| Nucleic Acids Research | Pages |
EcoKI with an amino acid substitution in any one of seven DEAD-box motifs has impaired ATPase and endonuclease activities
Introduction
Materials And Methods
Bacterial strains, plasmids and phages
DNA manipulation
Transfer of mutations to the bacterial chromosome
Reagents
Protein purification and analysis
Results
Mutations in the DEAD-box motifs
Phenotypes of mutants
The mutants specify EcoKI complexes of normal stoichiometry
The mutant EcoKI complexes bind DNA and undergo an ATP-dependent conformational change
Nuclease activity of the purified proteins
ATPase activities of proteins
Discussion
Acknowledgements
References
EcoKI with an amino acid substitution in any one of seven DEAD-box motifs has impaired ATPase and endonuclease activities
ABSTRACT
INTRODUCTION
Enzymes that bind ATP include a motif defined by two component sequences, `A' and `B' (1). Many of these enzymes comprise a `super family' within which members share a number of sequences including GxGKS/T, an abridged version of the A component of the ATP-binding motif, the sequence DEAD, DExH (e.g. DEAH and DECH) or DExx (e.g. DEYQ) as the B component of the ATP-binding motif, and five additional conserved sequences (2). Some of these proteins are known to be helicases (3,4). Mutational analyses of DEAD-box proteins have been reported, the most extensive being for an RNA helicase, the mammalian translation initiation factor eIF-4A (5). The results suggest that the seven highly conserved regions in the DEAD-box proteins are critical for helicase activity. Each of these conserved sequences will be referred to as a motif, identified as I, Ia and II-VI (6).
Some members of the DEAD-box proteins have been shown to be DNA, rather than RNA, helicases. These include Rad3 of yeast (7), and RecG (8) and RecQ (9) of Escherichia coli, all three of which are involved in recombination and repair processes. RecG includes the seven motifs characteristic of the DExH family (10) and catalyses branch migration of Holliday junctions in a reaction that requires ATP hydrolysis. A mutation affecting motif III, in which the sequence TAT was changed to TVT, was shown to reduce ATP hydrolysis and to block branch migration (8). For RecG, it is concluded that branch migration of Holliday junctions is related to an ATP-dependent helicase activity.
Type I restriction and modification (R-M) enzymes, like RecG, are members of the DExH subgroup of DEAD-box proteins, and although evidence for a helicase activity is lacking, like RecG, they are believed to translocate DNA (11-16). Type I R-M enzymes are complex proteins comprising three different subunits, HsdR, HsdM and HsdS. Each system recognizes its specific target sequence, the methylation status of which determines whether the R-M complex responds by modifying (methylating) or restricting (cutting) the DNA. Restriction, the response to unmethylated target sequences, is ATP dependent and involves the cutting of non-specific sequences remote from the target sequences. The restriction enzyme remains bound to its target sequence (17) and is believed to translocate DNA past the bound enzyme in an ATP-dependent process (12). Experiments using a linear DNA substrate led to a model (13) in which it was proposed that cutting is stimulated when two translocating complexes collide. This model is consistent with in vivo evidence that cutting occurs between target sequences (18).
The predicted amino acid sequence of EcoKI, the type I R-M enzyme found in E.coli K-12, includes the A and B components of the ATP-binding motif; motifs I and II of DEAD-box proteins. These sequences are in the HsdR subunit of the complex, consistent with the ATP dependence of the restriction reaction in vitro (19) and the demonstration that the HsdR subunit of type I R-M enzymes is essential for restriction but not modification (20,21). The B component of the ATP-binding motif in EcoKI is DExH; this, and the putative presence of the other sequences characteristic of the DExH subgroup of DEAD-box proteins in the HsdR subunits of EcoKI and EcoR124I, was recognized by Gorbalenya and Koonin (6). Sequence comparisons of the very dissimilar HsdR subunits of different type I R-M enzymes identified the DEAD-box motifs as their most noticeable common features, and this observation prompted the speculation that these motifs are relevant to the DNA translocation that is believed to precede restriction (14).
Conservative mutations in motifs I, II and III of EcoKI have been shown to impair DNA restriction in vivo (22). In this paper we extend the analysis to mutations in motifs Ia, IV, V and VI, as identified by Gorbalenya and Koonin (6), and an additional sequence (Y) located between motifs III and IV (23). The mutations in region Y and motifs Ia, V and VI, but not IV, confer a restriction-deficient phenotype (rK-) and we now propose that region Y identifies the correct motif IV. EcoKI has been purified from seven rK- mutants, one for each of the seven DEAD-box motifs including region Y as motif IV, and from two rK+ mutants. The proteins recognize their DNA target sequence, but those from rK- strains are unable to cut DNA and they have low or negligible levels of ATPase activity. These analyses identify seven regions of the HsdR polypeptide of EcoKI, which share sequence similarities with DEAD-box motifs, as being relevant to the restriction activity of EcoKI.
