- Email: [email protected]
The GTP-binding Domain of McrB: More Than Just a Variation on a Common Theme?
Article No. jmbi.1999.3103 available online at http://www.idealibrary.com on
J. Mol. Biol. (1999) 292, 547±556
The GTP-binding Domain of McrB: More Than Just a Variation on a Common Theme? Uwe Pieper*, Thomas Schweitzer, Detlef H. Groll, Frank-Ulrich Gast and Alfred Pingoud Institut fuÈr Biochemie, JustusLiebig-UniversitaÈt Giessen Heinrich-Buff-Ring 58 D-35392 Giessen, Germany
The methylation-dependent restriction endonuclease McrBC from Escherichia coli K12 cleaves DNA containing two RmC dinucleotides separated by about 40 to 2000 base-pairs. McrBC is unique in that cleavage is totally dependent on GTP hydrolysis. McrB is the GTP binding and hydrolyzing subunit, whereas MrC stimulates its GTP hydrolysis. The C-terminal part of McrB contains the sequences characteristic for GTP-binding proteins, consisting of the GxxxxGK(S/T) motif (position 201-208), followed by the DxxG motif (position 300-303). The third motif (NKxD) is present only in a non-canonical form (NTAD 333-336). Here we report a mutational analysis of the putative GTP-binding domain of McrB. Amino acid substitutions were initially performed in the three proposed GTP-binding motifs. Whereas substitutions in motif 1 (P203V) and 2 (D300N) show the expected, albeit modest effects, mutation in the motif 3 is at variance with the expectations. Unlike the corresponding EFTu and ras-p21 variants, the D336N mutation in McrB does not change the nucleotide speci®city from GTP to XTP, but results in a lack of GTPase stimulation by McrC. The ®nding that McrB is not a typical G protein motivated us to perform a search for similar sequences in DNA databases. Eight microbial sequences were found, mainly from un®nished sequencing projects, with highly conserved sequence blocks within a presumptive GTP-binding domain. From the ®ve sequences showing the highest homology, 17 invariant charged or polar residues outside the classical three GTP-binding motifs were identi®ed and subsequently exchanged to alanine. Several mutations speci®cally affect GTP af®nity and/or GTPase activity. Our data allow us to conclude that McrB is not a typical member of the superfamily of GTP-binding proteins, but de®nes a new subfamily within the superfamily of GTP-binding proteins, together with similar prokaryotic proteins of as yet unidenti®ed function. # 1999 Academic Press
*Corresponding author
Keywords: restriction enzymes; McrBC restriction; DNA binding; GTPase; site-directed mutagenesis
Introduction In Escherichia coli K12, the detection and destruction of foreign DNA is accomplished by at least Abbreviations used: BSA, bovine serum albumin; GST, glutathione S-transferase; GTP, guanosine 50 triphosphate; ITP, inosine 50 -triphosphate; Kass, association constant; KD, dissociation constant; NTP, nucleotide 50 -triphosphate; wt, wild-type; XTP, xanthosine 50 -triphosphate; ORF, open reading frame. E-mail address of the corresponding author: [email protected] 0022-2836/99/380547±10 $30.00/0
four restriction activities. One of them is the McrBC restriction endonuclease which acts against cytosine-modi®ed (methylated or hydroxymethylated) DNA. The McrBC system consists of two protein subunits McrB (53 kDa) and McrC (39 kDa) which are encoded by two slightly overlapping genes (Raleigh et al., 1989; Ross et al., 1989; Dila et al., 1990). The enzymatic properties of McrBC have been investigated in detail in recent years. McrBC recognizes and cleaves DNA containing two hemi or fully methylated RmC sites (regardless of orientation) in an optimal distance of about 40 to 80 # 1999 Academic Press
548 base-pairs (Sutherland et al., 1992) and to a lesser degree in a distance of up to 2000 base-pairs (Stewart & Raleigh, 1998). DNA binding, but not cleavage, can take place even at sites with more narrowly separated half-sites (Gast et al., 1997). The McrB subunit is responsible for the binding of DNA and GTP (KruÈger et al., 1995; Gast et al., 1997; Pieper et al., 1997), whereas the function of McrC, aside from the stimulation of the GTPase activity of McrB, is as yet unknown. In spite of its basic character, McrC does not bind directly to DNA, but to McrB, leading to complexes of higher molecular mass (KruÈger et al., 1995; Gast et al., 1997). The signature sequences of GTP-binding proteins (Dever et al., 1987; Bourne et al., 1991) which are involved in GTP interaction as con®rmed by numerous X-ray crystal structures (for a review, see Kjeldgaard et al., 1996), consist of: (1) the phosphate binding loop GxxxxGK(S/T); (2) the switch II region DxxG; and (3) the guanine recognition loop, NKxD. Based upon the amino acid sequence, Dila et al. (1990) proposed that McrB is the GTPbinding and hydrolyzing subunit, because its C-terminal part contains the characteristic motifs, starting at position 201 (numbered from the second start codon; see Materials and Methods) with the P-loop GxxxxGK(S/T), followed at position 300 by the DxxG motif. The third motif (NKxD) is present only in a non-canonical form (NTAD 333-336). By mutating Asn333 and Asn370 to alanine we asked whether NTAD (333-336) or an alternative candidate motif, NKKA (370-373) represents motif 3. We found that the exchange Asn333 to Ala completely abolished all McrB activities connected to GTP binding, whereas the variant N370A was indistinguishable from the wild-type enzyme. We therefore concluded that the NTAD (333-336) region is involved in GTP binding (Pieper et al., 1997). The DNA-binding domain of McrB has been shown by us to reside within the N-terminal 184 amino acid residues (Gast et al., 1997) and to be devoid of GTP-binding activity. Thus, the inference that the GTP-binding domain is in the C-terminal part of the molecule seemed to be justi®ed. We have proven this assumption experimentally by investigating a deletion mutant lacking the N-terminal 162 amino acid residues (Pieper et al., 1999) which is almost identical with McrBs, an additional product of McrB gene expression resulting from an internal translation start at codon 162. It has been demonstrated recently (Beary et al., 1997; Panne et al., 1998) that McrBs modulates McrBC restriction activity, presumably by sequestration of McrC. We have found that the McrB fragment consisting of amino acids 163 to 459 interacts with GTP and McrC in a manner identical to wild-type McrB (Pieper et al., 1999). Here we report the properties of a series of McrB variants with amino acid substitutions in the GTPbinding domain. The ®rst mutations were targeted to the three parts of the inferred GTP-binding motif in order to test the assumption that these motifs are involved in GTP binding, extending the
GTP-binding Domain of McrB
work described by Pieper et al. (1997). Database searches in microbial genomic sequences, especially un®nished ones, revealed a number of open reading frames (ORFs) which share several highly conserved sequence blocks with E. coli McrB. Interestingly, these homologous regions reside entirely in the C-terminal GTP-binding part of McrB. This ®nding allowed us to broaden our search for amino acid residues involved in guanine nucleotide and/or McrC interaction outside the classical G protein motifs. Based on a multiple alignment, we performed an alanine scanning mutagenesis of all 17 polar amino acid residues strictly conserved in E. coli McrB and the ®ve most similar sequences. Our results show that McrB is not a typical member of the superfamily of GTPbinding proteins, but may de®ne a new subfamily.
Results Amino acid substitutions in motifs 1 and 2: P203V and D300N In the phosphate-binding loop, Pro203 was exchanged to Val. We chose the variable position 3 in the GxxxxGK(S/T) motif because mutations in this region affect the GTPase activity in other GTPbinding proteins. V20G in E. coli elongation factor Tu (Jacquet & Parmeggiani, 1988) and G12V in rasp21 (Schmidt et al., 1996) reduce the GTP hydrolysis rate, leading to an oncogenic variant in the latter case. In the DxxG motif, we chose the invariant Asp300 which we exchanged rather conservatively to Asn. An analogous substitution in elongation factor Tu (D80N) weakens GTP binding and enhances GTP hydrolysis (Harmark et al., 1992). The McrB variants P203V and D300N still cleave DNA, albeit slightly slower than the wild-type enzyme. Speci®c DNA recognition, tested by gel retardation assays with methylated oligonucleotides, is unaffected by both mutations (data not shown). The GTP-related activities are listed in Table 1. The af®nities for GTP of both variants are reduced by a factor of three and ®ve, respectively. The intrinsic GTPase rate of the P203V variant is slowed down by a factor of four. The D300N substitution leaves the intrinsic GTPase rate unaffected. In both variants GTP hydrolysis is still stimulated by McrC, although slightly less than in wild-type McrB. Amino acid substitution in motif 3: D336N, an attempt to change nucleotide specificity As seen in several X-ray crystallographic studies (for a review, see Kjeldgaard et al., 1996), the invariant aspartate of the NKxD motif interacts via hydrogen bonding to the exocyclic 2-amino group and the N1-imino group of the purine ring, thereby determining guanine speci®city. Substitution of this aspartate residue by an asparagine residue changes speci®city from guanine to xanthine, as was demonstrated for EF-Tu (Hwang & Miller,
549
GTP-binding Domain of McrB Table 1. GTP-related properties of the McrB variants P203V and D300N, compared to wild-type McrB 6
ÿ1
GTP binding constant Kass (10 M ) % of wild-type GTP affinity Intrinsic GTPase activity (Pi released/McrB per h) % of wild-type GTPase activity McrC-stimulated GTPase activity (Pi released/McrB per hour) Factor of stimulation by McrC
McrB wt
P203V
D300N
2.09 100 41 100
0.74 36 10 24
0.42 20 53 130
607 14.8
82 8.3
288 5.4
1987) and several other GTP-binding proteins, e.g. ras-p21 (Schmidt et al., 1996), Rab5 (Hoffenberg et al., 1995), and FtsY (Powers & Walter, 1995). A similar speci®city change caused by substitution of Asp336 by Asn in McrB would be a clear indication that the proposed assignment of GTP-binding motif 3 is correct. The enzymatic properties of the corresponding McrB variant D336N are rather unexpected. Figure 1(a) shows DNA cleavage experiments with wild-type McrB and the D336N variant in the presence of GTP, XTP or both nucleotides. Unexpectedly, McrB D336N does not exhibit any DNA cleavage activity with either nucleotide. In contrast, wild-type McrB is almost as active with XTP as with GTP. As Figure 1(b) clearly demonstrates, the inability of the D336N variant to cleave DNA is not caused by a loss of GTP af®nity. Even the relative af®nities of both proteins to other purine nucleotides, including XTP, are very similar, as tested by competition experiments (Figure 2). The binding constants for ITP, XTP, and ATP are summarized in Table 2. XTP is bound two orders of magnitude weaker than GTP. This is still suf®cient for saturating McrB when the nucleotide is present in millimolar amounts, as is the case in the DNA cleavage experiments. ITP is bound even more strongly than XTP, i.e. only ten times more weakly than GTP, and hence also supports DNA cleavage by wild-type McrBC (Figure 3). ATP shows the lowest af®nity to McrB among the nucleotides tested, with a KD value in the millimolar range. Even this weak af®nity would not exclude utilization in DNA cleavage, if nucleotide binding was the only requirement for DNA cleavage. Figure 3, however, shows that even when present at 5 mM concentration, ATP does not support cleavage, con®rming earlier results by KruÈger et al. (1992) and Sutherland et al. (1992). Intriguingly, ATP is hydrolyzed by McrB with a rate comparable to that of other purine nucleotides (data not shown). This means that while GTP, ITP, XTP and ATP are all bound and hydrolyzed by McrB, only GTP, ITP and XTP, but not ATP, serve as allosteric effectors for DNA cleavage by McrBC. We tested whether GTP hydrolysis, rather than binding, might be affected by the D336N substitution, the results of which are shown in Figure 4. As reported previously (Pieper et al., 1997), wild-type McrB hydrolyzes GTP in a multiple-turnover reaction with a rate of about 40
per hour, which is enhanced by more than an order of magnitude when McrC is present in at least equimolar amounts. The variant D336N is still able to hydrolyze GTP (30 % of wild-type activity; the GTP hydrolysis rate is constant for over two hours which shows that D336N is not less stable). However, the GTPase activity of D336N cannot be stimulated by McrC, which may explain why this variant is inactive in DNA cleavage.
