Investigating the endonuclease activity of four ... - BioMedSearch

4158±4165

Nucleic Acids Research, 2002, Vol. 30 No. 19

ã 2002 Oxford University Press

Investigating the endonuclease activity of four Pyrococcus abyssi inteins Isabelle Saves1,*, CeÂcile Morlot1, Laurent Thion1, Jean-Luc Rolland2, Jacques DieÂtrich2 and Jean-Michel Masson1,3 1

Institut de Pharmacologie et Biologie Structurale (UMR5089), CNRS/UniversiteÂ Paul Sabatier Toulouse III, 205 Route de Narbonne, F-31077 Toulouse Cedex, France, 2IFREMER, Laboratoire de Biotechnologie, Centre de Brest, BP 70, F-29280 PlouzaneÂ, France and 3Institut National des Sciences AppliqueÂes, Complexe Scienti®que de Rangueil, F-31077 Toulouse Cedex, France Received June 26, 2002; Revised and Accepted August 19, 2002

ABSTRACT Among the 14 inteins of the Pyrococcus abyssi genome, 10 harbour the LAGLIDADG motifs of dodecapeptide endonucleases. Four of these were cloned, expressed in Escherichia coli and puri®ed to assay their potential endonuclease activity. PabRIR1-2 and PabRIR1-3 are speci®c endonucleases, named PI-PabI and PI-PabII, respectively, cleaving the sequence spanning their homing site. This is consistent with their size and with the relative positions and sequences of their endonuclease motifs. However, PI-PabI is 10-fold more active than PI-PabII and a discrepancy of the DNA recognition and cleavage mechanisms was observed between the two inteins. In particular, analysis of the DNA cleavage reactions by MALDI-TOF highlighted that while the cleavage of DNA by PI-PabI consists of two steps corresponding to the cleavage of each DNA strand, PI-PabII processes the two DNA strands simultaneously. Furthermore, the two inteins interact differently with DNA. In addition, we did not detect any endonuclease activity for PabLon and PabRIR1-1. Deletions in the intein sequences and mutations in the putative endonuclease motifs probably abolish this activity. Hence, inteins from the same archaebacteria, even if contained in the same host protein, did not evolve uniformly and are presumably at different stages of the invasion cycle. INTRODUCTION Inteins are protein introns that are embedded in-frame within a precursor protein and are post-translationally excised by a self-catalytic protein splicing process. The intein peptide sequences generally harbour eight conserved motifs: four motifs essential to the splicing process and four optional motifs conferring a potential homing endonuclease activity involved in the perpetuation and transfer of the intein coding sequence (1) (Inbase, the New England Biolabs intein

DDBJ/EMBL/GenBank accession no. AL096836

database, is available at www.neb.com/neb/inteins.htlm and the Inteins-Protein introns web site at bioinfo.weizmann.ac. il/~pietro/inteins/). Since the discovery of inteins in 1990 (2,3), 130 intein genes have been identi®ed, mostly in archaebacteria (www. neb.com/neb/inteins.htlm; bioinfo.weizmann.ac.il/~pietro/ inteins/). In 1999, sequencing of the Pyrococcus abyssi complete genome (the P.abyssi home page at Genoscope is available at www.genoscope.cns.fr/Pab/; GenBank accession no. AL096836) revealed 14 inteins inserted into 10 different host genes of this archaebacteria. These inteins differ greatly in size (164±608 amino acid residues) and in sequence. Ten of these 14 inteins harbour the LAGLIDADG motifs characteristic of the dodecapeptide (DOD) endonucleases (4±6), three are mini-inteins missing the endonuclease core within their peptide sequence and one, which has an intermediate size, does not present the intein motifs of the endonuclease domain. While sequence alignment analyses show that the majority of known inteins exhibit endonuclease motifs, doublestranded DNA cleavage activity has been demonstrated for only a few inteins (1,7±16), and it is still not clear whether all inteins containing the DOD motifs are homing endonucleases. In our goal of further understanding the intein invasion process and the evolution of intein endonuclease activity, we searched for the endonuclease activity of four P.abyssi inteins: PabLon spliced from the ATP-dependant protease LA, and PabRIR1-1, PabRIR1-2 and PabRIR1-3 found in the same extein, i.e. the ribonucleoside-diphosphate reductase. The four intein coding sequences, inserted at the sites lon-a, rir1-a, rir1-c and rir1-b of the P.abyssi genome, respectively, were cloned and expressed in Escherichia coli. The puri®ed recombinant inteins were then assayed for their ability to speci®cally bind and cleave DNA. MATERIALS AND METHODS Production and puri®cation of the inteins The coding sequences of PabLon, PabRIR1-1, PabRIR1-2 and PabRIR1-3 inteins were ampli®ed by PCR from Pab genomic DNA with oligonucleotide pairs IntA-ATG (5¢gacacgacccgcatatgtgctttagtggcg-3¢) and IntA-TAG (5¢-ggagttcctaaggatcctcagttcttgacg-3¢), IntB-ATG (5¢-cagaagatgggc-

