INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The RecJ exonuclease of Escherichia coli is a single-stranded DNA (ssDNA)-specific exonuclease that plays a role in DNA repair, recombination, and mutation avoidance (4, 6, 8, 9, 11). A comparison of the available sequenced genomes shows that RecJ is ubiquitous throughout the eubacteria. Sequences with a strong similarity to E. coli RecJ are found in at least nine bacterial divisions, indicating an ancient origin and important biological function for RecJ. Alignment of sequences from 11 eubacterial genera with strong similarity to RecJ has revealed seven conserved motifs among the eubacterial RecJ proteins (10) (Fig. 1). Mutational analysis of amino acids within six of these motifs in the E. coli RecJ protein (motifs 1, 2, 3, 4, 6, and 7) confirmed that these residues are essential for exonuclease activity in vitro and genetic function in vivo (10). Residues within these motifs are good candidates for interactions with the phosphates of the substrate DNA, or with Mg²⁺ ions, which are cofactors required for RecJ-mediated hydrolysis of DNA (9).

View larger version (53K): [in this window] [in a new window]   FIG. 1.   Comparison of RecJ motifs among archaeal proteins and PPX1 of S. cerevisiae. MJ, AF, MTH, and PHO prefixes indicate proteins of the archaea M. jannaschii, A. fulgidus, M. thermoautotrophicum, and P. horikoshii, respectively. Eco, E. coli sequence; Sce, S. cerevisiae sequence. NR, the motif is not recognizable within the sequence alignments. Each member of the defined groups has very strong sequence similarity to other members of the group, with E values in pairwise BLASTP alignments of <1025. (A) Conserved motifs from the E. coli RecJ sequence, with invariant residues among the eubacterial RecJ-related sequences (11) shown in bold. (B) Motifs from group B archaeal sequences with strongest similarity to RecJ. (C) AF0075 archaeal sequence with somewhat weaker similarity to RecJ and similarity to group B. (D) Archaeal RecJ-related sequences with both RecJ motifs and an N-terminal DnaJ domain. (E) AF0291, an outlying archaeal sequence with no strong homology to any of the other groups. (F) Archaeal sequences with similarity to group G. (G) Sequences with similarity to group F. (H) PPX1 of S. cerevisiae and related sequences from archaea.

The genomes of archaea are especially rich in sequences carrying RecJ motifs: Methanococcus jannaschii has six, Archaeoglobus fulgidus has seven, and Methanobacterium thermoautotrophicum and Pyrococcus horikoshii each have three. Each of these archaeal hypothetical proteins contains at least five of the seven motifs that are conserved among eubacterial RecJ proteins (Fig. 1). Although overall sequence similarity between these proteins and eubacterial RecJ is limited, the presence of the RecJ motifs strongly suggests that these proteins are related to RecJ and raises the possibility that they possess exonuclease activity.

The conserved motifs in RecJ define a large family of proteins in eubacteria, archaea, and eukaryotes (2, 10). Although the biochemical properties of this group, for the most part, have not been established, it includes two proteins that have been characterized extensively: the ssDNA-specific RecJ exonuclease of E. coli (9) and the PPX1 polyphosphatase of Saccharomyces cerevisiae (14, 15). Although RecJ and PPX1 have very little amino acid sequence similarity outside the conserved motifs, both have phosphoesterase activity, albeit on very different biological substrates. Although the presence of the RecJ motifs implicates phosphoesterase activity, it is difficult at present to assign phosphatase or exonuclease activity to individual members of this family. We have therefore chosen to examine the three open reading frames from M. jannaschii with strongest similarity to the sequences encoding eubacterial RecJ. We have tested these genes for genetic complementation of a UV-sensitive phenotype conferred by a recJ mutation in E. coli and for expression of nuclease activity on DNA. Two of the genes, MJ0977 and MJ0831, do indeed show partial genetic complementation, indicating that they likely share function with RecJ exonuclease. We also demonstrate that one of these genes, MJ0977, encodes a thermostable ssDNA nuclease activity with properties similar to those of E. coli RecJ exonuclease.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Database searches. The gapped BLASTP and PSI-BLAST programs (1) were used to generate sequence alignments by use of the website provided by the National Center for Biotechnology Information (http://www/ncbi/nlm.nih.gov/BLAST). In all cases, the BLOSUM62 matrix was used, with gap penalties of 11 (open) and 1 (per residue extension), with a lambda ratio of 0.85, and with the sequence filter off.