Figure 1. Properties of the D336N variant of McrB. (a) DNA cleavage experiments. Cleavage reactions of methylated plasmid DNA (EcoRI-linearized pBW201) were carried out for one hour at 25 C without nucleotide or in the presence of 1 mM GTP, XTP, or both. (b) GTP binding isotherms of D336N (*) and wild-type McrB (*) as determined by nitrocellulose ®lter binding. Active concentration of McrB was 0.55 mM. Binding constants were calculated by non-linear least-squares ®tting: Kass 2.1 106 Mÿ1 for D336N and 2.2 106 Mÿ1 for wild-type McrB.
550
GTP-binding Domain of McrB
Figure 2. Nucleotide speci®city of (a) wild-type and (b) D336N McrB. Relative af®nities for ATP (*), ITP ( & ) and XTP (~) were determined in competition experiments with 0.67 mM GTP and up to 6 mM competitor. Active protein concentrations were 0.28 mM for wild-type and 0.5 mM for D336N. Theoretical curves were calculated by non-linear least-squares ®tting. The order of af®nity for purine nucleotides is GTP > ITP > XTP > ATP for wild-type McrB and the D336N variant.
Figure 3. Nucleotide speci®city of DNA cleavage by McrB. The time course of the cleavage of pBW201 plasmid DNA (pre-cleaved with EcoRI) by 1 mM McrB with 5 mM of different purine nucleotides (ATP, GTP, ITP and XTP) in the time range of 0, 3, 10, 30, 100 minutes is shown. The right lanes contain substrate plasmid (S). While GTP, ITP, and XTP at 5 mM support DNA cleavage by McrB, ATP does not.
Sequence database search for McrB homologues The similar af®nity and preference of wild-type McrB and the D336N variant for nucleotide triphosphates allow us to conclude that the guanine speci®city of McrB is determined in a different manner than in ``canonical'' GTP-binding proteins. Either the consensus structure can also be formed with a non-canonical sequence instead of the canonical NKxD motif, or the GTP binding pocket is structured differently as compared to classical G proteins.
Table 2. Af®nities of wild-type McrB and McrB D336N for purine nucleotides NTP GTP ITP XTP ATP
Kass (Mÿ1)
wt 6
2 10 5.2 105 1.1 104 1.9 103
D336N 2 106 5.9 105 9.6 103 5.8 102
A Lactococcus lactis ORF with similarity to E. coli McrB has been reported by O'Sullivan et al. (1995). This sequence, named LlaI.2, is part of the plasmid-encoded restriction-modi®cation system LlaI. To test whether other proteins homologous to McrB exist, we performed a search of the SWISSPROT protein sequences database, which yielded no sequence with signi®cant homology to McrB. Depending on the search criteria, more than ten additional DNA sequences appear when the search is extended to ®nished or un®nished microbial genome sequencing projects with TBLASTN (Altschul et al., 1997). Figure 5 shows a multiple alignment of McrB and the ®ve sequences with the most signi®cant degree of homology to McrB (TBLASTN score > 100). The schematic view in Figure 5(a) indicates that the regions of high similarity cluster in the central to C-terminal part of McrB, i.e. in the GTP binding region, whereas no signi®cant similarities are found in the N-terminal DNA-binding domain. Figure 5(b) gives a detailed view of the highly conserved sequence blocks. GTP-binding motif 1 (GxxxxGK(S/T)) is perfectly conserved. In contrast,
551
GTP-binding Domain of McrB
Alignment-guided mutagenesis of McrB The lack of conservation of motif 2 in homologous sequences and the high conservation of blocks outside the classical GTP-binding motifs gives additional evidence that the nucleotide-binding mode of McrB differs from that of canonical GTPases. Some of these strictly conserved residues should be involved in nucleotide interaction. Therefore, we mutated all the conserved acidic (®ve), basic (eight), and uncharged polar (four) residues to alanine. The resulting variants listed in Table 3 were expressed as N-terminally His6tagged proteins. While most of the variants could be obtained as soluble protein in concentrations up to 10 mM, H233A, E292A, and R302A were completely insoluble, maybe due to incorrect folding. The remaining 14 variants were tested for their activities in GTP binding and hydrolysis as well as DNA binding and cleavage in the presence of McrC (Table 3 and Figure 6). Some of the variants (S287A, K288A, R337A) behave like wild-type McrB or are completely inactive in all aspects (D279A, E280A), which can be explained by drastic conformational changes induced by these mutations. Most interesting are the variants in which partial functions are affected. Some still bind to, but do not cleave DNA in the presence of McrC (R283A, R349A), probably due to their lost GTP binding capacity. N282A still binds GTP, but does not interact with DNA. This amino acid substitution seems to exert a long-range effect on the N-terminal DNA-binding domain. D343A and R347A appear to be inactive in the GTP binding assay, but nevertheless cleave DNA. This apparent contradiction can be explained by the different GTP concentrations in the respective assays (1 mM for DNA cleavage, 10 mM for GTP binding, which is 20-fold higher than the KD for wild-type McrB-GTP). The GTPase activity of McrB R347A, tested at 5 mM GTP, was more than half of the wild-type rate (data not shown), indicating
Figure 4. Time course of GTP hydrolysis by 0.29 mM active McrB wild-type (&, & ) and 0.16 mM D336N (*, *) in the absence (open symbols) and presence (®lled symbols) of 5 mM McrC. GTP concentration was 10 mM. While the relatively strong GTPase activity of wild-type McrB is stimulated by McrC, the relatively weak GTPase activity of the D336N variant is not.
motif 2 (DxxG), which is present in McrB, cannot be found in the homologous sequences. Instead, the two basic residues occupying the variable positions in E. coli McrB are invariant throughout the six sequences. The ®rst and the last residue of motif 3 (NKxD) are perfectly conserved, but the canonical lysine residue at the second position does not show up in any of the sequences. Instead, the threonine residue which occupies this position in McrB is found in four of the ®ve homologues. Thus, the functional signi®cance of residues N333 and D336 demonstrated by site-directed mutagenesis by Pieper et al. (1997) and in this communication is also re¯ected by their high degree of conservation.
Table 3. Summary of McrB alanine mutagenesis variants McrB variant N196A H233A D279A E280A N282A R283A S287A K288A E292A R302A N333A D336A R337A D343A R347A R348A R349A
Solubility
DNA binding
GTP binding
DNA cleavage
ÿ ÿ ÿ
n.d. ÿ ÿ ÿ n.d. n.d.
n.d. ÿ ÿ ÿ n.d. n.d. n.d. n.d. ÿ ÿ n.d. ÿ
n.d. ÿ ÿ ÿ ÿ n.d. n.d. ÿ ÿ ÿ ÿ
552
GTP-binding Domain of McrB
Figure 5. Multiple alignment of E. coli McrB with the ®ve most similar microbial ORF sequences: Eco, Escherichia coli K12; Smu, Streptococcus mutans; Cje, Campylobacter jejuni NCTC 11168; Ype, Yersinia pestis; Spn, Streptococcus pneumoniae; and Pgi, Porphyromonas gingivalis W83. (a) The positions of ®ve highly conserved sequence blocks in the respective amino acid sequences. (b) These regions are shown in detail. Invariant residues are indicated under the alignment in capital letters. Above the alignment the inferred GTP-binding motifs are shown.
that, although GTP binding is affected, the GTPase centre is largely intact. Variants with wild-type-like GTP af®nity were also tested for their GTPase activity in the absence
Figure 6. Intrinsic and McrC-stimulated GTP hydrolysis of selected McrB variants. Of the 17 variants listed in Table 3, N196A, N282A, S287A, K288A, and R337A show considerable GTP binding and were therefore tested for their GTPase activity. Also shown are the speci®c GTPase activities of P203V, D300N and D336N.
and presence of McrC. The results are shown in Figure 6, together with the results obtained for variants P203V, D300N and D336N mentioned above. N196A, S287A, and K288A hydrolyze GTP with a slightly reduced intrinsic rate, and GTPase activity is still stimulated by McrC, albeit to a lesser extent. Thus, these positions are not crucial for McrB function, as seen from their DNA cleavage activity. The most pronounced effects regarding GTP hydrolysis are shown by N282A and R337A. The intrinsic GTPase rate of N282A is unaltered, but cannot be stimulated by McrC. In this respect, N282A behaves similarly to D336N. The basal GTPase rate of R337A is already increased threefold, such that the small additional McrC stimulation leads to an almost wild-type GTPase rate. Like wild-type McrB and all other variants, the R337A variant needs McrC to cleave DNA (data not shown).
