6.1.1 DnaQ

DnaQ, also known as MutD or epsilon subunit of E. coli DNA polymerase III holoenzyme, encodes the 3′–5′ exonuclease (DEDDh) that performs proofreading in DNA replication (McHenry, Reference Mchenry1985). The 3′–5′ exonuclease may exist as a domain of DNA polymerase or a separate polypeptide chain in the replisome. As a domain of the Klenow fragment, the 3′–5′ exonuclease (DEDDy) was one of the earliest crystal structures of nucleic acid enzymes determined (Ollis et al. Reference Ollis, Brick, Hamlin, Xuong and Steitz1985) and the structures of Klenow-oligonucleotide complexes led to the proposal of the two-metal-ion mechanism (Beese & Steitz, Reference Beese and Steitz1991; Freemont et al. Reference Freemont, Friedman, Beese, Sanderson and Steitz1988). The structure of epsilon subunit is homologous to the 3′–5′ exonuclease domain of Klenow fragment, and only one metal ion was detected in the absence of ssDNA (Hamdan et al. Reference Hamdan, Carr, Brown, Ollis and Dixon2002). The structure of the Klenow fragment complexed with ssDNA substrate reveals the coordination of two-metal ions (Fig. 6a). Although the substrate DNA is single stranded, its conformation is highly similar to a strand in a double helix (Brautigam & Steitz, Reference Brautigam and Steitz1998; Brautigam et al. Reference Brautigam, Sun, Piccirilli and Steitz1999). These observations indicate that proper binding of two metal ions requires the presence of a cognate DNA substrate. The importance of the scissile phosphate in metal ion coordination is revealed by the sulfur substitution experiments. Replacement of the pro-Sp oxygen with sulfur essentially abolishes the metal-ion binding and the nuclease activity, and replacement of the 3′ bridging oxygen (leaving group) affects the metal-ion binding, particularly in the B site, whereas the replacement of pro-Rp oxygen moderately affects the metal ions, active site configuration and nuclease activity (Brautigam & Steitz, Reference Brautigam and Steitz1998; Brautigam et al. Reference Brautigam, Sun, Piccirilli and Steitz1999).

6.1.2 E. coli ExoI and ExoX

The 3′ ExoI is the first deoxyribonuclease purified and characterized from E. coli (Lehman & Nussbaum, Reference Lehman and Nussbaum1964). It specifically degrades ssDNA from the 3′-end and is implicated in various DNA repair pathways. The crystal structure of ExoI reveals a C-shaped monomer (PDB: 1FXX) (Breyer & Matthews, Reference Breyer and Matthews2000). The DEDDh catalytic core is followed by a C-terminal SH3-like domain and surrounded by α-helices. The structure suggests a mechanism for processive digestion of ssDNA. E. coli ExoX is highly similar to ExoI in the catalytic domain and ssDNA preference. But, without the C-terminal domain, ExoX is about a half of ExoI in size (Viswanathan & Lovett, Reference Viswanathan and Lovett1999). Homologs of E. coli ExoI and X are widely spread in bacteria.

6.1.3 TREX1 and TREX2

TREX1 and TREX2 are the major mammalian 3′–5′ exonucleases and also prefer ssDNA substrate. They can perform proofreading function for polymerases that lack an intrinsic 3′–5′ exonuclease activity (Shevelev et al. Reference Shevelev, Ramadan and Hubscher2002). Defective TREX1 (aka DNase III) causes the Aicardi–Goutieres syndrome in humans, a severe neurological disease, and systemic lupus erythematosus probably due to abnormal innate immune responses to the accumulation of intracellular nucleic acids (Crow et al. Reference Crow, Hayward, Parmar, Robins, Leitch, Ali, Black, Van Bokhoven, Brunner, Hamel, Corry, Cowan, Frints, Klepper, Livingston, Lynch, Massey, Meritet, Michaud, Ponsot, Voit, Lebon, Bonthron, Jackson, Barnes and Lindahl2006a). Mice lacking TREX1 develop inflammatory myocarditis (Morita et al. Reference Morita, Stamp, Robins, Dulic, Rosewell, Hrivnak, Daly, Lindahl and Barnes2004). It has recently been shown that TREX1 removes DNA of endogenous retroelements and prevents a type I interferon-mediated autoimmune response to self-nucleic acids (Stetson et al. Reference Stetson, Ko, Heidmann and Medzhitov2008). TREX1 is also a part of SET complex and degrades DNA during granzyme A-mediated caspase-independent cell death (Chowdhury et al. Reference Chowdhury, Beresford, Zhu, Zhang, Sung, Demple, Perrino and Lieberman2006). Structures of TREX1 and TREX2 have been determined and are highly similar to that of DnaQ (PDB: 2OA8, 2IOC, 2O4I and 1Y97) (Brucet et al. Reference Brucet, Querol-Audi, Serra, Ramirez-Espain, Bertlik, Ruiz, Lloberas, Macias, Fita and Celada2007; de Silva et al. Reference De Silva, Choudhury, Bailey, Harvey, Perrino and Hollis2007; Perrino et al. Reference Perrino, Harvey, Mcmillin and Hollis2005). TREX1 and TREX2 differ from DnaQ in being obligatory dimers, and the conserved residue in addition to DEDD in the active site is His rather than Tyr (Fig. 6b).