Plasmids. The following primers were used to amplify three recJ-related sequences from M. jannaschii genomic DNA by use of Pfu polymerase (Stratagene) under conditions described by the manufacturer: 5' TAATGTATCGATTGGAAAGGAGGTACAAGCGATGATGGAAAAACTTAAAGAAATA 3' (N terminal) and 5' TAATGGCTGCAGATATTTTTATCTCCTTAA 3' (C terminal) for MJ0831, 5' TAATGTATCGATTGGAAAGGAGGTACAAGCGATGGAAAATTGGGTAGAATTGAAA 3' (N terminal) and 5' TAATGGCTGCAGATATTTTTATCTCAAAGC 3' (C terminal) for MJ0977, and 5' TAATGTATCGAT TGGAAAGGAG GTACAAGCGATGATAG TAAAG TG TCCAAT T TG T 3' (N terminal) and 5' TAATGGCTGCAGTATTTTTTATTGCTCTTT 3' (C terminal) for MJ1198. The N-terminal primers contain a ClaI restriction site and a ribosome-binding site (underlined). The C-terminal primers contain a PstI restriction site. PCR products were digested with ClaI and PstI and ligated into compatible sites on the low-copy-number vector pWSK29 (12) to produce pLAR101, pLAR102, and pLAR103, respectively. Plasmid pSTL132 contains the ApaI-SacII fragment of pSTL175 (10) cloned into identical sites on pWSK29. A low-copy-number vector was chosen because of toxicity exhibited by E. coli recJ when present on a high-copy-number plasmid.

UV survival assays. Complementation experiments were carried out with plasmids pWSK29, pLAR101, pLAR102, pLAR103, and pSTL132, described above. Plasmids were introduced by electroporation (5) into the E. coli K-12 strain STL3266 [F DE3 recB21 recC22 sbcA23 recJ284::Tn10 thi-1 argE3 hisG4 (gpt-proA)62 thr-1 leuB6 kdgK51 rfbD1 ara-14 lacY1 galK2 xyl-5 mtl-1 tsx-33 supE44 rpsL31], a DE3 lysogen carrying an isopropyl--D-thiogalactopyranoside (IPTG)-inducible T7 RNA polymerase gene. Transformants were grown overnight on Luria-Bertani (LB) agar (1.5%) plates containing ampicillin (100 µg/ml) and tetracycline (15 µg/ml) (LB+Ap+Tc agar). Individual transformants were selected, and cells were grown in liquid LB+Ap+Tc medium to exponential phase (optical density at 600 nm, 0.6). Cultures were split, and one half was induced with 1 mM IPTG. Both sets of cultures were grown for 1 h, serially diluted in 56/2 buffer (13), and plated on LB+Ap+Tc agar plates. Plates were incubated at 45°C for 15 min, exposed to a 0-, 1-, 2-, or 5-J/m² dose of UV (254 nm) irradiation, and incubated at 45°C in the dark for an additional 15 min and then overnight at 37°C. Fractional survival was calculated as the number of CFU on irradiated plates relative to the number of CFU on unirradiated plates.

Protein expression. E. coli K-12 strain STL2329 [DE3 sbcB15 recJ284::Tn10 endA (xth-pncA gal thi)], a DE3 lysogen carrying an IPTG-inducible T7 RNA polymerase gene, was used to express the protein products of MJ0831, MJ0977, and MJ1198 from pLAR101, pLAR102, and pLAR103, respectively. Cells were grown at 37°C in liquid LB+Ap+Tc medium to exponential phase (optical density at 600 nm, 0.8). IPTG (2 mM final concentration) was added to induce plasmid-encoded gene expression, and the culture was incubated for an additional hour. Cells were harvested by centrifugation, resuspended in 0.01 volume of 50 mM Tris-HCl (pH 7.5)-10% sucrose (wt/vol), and stored at 70°C. To thawed cell suspensions, EDTA, lysozyme, and dithiothreitol were added to final concentrations of 4 mM, 1.5 mg/ml, and 0.8 mM, respectively. Cells were lysed by incubation at 4°C for 30 min followed by heat shock at 37°C for 5 min. The lysates were cleared by centrifugation, and the subsequent crude extracts were assayed as described below.