Discussion McrBC is the only restriction endonuclease known to be GTP-dependent (for reviews, see Raleigh, 1992; Bickle, 1993). GTP binding and
GTP-binding Domain of McrB
hydrolysis are accomplished by the McrB subunit which is also responsible for DNA recognition (KruÈger et al., 1995; Pieper et al., 1997; Gast et al., 1997). The McrC subunit might be responsible for DNA cleavage, but this has not yet been demonstrated unequivocally. We have shown recently that the DNA and GTP-binding functions of McrB are located on different domains which can be separated from each other without losing their respective partial functions. The N-terminal ca 160 amino acid residues harbor the DNA-binding region, and a deletion mutant lacking this part exhibits wild type-like GTP-binding and GTPase activity (Pieper et al., 1999). These ®ndings are in accordance with the identi®cation of GTP-binding motifs in this part of McrB by Dila et al. (1990). The assumption that these sequence motifs in McrB are involved in GTP binding in the same way as in other GTP-binding proteins implies that the GTP-binding domain of McrB is structurally similar to the GTP-binding domain of proteins belonging to the G protein superfamily whose members share an almost invariant core structure (for a review, see Kjeldgaard et al., 1996). There are two problems with this inference (both regarding the third motif, NKxD; Pieper et al., 1997). (1) The conserved lysine residue at position 2 is replaced by a threonine residue in McrB, and (2) the distance between motif 2 and 3 is unusually small in McrB (29 amino acid residues) compared to other GTP-binding proteins in which this distance is at least 49 amino acid residues. Thus, McrB must deviate somehow from the GTPase consensus structure in this region, even though N333 from motif 3 is absolutely required for GTP binding and hydrolysis. Here we present a much more thorough mutagenesis study of the GTP-binding domain of McrB. We started with amino acid substitutions in all three GTP-binding motifs in which distinct effects have already been described for other GTP-binding proteins. Mutagenesis of Pro203, at the third position in the GxxxxGK(S/T) motif, had the same effect as G12V in ras-p21 (Schmidt et al., 1996), and V20G in EF-Tu (Jaquet & Parmeggiani, 1988), namely a moderately reduced GTPase activity. Thus, motif 1 seems to have the canonical function in McrB. Conservative substitution of the conserved motif 2 aspartate to asparagine residue (D300N) does not lead to an increase in GTPase activity, as opposed to the analogous substitution D80N in EF-Tu (Harmark et al., 1992); instead, both McrB variants exhibit a reduced binding constant for GTP. Thus, DKRG (300-303) might not be the equivalent of motif 2 in McrB. In the third GTP-binding motif (NTAD 333-336 in McrB), no reversal of nucleotide speci®city from GTP to XTP is observed with McrB D336N, as opposed to several GTP binding proteins (e.g. D138N in E. coli EF-Tu; Hwang & Miller, 1987). D336N is identical with wild-type McrB with respect to GTP binding and relative af®nities to different purine nucleotides (XTP, ITP, ATP). This means that Asp336
553 does not determine nucleotide speci®city, but is rather functionally important in McrC interaction, such that the identi®cation of NTAD (333-336) as the guanine speci®city-determining GTP-binding motif 3 must be questioned. Thus, the proposal that McrB constitutes a distinct subclass of the GTPase superfamily is supported by the results presented here. Further con®rmation of this view comes from the open reading frames with signi®cant homology to E. coli McrB (although some of the un®nished sequences may not be correct in every detail). The homologous regions cluster in several blocks in the central to C-terminal part of McrB. No similarity whatsoever can be found in the DNA-binding domain. Apparently, the GTP-binding module of McrB was combined with a variety of insertions or N and C-terminal domains (see Figure 5(a)). As only motif 1 (the P-loop) is functionally signi®cant in McrB and conserved among the McrB-like proteins, other residues, most likely the strictly conserved amino acids revealed by the multiple alignment, must be involved in GTP binding and hydrolysis. Lacking any structural information, we chose all conserved polar residues for an alanine scanning mutagenesis. Of the functionally signi®cant mutations identi®ed thus far, the variants impaired in GTP binding (D279A, E280A, R283A, R347A, and R349A) are of special interest. The two acidic residues are obvious candidates for further mutational analysis because they potentially could have a similar function in guanine speci®city determination as the conserved aspartate of motif 3 in canonical GTP-binding proteins. DNA cleavage by McrB is dependent on McrCstimulated GTP hydrolysis. None of the variants tested change this strict coupling, not even R347A. This variant, with a GTP af®nity below the sensitivity of the ®lter binding assay, is active in DNA restriction in the presence of millimolar amounts of GTP. Two variants were altered in their GTPase properties, but were unaltered in their af®nity for GTP. (1) N282A has the same intrinsic GTPase activity as the wild-type enzyme, but cannot be stimulated in its GTPase activity by McrC and therefore cannot cleave DNA, much like D336N. (2) R337A is the only variant with a higher intrinsic GTPase activity, but on the expense of the stimulatory effect by McrC. Maybe this protein approaches the conformation which is normally induced upon McrC binding. In conclusion, we have shown that the GTPbinding domain of McrB is a structural and functional module which is different from the classic GTP-binding proteins and which is also present in several microbial proteins of presently unknown function. While our present mutagenesis study is purely phenomenological, only X-ray crystallography will tell whether the structure of the GTPbinding domain of McrB is a variation on a common theme or represents a fundamentally new fold.
554
GTP-binding Domain of McrB
Materials and Methods Bacterial strains, plasmids, and media For the expression of McrB and McrC as glutathione S-transferase (GST) fusion proteins and for the preparation of methylated substrate plasmids, the methylation-tolerant E. coli strain TC410 [mcrAÿ, (mrrhsdRMS-McrBC)201, minA, minB, rpsL, Sup] (NoyerWeidner et al., 1986; KruÈger et al., 1992) was used. GSTMcrB and GST-McrC were encoded on the expression plasmids pBN211 and pBN213, respectively (KruÈger, 1992; KruÈger et al., 1995). The vector pBBImcrB for the overexpression of His6-tagged McrB was constructed as follows. pBBInucA (Meiss et al., 1998), a descendant of pET-3d (Studier et al., 1990), was cleaved with NdeI and BamHI. The McrB gene, ampli®ed by PCR with primers introducing the appropiate restriction sites (NdeI N-terminally; primer: 50 -GATACATATGGAATCTATTCAACCCTG and BamHI C-terminally; primer: 50 -ATATGGATCCTATGAGTCCCCTAATAAT), was inserted into the large fragment. The resulting plasmid was transformed into E. coli BL21(DE3)pLysS (Studier et al., 1990). Cells were grown in LB medium with 75 mg/ml ampicillin. In the case of BL21(DE3)pLys, 20 mg/ml chloramphenicol was added. Site-directed mutagenesis and protein purification The mcrB gene contains two potential translation start codons 18 bp apart (Dila et al., 1990; Ross et al., 1990), leading to protein sizes of 465 or 459 amino acid residues. Because of a recent change (in release 37) of the SWISSPROT database entry (P15005) from the ®rst to the second start codon, McrB residue numbers given here are smaller by six compared to the numbering in previous publications (e.g. see Pieper et al., 1997). Three McrB variants were generated as GST fusion proteins expressed by the vector pBN211. D300N was generated by a three-primer PCR method described by Ito et al. (1991) and already used for the McrB variants described by Pieper et al. (1997). The mutagenesis primer was 50 -TGGCGAAGTCATGATGTTAATGGAACATAATAAACGA. P203V and D336N were generated by inverse PCR essentially as described by Meiss et al. (1998). The primers used to introduce the desired codon exchange and a silent marker restriction endonuclease site were: P203V/forward, 50 -CTATACCGGTCGGCGTTGGAAAAACCTTTG-30 ; P203V/reverse, 50 -ATATACCGGTCCCTGGAGGATAATATTTTTTTTGA-30 ; D336N/forward, 50 -CTTTAATCGATCTCTGGCCGT0 TGTTGACTATG-3 ; and D336N/reverse, 50 -CTCTATCGATTGGCAGTATTCATTAAACCG-30 . All other variants were generated by a protocol described by Kirsch & Joly (1998), starting from the His6-McrB expression vector pBBImcrB. The products were then directly transformed into E. coli LK111(l) cells. The mutagenic primers used were: N196A, 50 -CCATCAAAAAAGATATCATCCTCCAG; H233A, 50 -CAATATGGTCCAGTTCCCTCAATCTTAT; D279A, 50 -TATTATAGCTGAAATTAATCGTGCCAAT; E280A, 50 -TATTATCGATGCAATCAATCGTGCCAAT; N282A, 50 -TATTATCGATGAAATCGCTCGTGCCAAT; R283A, 50 -TATTATAGATGAAATTAATGCTGCCAAT; S287A, 5 0 -CAATCGTGCCAATCTCGCGAAAGTATT;
K288A, 5 0 -CAATCGAGCCAATCTCAGTGCAGTATT; E292A, 5 0 -GTATTTGGCGCAGTCATGATGTTAAT; 0 R302A, 5 -GATAAAGCAGGTGAAAACTGGTCTGTT; N333A, 5 0 -GGTTTAATGGCTACTGCAGATCGCTCT; D336A, 50 -GAATACTGCAGCTCGCTCTCTGGCCGTT; R337A, 50 -GAATACTGCAGATGCCTCTCTGGCCGTT; D343A, 50 -CCGTTGTTGCCTATGCCCTGCGCAGACGAT; R347A, 50 -CCGTTGTCGACTATGCCCTAGCCAGACGAT; R348A, 50 -GCCCTACGCGCGCGATTTTCTTTCAT; R349A, 50 -CGCAGAGCATTTTCTTTCATCGATATT; and The mcrB gene of positive clones was sequenced on both strands using the Taq cycle sequencing protocol (Perkin-Elmer) and an Applied Biosystems 373A sequencer. GST-McrB, GST-McrC and the McrB variants P203V, D300N and D336N were expressed and puri®ed as described by Pieper et al. (1997). For His6-tagged McrB (wild-type and variants) a small-scale (250 ml) Ni-NTAagarose (Qiagen) batch procedure from 50 ml cultures of E. coli BL21(DE3)pLysS cells was used according to manufacturer's instructions. The standard buffer contained 20 mM Hepes-KOH (pH 7.6), 1 mM EDTA, 50 mM KCl, and 10 % (v/v) glycerol. The His6-tagged proteins were eluted with 300 ml of 200 mM imidazole in standard buffer and stored at ÿ20 C. Protein concentration was determined by the method described by Bradford (1976) with bovine serum albumin (BSA) as standard or by absorbance at 280 nm with a molar extinction coef®cient of e 74,720 Mÿ1 cmÿ1 (wild-type McrB) calculated by the method described by Pace et al. (1995). GTP binding, hydrolysis and DNA cleavage assays Assays were performed as described by Pieper et al. (1997). Binding of [g-32P]GTP (DuPont-NEN) to McrB was measured by the nitrocellulose ®lter binding assay. If not indicated otherwise, the concentrations were 0.5 to 1 mM protein and 10 mM radioactive GTP in a ®nal volume of 20 ml assay buffer (20 mM Hepes (pH 7.6), 50 mM KCl, 5 mM MgCl2, 0.1 mg/ml BSA). GTP hydrolysis was assayed by the charcoal method which detects the amount of radioactive inorganic phosphate liberated after incubation with [g-32P]GTP. Standard conditions were 1 mM McrB, 10 mM GTP/[g-32P]GTP in a reaction volume of 20 ml assay buffer at 25 C. Alternatively, McrC-stimulated NTP hydrolysis was monitored by a colorimetric phosphate assay (Jenkins & Marshall, 1984). DNA cleavage activity of McrB was assayed with EcoRI-linearized pBW201, a plasmid harboring the BsuFI methylase gene leading to in vivo methylation at the ®rst cytosine of CCGG and thus creating McrB recognition sites (KruÈger et al., 1995). Electrophoretic mobility shift assay Gel shift experiments were performed essentially as described by KruÈger et al. (1995) and by Gast et al. (1997) in standard buffer. In a total volume of 10 ml, protein concentrations around 0.5 mM and either 0.5 mM (no radioactive labelling) or 0.01 mM (radioactive labelling) oligodeoxynucleotide was used. For the purpose of stabilization of McrB and McrC, all binding experiments were carried out in the presence of 0.1-0.2 mg/ml BSA. The DNA used in the non-radioactive assays was a 61mer double-strand with the sequence:
50 -AATTCTAAGACCGGTAGCGAGCTCGTATGATATCATATTAATCGGTAAGACCGGTAGCGAG GATTCTGGCCATCGCTCGAGCATACTATAGTATAATTAGCCATTCTGGCCATCGCTCTTAA
GTP-binding Domain of McrB
The 5-methylcytosine bases are underlined. In the radioactive assays a 44-mer double strand oligonucleotide was used (MH1 dimer; described by Gast et al., 1997). Database sequence search and alignment Sequence similarity searches were performed with the BLAST program (Altschul et al., 1997). For nucleotide sequence databases, TBLASTN, and for protein sequences, BLASTP was used with standard parameters. Searches were done on the web sites of NCBI (www.ncbi.nlm.nih.gov) and Infobiogen (www.infobiogen.fr). Preliminary sequence data for Streptococcus pneumoniae and Porphyromonas gingivalis W83 was obtained from the Institute for Genomic Research (TIGR; www.tigr.org), for Campylobacter jejuni NCTC 11168 and Yersinia pestis from the Sanger Centre (www.sanger.ac.uk), and for Streptococcus mutans UAB159 from the University of Oklahoma's Advanced Center for Genome Technology (ACGT; www.genome.ou.edu). Multiple alignments were performed with the program MAP (Huang, 1994).