*To whom correspondence should be addressed. Tel: +33 561 175 471; Fax: +33 561 175 994; Email: [email protected]

Nucleic Acids Research, 2002, Vol. 30 No. 19 catatgtgtatagacggaaacgc-3¢) and IntB-TAG (5¢-gttttgagaaattggatcctcaattgtggacgaag-3¢), IntC-ATG (5¢-caactacggggcatatgtgttttactgggg-3¢) and IntC-TAG (5¢-catgaacgaaacggatcctcaattggatgtg-3¢), and IntD-ATG (5¢-ccaataagggctactcatatgtgtgtcgtagggg-3¢) and IntD-TAG (5¢-catagaggggctcggatcctcagttatggctcatg-3¢), respectively. The ampli®ed fragments were digested with NdeI and BamHI and cloned into a NdeI±BamHI (New England Biolabs) digested pET26b vector (Novagen). The resulting constructs, named pET-A, pET-B, pET-C and pET-D, allowing the expression of PabLon, PabRIR1-1, PabRIR1-2 and PabRIR1-3 inteins under the control of T7 promoter, respectively, were sequenced. Escherichia coli BL21(DE3)[pLysS] bacteria transformed with these expression vectors were grown at 37°C in Luria Broth culture medium supplemented with 50 mg/ml kanamycin (Sigma Chemical Co.). A 3 h induction with 1 mM isopropyl-b-Dthiogalactopyranoside (IPTG; Sigma) was performed and cells were lysed in 50 mM sodium citrate pH 5.5 by six cycles of freezing and thawing. The lysate was centrifuged at 10 000 g for 60 min and the supernatant was incubated for 10 min at 70°C. Denatured proteins were discarded by centrifugation at 10 000 g for 15 min. The fractions of PabLon and PabRIR1-1 were dialysed against 10 mM Tris±HCl pH 7.5, 50% glycerol, 0.1 mM EDTA, 1 mM DTT, 200 mg/ml BSA, 50 mM NaCl for storage. PabRIR1-2 and PabRIR1-3 have the best stability in 10 mM Tris±HCl pH 8.5, 50% glycerol, 50 mM NaCl and in 10 mM Tris±HCl pH 7.5, 50% glycerol, 50 mM NaCl, respectively. Endonuclease activity assay To assay the endonuclease activity of the four inteins, 39±41 bp sequences were inserted between XbaI and HindIII restriction sites of the plasmid pUC19 in order to clone the 40 bp DNA sequences spanning their homing sites. Partially complementary oligonucleotides SiteA±Xba (5¢-ctagagagttcctaatccacctgactggaacggggtcgtgtcta-3¢) and SiteA±Hind (5¢-agcttagacacgacccgttccagtcaggtggattaggaactct-3¢), SiteB± Xba (5¢-ctagatacagaagatgggcggtggaactggccttaatttctca-3¢) and SiteB±Hind (5¢-agcttgagaaattaaggccagttccaccgcccatcttctgtat3¢), SiteC±Xba (5¢-ctagagaacaactacgggggcagccagtggtcccgtttcgttca-3¢) and SiteC±Hind (5¢-agcttgaacgaaacgggaccactggctgcccccgtagttgttct-3¢), and SiteD±Xba (5¢-ctagacaataagggctactaacccctgtggagaggagcccctcta-3¢) and SiteD±Hind (5¢-agcttagaggggctcctctccacaggggttagtagcccttattgt-3¢) were annealed by boiling a mix of 1 nmol of each oligonucleotide in 10 mM Tris±HCl pH 7.5, 100 mM NaCl, for 5 min and slow cooling to room temperature. The annealed oligonucleotides were then ligated in pUC19 overdigested with HindIII and XbaI (New England Biolabs). The resulting constructs, namely pSiteA, pSiteB, pSiteC and pSiteD, respectively, were sequenced. These plasmids, containing the potential cleavage sites of PabLon, PabRIR1-1, PabRIR1-2 and PabRIR1-3, respectively, were either linearised by ScaI and puri®ed (Qiaquick gel extraction kit; Qiagen) or puri®ed on the supercoiled form from a 1% agarose gel in 0.53 TBE (90 mM Tris±borate, 2 mM EDTA) buffer. Finally, linear and circular forms of these substrates were diluted in water to a concentration of 100 ng/ml. Endonuclease activity assays were performed with 100 ng of plasmid substrate in a ®nal volume of 10 ml, in various reaction buffers and temperatures ranging from 37 to 90°C. The reaction mixtures were analysed on a 1% agarose gel in