Nuclease assays. Uniformly labeled bacteriophage T7 ³H-DNA was prepared as previously described (7) with [³H]thymidine (Dupont NEN). End-labeled substrates were prepared from linearized 3-kb pBSSK plasmid DNA and purified with Qiagen QiaExII kits. 3'-End-labeled substrate was generated from XbaI-digested pBSSK by "fill-in" synthesis with the E. coli DNA polymerase I large fragment (New England Biolabs) in the presence of [-³²P]dATP (Dupont NEN). 5'-End-labeled substrate was generated from HindIII-digested pBSSK DNA, treated with shrimp alkaline phosphatase (United States Biochemical Corp.), and phosphorylated with T4 polynucleotide kinase (New England Biolabs) and [-³²P]ATP (Dupont NEN).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Sequence comparison of archaeal hypothetical proteins with eubacterial RecJ and polyphosphatase PPX1 of yeast. Database searches of eubacterial genomes with E. coli RecJ and the gapped BLASTP program (1) revealed 14 sequences that define seven conserved sequence motifs (10) (Fig. 1A). Individual pairwise alignments from such searches yielded E values (the probability of matching by chance) of 10²¹ to 10⁶⁹, indicating very strong similarity with E. coli RecJ exonuclease, including open reading frames from the diverse genera Aquifex, Bacillus, Borrelia, Chlamydia, Chlamydophila, Erwinia, Haemophilus, Helicobacter, Rickettsia, Synechocystis, and Treponema. Both Bacillus subtilis and Helicobacter pylori carry two RecJ-related sequences, one with somewhat less similarity to E. coli RecJ. (The second Bacillus sequence is encoded on an SP2 prophage.)

25

3

9

5

92

159

5

45

84

87

Cloning of RecJ-related genes from Methanococcus and genetic complementation of recJ in E. coli. To assess the ability of the methanococcal genes to complement a recJ mutation, the open reading frames MJ0977, MJ0831, and MJ1198 were amplified by PCR from M. jannaschii chromosomal DNA, fused to a ribosome-binding site sequence, and cloned into a low-copy-number expression vector of E. coli. The plasmid derivatives were introduced by transformation into an E. coli recJ mutant strain, STL3266, and tested for their ability to enhance UV survival. A T7 phage promoter on the vector controls expression of the genes. The addition of IPTG to the growth medium triggers the expression of the genes by induction of T7 RNA polymerase from a resident DE3 lysogenic phage.

View larger version (17K): [in this window] [in a new window]   FIG. 2.   UV survival of E. coli STL3266 transformed with plasmids containing M. jannaschii genes, E. coli recJ, or the vector alone. Symbols: , vector; , E. coli recJ; , MJ0831; , MJ0977; , MJ1198. Values are the means of five or more experiments.

Overexpression and characterization of MJ0977 nuclease activity. Plasmids pLAR101, pLAR102, and pLAR103 were introduced into STL2329 for protein expression. Crude extracts were prepared from cells induced for expression of each of the three methanococcal genes. Extracts were tested for the ability to degrade a uniformly ³H-labeled T7 DNA substrate. At 37°C, extracts from cells expressing the MJ0977-encoded protein exhibited a sixfold increase in nuclease activity over cells carrying the vector alone (Fig. 3). No increased DNase activity was evident in extracts from cells expressing the product of MJ0831 or MJ1198 (data not shown).

View larger version (41K): [in this window] [in a new window]   FIG. 3.   Thermostability of ssDNA nuclease activity in crude extracts of cells induced for the expression of the MJ0977 product or E. coli RecJ. Nuclease assays were performed under standard conditions (as described in Materials and Methods) with noted temperature variations. The specific ssDNA nuclease activity of extracts from cells containing the vector alone was <35 U/mg of protein.

View larger version (26K): [in this window] [in a new window]   FIG. 4.   Optimization of nuclease activity on ssDNA in crude extracts of cells induced for the expression of the MJ0977 product. T7 nuclease assays were performed at 65°C under standard conditions (as described in Materials and Methods) with noted variations in Tris buffer pH (measured at 65°C) (A), NaCl concentration (B), Mg2+ concentration (C), or Mn2+ concentration in the absence of Mg2+ (D).