MATERIALS AND METHODS
Bacterial strains, plasmids and phages
AB1157 (24) was used as a restriction-proficient (rK+ mK+) E.coli K-12 strain. NM795 and NM802 are rK- derivatives of AB1157; NM795 has missense mutations in hsdR (22) and NM802 an internal deletion, hsdR[Delta]4 (21), that removes the DNA specifying the DEAD-box motifs, with the possible exception of motif I. A strain deleted for the hsd genes, NM679 (25), was the host for propagating hsdR-M+S+ plasmids for DNA and the production of mutant EcoKI for purification.
The origins of the mutations in motifs I, II and III have been described previously (22). pSB2, a plasmid including hsdR but not hsdM and S (22), was the template for amplification by PCR to obtain mutations in motif Ia. This site-directed mutagenesis used the recombinant PCR technique as described by Higuchi (26). Amplified DNA fragments were substituted for the relevant wild-type sequence of the hsdR gene in pSB2 and the complete nucleotide sequences of inserts were determined to identify the anticipated base changes and demonstrate the absence of additional mutations. Mutations in motifs IV, V, VI and region Y were made directly in pSB2 using the Quickchange[trade] procedure of Stratagene. The mutations were identified from the nucleotide sequence. For each base change, duplicate isolates were selected, one from each of two separate reactions to ensure that each mutant was of independent origin. The nucleotide sequences of oligonucleotides used to make changes in the motifs are given in Table 2. Some mutations were transferred to phsd+ to make hsdR-M+S+ plasmids, the source of mutant enzymes; the 2.2 kb fragment (Asp718-SmaI; Fig.
Figure 1. Restriction map of the hsd region cloned in phsd+. mrr is a gene encoding a methylation-dependent restriction system. The locations of the coding sequences for the DEAD-motifs are indicated below the map. The amino acid sequences of the motifs in EcoKI are given; bold type identifies the amino acids for which substitutions were made. The hsdR- derivatives of pSB2 were used for complementation tests in the hsdR- strain NM795, and for dominant negative complementation tests in the hsd+ strain AB1157. [lambda] virulent (laboratory collection), either unmodified ([lambda]vir.0) following propagation on NM679 or modified ([lambda]vir.K) following propagation on AB1157, was used to measure restriction phenotypes. Phages [lambda]NM1265 (27) and [lambda]NM1347 (22) were used to clone mutant hsdR genes within a SmaI-EcoRI DNA fragment and facilitate their transfer from the plasmid to the chromosome of AB1157. [lambda]NM1266 retains one target for SmaI, the leftmost, and [lambda]NM1347 retains only the leftmost EcoRI site.
DNA manipulation
Preparation and manipulation of DNA and nucleotide sequencing were carried out as described previously (25). The nucleotide sequence for Ia is between unique KpnI (Asp718) and BamHI targets in the hsdR gene of plasmid pSB2 (Fig.
The mutant hsdR coding sequences were excised from the pSB2 derivatives and inserted in a [lambda] genome to facilitate their transfer to the E.coli chromosome. The 3.6 kb EcoRI-SmaI fragment excised from each pSB2 derivative contained all but the first few codons of hsdR (Fig.
Transfer of mutations to the bacterial chromosome
The [lambda] phages including hsdR encode a temperature-sensitive repressor (cI857) and are defective in their attachment site (b527). Lysogens selected as immune colonies at 32°C frequently result from homology-dependent recombination but they occasionally lose their prophage by homologous recombination. These `cured' clones may be selected at 42°C. The b527 mutation greatly reduces the frequency of site-specific integration but does not entirely prevent it. Lysogens resulting from site-specific recombination only rarely produce cured clones and liberate no or few progeny phage.
[lambda] phages in which the hsdR mutation has conferred restriction deficiency gave both rK+ and rK- lysogens of AB1157, depending on the location of the integrative cross-over with respect to the mutation. Approximately 50% of the cured colonies selected at 42°C had replaced the chromosomal hsdR allele with that of the phage. As expected, the two restriction-proficient mutants (A619G and F730S) failed to give rK- lysogens in an hsdR+ host; lysogens were made, therefore, in an hsdR- host (NM802) and on curing these lysogens yielded rK+ derivatives.
Reagents
DNA sequencing kits, sequenase and [[gamma]-32P]ATP were obtained from Amersham International, type II restriction enzymes from Boehringer Mannheim and New England BioLabs, DNA ligase and Vent DNA polymerase from New England BioLabs and Red Hot DNA polymerase from Advanced Biotechnologies. Polynucleotide kinase was supplied by S.Bruce and K.Murray. ATP was obtained from Sigma and AdoMet from New England BioLabs. QuikChange[trade] Site-Directed Mutagenesis kits were bought from Stratagene. Synthetic oligonucleotides were supplied by OSWEL DNA, University of Southampton.