4159

0.53 TBE buffer. The amounts of undigested substrates and products were quanti®ed with the ImageQUANT program (Molecular Dynamics Inc.). One unit of PI-PabI or PI-PabII endonuclease is required to digest 1 mg of substrate pSiteC or pSiteD, respectively, in 1 h at 70°C, in the speci®c optimal buffers. Speci®c activities of PI-PabI and PI-PabII were measured by incubating known amounts of DNA substrates with known amounts of puri®ed endonucleases. Alternatively, the DNA cleavage activity of the four inteins was researched using matrix-assisted laser desorption/ionisation mass spectrometry as recently described by L.Thion, E.Laurine, M.Erard, O.Burlet-Schiltz, B.Monsarrat, J.-M. Masson and I.Saves (manuscript submitted). Synthetic substrates were obtained by hybridisation of the 45mer oligonucleotide pairs SiteA±Xba and SiteA±Hind, SiteB±Xba and SiteB±Hind, SiteC±Xba and SiteC±Hind, and SiteD±Xba and SiteD±Hind. An aliquot of 50 pmol of these substrates was incubated at 70°C with various amounts of inteins either in the optimal buffers de®ned for PI-PabI and PI-PabII or in various buffers in the cases of PabLon and PabRIR1-1. After the desalting of the cleavage reactions, 1 pmol of DNA was analysed as described (L.Thion et al., manuscript submitted). De®nition of the endonuclease cleavage site and delineation of the minimal recognition sequence The endonuclease cleavage sites of PI-PabI and PI-PabII were ®rstly determined by matrix-assisted laser desorption ionization±time of ¯ight (MALDI-TOF) mass spectrometry, their exact location being deduced from the measured masses of the cleavage fragments. The endonuclease minimal recognition sequences for PIPabI and PI-PabII were determined by a primer extension method (17) using plasmids pSiteC and pSiteD as DNA matrix. The sequencing and digestion procedures, using the T7 polymerase sequencing kit (Pharmacia) and universal primers ±21M13Forward (5¢-tgtaaaacgacggccagt-3¢) and ±21M13Reverse (5¢-ggaaacagctatgaccatg-3¢) were as previously described (12). Electrophoretic mobility shift assays DNA probes containing the 40 bp sequence spanning the homing site of each intein were obtained by PCR using the universal reverse (5¢-caggaaacagctatgac-3¢) and forward (5¢-gttttcccagtcacgac-3¢) primers and 100 ng of plasmids pSiteA, pSiteB, pSiteC and pSiteD as matrix in standard PCR ampli®cation reactions. These 139 bp fragments were 33P labelled by phosphorylation with [g-33P]ATP (NEN), puri®ed and diluted in water to a concentration of 30 nM. Aliquots of 5±10 fmol of the DNA probes were incubated in different buffers, either with or without intein, in a ®nal volume of 20 ml, for 15 min at 70°C. These reaction mixtures were then subjected to electrophoresis in a 7.5% non-denaturing polyacrylamide (30:1) gel in 0.253 TBE buffer at 220 V for 2 h. After electrophoresis, the gel was dried and radioactive bands were detected using a phosphorimager (Molecular Dynamics Inc.). The speci®city of the DNA±protein interactions was controlled using 7.5 pmol of herring sperm DNA as a non-speci®c competitor in the binding reactions.

4160


RESULTS AND DISCUSSION Expression and puri®cation of the four P.abyssi inteins The four expression vectors pET-A, pET-B, pET-C and pET-D were used to transform the BL21(DE3)[pLysS] E.coli strain and various conditions of expression of the four intein genes were assayed. In optimal conditions of expression, bacteria were grown at 37°C and 1 mM IPTG was added during the exponential growth phase to induce the expression of the intein gene. The comparison between proteins contained in soluble crude extracts of cultures treated or not with IPTG (Fig. 1) showed that the four recombinant inteins PabLon (37 kDa), PabRIR1-1 (44 kDa), PabRIR1-2 (49 kDa) and PabRIR1-3 (42 kDa) were expressed upon induction, the synthesis of PabLon and PabRIR1-1 inteins being substantial. Since P.abyssi is a thermophilic archaebacteria, pyrococcal inteins were puri®ed from bacterial proteins by heating the crude extracts for 10 min at 70°C. This treatment resulted in the denaturation of most E.coli proteins while marginally affecting the intein (Fig. 1). It can be seen that the yield of puri®cation of the inteins is higher for PabLon and PabRIR1-1 than for the other two inteins; this may be due to the higher initial proportion of intein in the crude extract. Nevertheless, the four inteins were obtained in soluble homogenous fractions that were used for the functional assays. PabRIR1-2 and PabRIR1-3 are active endonucleases, named PI-PabI and PI-PabII In order to assay the endonuclease activity of the four pyrococcal inteins, their putative substrates were constructed by cloning the 40 bp sequences of the host genes, which extend across the intein homing sites, in the plasmid pUC19. Since known inteins exhibiting an endonuclease activity cleave 16±31 bp target sequences (1,7,9±16), these 40 bp sequences should be long enough to be recognised and cleaved by the inteins. Each 2710 bp plasmid was puri®ed either as a linear substrate after a ScaI overdigestion or as a supercoiled substrate and was incubated with the corresponding intein in various conditions. Both supercoiled and linear plasmids pSiteC and pSiteD were cleaved by the recombinant inteins PabRIR1-2 (Fig. 2A and B) and PabRIR1-3 (Fig. 2D and E), respectively. Indeed, in both cases, the supercoiled DNA was processed into linear DNA (Fig. 2A and D) and the 2710 bp linear DNA was cleaved in two fragments of 960 and 1750 bp (Fig. 2B and E) in the presence of intein, while these substrates remained unmodi®ed in the absence of intein (data not shown). As additional negative controls, the wild-type plasmid pUC19 missing the target sequence was not cleaved by the inteins (Fig. 2C and F). Both inteins, PabRIR1-2 and PabRIR1-3, are thus endonucleases that speci®cally cleave the sequence spanning their respective homing site. They were named PIPabI and PI-PabII, respectively, according to the current nomenclature of homing endonucleases. The ®nding of endonuclease activity for these inteins was not surprising in view of their peptide sequences. Their size (438 and 382 amino acids, respectively) and the relative positions of the endonuclease motifs are indeed similar to those of known intein endonucleases and the sequences of the two LAGLIDADG peptides are well conserved with a close