View larger version (13K): [in this window] [in a new window]   FIG. 5.   DNA end preference of the expressed MJ0977 product. Values were derived by subtracting the fraction of acid-soluble radioactivity released by extracts from cells expressing MJ0977 from that released by extracts from cells containing the vector alone. Symbols: , 5'-end-labeled ssDNA substrate; , 3'-end-labeled substrate.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

RecJ-related sequences appear to be especially abundant in archaeal genomes. Alignments of these diverged archaeal sequences define several groups that share sequence motifs with eubacterial RecJ exonucleases. We sought to determine which, if any, of these sequences had functional similarity to RecJ. We have demonstrated that two sequences from M. jannaschii, MJ0831 and MJ0977, members of the group with the strongest similarity to eubacterial RecJ sequences, can partially complement a recJ mutation in E. coli K-12. This result indicates a conservation of function between these proteins and E. coli RecJ. A third opening reading frame, MJ1198, yielded complementation too weak to assess its functional relevance.

The expression of one of the genes, MJ0977, resulted in high levels of a thermostable ssDNA nuclease activity. This nuclease had many properties in common with E. coli RecJ (9): it was highly specific for ssDNA, required Mg²⁺ or Mn²⁺ as a cofactor, and attacked 5' ends with a higher avidity. Despite overall weak sequence similarity between the MJ0977 product and E. coli RecJ (Table 1), these genetic and biochemical data support the contention that these proteins are indeed functionally related.

Failure to achieve complete genetic complementation with MJ0977 or MJ0831 was not unexpected. Because M. jannaschii is a thermophilic organism, its proteins may be less apt to function in a mesophilic organism such as E. coli. In addition, different biochemical properties, such as a lower affinity for substrate, catalytic rate, processivity, or a failure to interact properly with other E. coli proteins, may account for incomplete functionality in E. coli.

Although both MJ0977 and MJ0831 demonstrated similar abilities to complement an E. coli recJ mutation, we detected ssDNA nuclease activity associated only with the expression of MJ0977. A failure to detect nuclease activity for MJ0831 could have been due to poorer stability in E. coli or to enzymatic requirements not supplied by our reaction conditions. The two proteins might also differ in their thermostability: if the MJ0831 product is considerably less thermostable than the MJ0977 product, it might be difficult to detect over the background of nucleases in E. coli crude extracts. Further purification of the MJ0831-encoded peptide should clarify this issue.

The aspartate, histidine, and arginine residues found in conserved motifs 1, 2, 3, 4, 6, and 7 of the E. coli RecJ exonuclease have been shown to be important for its biological and biochemical functions (10). Most of these residues are also conserved among the archaeal groups (Fig. 1). Some of these motifs may define residues that interact with the catalytic divalent metal ion(s) required for phophoesterase activity. In E. coli RecJ, mutation of the aspartate residues in motifs 1, 3, and 4 results in a dominant-negative phenotype in vivo and a complete loss of nuclease activity in vitro (10). In the archaeal proteins, one aspartate (corresponding to D79 of motif 1 of E. coli RecJ) has been replaced by histidine and displaced in its position within motif 1 for groups F, G, and H. The DHH of motif 3 is probably the most diagnostic of this family of proteins (2) and, accordingly, all three residues are essential for E. coli RecJ activity (10). The NP of motif 3A, invariant in the eubacterial RecJ proteins, is reduced to a single proline or absent in most of the archaeal sequences, with groups B and D as the exceptions. Motif 5, barely recognizable among some members of group B, is missing or not recognizable (perhaps reduced to a single lysine or arginine) in the other groups. Mutations in this motif have not yet been assayed for their impact on the activity of E. coli RecJ exonuclease. In motif 6, only the arginine is invariant; in E. coli RecJ, mutation of this residue to alanine results in a complete loss of exonuclease activity and a recessive mutant phenotype. Although the glycine richness of motif 7 is conserved among archaeal groups, the histidine often is not, and the motif is entirely missing in the methanococcal Ppx1 sequence. This lack of conservation may not be so surprising, since the mutation of motif 7 histidine to alanine in RecJ exonuclease produced a somewhat leaky phenotype (10).

Because of the high degree of sequence similarity among E. coli RecJ nuclease, the MJ0977 product, and the products of the group B sequences (Fig. 1B), we think it is likely that all group B proteins are nucleases for DNA or RNA. Although it is tempting to speculate, based on sequence similarity, that group C and D proteins are also nucleases and that group F, G, and H proteins are phosphatases, further experimentation is required to assign a biological substrate to these proteins.