHI. The cleavage activity wasreduced fivefold over 4 months of storage at -20°C in 50%glycerol. The activity of the wild-type enzyme did not dropsignificantly during the same period of time.To test whether R76K-E77K and E77K-P79T have in-

creased star activity, the purified mutant endonucleaseswere used to cleave X DNA. Under the in vitro conditionsused, no noncanonical cleavage was observed (Fig. 2E,lanes 2 and 3). Neither nicking nor cleavage was found whenthe BamHI site was filled in or methylated (data not shown).

DISCUSSION

We have purified the BamHI endonuclease catalytic mu-tant E77K to homogeneity. We found that E77K is moreactive in the presence of Mn2+ than in the presence of Mg2+,while the opposite is true for wild-type BamHI activity. Astrategy utilizing a DNA damage (dinD: :lacZ indicatorstrain) phenotype was successful in isolating two suppressormutations of E77K. The two suppressor mutations arelocated (positions 76 and 79) proximal to the primary E77Kmutation. The suppressors not only increased activity 50-fold but also changed the cofactor preference of E77K to thatof the wild type.The result that E77K differs from the wild-type enzyme in

cofactor requirements suggests that Glu-77 may be involvedin metal ion binding. The E77K mutation is an alteration of

VOL. 173, 1991

on May 6, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

5034 XU AND SCHILDKRAUT

the charge of the amino acid residue. It is possible that thechange in cofactor preference reflects a change in the localstructure for metal ion binding. Suppressing mutations thatrestore partial cleavage activity carry an amino acid substi-tution in neighboring positions (residues 76 and 79, respec-tively). In both cases (Arg-76 to Lys and Pro-79 to Thr), thesubstitutions did not return the charge to that of the wildtype but did decrease the volume of amino acid residues (Arg[173.4 A3; 1 A = 0.1 nm], Lys [168.6 A3]; Pro [122.7 A3], andThr [116.1 A3]; 4). Apparently this partially compensates forthe original substitution (Glu-77 to Lys), which increased theresidue volume (Glu [138.4 A3] and Lys [168.6 A3]; 4). Themechanism of Mn2+ stimulation on E77K endonucleaseactivity is not understood. It may reflect a preferred coordi-nation geometry of Mn2+ with E77K endonuclease. E77Kalso shows an enhanced DNA-binding activity (28), whichsuggests that Glu-77 may be located near the DNA-proteininterface. In the EcoRI restriction-modification system, anEcoRI mutant endonuclease ElliG was isolated and shownto increase activity in a high-pH star buffer (13). ElliG alsodisplays an enhanced DNA-binding activity (13). In theEcoRI crystal structure, the Glu-111 residue is located nearthe DNA scissile bond (12a).The dinD: :lacZ fusion strain has been used successfully to

isolate mutant proteins which interact in one way or theother with DNA (5-8). However, this is the first time that thedinD::lacZ fusion has been used to screen for revertantsfrom a cleavage-deficient endonuclease mutant. The screen-ing method used here to isolate suppressing mutations isbased on the triggering of the SOS response by endonucleasenicking or cleavage in the absence of cognate methylase.This method can isolate only partial suppressors, since highendonuclease activity in vivo would be lethal to the host inthe absence of cognate methylase. However, it may bepossible to isolate suppressors which yield full activity byusing a temperature-sensitive BamHI methylase that canprotect the host chromosome at a low temperature from theDNA damage caused by a high-activity revertant and thenscreen for those colonies that do not grow at high tempera-ture. An alternative method is to establish the level ofexpression for the wild-type protein at which the hostsurvives and SOS response is induced. Once this lower levelof expression is established, it can be used to isolate fullyactive revertants of a cleavage-deficient mutant. Thesescreening methods can also be applied to other restrictionand modification systems to isolate suppressing mutationsfrom endonuclease null mutants. Suppressors of endonucle-ase null mutants may help to identify critical amino acidresidues in the protein. Recently, the dinD::lacZ indicatorstrain has been used for cloning methylase genes from typeII restriction and modification systems (21). The same indi-cator strain has been used to isolate BamHI variants whichconfer a Lac' phenotype to the host despite the presence ofthe cognate methylase (30).The question arises as to the nature of the induction of the