Figure 1. Expression in E.coli and puri®cation of Pab inteins. Proteins contained in induced (I) or non-induced (NI) soluble crude extracts and in puri®ed fractions (P) of recombinant PabLon, PabRIR1-1, PabRIR1-2 and PabRIR1-3 inteins were separated in a 10% SDS±PAGE and detected by Coomasie blue staining.

®tting to the consensus sequence de®ned by Pietrokovski (18; bioinfo.weizmann.ac.il/~pietro/inteins/). Moreover, the whole sequence of PabRIR1-3 intein (PI-PabII) is 84.2% identical to its allele PfuRIR1-2, known to possess endonuclease activity (PI-PfuII) (11), with 100% identity within the endonuclease motifs sequences. The optimal conditions of cleavage regarding the reaction temperature, buffer pH and composition, and the mono and bivalent ions used as cofactors were determined for both inteins. The inteins are active within a wide range of saline conditions between 50 and 90°C. Nevertheless, PI-PabI activity is optimal in a 5 mM Tris±acetate pH 9.5 buffer containing 5 mM KCl and 5 mM MgCl2 and PI-PabII is most active in a 10 mM Tris±HCl pH 8 buffer containing 25 mM NH4OAc and 2 mM MgCl2. It is noticeable that, while most homing endonucleases display higher cleavage activity in the presence of manganese ions, these two inteins, as was the case with PI-TfuII (12), are most active in the presence of magnesium, with a 2-fold increase in activity whatever the DNA substrate conformation. The most favourable reaction temperature is 70°C for both enzymes since they are rapidly inactivated at higher temperatures (data not shown). Under these optimal conditions, the enzymes do not discriminate between the DNA substrate conformations and their speci®c activities are 700,000 6 50,000 U/mg and 72,000 6 6,000 U/mg, respectively, either with linear or supercoiled substrates. Further studies of the cleavage reaction by both inteins highlighted that the lower activity level of PI-PabII is associated with an apparent discrepancy of the mechanisms of DNA recognition and cleavage between the two inteins. An open-circular DNA form was detected during the cleavage of supercoiled substrates by both inteins PI-PabI and PI-PabII. A similar observation had previously been made


4161

Figure 2. Standard cleavage assays for PI-PabI (PabRIR1-2) and PI-PabII (PabRIR1-3). Cleavage assays for PI-PabI on its circular substrate (A), linear substrate (B) and on linear plasmid pUC19 (C). 100 ng of puri®ed pSiteC plasmid was incubated with 0.10 ng (A), 0.40 ng (B) or 15 ng (C) of PI-PabI for different times at 70°C, in a 5 mM Tris±acetate pH 9.5 buffer containing 5 mM KCl and 5 mM MgCl2. Cleavage assays for PI-PabII on its circular substrate (D), linear substrate (E) and on linear plasmid pUC19 (F). 100 ng of puri®ed pSiteD plasmid were incubated with 1 ng [(D) and (E)] or 15 ng (F) of PI-PabII for different times at 70°C, in a 10 mM Tris±HCl pH 8 buffer containing 25 mM NH4OAc and 2 mM MgCl2. Either supercoiled (sc), open circular (oc) and linear (lin) forms of DNA or linear substrate (S, 2710 bp) and fragment products (P, 960 and 1750 bp) were separated on a 1% agarose gel in 0.53 TBE buffer.