Protein purification and analysis
The enzymes were purified as described by Dryden et al. (28). DNA binding and the stability of the nuclease (R2M2S1) complex were checked by gel retardation of the unmodified specific 45 bp duplex DNA containing the EcoKI recognition site (0.1 nM) by EcoKI (0, 5, 10, 20, 30, 40, 50, 60, 70, 80, 90 and 100 nM). Synthetic oligonucleotide (45mers) were end-labelled and hybridized to yield the 45 bp duplex as described by Powell et al. (29) except [[gamma]-33P]ATP (ICN) was used instead of [[gamma]-32P]ATP. The protein was preincubated on ice with AdoMet (100 µM), or AdoMet (100 µM) and ATP (2 mM), for 10 min prior to DNA addition. The binding buffer was 20 mM Tris-HCl pH 8.0, 100 mM NaCl, 6 mM MgCl2, 7 mM [beta]-mercaptoethanol and 5% glycerol. After DNA addition, the samples were incubated for 10 min at 22°C prior to running on a 5% non-denaturing gel for 2 h at 35 mA. Gels were dried for 75 min at 80°C and exposed to Kodak Biomax MR film for 24-48 h.
Exonuclease III footprinting was carried out essentially as described by Powell et al. (30). Protein (100 nM) was preincubated for 10 min on ice with AdoMet (100 µM) or AdoMet and ATP (2 mM), prior to DNA addition. The samples were incubated for a further 10 min at 22°C in the buffer used for gel retardation, to allow DNA-protein complex formation. After this incubation, 1.5 U exonuclease III was added to 60 µl of each reaction, and digestion was for 10 min at 37°C. The samples were processed as described by Powell et al. (30) and run on a 12% urea denaturing gel which was dried prior to autoradiography. Gel retardation confirmed that the EcoKI complex (wild-type or mutant) used in each footprinting reaction remained intact, with no dissociation of the R subunit to yield R1M2S1 and M2S1.
Endonuclease activity was assayed by a modification of the method of Dryden et al. (28). The reactions were stopped by the addition of 0.5 vol `stop' solution [0.1 M EDTA, 0.1 M Tris-HC1 pH 8.0, 40% (w/v) sucrose], and the products of the reaction were separated by electrophoresis through a 1.2% agarose gel in TAE buffer (31).
Two assays were used for ATPase. One followed the release of inorganic phosphate from [gamma]-labelled ATP directly (32), the second was the pyruvate kinase/lactate dehydrogenase-linked assay of Ali et al. (33). Both assays were done at 37°C in buffer A from Boehringer Mannheim (33 mM Tris-acetate, 10 mM magnesium acetate, 66 mM potassium acetate and 0.5 mM dithiothreitol, pH 7.9) and were initiated by the addition of ATP. The rates are initial velocities based on a period of linear reduction of A340, within 3 min of starting the reaction for active enzymes and within 15 min for impaired enzymes. All assays were done in triplicate, and corrected for background activity in the absence of AdoMet. The two assays gave very similar results, the second being preferable for the determination of rates, since ADP does not accumulate and inhibit the reaction.
RESULTS
Mutations in the DEAD-box motifs
The mutations affecting motifs I, II and III were made in plasmid pSB2, a plasmid in which the entire hsdR gene is located downstream of the T7 promoter in pT7-7, and have been described previously (22). The mutations in motif Ia were made by site-directed mutagenesis (SDM) using the polymerase chain reaction (PCR); a 550 bp Asp718-BamHI fragment was amplified (Fig.
The amino acid substitutions made are listed in Table 1. Motifs I, Ia, II, IV, V and VI are as identified by Gorbalenya and Koonin (6). Motif III (22) and region Y (23) are within the corrected sequence for hsdR in which a short frame-shifted segment in the original hsdR sequence was identified (34).
Phenotypes of mutants
Tests dependent on multicopy plasmids. Plasmid pSB2 includes hsdR cloned in the absence of hsdM and S (22). This hsdR+ parent plasmid, when maintained in a strain encoding a defective HsdR subunit (e.g. NM795), compensates for the chromosomal hsdR mutation and the restriction-proficient phenotype is recognized by the low efficiency of plaque formation (e.o.p.) of [lambda]vir.0. The phenotypes associated with the presence of the mutant derivatives of pSB2 were tested in the same way; a low e.o.p. of unmodified phages would be indicative of a functional HsdR polypeptide. On the basis of this test, one of three substitutions in motif III (22) and all three of those in motif IV failed to inactivate the HsdR polypeptide (Table 1). The phenotypes of the other mutants are consistent with defects in restriction. The test, however, is not reliable, because any cell lacking the plasmid will fail to restrict [lambda]vir.0 and progeny phages will be modified by the methyltransferase component of EcoKI. It is, therefore, critical that the vast majority of cells retain the plasmid, otherwise a false indication of restriction deficiency will result.Table 1.