dinD::lacZ fusion. Do low levels of single/double-strandedDNA cleavages or BamHI/DNA complexes induce thedinD::lacZ fusion? We argue here that the low level of SOSinduction by the E77K endonuclease is probably caused bythe residual cleavage activity of the enzyme. Another cata-lytic BamHI mutant, E113K, also shows enhanced DNA-binding activity (28). The purified E113K endonucleasedisplays no detectable nicking or cleavage activity in vitronor does E113K induce a detectable SOS response in vivodespite its enhanced binding activity (28). Another BamHIvariant, T1531, displays about 1% of the wild-type cleavage

activity (28). This mutant endonuclease is deficient in DNAbinding, as determined by gel retardation assay (presumablyits lowered cleavage activity can be accounted for by itslowered affinity for DNA). This mutant endonuclease in-duces the SOS response in AP1-200 cells, as evidenced bydark blue colony formation on indicator plates.We have found three amino acid residues (Glu-77, Asp-94,

and Glu-113), all containing carboxylate groups, whereamino acid substitutions inactivate cleavage but do notinterfere with DNA binding. Interestingly, Beese and Steitz(1) have determined from the crystal structure of the 3'-5'exonuclease of the Klenow fragment of E. coli DNA poly-merase I that the two metal ions are coordinated by thecarboxylate groups of amino acid residue Asp-355, Asp-357,and Glu-501 in the enzyme and the 5' phosphate of thedeoxynucleoside monophosphate product. It was proposedthat a proton in a water molecule is displaced by metal A andthus creates an activated hydroxide ion which attacks theP-O bond. Metal B plays a role in facilitating the leaving ofthe 3'-OH group. It will be interesting to determine whetherBamHI amino acid residues Glu-77, Asp-94, and Glu-113,identified as critical residues in cleavage by genetic analysis,play a similar catalytic function by holding and orientingmetal ions in the active site.

ACKNOWLEDGMENTS

We thank Joan Brooks, William Jack, Donald Nwankwo, andGeoffrey Wilson for cloning and overexpression of the bamHIR andbamHIM genes, which made this work possible, and for their adviceand encouragement. We also thank Elisabeth Raleigh for strains,Lydia Domer and Rebecca Kucera for help with protein purifica-tion, Paul Evans and Julia Kelleher for discussion, and Joan Brooks,William Jack, Elisabeth Raleigh, and Paul Riggs for critical readingof the manuscript. Special thanks to Don Comb of New EnglandBiolabs for his support.

REFERENCES1. Beese, L. S., and T. A. Steitz. 1991. Structural basis for the 3'-5'

exonuclease activity of Escherichia coli DNA polymerase I: atwo metal ion mechanism. EMBO J. 10:25-33.

2. Brooks, J. E., J. S. Benner, D. F. Heiter, K. R. Silber, L. A.Sznyter, T. Jager-Quinton, L. S. Moran, B. E. Slatko, G. G.Wilson, and D. 0. Nwankwo. 1989. Cloning the BamHI restric-tion-modification system. Nucleic Acids Res. 17:979-997.

3. Brooks, J. E., P. D. Nathan, D. Landry, L. A. Sznyter, P.Waite-Rees, C. C. Ives, L. M. Mazzola, B. E. Slatko, and J. S.Benner. 1991. Characterization of the cloned BamHI system: itsnucleotide sequence, properties of the methylase, and expres-sion in heterologous hosts. Nucleic Acids Res. 19:841-850.

4. Creighton, T. E. 1983. Proteins: structure and molecular prin-ciples, p. 6-8. W. H. Freeman & Co., New York.

5. Heitman, J., and P. Model. 1990. Substrate recognition by theEcoRI endonuclease. Proteins Struct. Funct. Genet. 7:185-197.

6. Heitman, J., and P. Model. 1990. Mutants of the EcoRI endo-nuclease with promiscuous substrate specificity implicate resi-dues involved in substrate recognition. EMBO J. 9:3369-3378.

7. Heitman, J., and P. Model. 1991. SOS induction as an in vivoassay of DNA-protein interactions. Gene, in press.

8. Heitman, J., N. D. Zinder, and P. Model. 1989. Repair of theEscherichia coli chromosome after in vivo scission by the EcoRIendonuclease. Proc. Natl. Acad. Sci. USA 86:2281-2285.