when studying the DNA cleavage activity of one intein spliced from the Thermococcus fumicolans DNA polymerase precursor (12). In this case, the open-circular form is produced at the beginning of the cleavage reaction and progressively disappears from the reaction mix. It has been demonstrated that the cleavage of supercoiled DNA by PI-TfuI consists of a two-step reaction, which transiently generates open-circular DNA with a nicked bottom strand, the cleavage of the top strand being rate limiting (L.Thion et al, manuscript submitted). While the amount of open-circular DNA formed during cleavage by PI-PabI evolves in a comparable way (Fig. 2A), the evolution of this DNA form during the reaction with PIPabII is rather different. First, a non-negligible amount of open-circular DNA contaminates the supercoiled DNA substrate and, secondly, the additional open-circular DNA formed during the cleavage by PI-PabI is rather stable (Fig. 2D). In order to characterise the potential intermediates of the cleavage reactions by PI-PabI and PI-PabII, we kinetically analysed the DNA cleavage reaction by both inteins using MALDI-TOF mass spectrometry as described by Thion et al. This technology allowing the detection of as little as 250 fmol of 20mer to 40mer oligonucleotides, we used two synthetic double-stranded substrates containing the 40 bp target sequences to monitor the cleavage reactions. These substrates were obtained through the annealing of 45mer complementary oligonucleotide pairs. They were then incubated with the two inteins in optimal buffers for different reaction times and the cleavage reactions were subsequently puri®ed from salts and analysed by MALDI-TOF mass spectrometry to determine the mass and to follow the appearance of each DNA fragment generated. Figure 3 shows the mass spectrometry pro®les obtained after cleavage of DNA by PI-PabI (Fig. 3A±D) and PI-PabII (Fig. 3E±G) and their interpretation in term of cleavage sites

appears in Figure 4A and B, respectively. Indeed, the measured masses of the cleavage fragments (Table 1) allowed us to locate the cleavage site on each DNA strand. As previously observed for all intein endonucleases, cleavage by PI-PabI and PI-PabII generates 3¢ overhangs of four bases. While these overhangs spanned the rir1-c insertion site in the case of PI-PabI, they are located 3¢ to the rir1-b site in the case of PI-PabII. As expected, the cleavage site of PI-PabII is identical to that of PI-PfuII (11); that suggests that these inteins are isoschizomers as has been shown for PI-TfuII, PI-TliI and PI-ThyI inteins (13). The MALDI-TOF analyses of the reaction mixtures after short incubation times pointed to an apparent dissimilarity between the mechanisms of double-stranded DNA cleavage by PI-PabI and PI-PabII. Effectively, only two cleavage fragments (F3 and F4), generated by the cleavage of the bottom strand of the DNA substrate, were detected after a 2 min incubation with PI-PabI (Fig. 3A). The two other cleavage fragments (F1 and F2), generated by cleavage of the top strand, appeared after progressively longer incubation times (Fig. 3B±D), meaning that the cleavage of the top strand was rate-limiting in a two-step cleavage reaction, which is consistent with the detection of an open-circular intermediate of cleavage of supercoiled DNA (Fig. 2A). In contrast, the four cleavage products (f1±f4) of the synthetic substrate of PIPabII appeared together in the reaction mixture (Fig. 3E±H); a slight difference between the rate of cleavage of both strands can certainly account for the low production of open-circular DNA during the cleavage of supercoiled DNA. In conclusion, while PI-PabI, as PI-TfuI, clearly cleaves DNA in two steps consisting of the cleavage of each strand (L.Thion et al., manuscript submitted), PI-PabII, as well as PI-TfuII, PI-SceI, PI-PfuI and PI-PfuII (1,11,12), simultaneously processes both DNA strands. However, it is conceivable that in all cases, as

4162


Figure 3. MALDI-TOF analyses of DNA cleavage reaction by PI-PabI (A±D) and PI-PabII (E±H). Aliquots of 50 pmol of synthetic DNA were incubated with 1 mg of PI-PabI or 10 mg of PI-PabII in the corresponding optimal buffers at 70°C, for 2 min [(A) and (E)], 5 min [(B) and (F)], 15 min [(C) and (G)] and 60 min [(D) and (H)]. Digested DNA (1 pmol) was analysed as described.

Figure 4. Interpretation of MALDI-TOF analyses in terms of cleavage sites for PI-PabI (A) and PI-PabII (B). Oligonucleotides (F1±F4 and f1±f4) generated by the cleavage reaction by PI-PabI and PI-PabII, respectively, are identi®ed. Arrows indicate the position of the cleavage sites on each DNA strand. The location of the intein insertion sites rir1-c and rir1-b are indicated.

was shown for PI-SceI (19), the double-stranded DNA is cleaved by these enzymes through a similar two-step mechanism, the ef®ciencies of the two steps being equivalent.

Furthermore, we compared the binding of PI-PabI and PIPabII to their respective target sequences and showed that the two inteins interact with DNA in different ways (Fig. 5).


4163

Table 1. Characteristics of the cleavage fragments generated by DNA cleavage by PI-PabI (F1±F4) and PI-PabII (f1±f4)

Figure 5. Electrophoretic mobility shift assays for PI-PabI (A) and PI-PabII (B). (A) The DNA probe (0.5 nM) was incubated for 15 min with 0, 10, 25, 50, 100 and 250 nM of PI-PabI at 70°C, in a 20 mM Tris±acetate pH 8 buffer containing 2 mM NH4OAc, 0.1% Triton X-100 and 2% glycerol. (B) The DNA probe (0.25 nM) was incubated for 15 min at 70°C with 0, 0.25 or 2.5 mM of PI-PabII in a 20 mM Tris±acetate pH 8 buffer containing 2 mM NH4OAc, 0.1% Triton X-100 and 2% glycerol and with 10 pmol of PI-PabII in a 20 mM Tris±acetate pH 8 buffer containing 2 mM Mg(OAc)2, 2 mM NH4OAc, 0.1% Triton X-100 and 2% glycerol (Mg2+). Black and open arrows indicate the 119 bp DNA probe and the probe±intein complexes, respectively.