6.1.4 WRN exonuclease

Defective WRN protein is responsible for the Werner Syndrome, or premature aging (Muftuoglu et al. Reference Muftuoglu, Oshima, Von Kobbe, Cheng, Leistritz and Bohr2008). WRN contains both a DNA helicase and 3′–5′ exonuclease activity. The isolated WRN exonuclease domain is monomeric and can degrade the 3′-ends of dsDNA with a 5′ overhang (Perry et al. Reference Perry, Yannone, Holden, Hitomi, Asaithamby, Han, Cooper, Chen and Tainer2006a). The structure of the WRN exonuclease domain contains the DEDDy motif and is similar to other members of the DnaQ family. Two divalent ions were found in the active site in the presence of 50 mM metal ions even without DNA substrate (Perry et al. Reference Perry, Yannone, Holden, Hitomi, Asaithamby, Han, Cooper, Chen and Tainer2006a).

6.1.5 RNase T and Orn

Bacterial RNase T and Orn (oligoribonuclease) are homodimeric 3′–5′ exonucleases with the DEDDh motif. Both prefer ssRNA (Zuo et al. Reference Zuo, Zheng, Wang, Chruszcz, Cymborowski, Skarina, Savchenko, Malhotra and Minor2007) (PDB: 2F96, 2IS3, 1YTA, 2IGI, 2GBZ and 1J9A). RNase T is essential for 5S and 23S rRNA maturation and for processing of precursor tRNA (Deutscher & Li, Reference Deutscher and Li2001). Orn can also hydrolyze DNA in the 3′–5′ direction (Fiedler et al. Reference Fiedler, Vincent, Zuo, Gavrialov and Malhotra2004). In the absence of Orn, E. coli cells grow slowly. Interestingly, the two active sites of the dimeric RNase T and Orn are on opposite surfaces and appear to be independent of each other.

6.1.6 RNase D and Rrp6

RNase D has the DEDDy motif and processes the 3′ end of structured RNA (Zuo et al. Reference Zuo, Wang and Malhotra2005). It contains two HRDC (helicase RNase DC-terminal) nucleic acid binding domains in addition to the DnaQ-like catalytic domain. The circular substrate-binding tunnel suggests processivity of RNase D (PDB: 1YT3). One Zn²⁺ ion is tightly bound in the active site and the second Zn²⁺ ion has low occupancy (Zuo et al. Reference Zuo, Wang and Malhotra2005). The yeast homologue of RNase D, Rrp6p, is a component of the nuclear exosome for the degradation of unstable mRNA transcripts and maturation of rRNA, small nuclear RNA (snRNA) and small nucleolar RNAs (snoRNAs) (for review, see Houseley et al. Reference Houseley, Lacava and Tollervey2006). In the presence of 10 mM MnCl₂, two-metal ions are also found in the active site of yeast Rrp6 (PDB: 2HBK) (Midtgaard et al. Reference Midtgaard, Assenholt, Jonstrup, Van, Jensen and Brodersen2006).