Acknowledgments We are grateful to Drs M. Noyer-Weidner and T. KruÈger for expression plasmids for McrB and McrC. This work was supported by the Deutsche Forschungsgemeinschaft (Pi 122/11-2).
References Altschul, S. F., Madden, Z., SchaÈffer, A. A., Zhang, J., Zhang, Z., Miller, W. & Lipman, D. J. (1997). Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucl. Acids Res. 25, 3389-3402. Beary, T. P., Braymer, H. D. & Achberger, E. C. (1997). Evidence of participation of McrB(S) in McrBC restriction in Escherichia coli K-12. J. Bacteriol. 179, 7768-7775. Bickle, T. (1993). The ATP-dependent restriction enzymes. In Nucleases (Linn, S. M., Lloyd, R. S. & Roberts, R. J., eds), pp. 89-109, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Bourne, H. R., Sanders, D. A. & McCormick, F. (1991). The GTPase superfamily: conserved structure and molecular mechanism. Nature, 349, 117-127. Bradford, M. M. (1976). A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein dye binding. Anal. Biochem. 72, 248-254. Dever, T. E., Glynias, M. J. & Merrick, W. C. (1987). GTP-binding domain: three consensus sequence elements with distinct spacing. Proc. Natl Acad. Sci. USA, 84, 1814-1818. Dila, D., Sutherland, E., Moran, L., Slatko, B. & Raleigh, E. A. (1990). Genetic and sequence organization of the mcrBC locus of Escherichia coli K-12. J. Bacteriol. 172, 4888-4900. Gast, F. U., Brinkmann, T., Pieper, U., KruÈger, T., Noyer-Weidner, M. & Pingoud, A. (1997). The
555 recognition of methylated DNA by the GTP-dependent restriction endonuclease McrBC resides in the N-terminal domain of McrB. Biol. Chem. 378, 975982. Harmark, K., Anborgh, P. H., Merola, M., Clark, B. F. C. & Parmeggiani, A. (1992). Substitution of aspartic acid-80, a residue involved in coordination of magnesium, weakens the GTP binding and strongly enhances the GTPase of the G domain of elongation factor Tu. Biochemistry, 31, 7367-7372. Hoffenberg, S., Nikolova, L., Pan, J. Y., Daniel, D. S., Wessling-Resnick, M., Knoll, B. J. & Dickey, B. F. (1995). Functional and structural interactions of the Rab5 D136N mutant with xanthine nucleotides. Biochem. Biophys. Res. Commun. 215, 241-249. Huang, X. (1994). On global sequence alignment. Comput. Appl. Biosci. 10, 227-235. Hwang, Y. W. & Miller, D. L. (1987). A mutation that alters the nucleotide speci®city of elongation factor Tu, a GTP regulatory protein. J. Biol. Chem. 262, 1308-13085. Ito, W., Ishiguro, H. & Kurosawa, Y. (1991). A general method for introducing a series of mutations into cloned DNA using the polymerase chain reaction. Gene, 102, 67-70. Jacquet, E. & Parmeggiani, A. (1988). Structure-function relationships in the GTP binding domain of EF-Tu: mutation of Val20, the residue homologous to position 12 in p21. EMBO J. 7, 2861-2867. Jenkins, W. T. & Marshall, A. M. (1984). A modi®ed direct phosphate assay for studying ATPases. Anal. Biochem. 141, 155-160. Kirsch, R. D. & Joly, E. (1998). An improved PCR-mutagenesis strategy for two-site mutagenesis or sequence swapping between related genes. Nucl. Acids Res. 26, 1848-1850. Kjeldgaard, M., Nyborg, J. & Clark, B. F. C. (1996). The GTP binding motif: variations on a theme. FASEB J. 10, 1347-1368. È berproduktion, Reinigung und CharKruÈger, T. (1992). U akterisierung der Untereinheiten des McrBCRestriktionssystems von Escherichia coli K-12, PhD thesis, Freie UniversitaÈt, Berlin. KruÈger, T., Grund, C., Wild, C. & Noyer-Weidner, M. (1992). Characterization of the mcrBC region of Escherichia coli K-12 wild-type and mutant strains. Gene, 114, 1-12. KruÈger, T., Wild, C. & Noyer-Weidner, M. (1995). McrB: a prokaryotic protein speci®cally recognizing DNA containing modi®ed cytosine residues. EMBO J. 14, 2661-2669. Meiss, G., Franke, I., Gimadutdinow, O., Urbanke, C. & Pingoud, A. (1998). Biochemical characterization of Anabaena sp. strain PCC 7120 non-speci®c nuclease NucA and its inhibitor NuiA. Eur. J. Biochem. 251, 924-934. Noyer-Weidner, M., Diaz, R. & Reiners, L. (1986). Cytosine-speci®c DNA modi®cation interferes with plasmid establishment in Escherichia coli K12: involvement of rgl B. Mol. Gen. Genet. 205, 469-475. O'Sullivan, D. J., Zagula, K. & Klaenhammer, T. R. (1995). In vivo restriction by LlaI is encoded by three genes, arranged in an operon with llaIM, on the conjugative Lactococcus plasmid pTR2030. J. Bacteriol. 177, 134-143.
556
GTP-binding Domain of McrB
Pace, C. N., Vajdos, F., Fee, L., Grimsley, G. & Gray, T. (1995). How to measure and predict the molar absorption coef®cient of a protein. Protein Sci. 4, 2411-2425. Panne, D., Raleigh, E. A. & Bickle, T. A. (1998). McrBs, a modulator peptide for McrBC activity. EMBO J. 17, 5477-5483. Pieper, U., Brinkmann, T., KruÈger, T., Noyer-Weidner, M. & Pingoud, A. (1997). Characterization of the interaction between the restriction endonuclease McrBC from E. coli and its cofactor GTP. J. Mol. Biol. 272, 190-199. Pieper, U., Schweitzer, T., Groll, D. H. & Pingoud, A. (1999). De®ning the location and function of domains of McrB by deletion mutagenesis. Biol. Chem. In the press. Powers, T. & Walter, P. (1995). Reciprocal stimulation of GTP hydrolysis by two directly interacting GTPases. Science, 269, 1422-1424. Raleigh, E. A. (1992). Organization and function of the mcrBC genes of Escherichia coli K-12. Mol. Microbiol. 6, 1079-1086. Raleigh, E. A., Trimarchi, R. & Revel, H. (1989). Genetic and physical mapping of the mcrA (rglA), and mcrB
(rglB) loci of Escherichia coli K-12. Genetics, 122, 279296. Ross, T. K., Achberger, E. C. & Braymer, H. D. (1989). Nucleotide sequence of the McrB region of Escherichia coli K-12 and evidence for two independent translational initiation sites at the mcrB locus. J. Bacteriol. 171, 1974-1981. Schmidt, G., Lenzen, C., Simon, I., Deuter, R., Cool, R. H., Goody, R. S. & Wittinghofer, A. (1996). Biochemical and biological consequences of changing the speci®city of p21ras from guanosine to xanthosine nucleotides. Oncogene, 12, 87-96. Stewart, F. J. & Raleigh, A. (1998). Dependence of McrBC cleavage on distance between recognition elements. Biol. Chem. 379, 611-616. Studier, F. W., Rosenberg, A. H., Dunn, J. J. & Dubendorff, J. W. (1990). Use of T7 RNA polymerase to direct expression of cloned genes. Methods Enzymol. 185, 60-89. Sutherland, E., Coe, L. & Raleigh, E. A. (1992). McrBC: a multisubunit GTP-dependent restriction endonuclease. J. Mol. Biol. 225, 327-348.