Electrophoretic mobility shift assays were performed in the absence of Mg2+ ions to prevent the cleavage of the DNA probes. The speci®c DNA±PI-PabI complex observed at lower intein concentrations smears to the top of the gel when the intein concentration increased to 50 nM (Fig. 5A). This may be due to an aggregation of the complex after all the DNA is bound. In contrast, the complex between PI-PabII and its target sequence was only observed at extremely high concentrations

of intein (Fig. 5B). Indeed, the af®nity of this intein for the DNA appeared really poor, an intein concentration of 2.5 mM being necessary to detect a shift of the DNA probe. It is thus possible that an additional complex could appear when all DNA is bound to the intein, as was described for PI-SceI, which interacts with its target sequence in a biphasic pathway (20,21). Curiously, the complex was not observed in the absence of magnesium at intein concentrations suf®cient to

4164


cleave the probe while it was observed at low concentrations in the presence of magnesium ions with the same intein concentration (0.25 mM). This shows that magnesium ions play a role in the recognition of DNA by PI-PabII as it was shown for PI-PfuII (22). This low af®nity of PI-PabII for its target sequence could yet re¯ect the low speci®c activity of this enzyme. Since this sequence was shown to be cleaved near its 3¢ extremity (Fig. 4B), we wondered whether the cloned target sequence was not interrupted in its 3¢ part. We thus completed the characterisation of the intein endonucleases by de®ning the minimal cleavable DNA sequences. Although the Wenzlau procedure is not the most accurate way to delineate the recognition site, the results depending on DNA conformation and enzyme concentration (12,20), the comparison of intein-digested and -undigested DNA patterns (Fig. 6A for PI-PabII and not shown for PI-PabI) con®rmed the location of the cleavage sites and allowed us to de®ne 25 and 20 bp minimal recognition sequences for PI-PabI and PIPabII (Fig. 6B), respectively. That is in agreement with the known speci®city of inteins. Moreover, it highlighted that the cloned 40 bp target sequence contained the whole minimal sequence cleaved by PI-PabII. Hence, as in the case of PIMgaI intein (14), the recognition site of PI-PabII is atypically short on the 3¢ side of the cleavage site, with only seven bases 3¢ to this site on the upper DNA strand, compared with 13 bases on the 5¢ side. Hence, if the intein speci®cally interacts with 5¢ part of the cleavage site, as in the case of PI-SceI and PI-TfuII (12,23), the shortness of this sequence may account for the low af®nity of PI-PabII for its DNA target sequence. However, it is probable that the DNA recognition way of PIPabII vary, as described for PI-TfuI (12). PabLon and PabRIR1-1 do not possess homing endonuclease activity Standard cleavage assays were performed with puri®ed PabLon and PabRIR1-1 inteins and supercoiled or linear pSiteA and pSiteB plasmids as DNA substrates under various reaction conditions. The wide range of experimental conditions included 2±200 mM Tris±HCl or Tris±acetate buffers pH 7±9, containing 0±100 mM of Mg(OAc)2, MgCl2, ZnCl2, CaCl2, MnSO4 or MnCl2, 0±400 mM of NH4OAc, NaCl or KCl, 0±30% glycerol and the assay temperature was varied from 37 to 80°C. We did not detect any endonuclease activity for either intein under any of the above conditions. Moreover, no cleavage fragments have ever been detected by MALDITOF mass spectrometry when 40 bp synthetic DNA substrates were incubated with the two inteins in various conditions (data not shown). Since a wide range of conditions with regard to buffer pH and composition, mono and bivalent ions used as cofactors, and temperature were investigated, we concluded that PabLon and PabRIR1-1 inteins do not possess any endonuclease activity. This absence of enzymatic activity can probably explain the high level of expression of these two recombinant inteins. Indeed, our previous unpublished results tend to show that the expression of homing endonucleases is rather toxic for the bacteria, perhaps due to non-speci®c DNA binding and/or cleavage at high intracellular enzyme concentration. The small intein PabLon (333 amino acids) has a low homology with the other inteins of its allelic family, namely

Figure 6. Minimal recognition site of PI-PabI and PI-PabII, determined using the primer extension procedure. (A) Autoradiogram of the sequencing gel for PI-PabII site determination. The sequencing reactions were performed in direct (M13Forw) or reverse (M13Rev) orientations. PI-PabII digested (+ lanes) and undigested (± lanes) reactions were loaded side by side on a 6% denaturing polyacrylamide gel. A large arrow indicates the cleavage site on each DNA strand. Boxes represent bases belonging to the minimal site, in each direction. (B) Minimal nucleotide sequence necessary for recognition and cleavage by PI-PabI and PI-PabII. The dashed boxes indicate the minimal recognition sequences and arrowheads designate the cleavage points on each DNA strand. The dashed lines indicate the intein insertion sites rir1-c and rir1-b in the rir1 gene.