| Motif | Sequence | Mutation on plasmid | Mutation on chromosomeb |
|
| in hsdR- hosta | in hsdR+ hosta | |||
| (I)c | ||||
| GTGKT | K477I | rk- | rk- | * |
| K477R | rk- | rk- | rk- | |
| K477T | rk- | rk- | * | |
| (Ia) | ||||
| FLVDRR | D502H | rk- | rk- | * |
| D502N | rk- | rk- | * | |
| D502Y | rk- | rk- | rk- | |
| (II)c | ||||
| DEAH | H577D | rk- | rk- | rk- |
| H577N | rk- | rk- | * | |
| H577Y | rk- | rk+ | rk- | |
| (III)c | ||||
| LTATP | A619D | rk- | rk- | * |
| A619V | rk- | rk- | rk- | |
| A619G | rk+ | rk+ | rk+/- | |
| (IV)d | ||||
| LVFCVT | F730C | rk+ | rk+ | * |
| F730S | rk+ | rk+ | rk+ | |
| F730T | rk+ | rk+ | * | |
| (V) | ||||
| LTTGVD | G799C | rk- | rk- | rk- |
| G799R | rk- | rk- | * | |
| G799S | rk- | rk- | * | |
| (VI) | ||||
| QMKGRAT | R826H | rk- | rk- | rk- |
| R826L | rk- | rk- | * | |
| R826P | rk- | rk- | * | |
| (Y) | ||||
| FGEPVYRY | F629C | rk- | rk+ | * |
| F629Y | rk- | rk- | rk- | |
Positive evidence of a restriction-deficient phenotype is obtained if the mutant HsdR polypeptide competes with wild-type for HsdM and HsdS and, particularly when present in excess, converts an hsd+ host into an rK- phenocopy. Therefore, each of the plasmids carrying a mutant hsdR gene was used to transform the rK+ strain AB1157, and the restriction phenotypes of the transformed strains were tested. In this test an rK+ phenotype is consistent with either a mutant HsdR polypeptide that remains functional, or an inactive polypeptide that fails to displace the wild-type HsdR subunit. An rK- phenotype is good evidence of a defective HsdR subunit that retains the ability to interact with the methylase, and mutations conferring this dominant negative phenotype were isolated in each of the regions tested, other than the original motif IV. In this motif none of three different substitutions for the phenylalanine residue prevented restriction (Table 1).
Test dependent on mutations in the bacterial chromosome. Quantification of the restriction phenotype of any hsdR mutant is only possible if the mutation is stably maintained in single copy on the bacterial chromosome, preferably in its normal location under the control of its resident promoter. Some mutations, including those for which EcoKI has been purified, have been transferred to the chromosome of AB1157 and restriction quantified using [lambda]vir.0. The mutations in region Y and changes in all the previously designated motifs other than IV can confer an rK- phenotype identical to that found for a strain lacking the hsdR gene (Table 1). The most conservative mutation in motif III had only a small effect on the restriction phenotype (22).The mutants specify EcoKI complexes of normal stoichiometry
EcoKI was purified from nine mutants. The mutations chosen included one within each conserved sequence other than motif III, for which enzyme was purified from one mutant (A619V) with an rK- phenotype and one (A619G) that retained almost full restriction capacity. Each of the nine mutations was transferred from the pSB2 derivative, which lacks hsdM and hsdS, to phsd+, a plasmid that includes the three hsd genes (35; Fig.
For each of the mutants, the elution profiles obtained during purification of the EcoKI complex, including that from the size-exclusion chromatography column (results not shown), were indistinguishable from those obtained for the wild-type enzyme (28). This similarity supports the correlation of the dominant negative phenotype of the mutant strains with localized defects in the HsdR subunit of the EcoKI complex rather than merely the absence of a stable HsdR polypeptide or the failure to maintain the EcoKI complex.
It has been shown recently by gel retardation that wild-type EcoKI has a tendency to dissociate at very low protein concentration, particularly in the presence of ATP (36). This dissociation was shown by the detection of R1M2S1-DNA and M2S1-DNA complexes as well as the major species in which the active nuclease, R2M2S1, is associated with DNA. Each of the three complexes was excised from a non-denaturing gel and the component subunits separated on an SDS-polyacrylamide gel to permit their identification in western blots by antibody to EcoKI (36).
In order to check the integrity of the mutant nuclease complexes over a range of protein concentrations (5-100 nM), we examined the gel retardation of a 45 bp duplex that includes an unmodified target sequence for EcoKI. In the presence of 100 µM AdoMet, but in the absence of ATP, the only DNA-protein complex observed for wild-type EcoKI was that for intact R2M2S1, even at protein concentrations as low as 5 nM. In the presence of ATP (2 mM), faint bands representing R1M2S1-DNA and M2S1-DNA were apparent at 5 and 10 nM protein, but the major complex was R2M2S1-DNA. Gel retardation assays in the presence and absence of ATP were carried out on all the proteins made from restriction-deficient mutants (data not shown). The assays in the absence of ATP showed that the R2M2S1-DNA complex is the major species across the 5-100 nM range of protein. In the presence of ATP, R2M2S1 generally remained the main species across the 5-100 nM range. However, the enzyme preparations with substitutions in either motif II or V required 10 nM protein in order that the main species was R2M2S1, while the concentration of the protein with a substitution in motif VI had to be raised to 20 nM to achieve the EcoKI complex as the major species. For this reason, the enzyme assays on EcoKI with a substitution in motif VI were performed using 40 nM protein as well as the 10 nM concentration used in other assays.