Edited by J. Karn (Received 13 April 1999; received in revised form 9 August 1999; accepted 9 August 1999)
J. Mol. Biol. (1999) 292, 547±556
The GTP-binding Domain of McrB: More Than Just a Variation on a Common Theme? Uwe Pieper*, Thomas Schweitzer, Detlef H. Groll, Frank-Ulrich Gast and Alfred Pingoud Institut fuÈr Biochemie, JustusLiebig-UniversitaÈt Giessen Heinrich-Buff-Ring 58 D-35392 Giessen, Germany
The methylation-dependent restriction endonuclease McrBC from Escherichia coli K12 cleaves DNA containing two RmC dinucleotides separated by about 40 to 2000 base-pairs. McrBC is unique in that cleavage is totally dependent on GTP hydrolysis. McrB is the GTP binding and hydrolyzing subunit, whereas MrC stimulates its GTP hydrolysis. The C-terminal part of McrB contains the sequences characteristic for GTP-binding proteins, consisting of the GxxxxGK(S/T) motif (position 201-208), followed by the DxxG motif (position 300-303). The third motif (NKxD) is present only in a non-canonical form (NTAD 333-336). Here we report a mutational analysis of the putative GTP-binding domain of McrB. Amino acid substitutions were initially performed in the three proposed GTP-binding motifs. Whereas substitutions in motif 1 (P203V) and 2 (D300N) show the expected, albeit modest effects, mutation in the motif 3 is at variance with the expectations. Unlike the corresponding EFTu and ras-p21 variants, the D336N mutation in McrB does not change the nucleotide speci®city from GTP to XTP, but results in a lack of GTPase stimulation by McrC. The ®nding that McrB is not a typical G protein motivated us to perform a search for similar sequences in DNA databases. Eight microbial sequences were found, mainly from un®nished sequencing projects, with highly conserved sequence blocks within a presumptive GTP-binding domain. From the ®ve sequences showing the highest homology, 17 invariant charged or polar residues outside the classical three GTP-binding motifs were identi®ed and subsequently exchanged to alanine. Several mutations speci®cally affect GTP af®nity and/or GTPase activity. Our data allow us to conclude that McrB is not a typical member of the superfamily of GTP-binding proteins, but de®nes a new subfamily within the superfamily of GTP-binding proteins, together with similar prokaryotic proteins of as yet unidenti®ed function. # 1999 Academic Press
*Corresponding author
Keywords: restriction enzymes; McrBC restriction; DNA binding; GTPase; site-directed mutagenesis
Introduction In Escherichia coli K12, the detection and destruction of foreign DNA is accomplished by at least Abbreviations used: BSA, bovine serum albumin; GST, glutathione S-transferase; GTP, guanosine 50 triphosphate; ITP, inosine 50 -triphosphate; Kass, association constant; KD, dissociation constant; NTP, nucleotide 50 -triphosphate; wt, wild-type; XTP, xanthosine 50 -triphosphate; ORF, open reading frame. E-mail address of the corresponding author: [email protected] 0022-2836/99/380547±10 $30.00/0
four restriction activities. One of them is the McrBC restriction endonuclease which acts against cytosine-modi®ed (methylated or hydroxymethylated) DNA. The McrBC system consists of two protein subunits McrB (53 kDa) and McrC (39 kDa) which are encoded by two slightly overlapping genes (Raleigh et al., 1989; Ross et al., 1989; Dila et al., 1990). The enzymatic properties of McrBC have been investigated in detail in recent years. McrBC recognizes and cleaves DNA containing two hemi or fully methylated RmC sites (regardless of orientation) in an optimal distance of about 40 to 80 # 1999 Academic Press
548 base-pairs (Sutherland et al., 1992) and to a lesser degree in a distance of up to 2000 base-pairs (Stewart & Raleigh, 1998). DNA binding, but not cleavage, can take place even at sites with more narrowly separated half-sites (Gast et al., 1997). The McrB subunit is responsible for the binding of DNA and GTP (KruÈger et al., 1995; Gast et al., 1997; Pieper et al., 1997), whereas the function of McrC, aside from the stimulation of the GTPase activity of McrB, is as yet unknown. In spite of its basic character, McrC does not bind directly to DNA, but to McrB, leading to complexes of higher molecular mass (KruÈger et al., 1995; Gast et al., 1997). The signature sequences of GTP-binding proteins (Dever et al., 1987; Bourne et al., 1991) which are involved in GTP interaction as con®rmed by numerous X-ray crystal structures (for a review, see Kjeldgaard et al., 1996), consist of: (1) the phosphate binding loop GxxxxGK(S/T); (2) the switch II region DxxG; and (3) the guanine recognition loop, NKxD. Based upon the amino acid sequence, Dila et al. (1990) proposed that McrB is the GTPbinding and hydrolyzing subunit, because its C-terminal part contains the characteristic motifs, starting at position 201 (numbered from the second start codon; see Materials and Methods) with the P-loop GxxxxGK(S/T), followed at position 300 by the DxxG motif. The third motif (NKxD) is present only in a non-canonical form (NTAD 333-336). By mutating Asn333 and Asn370 to alanine we asked whether NTAD (333-336) or an alternative candidate motif, NKKA (370-373) represents motif 3. We found that the exchange Asn333 to Ala completely abolished all McrB activities connected to GTP binding, whereas the variant N370A was indistinguishable from the wild-type enzyme. We therefore concluded that the NTAD (333-336) region is involved in GTP binding (Pieper et al., 1997). The DNA-binding domain of McrB has been shown by us to reside within the N-terminal 184 amino acid residues (Gast et al., 1997) and to be devoid of GTP-binding activity. Thus, the inference that the GTP-binding domain is in the C-terminal part of the molecule seemed to be justi®ed. We have proven this assumption experimentally by investigating a deletion mutant lacking the N-terminal 162 amino acid residues (Pieper et al., 1999) which is almost identical with McrBs, an additional product of McrB gene expression resulting from an internal translation start at codon 162. It has been demonstrated recently (Beary et al., 1997; Panne et al., 1998) that McrBs modulates McrBC restriction activity, presumably by sequestration of McrC. We have found that the McrB fragment consisting of amino acids 163 to 459 interacts with GTP and McrC in a manner identical to wild-type McrB (Pieper et al., 1999). Here we report the properties of a series of McrB variants with amino acid substitutions in the GTPbinding domain. The ®rst mutations were targeted to the three parts of the inferred GTP-binding motif in order to test the assumption that these motifs are involved in GTP binding, extending the
GTP-binding Domain of McrB
work described by Pieper et al. (1997). Database searches in microbial genomic sequences, especially un®nished ones, revealed a number of open reading frames (ORFs) which share several highly conserved sequence blocks with E. coli McrB. Interestingly, these homologous regions reside entirely in the C-terminal GTP-binding part of McrB. This ®nding allowed us to broaden our search for amino acid residues involved in guanine nucleotide and/or McrC interaction outside the classical G protein motifs. Based on a multiple alignment, we performed an alanine scanning mutagenesis of all 17 polar amino acid residues strictly conserved in E. coli McrB and the ®ve most similar sequences. Our results show that McrB is not a typical member of the superfamily of GTPbinding proteins, but may de®ne a new subfamily.
Results Amino acid substitutions in motifs 1 and 2: P203V and D300N In the phosphate-binding loop, Pro203 was exchanged to Val. We chose the variable position 3 in the GxxxxGK(S/T) motif because mutations in this region affect the GTPase activity in other GTPbinding proteins. V20G in E. coli elongation factor Tu (Jacquet & Parmeggiani, 1988) and G12V in rasp21 (Schmidt et al., 1996) reduce the GTP hydrolysis rate, leading to an oncogenic variant in the latter case. In the DxxG motif, we chose the invariant Asp300 which we exchanged rather conservatively to Asn. An analogous substitution in elongation factor Tu (D80N) weakens GTP binding and enhances GTP hydrolysis (Harmark et al., 1992). The McrB variants P203V and D300N still cleave DNA, albeit slightly slower than the wild-type enzyme. Speci®c DNA recognition, tested by gel retardation assays with methylated oligonucleotides, is unaffected by both mutations (data not shown). The GTP-related activities are listed in Table 1. The af®nities for GTP of both variants are reduced by a factor of three and ®ve, respectively. The intrinsic GTPase rate of the P203V variant is slowed down by a factor of four. The D300N substitution leaves the intrinsic GTPase rate unaffected. In both variants GTP hydrolysis is still stimulated by McrC, although slightly less than in wild-type McrB. Amino acid substitution in motif 3: D336N, an attempt to change nucleotide specificity As seen in several X-ray crystallographic studies (for a review, see Kjeldgaard et al., 1996), the invariant aspartate of the NKxD motif interacts via hydrogen bonding to the exocyclic 2-amino group and the N1-imino group of the purine ring, thereby determining guanine speci®city. Substitution of this aspartate residue by an asparagine residue changes speci®city from guanine to xanthine, as was demonstrated for EF-Tu (Hwang & Miller,
549
GTP-binding Domain of McrB Table 1. GTP-related properties of the McrB variants P203V and D300N, compared to wild-type McrB 6
ÿ1
GTP binding constant Kass (10 M ) % of wild-type GTP affinity Intrinsic GTPase activity (Pi released/McrB per h) % of wild-type GTPase activity McrC-stimulated GTPase activity (Pi released/McrB per hour) Factor of stimulation by McrC
McrB wt
P203V
D300N
2.09 100 41 100
0.74 36 10 24
0.42 20 53 130
607 14.8
82 8.3
288 5.4
1987) and several other GTP-binding proteins, e.g. ras-p21 (Schmidt et al., 1996), Rab5 (Hoffenberg et al., 1995), and FtsY (Powers & Walter, 1995). A similar speci®city change caused by substitution of Asp336 by Asn in McrB would be a clear indication that the proposed assignment of GTP-binding motif 3 is correct. The enzymatic properties of the corresponding McrB variant D336N are rather unexpected. Figure 1(a) shows DNA cleavage experiments with wild-type McrB and the D336N variant in the presence of GTP, XTP or both nucleotides. Unexpectedly, McrB D336N does not exhibit any DNA cleavage activity with either nucleotide. In contrast, wild-type McrB is almost as active with XTP as with GTP. As Figure 1(b) clearly demonstrates, the inability of the D336N variant to cleave DNA is not caused by a loss of GTP af®nity. Even the relative af®nities of both proteins to other purine nucleotides, including XTP, are very similar, as tested by competition experiments (Figure 2). The binding constants for ITP, XTP, and ATP are summarized in Table 2. XTP is bound two orders of magnitude weaker than GTP. This is still suf®cient for saturating McrB when the nucleotide is present in millimolar amounts, as is the case in the DNA cleavage experiments. ITP is bound even more strongly than XTP, i.e. only ten times more weakly than GTP, and hence also supports DNA cleavage by wild-type McrBC (Figure 3). ATP shows the lowest af®nity to McrB among the nucleotides tested, with a KD value in the millimolar range. Even this weak af®nity would not exclude utilization in DNA cleavage, if nucleotide binding was the only requirement for DNA cleavage. Figure 3, however, shows that even when present at 5 mM concentration, ATP does not support cleavage, con®rming earlier results by KruÈger et al. (1992) and Sutherland et al. (1992). Intriguingly, ATP is hydrolyzed by McrB with a rate comparable to that of other purine nucleotides (data not shown). This means that while GTP, ITP, XTP and ATP are all bound and hydrolyzed by McrB, only GTP, ITP and XTP, but not ATP, serve as allosteric effectors for DNA cleavage by McrBC. We tested whether GTP hydrolysis, rather than binding, might be affected by the D336N substitution, the results of which are shown in Figure 4. As reported previously (Pieper et al., 1997), wild-type McrB hydrolyzes GTP in a multiple-turnover reaction with a rate of about 40
per hour, which is enhanced by more than an order of magnitude when McrC is present in at least equimolar amounts. The variant D336N is still able to hydrolyze GTP (30 % of wild-type activity; the GTP hydrolysis rate is constant for over two hours which shows that D336N is not less stable). However, the GTPase activity of D336N cannot be stimulated by McrC, which may explain why this variant is inactive in DNA cleavage.