PfuLon and PhoLon, which contain 401 and 474 residues, respectively. Even if nothing is known about the enzymology of these last two inteins, it is clear that deletions in the P.abyssi intein had occurred during evolution. A hypothetical location of the endonuclease motifs in PabLon sequence was nevertheless proposed (Inbase) but the existence of the second LAGLIDADG motif is rather uncertain since sequence alignment with PhoLon suggest that this motif is located in a 20±25 residue central peptide which is deleted in the PabLon sequence. In addition, a deletion of roughly 60 residues is located 20 residues upstream from the ®rst LAGLIDADG motif and a third one, consisting of approximately 60 residues, is located at the C-terminal end of the intein. It is thus probable that these deletions in the intein sequence hinder the correct folding of the endonuclease domain and consequently abolish the endonuclease activity of the intein. Moreover, the sequence of the putative endonuclease motifs could also explain the absence of activity. In particular, a glycine is found in place of the catalytic aspartic acid of the ®rst LAGLIDADG

Nucleic Acids Research, 2002, Vol. 30 No. 19 motif. Hence, the loss of the endonuclease activity of PabLon intein may result from successive mutational events such as deletions and point mutations along the intein coding sequence. Similar evolution can be proposed for PabRIR1-1. This 399 residue intein is an allele of PfuRIR1-1 (454 amino acids) known to possess endonuclease activity (PI-PfuI). The alignment of the two peptide sequences highlights good homology between the inteins (80% identity) but also a central deletion of 55 residues located only 10 residues downstream of the ®rst LAGLIDADG motif in PabRIR1-1. Moreover, the catalytic aspartic acid of this block is replaced by an asparagine. In order to verify that the absence of endonuclease activity was solely due to the mutation of the catalytic aspartic acids, we searched for a binding activity of both inteins to their target DNA sequences. Mobility shift assays were performed in a large set of reaction buffers but no shift of the DNA probes was detected. The DNA mobility was unaffected even at extremely high concentrations of inteins (data not shown). Mutational analyses on PI-SceI, PI-PfuI and PI-PfuII (22,23) showed that the substitution of the catalytic residues does not disrupt the interaction between the intein and its target DNA. We thus concluded that the absence of endonuclease activity of PabLon and PabRIR1-1 inteins also implies a global misfolding of the endonuclease domain. CONCLUSION In our goal to further understand the evolution of the homing endonuclease activity of inteins, we showed that among four full-size inteins of P.abyssi only two, namely PabRIR1-2 and PabRIR1-3, exhibit a speci®c double-stranded DNA cleavage activity, named PI-PabI and PI-PabII, respectively. However, these two enzymes harbour variable enzymatic behaviours, their way of interacting with DNA and of cleaving the target sequence being divergent. The two other inteins, namely PabLon and PabRIR1-1, do not cleave their homing site, both DNA binding and cleavage activities being lost. Peptide sequence analyses highlight that additive mutational events such as amino acid deletions and/or substitutions are responsible for this loss of activity. Hence, it appears that inteins from the same archaebacteria, even if contained in the same host protein, did not evolve uniformly and are at different stages of the invasion cycle proposed by Goddard and Burt (24). ACKNOWLEDGEMENT This work was supported in part by a grant (#010012) by IFREMER to J.-M.M. REFERENCES 1. Gimble,F.S. and Thorner,J. (1992) Homing of a DNA endonuclease gene by meiotic gene conversion in Saccharomyces cerevisiae. Nature, 357, 301±306. 2. Kane,P.M., Yamashiro,C.T., Wolczyk,D.F., Neff,N., Goebl,M. and Stevens,T.H. (1990) Protein splicing converts the yeast TFP1 gene product to the 69-kD subunit of the vacuolar H(+)-adenosine triphosphatase. Science, 250, 651±657. 3. Hirata,R., Ohsumk,Y., Nakano,A., Kawasaki,H., Suzuki,K. and Anraku,Y. (1990) Molecular structure of a gene, VMA1, encoding the

4.

5. 6. 7.

8. 9. 10.

11.

12.

13. 14. 15.

16. 17. 18. 19.

20. 21. 22.

23.

24.