The mutant EcoKI complexes bind DNA and undergo an ATP-dependent conformational change
Exonuclease III footprinting has recently been used to investigate the effects of the cofactors AdoMet and ATP on DNA binding by the wild-type EcoKI endonuclease (36). With DNA containing an unmodified EcoKI recognition sequence, protection of 42-45 bp was observed either in the absence of any cofactor or in the presence of AdoMet alone. In the presence of ATP and AdoMet, a shorter footprint of 30 bp was observed, while in the presence of ATP alone there was no protection of the DNA. These results are consistent with an ATP-dependent change in the conformation of EcoKI, as previously observed by electron microscopy (17), and with AdoMet being essential for the recognition of specific unmodified DNA in the presence of ATP. The hydrolysis of ATP is not necessary for this change in the footprint as both ATP[gamma]S and ADP were able to substitute for ATP in producing this smaller footprint. No protection was observed in the presence of both ATP and AdoMet when the oligonucleotide was fully modified or when it lacked a target sequence; in the presence of both cofactors protection requires an unmodified or hemimethylated EcoKI target sequence.
We have used the exonuclease III footprinting assay as an indication of whether the mutant proteins, like wild-type EcoKI, can undergo the ATP-dependent change in conformation in response to unmodified target sequences, the step that precedes ATP hydrolysis and presumably translocation and cutting. The top strand of the 45 bp duplex that includes an unmodified target for EcoKI was labelled and used as substrate for footprinting experiments. The cofactor AdoMet was always present and the effect of ATP was monitored.
The footprints obtained in the absence of ATP are shown in the first 12 lanes of Figure
Table 2.
| Motif | Oligonucleotide (5[prime]->3[prime]) | Mutation | Substitution | Plasmid |
| I | G ACC GGT ACC GGT ABA ACC CG | A T A | K477I | - |
| A G A | K477R | pJW17a | ||
| A C A | K477T | - | ||
| Ia | C TTC CTT GTC HAC CGC GGT TCb | C A C | D502H | pJW18 |
| A A C | D502N | pJW19 | ||
| T A C | D502Y | pJW20a | ||
| II | GA CGA AGC GDT TCG CGG Cb | G A T | H577D | pJW21a |
| A A T | H577N | - | ||
| T A T | H577Y | - | ||
| III | C GCT CTC ACC GDC ACC CCG Gb | G A C | A619D | - |
| G G C | A619G | pJW22 | ||
| G T C | A619V | phsdR-a,c | ||
| IV | CCG GAT CGC AAA AAA CGC TGG TCT TVG TGC GTC ACCb |
T G C | F730C | - |
| T C C | F730S | pJW23a | ||
| T A C | F730T | - | ||
| V | C GAC CTG CTG ACG ACC HGC GTC GAT ATT CCGb |
T G C | G799C | pJW24a |
| C G C | G799R | pJW25 | ||
| A G C | G799S | pJW26 | ||
| VI | C GAA CAG ATG AAA GGC CHC GCC ACG CGC TTA TGCb |
C A C | R826H | pJW27a |
| C T C | R826L | - | ||
| C C C | R826P | - | ||
| Y | GCG CTA CAT ACT GTG CAG ATT TVC GGC GAG CCG Gb |
T G C | F629C | pGDY12 |
| T A C | F629Y | pGDY9a |
For wild-type protein, we have identified previously the DNA fragments protected for both strands of this substrate (36). For the top strand the length of the major protected fragment was 45 bases in the absence of ATP and the presence of 100 µM AdoMet, and 38 bases in the presence of both ATP and AdoMet. For the bottom strand the protected fragments were 42-45 bases in the presence of AdoMet and absence of ATP, and 37 bases in the presence of both AdoMet and ATP. In the present experiments in which the top strand was labelled, it is striking that, in the presence of AdoMet and ATP, all of the mutants gave the same protected 38mer species as the wild-type protein. This implies that all of the mutants are able to bind ATP, although a reduction in the affinity of ATP binding would not necessarily be detected in this assay as we aimed to use an excess (2 mM), so that ATP would not be limiting. These results also confirm that the binding specificity of the mutants is not lost, as a non-specific interaction of EcoKI results in no protection of the DNA (36).