Figure 1. Properties of the D336N variant of McrB. (a) DNA cleavage experiments. Cleavage reactions of methylated plasmid DNA (EcoRI-linearized pBW201) were carried out for one hour at 25 C without nucleotide or in the presence of 1 mM GTP, XTP, or both. (b) GTP binding isotherms of D336N (*) and wild-type McrB (*) as determined by nitrocellulose ®lter binding. Active concentration of McrB was 0.55 mM. Binding constants were calculated by non-linear least-squares ®tting: Kass 2.1 106 Mÿ1 for D336N and 2.2 106 Mÿ1 for wild-type McrB.
550
GTP-binding Domain of McrB
Figure 2. Nucleotide speci®city of (a) wild-type and (b) D336N McrB. Relative af®nities for ATP (*), ITP ( & ) and XTP (~) were determined in competition experiments with 0.67 mM GTP and up to 6 mM competitor. Active protein concentrations were 0.28 mM for wild-type and 0.5 mM for D336N. Theoretical curves were calculated by non-linear least-squares ®tting. The order of af®nity for purine nucleotides is GTP > ITP > XTP > ATP for wild-type McrB and the D336N variant.
Figure 3. Nucleotide speci®city of DNA cleavage by McrB. The time course of the cleavage of pBW201 plasmid DNA (pre-cleaved with EcoRI) by 1 mM McrB with 5 mM of different purine nucleotides (ATP, GTP, ITP and XTP) in the time range of 0, 3, 10, 30, 100 minutes is shown. The right lanes contain substrate plasmid (S). While GTP, ITP, and XTP at 5 mM support DNA cleavage by McrB, ATP does not.
Sequence database search for McrB homologues The similar af®nity and preference of wild-type McrB and the D336N variant for nucleotide triphosphates allow us to conclude that the guanine speci®city of McrB is determined in a different manner than in ``canonical'' GTP-binding proteins. Either the consensus structure can also be formed with a non-canonical sequence instead of the canonical NKxD motif, or the GTP binding pocket is structured differently as compared to classical G proteins.
Table 2. Af®nities of wild-type McrB and McrB D336N for purine nucleotides NTP GTP ITP XTP ATP
Kass (Mÿ1)
wt 6
2 10 5.2 105 1.1 104 1.9 103
D336N 2 106 5.9 105 9.6 103 5.8 102
A Lactococcus lactis ORF with similarity to E. coli McrB has been reported by O'Sullivan et al. (1995). This sequence, named LlaI.2, is part of the plasmid-encoded restriction-modi®cation system LlaI. To test whether other proteins homologous to McrB exist, we performed a search of the SWISSPROT protein sequences database, which yielded no sequence with signi®cant homology to McrB. Depending on the search criteria, more than ten additional DNA sequences appear when the search is extended to ®nished or un®nished microbial genome sequencing projects with TBLASTN (Altschul et al., 1997). Figure 5 shows a multiple alignment of McrB and the ®ve sequences with the most signi®cant degree of homology to McrB (TBLASTN score > 100). The schematic view in Figure 5(a) indicates that the regions of high similarity cluster in the central to C-terminal part of McrB, i.e. in the GTP binding region, whereas no signi®cant similarities are found in the N-terminal DNA-binding domain. Figure 5(b) gives a detailed view of the highly conserved sequence blocks. GTP-binding motif 1 (GxxxxGK(S/T)) is perfectly conserved. In contrast,
551
GTP-binding Domain of McrB
Alignment-guided mutagenesis of McrB The lack of conservation of motif 2 in homologous sequences and the high conservation of blocks outside the classical GTP-binding motifs gives additional evidence that the nucleotide-binding mode of McrB differs from that of canonical GTPases. Some of these strictly conserved residues should be involved in nucleotide interaction. Therefore, we mutated all the conserved acidic (®ve), basic (eight), and uncharged polar (four) residues to alanine. The resulting variants listed in Table 3 were expressed as N-terminally His6tagged proteins. While most of the variants could be obtained as soluble protein in concentrations up to 10 mM, H233A, E292A, and R302A were completely insoluble, maybe due to incorrect folding. The remaining 14 variants were tested for their activities in GTP binding and hydrolysis as well as DNA binding and cleavage in the presence of McrC (Table 3 and Figure 6). Some of the variants (S287A, K288A, R337A) behave like wild-type McrB or are completely inactive in all aspects (D279A, E280A), which can be explained by drastic conformational changes induced by these mutations. Most interesting are the variants in which partial functions are affected. Some still bind to, but do not cleave DNA in the presence of McrC (R283A, R349A), probably due to their lost GTP binding capacity. N282A still binds GTP, but does not interact with DNA. This amino acid substitution seems to exert a long-range effect on the N-terminal DNA-binding domain. D343A and R347A appear to be inactive in the GTP binding assay, but nevertheless cleave DNA. This apparent contradiction can be explained by the different GTP concentrations in the respective assays (1 mM for DNA cleavage, 10 mM for GTP binding, which is 20-fold higher than the KD for wild-type McrB-GTP). The GTPase activity of McrB R347A, tested at 5 mM GTP, was more than half of the wild-type rate (data not shown), indicating
Figure 4. Time course of GTP hydrolysis by 0.29 mM active McrB wild-type (&, & ) and 0.16 mM D336N (*, *) in the absence (open symbols) and presence (®lled symbols) of 5 mM McrC. GTP concentration was 10 mM. While the relatively strong GTPase activity of wild-type McrB is stimulated by McrC, the relatively weak GTPase activity of the D336N variant is not.
motif 2 (DxxG), which is present in McrB, cannot be found in the homologous sequences. Instead, the two basic residues occupying the variable positions in E. coli McrB are invariant throughout the six sequences. The ®rst and the last residue of motif 3 (NKxD) are perfectly conserved, but the canonical lysine residue at the second position does not show up in any of the sequences. Instead, the threonine residue which occupies this position in McrB is found in four of the ®ve homologues. Thus, the functional signi®cance of residues N333 and D336 demonstrated by site-directed mutagenesis by Pieper et al. (1997) and in this communication is also re¯ected by their high degree of conservation.
Table 3. Summary of McrB alanine mutagenesis variants McrB variant N196A H233A D279A E280A N282A R283A S287A K288A E292A R302A N333A D336A R337A D343A R347A R348A R349A
Solubility
DNA binding
GTP binding
DNA cleavage
ÿ ÿ ÿ
n.d. ÿ ÿ ÿ n.d. n.d.
n.d. ÿ ÿ ÿ n.d. n.d. n.d. n.d. ÿ ÿ n.d. ÿ
n.d. ÿ ÿ ÿ ÿ n.d. n.d. ÿ ÿ ÿ ÿ
552
GTP-binding Domain of McrB
Figure 5. Multiple alignment of E. coli McrB with the ®ve most similar microbial ORF sequences: Eco, Escherichia coli K12; Smu, Streptococcus mutans; Cje, Campylobacter jejuni NCTC 11168; Ype, Yersinia pestis; Spn, Streptococcus pneumoniae; and Pgi, Porphyromonas gingivalis W83. (a) The positions of ®ve highly conserved sequence blocks in the respective amino acid sequences. (b) These regions are shown in detail. Invariant residues are indicated under the alignment in capital letters. Above the alignment the inferred GTP-binding motifs are shown.
that, although GTP binding is affected, the GTPase centre is largely intact. Variants with wild-type-like GTP af®nity were also tested for their GTPase activity in the absence
Figure 6. Intrinsic and McrC-stimulated GTP hydrolysis of selected McrB variants. Of the 17 variants listed in Table 3, N196A, N282A, S287A, K288A, and R337A show considerable GTP binding and were therefore tested for their GTPase activity. Also shown are the speci®c GTPase activities of P203V, D300N and D336N.
and presence of McrC. The results are shown in Figure 6, together with the results obtained for variants P203V, D300N and D336N mentioned above. N196A, S287A, and K288A hydrolyze GTP with a slightly reduced intrinsic rate, and GTPase activity is still stimulated by McrC, albeit to a lesser extent. Thus, these positions are not crucial for McrB function, as seen from their DNA cleavage activity. The most pronounced effects regarding GTP hydrolysis are shown by N282A and R337A. The intrinsic GTPase rate of N282A is unaltered, but cannot be stimulated by McrC. In this respect, N282A behaves similarly to D336N. The basal GTPase rate of R337A is already increased threefold, such that the small additional McrC stimulation leads to an almost wild-type GTPase rate. Like wild-type McrB and all other variants, the R337A variant needs McrC to cleave DNA (data not shown).