4165

catalytic subunit of H(+)-translocating adenosine triphosphatase from vacuolar membranes of Saccharomyces cerevisiae. J. Biol. Chem., 265, 6726±6733. Dalgaard,J.Z., Klar,A.J., Moser,M.J., Holley,W.R., Chatterjee,A. and Mian,I.S. (1997) Statistical modeling and analysis of the LAGLIDADG family of site-speci®c endonucleases and identi®cation of an intein that encodes a site-speci®c endonuclease of the HNH family. Nucleic Acids Res., 25, 4626±4638. Mueller,J.E., Bryk,M., Loizos,N. and Belfort,M. (1993) Homing endonucleases. Nucleases, 111±143. Pietrokovski,S. (1994) Conserved sequence features of inteins (protein introns) and their use in identifying new inteins and related proteins. Protein Sci., 3, 2340±2350. Perler,F.B., Comb,D.G., Jack,W.E., Moran,L.S., Qiang,B., Kucera,R.B., Benner,J., Slatko,B.E., Nwankwo,D.O., Hempstead,S.K. et al. (1992) Intervening sequences in an Archaea DNA polymerase gene. Proc. Natl Acad. Sci. USA, 89, 5577±5581. Xu,M.Q., Southworth,M.W., Mersha,F.B., Hornstra,L.J. and Perler,F.B. (1993) In vitro protein splicing of puri®ed precursor and the identi®cation of a branched intermediate. Cell, 75, 1371±1377. Perler,F.B., Olsen,G.J. and Adam,E. (1997) Compilation and analysis of intein sequences. Nucleic Acids Res., 25, 1087±1093. Nishioka,M., Fujiwara,S., Takagi,M. and Imanaka,T. (1998) Characterization of two intein homing endonucleases encoded in the DNA polymerase gene of Pyrococcus kodakaraensis strain KOD1. Nucleic Acids Res., 26, 4409±4412. Komori,K., Fujita,N., Ichiyanagi,K., Shinagawa,H., Morikawa,K. and Ishino,Y. (1999) PI-PfuI and PI-PfuII, intein-coded homing endonucleases from Pyrococcus furiosus. I. Puri®cation and identi®cation of the homing-type endonuclease activities. Nucleic Acids Res., 27, 4167±4174. Saves,I., Ozanne,V., Dietrich,J. and Masson,J.M. (2000) Inteins of Thermococcus fumicolans DNA polymerase are endonucleases with distinct enzymatic behaviors. J. Biol. Chem., 275, 2335±2341. Saves,I., Eleaume,H., Dietrich,J. and Masson,J.-M. (2000) The Thy Pol-2 intein of Thermococcus hydrothermalis is an isoschizomer of PI-TliI and PI-TfuII endonucleases. Nucleic Acids Res., 28, 4391±4396. Saves,I., Westrelin,F., DaffeÂ,M. and Masson,J.-M. (2001) Identi®cation of the ®rst eubacterial endonuclease coded by an intein allele in the pps1 gene of mycobacteria. Nucleic Acids Res., 29, 4310±4318. Saves,I., Lewis,L.-A., Westrelin,F., Warren,R., DaffeÂ,M. and Masson,J.-M. (2002) Speci®cities and functions of the recA and pps1 intein genes of Mycobacterium tuberculosis and application for diagnosis of Tuberculosis. J. Clin. Microbiol., 40, 943±950. Guhan,N. and Muniyappa,K. (2002) Mycobacterium tuberculosis RecA intein possesses a novel ATP-dependent site-speci®c double-stranded DNA endonuclease activity. J. Biol. Chem., 277, 16257±16264. Wenzlau,J.M., Saldanha,R.J., Butow,R.A. and Perlman,P.S. (1989) A latent intron-encoded maturase is also an endonuclease needed for intron mobility. Cell, 56, 421±430. Pietrokovski,S. (1998) Modular organization of inteins and C-terminal autocatalytic domains. Protein Sci., 7, 64±71. Christ,F., Schoettler,S., Wende,W., Steuer,S., Pingoud,A. and Pingoud,V. (1999) The monomeric homing endonuclease PI-SceI has two catalytic centres for cleavage of the two strands of its DNA substrate. EMBO J., 18, 6908±6916. Wende,W., Grindl,W., Christ,F., Pingoud,A. and Pingoud,V. (1996) Binding, bending and cleavage of DNA substrates by the homing endonuclease PI-SceI. Nucleic Acids Res., 24, 4123±4132. Gimble,F.S. and Wang,J. (1996) Substrate recognition and induced DNA distortion by the PI-SceI endonuclease, an enzyme generated by protein splicing. J. Mol. Biol., 263, 163±180. Komori,K., Ichiyanagi,K., Morikawa,K. and Ishino,Y. (1999) PI-PfuI and PI-PfuII, intein-coded homing endonucleases from Pyrococcus furiosus. II. Characterization of the binding and cleavage abilities by sitedirected mutagenesis. Nucleic Acids Res., 27, 4175±4182. Gimble,F.S. and Stephens,B.W. (1995) Substitutions in conserved dodecapeptide motifs that uncouple the DNA binding and DNA cleavage activities of PI-SceI endonuclease. J. Biol. Chem., 270, 5849±5856. Goddard,M.R. and Burt,A. (1999) Recurrent invasion and extinction of a sel®sh gene. Proc. Natl Acad. Sci. USA, 96, 13880±13885.