Figure 2. DNA protection defined by exonuclease III footprints of the top strand of the unmodified 45 bp substrate. 45 and 38 indicate the lengths of the major species of oligonucleotides, as identified from a sequencing ladder. See Table 1 and text for further details of the motifs. The major species protected from exonuclease III by the EcoKI methyltransferase (M2S1), using the 45 bp duplex with the top strand labelled as substrate, is the same length as that protected by EcoKI (R2M2S1) in the presence of both cofactors (30,36). For this reason, the composition of the protein in the DNA-protein complex was validated for each of the mutant enzymes. Samples from the footprinting reactions, before and after exonuclease III treatment, were run on 5% non-denaturing polyacrylamide gels; the mobilities of the complexes confirmed that the R subunits had not dissociated from the nuclease (R2M2S1) in the protein-DNA complexes isolated during footprinting (results not shown). The footprinting experiments, therefore, indicate that the DEAD-box mutations are exerting their effects on a step, or steps, following the conformational change of the R2M2S1 complex associated with ATP binding.
Nuclease activity of the purified proteins
The nine proteins from mutant strains were compared with wild-type enzyme for nuclease activity. The substrate used in the assay was an unmodified plasmid 6.2 kb in length (pRH3) that has two target sequences for EcoKI, 2.3 kb apart. Protein (10 nM) was added to provide one molecule of EcoKI per target sequence (28). Endonuclease activity is indicated by conversion of the covalently-closed circular DNA to a full-length linear molecule and the subsequent degradation to DNA fragments of variable sizes (Fig.
a
b  |
Figure 3. Nuclease assays. (a) Assay for wild-type enzyme. (b) Assays for enzymes purified from one restriction-proficient mutant (F730S), and from one restriction-deficient mutant (K477R). The substrate DNA was plasmid pRH3, either unmodified (.0) or modified (.K). Reactions were carried out as described in Materials and Methods with samples removed at the times indicated. The positions of supercoiled (ccc), linear (lin) and open-circle (oc) forms of the plasmid are shown. DNA digested with HindIII was used as marker DNA (lane M); marker sizes are given in kb.
DNA samples taken a few seconds after the addition of the restriction-proficient enzyme indicate that the plasmid DNA is nicked prior to linearization (Fig.
ATPase activities of proteins
The ATPase activities of the proteins purified from the nine mutants were compared with that of wild-type EcoKI. pRH3, the plasmid with two unmodified EcoKI recognition sites, was used as substrate for all the enzymes. Wild-type enzyme and protein from other mutants that retained appreciable ATPase and endonuclease activities with pRH3 as substrate were tested for activity on the 45 bp duplex containing one EcoKI site. Both substrates were used in an attempt to separate different ATPase activities; activities that might precede, accompany or follow DNA cleavage. The 45 bp duplex, in contrast to the plasmid, is not a substrate for either the endonuclease or nicking activity of EcoKI (36 and unpublished results). pRH3 was used at half the molar concentration of the 45 bp duplex to maintain a ratio of one enzyme molecule per target sequence.
ATPase activity was assayed directly by the release of radiolabelled inorganic phosphate from [gamma]-labelled ATP (32), and indirectly by a pyruvate kinase/lactate dehydrogenase linked assay (33). For both assays a correction was made for a very small amount of activity observed in the absence of the cofactor AdoMet. The two assays gave consistent results, but the enzyme-linked assay provides better quantification. The rates shown (Fig.
Figure 4. Rates of ATP hydrolysis for EcoKI. Assay carried out using pyruvate kinase/lactate dehydrogenase linked coupled assay (33). Error bars indicate standard deviation of at least three experiments. Using the plasmid as substrate for ATPase activity, an enzyme from the restriction-proficient mutant with a substitution in the original, though suspect, motif IV was indistinguishable from wild-type; that with the A619G substitution in motif III was only slightly impaired (Fig. 
DISCUSSION
Type I R-M enzymes comprise three polypeptides, and whenever the amino acid sequences of two type I systems have been compared, the component polypeptides have been found to be either very similar (e.g. EcoKI and EcoBI) or quite dissimilar (EcoKI and EcoR124I). Enzymes of very similar sequences have been grouped together as members of one family, those of very dissimilar sequences as representatives of different families (37,38). Pair-wise comparisons of the sequences of HsdR polypeptides from different families, usually detect between 20 and 25% amino acid identity (14,23). Those motifs that identify HsdR polypeptides as members of the super family of DEAD-box proteins make the major contribution to the conserved amino acids. DEAD-box proteins include RNA and DNA helicases (2,4) though all members of this family are not necessarily helicases (39,40). There is no experimental evidence that type I R-M systems are ATP-dependent helicases, but the prominence of the DEAD-box motifs has tempted the speculation that these motifs are relevant to the ATP-dependent translocation of DNA that has been proposed to precede DNA cutting (14). In the present paper we reconsider the identification of DEAD-box motif IV in HsdR and test the hypothesis that each DEAD-box motif in this subunit is associated with some step in the pathway of events that leads to restriction of unmodified DNA.