9. Holmes, D. S., and M. Quigley. 1981. A rapid boiling method forthe preparation of bacterial plasmids. Anal. Biochem. 114:193-197.

10. Jack, W. E., L. Greenough, L. F. Dorner, S. Y. Xu, T.Strzelecka, A. K. Aggarwal, and I. Schildkraut. 1991. Overex-pression, purification, and crystallization of BamHI endonucle-ase. Nucleic Acids Res. 19:1825-1829.

11. Jack, W. E. Unpublished data.

J. BACTERIOL.

on May 6, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

COFACTOR REQUIREMENTS OF BamHI MUTANT E77K 5035

12. Kenyon, C., and G. C. Walker. 1980. DNA-damaging agentsstimulate gene expression at specific loci in Escherichia coli.Proc. Natl. Acad. Sci. USA 77:2819-2823.

12a.Kim, Y. C., J. C. Grable, R. Love, P. J. Greene, and J. M.Rosenberg. 1990. Refinement of EcoRI endonuclease crystalstructure: a revised protein chain tracing. Science 249:1307-1309.

13. King, K., S. J. Benkovic, and P. Modrich. 1989. Glu-111 isrequired for activation of the DNA cleavage center of EcoRIendonuclease. J. Biol. Chem. 264:11807-11815.

14. Malygin, E. G., and V. V. Zinoviev. 1989. Studies on the role ofsymmetry in the specific recognition on natural and syntheticDNA by type II restriction and modification enzymes. Sov. Sci.Rev. Physiochem. Biol. 9:87-142.

15. Maniatis, T., E. F. Fritsch, and J. Sambrook. 1982. Molecularcloning: a laboratory manual. Cold Spring Harbor Laboratory,Cold Spring Harbor, N.Y.

16. Miller, J. H. 1972. Experiments in molcular genetics. ColdSpring Harbor Laboratory, Cold Spring Harbor, N.Y.

17. Nardone, G., and J. G. Chirikjian. 1987. The enzyme of theBamHI restriction-modification system, p. 148-184. In J.Chirikjian (ed.), Gene amplification and analysis. Elsevier Sci-ence Publishers, New York.

18. New England Biolabs Inc. 1987. M13 cloning and sequencingsystem: a laboratory manual. New England Biolabs Inc., Bev-erly, Mass.

19. Ossanna, N., K. R. Peterson, and D. W. Mount. 1986. Geneticsof DNA repair in bacteria. Trends Genet. 2:55-58.

20. Panayotatos, N., and A. Fontaine. 1985. An endonuclease spe-cific for single-stranded DNA selectively damages the genomicDNA and induces the SOS response. J. Biol. Chem. 260:3173-3177.

21. Piekarowicz, A., R. Yuan, and D. C. Stein. 1991. A new methodfor the rapid identification of genes encoding restriction andmodification enzymes. Nucleic Acids Res. 19:1831-1835.

22. Sancar, A., and G. B. Sancar. 1988. DNA repair enzymes.Annu. Rev. Biochem. 57:29-67.

23. Sanger, F., S. Nicklen, and A. R. Coulson. 1977. DNA sequenc-ing with chain-terminating inhibitors. Proc. Natl. Acad. Sci.USA 74:5463-5467.

24. Smith, L. A., and J. G. Chirikjian. 1979. Purification andcharacterization of the sequence-specific endonuclease BamHI.J. Biol. Chem. 254:1003-1006.

25. Walker, G. C. 1984. Mutagenesis and inducible response todeoxyribonucleic acid damage in Escherichia coli. Microbiol.Rev. 48:60-93.

26. Wood, W. B. 1966. Host specificity of DNA produced byEscherichia coli bacterial mutations affecting the restriction andmodification of DNA. J. Mol. Biol. 16:118-133.

27. Wu, T. H., C. H. Clarke, and M. G. Marinus. 1990. Specificityof Escherichia coli mutD and mutL mutator strains. Gene87:1-5.

28. Xu, S. Y., and I. Schildkraut. 1991. Isolation ofBamHI variantswith reduced cleavage activity. J. Biol. Chem. 266:4425-4429.

29. Xu, S. Y., and I. Schildkraut. 1991. Submitted for publication.30. Xu, S. Y., and I. Schildkraut. Unpublished data.

VOL. 173, 1991

on May 6, 2018 by guest

http://jb.asm.org/

Dow

nloaded from