Discussion McrBC is the only restriction endonuclease known to be GTP-dependent (for reviews, see Raleigh, 1992; Bickle, 1993). GTP binding and
GTP-binding Domain of McrB
hydrolysis are accomplished by the McrB subunit which is also responsible for DNA recognition (KruÈger et al., 1995; Pieper et al., 1997; Gast et al., 1997). The McrC subunit might be responsible for DNA cleavage, but this has not yet been demonstrated unequivocally. We have shown recently that the DNA and GTP-binding functions of McrB are located on different domains which can be separated from each other without losing their respective partial functions. The N-terminal ca 160 amino acid residues harbor the DNA-binding region, and a deletion mutant lacking this part exhibits wild type-like GTP-binding and GTPase activity (Pieper et al., 1999). These ®ndings are in accordance with the identi®cation of GTP-binding motifs in this part of McrB by Dila et al. (1990). The assumption that these sequence motifs in McrB are involved in GTP binding in the same way as in other GTP-binding proteins implies that the GTP-binding domain of McrB is structurally similar to the GTP-binding domain of proteins belonging to the G protein superfamily whose members share an almost invariant core structure (for a review, see Kjeldgaard et al., 1996). There are two problems with this inference (both regarding the third motif, NKxD; Pieper et al., 1997). (1) The conserved lysine residue at position 2 is replaced by a threonine residue in McrB, and (2) the distance between motif 2 and 3 is unusually small in McrB (29 amino acid residues) compared to other GTP-binding proteins in which this distance is at least 49 amino acid residues. Thus, McrB must deviate somehow from the GTPase consensus structure in this region, even though N333 from motif 3 is absolutely required for GTP binding and hydrolysis. Here we present a much more thorough mutagenesis study of the GTP-binding domain of McrB. We started with amino acid substitutions in all three GTP-binding motifs in which distinct effects have already been described for other GTP-binding proteins. Mutagenesis of Pro203, at the third position in the GxxxxGK(S/T) motif, had the same effect as G12V in ras-p21 (Schmidt et al., 1996), and V20G in EF-Tu (Jaquet & Parmeggiani, 1988), namely a moderately reduced GTPase activity. Thus, motif 1 seems to have the canonical function in McrB. Conservative substitution of the conserved motif 2 aspartate to asparagine residue (D300N) does not lead to an increase in GTPase activity, as opposed to the analogous substitution D80N in EF-Tu (Harmark et al., 1992); instead, both McrB variants exhibit a reduced binding constant for GTP. Thus, DKRG (300-303) might not be the equivalent of motif 2 in McrB. In the third GTP-binding motif (NTAD 333-336 in McrB), no reversal of nucleotide speci®city from GTP to XTP is observed with McrB D336N, as opposed to several GTP binding proteins (e.g. D138N in E. coli EF-Tu; Hwang & Miller, 1987). D336N is identical with wild-type McrB with respect to GTP binding and relative af®nities to different purine nucleotides (XTP, ITP, ATP). This means that Asp336
553 does not determine nucleotide speci®city, but is rather functionally important in McrC interaction, such that the identi®cation of NTAD (333-336) as the guanine speci®city-determining GTP-binding motif 3 must be questioned. Thus, the proposal that McrB constitutes a distinct subclass of the GTPase superfamily is supported by the results presented here. Further con®rmation of this view comes from the open reading frames with signi®cant homology to E. coli McrB (although some of the un®nished sequences may not be correct in every detail). The homologous regions cluster in several blocks in the central to C-terminal part of McrB. No similarity whatsoever can be found in the DNA-binding domain. Apparently, the GTP-binding module of McrB was combined with a variety of insertions or N and C-terminal domains (see Figure 5(a)). As only motif 1 (the P-loop) is functionally signi®cant in McrB and conserved among the McrB-like proteins, other residues, most likely the strictly conserved amino acids revealed by the multiple alignment, must be involved in GTP binding and hydrolysis. Lacking any structural information, we chose all conserved polar residues for an alanine scanning mutagenesis. Of the functionally signi®cant mutations identi®ed thus far, the variants impaired in GTP binding (D279A, E280A, R283A, R347A, and R349A) are of special interest. The two acidic residues are obvious candidates for further mutational analysis because they potentially could have a similar function in guanine speci®city determination as the conserved aspartate of motif 3 in canonical GTP-binding proteins. DNA cleavage by McrB is dependent on McrCstimulated GTP hydrolysis. None of the variants tested change this strict coupling, not even R347A. This variant, with a GTP af®nity below the sensitivity of the ®lter binding assay, is active in DNA restriction in the presence of millimolar amounts of GTP. Two variants were altered in their GTPase properties, but were unaltered in their af®nity for GTP. (1) N282A has the same intrinsic GTPase activity as the wild-type enzyme, but cannot be stimulated in its GTPase activity by McrC and therefore cannot cleave DNA, much like D336N. (2) R337A is the only variant with a higher intrinsic GTPase activity, but on the expense of the stimulatory effect by McrC. Maybe this protein approaches the conformation which is normally induced upon McrC binding. In conclusion, we have shown that the GTPbinding domain of McrB is a structural and functional module which is different from the classic GTP-binding proteins and which is also present in several microbial proteins of presently unknown function. While our present mutagenesis study is purely phenomenological, only X-ray crystallography will tell whether the structure of the GTPbinding domain of McrB is a variation on a common theme or represents a fundamentally new fold.
554
GTP-binding Domain of McrB
Materials and Methods Bacterial strains, plasmids, and media For the expression of McrB and McrC as glutathione S-transferase (GST) fusion proteins and for the preparation of methylated substrate plasmids, the methylation-tolerant E. coli strain TC410 [mcrAÿ, (mrrhsdRMS-McrBC)201, minA, minB, rpsL, Sup] (NoyerWeidner et al., 1986; KruÈger et al., 1992) was used. GSTMcrB and GST-McrC were encoded on the expression plasmids pBN211 and pBN213, respectively (KruÈger, 1992; KruÈger et al., 1995). The vector pBBImcrB for the overexpression of His6-tagged McrB was constructed as follows. pBBInucA (Meiss et al., 1998), a descendant of pET-3d (Studier et al., 1990), was cleaved with NdeI and BamHI. The McrB gene, ampli®ed by PCR with primers introducing the appropiate restriction sites (NdeI N-terminally; primer: 50 -GATACATATGGAATCTATTCAACCCTG and BamHI C-terminally; primer: 50 -ATATGGATCCTATGAGTCCCCTAATAAT), was inserted into the large fragment. The resulting plasmid was transformed into E. coli BL21(DE3)pLysS (Studier et al., 1990). Cells were grown in LB medium with 75 mg/ml ampicillin. In the case of BL21(DE3)pLys, 20 mg/ml chloramphenicol was added. Site-directed mutagenesis and protein purification The mcrB gene contains two potential translation start codons 18 bp apart (Dila et al., 1990; Ross et al., 1990), leading to protein sizes of 465 or 459 amino acid residues. Because of a recent change (in release 37) of the SWISSPROT database entry (P15005) from the ®rst to the second start codon, McrB residue numbers given here are smaller by six compared to the numbering in previous publications (e.g. see Pieper et al., 1997). Three McrB variants were generated as GST fusion proteins expressed by the vector pBN211. D300N was generated by a three-primer PCR method described by Ito et al. (1991) and already used for the McrB variants described by Pieper et al. (1997). The mutagenesis primer was 50 -TGGCGAAGTCATGATGTTAATGGAACATAATAAACGA. P203V and D336N were generated by inverse PCR essentially as described by Meiss et al. (1998). The primers used to introduce the desired codon exchange and a silent marker restriction endonuclease site were: P203V/forward, 50 -CTATACCGGTCGGCGTTGGAAAAACCTTTG-30 ; P203V/reverse, 50 -ATATACCGGTCCCTGGAGGATAATATTTTTTTTGA-30 ; D336N/forward, 50 -CTTTAATCGATCTCTGGCCGT0 TGTTGACTATG-3 ; and D336N/reverse, 50 -CTCTATCGATTGGCAGTATTCATTAAACCG-30 . All other variants were generated by a protocol described by Kirsch & Joly (1998), starting from the His6-McrB expression vector pBBImcrB. The products were then directly transformed into E. coli LK111(l) cells. The mutagenic primers used were: N196A, 50 -CCATCAAAAAAGATATCATCCTCCAG; H233A, 50 -CAATATGGTCCAGTTCCCTCAATCTTAT; D279A, 50 -TATTATAGCTGAAATTAATCGTGCCAAT; E280A, 50 -TATTATCGATGCAATCAATCGTGCCAAT; N282A, 50 -TATTATCGATGAAATCGCTCGTGCCAAT; R283A, 50 -TATTATAGATGAAATTAATGCTGCCAAT; S287A, 5 0 -CAATCGTGCCAATCTCGCGAAAGTATT;
K288A, 5 0 -CAATCGAGCCAATCTCAGTGCAGTATT; E292A, 5 0 -GTATTTGGCGCAGTCATGATGTTAAT; 0 R302A, 5 -GATAAAGCAGGTGAAAACTGGTCTGTT; N333A, 5 0 -GGTTTAATGGCTACTGCAGATCGCTCT; D336A, 50 -GAATACTGCAGCTCGCTCTCTGGCCGTT; R337A, 50 -GAATACTGCAGATGCCTCTCTGGCCGTT; D343A, 50 -CCGTTGTTGCCTATGCCCTGCGCAGACGAT; R347A, 50 -CCGTTGTCGACTATGCCCTAGCCAGACGAT; R348A, 50 -GCCCTACGCGCGCGATTTTCTTTCAT; R349A, 50 -CGCAGAGCATTTTCTTTCATCGATATT; and The mcrB gene of positive clones was sequenced on both strands using the Taq cycle sequencing protocol (Perkin-Elmer) and an Applied Biosystems 373A sequencer. GST-McrB, GST-McrC and the McrB variants P203V, D300N and D336N were expressed and puri®ed as described by Pieper et al. (1997). For His6-tagged McrB (wild-type and variants) a small-scale (250 ml) Ni-NTAagarose (Qiagen) batch procedure from 50 ml cultures of E. coli BL21(DE3)pLysS cells was used according to manufacturer's instructions. The standard buffer contained 20 mM Hepes-KOH (pH 7.6), 1 mM EDTA, 50 mM KCl, and 10 % (v/v) glycerol. The His6-tagged proteins were eluted with 300 ml of 200 mM imidazole in standard buffer and stored at ÿ20 C. Protein concentration was determined by the method described by Bradford (1976) with bovine serum albumin (BSA) as standard or by absorbance at 280 nm with a molar extinction coef®cient of e 74,720 Mÿ1 cmÿ1 (wild-type McrB) calculated by the method described by Pace et al. (1995). GTP binding, hydrolysis and DNA cleavage assays Assays were performed as described by Pieper et al. (1997). Binding of [g-32P]GTP (DuPont-NEN) to McrB was measured by the nitrocellulose ®lter binding assay. If not indicated otherwise, the concentrations were 0.5 to 1 mM protein and 10 mM radioactive GTP in a ®nal volume of 20 ml assay buffer (20 mM Hepes (pH 7.6), 50 mM KCl, 5 mM MgCl2, 0.1 mg/ml BSA). GTP hydrolysis was assayed by the charcoal method which detects the amount of radioactive inorganic phosphate liberated after incubation with [g-32P]GTP. Standard conditions were 1 mM McrB, 10 mM GTP/[g-32P]GTP in a reaction volume of 20 ml assay buffer at 25 C. Alternatively, McrC-stimulated NTP hydrolysis was monitored by a colorimetric phosphate assay (Jenkins & Marshall, 1984). DNA cleavage activity of McrB was assayed with EcoRI-linearized pBW201, a plasmid harboring the BsuFI methylase gene leading to in vivo methylation at the ®rst cytosine of CCGG and thus creating McrB recognition sites (KruÈger et al., 1995). Electrophoretic mobility shift assay Gel shift experiments were performed essentially as described by KruÈger et al. (1995) and by Gast et al. (1997) in standard buffer. In a total volume of 10 ml, protein concentrations around 0.5 mM and either 0.5 mM (no radioactive labelling) or 0.01 mM (radioactive labelling) oligodeoxynucleotide was used. For the purpose of stabilization of McrB and McrC, all binding experiments were carried out in the presence of 0.1-0.2 mg/ml BSA. The DNA used in the non-radioactive assays was a 61mer double-strand with the sequence:
50 -AATTCTAAGACCGGTAGCGAGCTCGTATGATATCATATTAATCGGTAAGACCGGTAGCGAG GATTCTGGCCATCGCTCGAGCATACTATAGTATAATTAGCCATTCTGGCCATCGCTCTTAA
GTP-binding Domain of McrB
The 5-methylcytosine bases are underlined. In the radioactive assays a 44-mer double strand oligonucleotide was used (MH1 dimer; described by Gast et al., 1997). Database sequence search and alignment Sequence similarity searches were performed with the BLAST program (Altschul et al., 1997). For nucleotide sequence databases, TBLASTN, and for protein sequences, BLASTP was used with standard parameters. Searches were done on the web sites of NCBI (www.ncbi.nlm.nih.gov) and Infobiogen (www.infobiogen.fr). Preliminary sequence data for Streptococcus pneumoniae and Porphyromonas gingivalis W83 was obtained from the Institute for Genomic Research (TIGR; www.tigr.org), for Campylobacter jejuni NCTC 11168 and Yersinia pestis from the Sanger Centre (www.sanger.ac.uk), and for Streptococcus mutans UAB159 from the University of Oklahoma's Advanced Center for Genome Technology (ACGT; www.genome.ou.edu). Multiple alignments were performed with the program MAP (Huang, 1994).