The definition of type I R-M systems as DEAD-box proteins was based originally on the sequences of the HsdR polypeptides of EcoKI and EcoR124I (6). Motif III in EcoKI and motif IV in EcoR124I were difficult to identify. In 1995, a short frame-shifted sequence was found in the hsdR gene of E.coli K-12 (34) and an obvious candidate for motif III was readily identified within the new segment of amino acid sequence (22). When this same segment of polypeptide sequence was compared with those of six other, known or putative, HsdR sequences, a conserved region (Y) was located between motifs III and V (23). Region Y became the preferred candidate for motif IV. This preference is consistent with evidence from the crystal structure of PcrA, a DExx helicase isolated from Bacillus stearothermophilus (41). In this structure a tyrosine residue in motif IV was shown to be involved in stacking the bound nucleotide. The original motif IV identified by Gorbalenya and Koonin (6) in EcoKI included a phenylalanine residue in the position where the consensus residue is either phenylalanine or tyrosine, but neither phenylalanine nor tyrosine was found at this position in most of the HsdR polypeptide sequences (23). In this paper we show that substitution of serine, threonine or cysteine for the tyrosine residue in the original motif IV does not confer a restriction-deficient phenotype. Region Y, as defined by the alignment of the seven HsdR polypeptides, includes a conserved tyrosine residue (23). We show that mutations in region Y, and not motif IV as originally defined, confer a restriction-deficient phenotype. In our following discussion we consider region Y to be the correct motif IV.
We find that mutations in each of the seven motifs (Table 1) can confer a restriction-deficient phenotype. In support of the relevance of these phenotypes, we show for one mutation in each of the seven motifs that the resulting proteins retain the gross configuration of the wild-type enzyme but fail to degrade unmodified plasmid DNA.
Previous knowledge of the steps in the pathway that leads to the cutting of unmodified DNA by EcoKI indicates that the enzyme binds to DNA in the presence of the cofactor AdoMet, but requires ATP to recognize and respond to unmethylated target sequences (17). Recognition of an unmodified target sequence elicits a conformational change, first visualized by electron microscopy (17) and recently documented by footprinting experiments (36). The hydrolysis of ATP, rather than the binding of ATP, is only essential for subsequent steps, including the formation of DNA loops that have been taken as evidence of DNA translocation (12). For EcoKI and EcoR124I or II it has been proposed that DNA cutting is elicited either when translocating complexes collide (13,15,22) or when a change in DNA topology leads to stalling of DNA translocation (13,15,16,22). There is no evidence to implicate ATP hydrolysis in the breakage of the phosphodiester bonds of the DNA chains but ATP hydrolysis has been shown to continue after DNA breakage (42,43).
All the mutant enzymes retain the ability to bind an unmodified target sequence and to protect a sequence of 43-45 bases on the top strand of a 45 bp duplex. In response to ATP each protein undergoes a conformational change and in the resulting complex, like that generated by the wild-type enzyme (36), a much shorter sequence (38 bases) of DNA is protected. The ATP-dependence of the conformational change shows that none of the mutations has prevented binding of the cofactor. It might have been anticipated that even the very conservative substitution of arginine for lysine in the ATP-binding component of the Walker motif (1) would decrease the affinity of the enzyme for ATP. This may be so, since our experiments have not quantified the affinity of the protein for ATP but, under the experimental conditions we have used (2 mM ATP), they show that ATP is able to elicit the conformational change associated with the recognition of a restriction target.
Mutations that conferred a restriction-deficient phenotype resulted in enzyme that was deficient in nuclease activity. The assays for endonuclease activity demonstrate that a covalently-closed circular DNA substrate is nicked prior to linearization. Earlier evidence for nicking has supported the simple concept that two nicks generate a break (19). This is consistent with the stimulation of nicking when two translocating complexes collide (13). Substitutions in the DEAD-box motifs block endonuclease activity but, although the rate of nicking is much reduced, none prevents nicking (compare Fig.
All the endonuclease-deficient mutants retained little or no ATPase activity. Mutations that failed to block restriction in vivo, including that in the original motif IV, yielded enzyme that resembled wild-type for both activities. The assays for ATP hydrolysis used two substrates, pRH3, a plasmid that would permit DNA translocation and cleavage, and a duplex oligonucleotide 45 bp in length which might permit some limited translocation but was known to be too short a substrate for DNA cleavage or nicking (36 and unpublished observations). For wild-type enzyme the short substrate elicits appreciable levels of ATPase activity. The ATP hydrolysis associated with the oligomer is dependent on an unmodified target sequence and occurs in the absence of cutting or nicking of the DNA. If we suppose that ATPase activity is a manifestation of DNA translocation, then the activity seen with the 45 bp oligonucleotide implies that translocation is possible on this short substrate. Alternative explanations are possible. For EcoR124II, unmodified oligonucleotides of 35 bp support ATPase activity in the absence of DNA cleavage, with a level of activity similar to that seen with a plasmid substrate (43).
Mutations in each of the seven DEAD-box motifs have resulted in bacteria defective in restriction by the EcoKI system. All of the mutations impair both the ATPase and endonuclease activities of the purified enzyme. Mutants with substitutions in motifs I, II, III (A619V), IV, V and VI retain little or no ATPase activity. The substitution in motif Ia leaves appreciable (Fig.
Other mutational analyses of DEAD-box motifs, most commonly focusing on motifs I through IV, have demonstrated that changes in the motifs affect ATPase and related enzyme activities e.g. helicase, branch migration and endonuclease (5,8,46). Some mutations have been found to enhance rather than impair ATPase activity (5). Mutations in motif II of EcoP1I, like those in EcoKI, reduce ATPase activity, but the decrease is not necessarily associated with a block in restriction (46). As predicted from structural studies (47,48), mutations in motif I, the ATP-binding component of the Walker motif (1), can prevent the binding of ATP (49). The conservative change made in EcoKI did not prevent ATP binding, but resulted in much reduced ATPase activity. The same amino acid change in Rad3 of yeast abolished ATPase and helicase activities but, as with EcoKI, the mutant protein retained ATP-binding activity (7). Recent structures of two helicases, both DEAD-box proteins of the DExx-family (41,50), now locate the motifs in key positions with respect to each other, the cofactor and the substrate, providing a framework for functional models (51).
ACKNOWLEDGEMENTS
We thank Tatyana Prokhorova and Iain Cheeseman for the isolation of some mutants, our colleagues, particularly David Dryden, for discussion and constructive criticism of the manuscript, and Natalie Honeyman for its careful preparation. The work was supported by the Medical Research Council and Studentships to G.P.D. (MRC) and J.L.W. (BBSRC).
REFERENCES
This article has been cited by other articles:
This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals
Comments and feedback: www-admin{at}oup.co.uk
Last modification: 17 Oct 1998
Copyright©Oxford University Press, 1998.
CiteULike 
Connotea 
Del.icio.us What's this?
F. Zhao, X. Zhang, C. Liang, J. Wu, Q. Bao, and S. Qin
Genome-wide analysis of restriction-modification system in unicellular and filamentous cyanobacteria
Physiol Genomics,
February 23, 2006;
24(3):
181 - 190.
[Abstract]
[Full Text]
[PDF]
S. Chen, A. A. Davies, D. Sagan, and H. D. Ulrich
The RING finger ATPase Rad5p of Saccharomyces cerevisiae contributes to DNA double-strand break repair in a ubiquitin-independent manner
Nucleic Acids Res.,
October 13, 2005;
33(18):
5878 - 5886.
[Abstract]
[Full Text]
[PDF]
S. A. Keatch, T.-J. Su, and D. T. F. Dryden
Alleviation of restriction by DNA condensation and non-specific DNA binding ligands
Nucleic Acids Res.,
November 1, 2004;
32(19):
5841 - 5850.
[Abstract]
[Full Text]
[PDF]
W. A. M. Loenen
Tracking EcoKI and DNA fifty years on: a golden story full of surprises
Nucleic Acids Res.,
December 15, 2003;
31(24):
7059 - 7069.
[Abstract]
[Full Text]
[PDF]
N. E. Murray
Immigration control of DNA in bacteria: self versus non-self
Microbiology,
January 1, 2002;
148(1):
3 - 20.
[Full Text]
[PDF]
D. T. F. Dryden, N. E. Murray, and D. N. Rao
Nucleoside triphosphate-dependent restriction enzymes
Nucleic Acids Res.,
September 15, 2001;
29(18):
3728 - 3741.
[Abstract]
[Full Text]
[PDF]
N. E. Murray
Type I Restriction Systems: Sophisticated Molecular Machines (a Legacy of Bertani and Weigle)
Microbiol. Mol. Biol. Rev.,
June 1, 2000;
64(2):
412 - 434.
[Abstract]
[Full Text]
[PDF]
S. Makovets, V. A. Doronina, and N. E. Murray
Regulation of endonuclease activity by proteolysis prevents breakage of unmodified bacterial chromosomes by type I restriction enzymes
PNAS,
August 17, 1999;
96(17):
9757 - 9762.
[Abstract]
[Full Text]
[PDF]
R. Rimseliene and A. Janulaitis
Mutational Analysis of Two Putative Catalytic Motifs of the Type IV Restriction Endonuclease Eco57I
J. Biol. Chem.,
March 23, 2001;
276(13):
10492 - 10497.
[Abstract]
[Full Text]
[PDF]
This Article 
Abstract
Print PDF (290K)
Alert me when this article is cited
Alert me if a correction is posted
Services
Email this article to a friend
Similar articles in this journal
Similar articles in ISI Web of Science
Similar articles in PubMed