Acknowledgments We are grateful to Drs M. Noyer-Weidner and T. KruÈger for expression plasmids for McrB and McrC. This work was supported by the Deutsche Forschungsgemeinschaft (Pi 122/11-2).
References Altschul, S. F., Madden, Z., SchaÈffer, A. A., Zhang, J., Zhang, Z., Miller, W. & Lipman, D. J. (1997). Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucl. Acids Res. 25, 3389-3402. Beary, T. P., Braymer, H. D. & Achberger, E. C. (1997). Evidence of participation of McrB(S) in McrBC restriction in Escherichia coli K-12. J. Bacteriol. 179, 7768-7775. Bickle, T. (1993). The ATP-dependent restriction enzymes. In Nucleases (Linn, S. M., Lloyd, R. S. & Roberts, R. J., eds), pp. 89-109, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Bourne, H. R., Sanders, D. A. & McCormick, F. (1991). The GTPase superfamily: conserved structure and molecular mechanism. Nature, 349, 117-127. Bradford, M. M. (1976). A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein dye binding. Anal. Biochem. 72, 248-254. Dever, T. E., Glynias, M. J. & Merrick, W. C. (1987). GTP-binding domain: three consensus sequence elements with distinct spacing. Proc. Natl Acad. Sci. USA, 84, 1814-1818. Dila, D., Sutherland, E., Moran, L., Slatko, B. & Raleigh, E. A. (1990). Genetic and sequence organization of the mcrBC locus of Escherichia coli K-12. J. Bacteriol. 172, 4888-4900. Gast, F. U., Brinkmann, T., Pieper, U., KruÈger, T., Noyer-Weidner, M. & Pingoud, A. (1997). The
555 recognition of methylated DNA by the GTP-dependent restriction endonuclease McrBC resides in the N-terminal domain of McrB. Biol. Chem. 378, 975982. Harmark, K., Anborgh, P. H., Merola, M., Clark, B. F. C. & Parmeggiani, A. (1992). Substitution of aspartic acid-80, a residue involved in coordination of magnesium, weakens the GTP binding and strongly enhances the GTPase of the G domain of elongation factor Tu. Biochemistry, 31, 7367-7372. Hoffenberg, S., Nikolova, L., Pan, J. Y., Daniel, D. S., Wessling-Resnick, M., Knoll, B. J. & Dickey, B. F. (1995). Functional and structural interactions of the Rab5 D136N mutant with xanthine nucleotides. Biochem. Biophys. Res. Commun. 215, 241-249. Huang, X. (1994). On global sequence alignment. Comput. Appl. Biosci. 10, 227-235. Hwang, Y. W. & Miller, D. L. (1987). A mutation that alters the nucleotide speci®city of elongation factor Tu, a GTP regulatory protein. J. Biol. Chem. 262, 1308-13085. Ito, W., Ishiguro, H. & Kurosawa, Y. (1991). A general method for introducing a series of mutations into cloned DNA using the polymerase chain reaction. Gene, 102, 67-70. Jacquet, E. & Parmeggiani, A. (1988). Structure-function relationships in the GTP binding domain of EF-Tu: mutation of Val20, the residue homologous to position 12 in p21. EMBO J. 7, 2861-2867. Jenkins, W. T. & Marshall, A. M. (1984). A modi®ed direct phosphate assay for studying ATPases. Anal. Biochem. 141, 155-160. Kirsch, R. D. & Joly, E. (1998). An improved PCR-mutagenesis strategy for two-site mutagenesis or sequence swapping between related genes. Nucl. Acids Res. 26, 1848-1850. Kjeldgaard, M., Nyborg, J. & Clark, B. F. C. (1996). The GTP binding motif: variations on a theme. FASEB J. 10, 1347-1368. È berproduktion, Reinigung und CharKruÈger, T. (1992). U akterisierung der Untereinheiten des McrBCRestriktionssystems von Escherichia coli K-12, PhD thesis, Freie UniversitaÈt, Berlin. KruÈger, T., Grund, C., Wild, C. & Noyer-Weidner, M. (1992). Characterization of the mcrBC region of Escherichia coli K-12 wild-type and mutant strains. Gene, 114, 1-12. KruÈger, T., Wild, C. & Noyer-Weidner, M. (1995). McrB: a prokaryotic protein speci®cally recognizing DNA containing modi®ed cytosine residues. EMBO J. 14, 2661-2669. Meiss, G., Franke, I., Gimadutdinow, O., Urbanke, C. & Pingoud, A. (1998). Biochemical characterization of Anabaena sp. strain PCC 7120 non-speci®c nuclease NucA and its inhibitor NuiA. Eur. J. Biochem. 251, 924-934. Noyer-Weidner, M., Diaz, R. & Reiners, L. (1986). Cytosine-speci®c DNA modi®cation interferes with plasmid establishment in Escherichia coli K12: involvement of rgl B. Mol. Gen. Genet. 205, 469-475. O'Sullivan, D. J., Zagula, K. & Klaenhammer, T. R. (1995). In vivo restriction by LlaI is encoded by three genes, arranged in an operon with llaIM, on the conjugative Lactococcus plasmid pTR2030. J. Bacteriol. 177, 134-143.
556
GTP-binding Domain of McrB
Pace, C. N., Vajdos, F., Fee, L., Grimsley, G. & Gray, T. (1995). How to measure and predict the molar absorption coef®cient of a protein. Protein Sci. 4, 2411-2425. Panne, D., Raleigh, E. A. & Bickle, T. A. (1998). McrBs, a modulator peptide for McrBC activity. EMBO J. 17, 5477-5483. Pieper, U., Brinkmann, T., KruÈger, T., Noyer-Weidner, M. & Pingoud, A. (1997). Characterization of the interaction between the restriction endonuclease McrBC from E. coli and its cofactor GTP. J. Mol. Biol. 272, 190-199. Pieper, U., Schweitzer, T., Groll, D. H. & Pingoud, A. (1999). De®ning the location and function of domains of McrB by deletion mutagenesis. Biol. Chem. In the press. Powers, T. & Walter, P. (1995). Reciprocal stimulation of GTP hydrolysis by two directly interacting GTPases. Science, 269, 1422-1424. Raleigh, E. A. (1992). Organization and function of the mcrBC genes of Escherichia coli K-12. Mol. Microbiol. 6, 1079-1086. Raleigh, E. A., Trimarchi, R. & Revel, H. (1989). Genetic and physical mapping of the mcrA (rglA), and mcrB
(rglB) loci of Escherichia coli K-12. Genetics, 122, 279296. Ross, T. K., Achberger, E. C. & Braymer, H. D. (1989). Nucleotide sequence of the McrB region of Escherichia coli K-12 and evidence for two independent translational initiation sites at the mcrB locus. J. Bacteriol. 171, 1974-1981. Schmidt, G., Lenzen, C., Simon, I., Deuter, R., Cool, R. H., Goody, R. S. & Wittinghofer, A. (1996). Biochemical and biological consequences of changing the speci®city of p21ras from guanosine to xanthosine nucleotides. Oncogene, 12, 87-96. Stewart, F. J. & Raleigh, A. (1998). Dependence of McrBC cleavage on distance between recognition elements. Biol. Chem. 379, 611-616. Studier, F. W., Rosenberg, A. H., Dunn, J. J. & Dubendorff, J. W. (1990). Use of T7 RNA polymerase to direct expression of cloned genes. Methods Enzymol. 185, 60-89. Sutherland, E., Coe, L. & Raleigh, E. A. (1992). McrBC: a multisubunit GTP-dependent restriction endonuclease. J. Mol. Biol. 225, 327-348.
Edited by J. Karn (Received 13 April 1999; received in revised form 9 August 1999; accepted 9 August 1999)