6.1.7 Poly A-specific ribonuclease (PARN), Pan2 and Pop2

PARN is a dimeric processive 3′-exonuclease with the DEDDh motif. In addition to the DnaQ-like catalytic domain, PARN has RNA-binding domains (PDB: 2A1S) (Wu et al. Reference Wu, Reuter, Lilie, Liu, Wahle and Song2005). Related to PARN, Pop2 is a component of the Ccr4–NOT complex and functions in mRNA degradation. Pop2 is likely to participate in mRNA deadenylation both directly with its 3′–5′-exonuclease activity and indirectly by facilitating the Ccr4 nuclease activity (Parker & Song, Reference Parker and Song2004). Curiously, although the DEDDh motif is replaced by SEDQt, yeast Pop2 (yPop2) still has the RNase activity (Thore et al. Reference Thore, Mauxion, Seraphin and Suck2003). It is unknown whether the active site still uses two-metal-ion catalysis. Fission Yeast Pop2p, however, has the normal DEDDh motifs, and its structure has been determined at high resolution and two-metal ions are found in the active site (Andersen et al. Reference Andersen, Jonstrup, Van and Brodersen2009; Jonstrup et al. Reference Jonstrup, Andersen, Van and Brodersen2007). Pan2 (poly-A nuclease), which is a component of the third 3′–5′ exoribonuclease complex for mRNA degradation and maturation, is also predicted to have a DnaQ-like exonuclease domain with the DEDDh motif.

6.1.8 ERI-1, 3′hExo and CRN-4 exonucleases

Eukaryotic ERI-1 (stands for exoribonuclease 1), human 3′-Exo (3′hExo) and C. elegans CRN-4 nucleaes share the conserved catalytic domain and the DEDDh motif. However, they are not functional orthologs perhaps owing to different additional structural domains. 3′hExo is responsible for histone mRNA degradation (Cheng & Patel, Reference Cheng and Patel2004), ERI-1 is implicated in 5·8S rRNA processing and production of small RNA for RNAi (Gabel & Ruvkun, Reference Gabel and Ruvkun2008), whereas CRN-4 is implicated in apoptotic DNA degradation (Parrish & Xue, Reference Parrish and Xue2006). The crystal structure of the full-length Caenorhabditis elegans CRN-4 has been reported (Hsiao et al. Reference Hsiao, Nakagawa, Shi, Mitani, Xue and Yuan2009). It is a dimeric nuclease and cleaves both RNA and DNA. A C-terminal Zn-domain is appended to the catalytic core and may influence CRN-4's preference to exonucleolytically degrade dsDNA over ssDNA.

6.2.1 RNase H1

RNase H1 (also referred to as RNase H) is the founding member of this large eclectic superfamily of endonucleases. It is conserved from bacteria, retroviruses to humans and cleaves an RNA strand when it is hybridized to DNA. Cellular RNase H1 primarily degrades RNA primers in DNA replication and RNAs in random RNA–DNA hybrids (R loops) (Cerritelli & Crouch, Reference Cerritelli and Crouch2009; Tadokoro & Kanaya, Reference Tadokoro and Kanaya2009). RNase H activity is essential for retroviral reverse transcription (Champoux & Schultz, Reference Champoux and Schultz2009). Although not sequence-specific, RNase H1 requires a minimum of four ribonucleotides forming RNA/DNA hybrid for cleavage to occur (Tadokoro & Kanaya, Reference Tadokoro and Kanaya2009). Substrate binding and specificity can be further strengthened by one or two RNA/DNA hybrid-binding domains (HBD) found in many cellular RNase H1 and by the DNA polymerase domain in retroviral reverse transcriptase (Champoux & Schultz, Reference Champoux and Schultz2009; Nowotny et al. Reference Nowotny, Cerritelli, Ghirlando, Gaidamakov, Crouch and Yang2008).

RNase H catalysis by the two-metal-ion mechanism has been established in a series of enzyme–nucleic acid complexes along the reaction pathway by X-ray crystallography (Nowotny et al. Reference Nowotny, Gaidamakov, Crouch and Yang2005; Nowotny & Yang, Reference Nowotny and Yang2006). Four carboxylates (D10, E48, D70 and D134 according to E. coli RNase H) form the catalytic core (Fig. 7a). The first and third carboxylates coordinate the two metal ions (Fig. 7b). The second carboxylate, Glu located on helix αA, is unique to RNase H1 and binds both the 2′-OH of RNA substrate and metal ion B, thereby enhancing the catalysis specificity (Nowotny et al. Reference Nowotny, Gaidamakov, Crouch and Yang2005). The fourth carboxylate coordinates metal ion A only. Although its replacement by Asn significantly reduces catalytic activity, it can be restored by the substitution of Mn²⁺ for Mg²⁺ (Nowotny & Yang, Reference Nowotny and Yang2006; Nowotny et al. Reference Nowotny, Gaidamakov, Crouch and Yang2005). In addition, the substrate actively participates in cleavage. The pro-Sp oxygen of the scissile phosphate coordinates both metal ions and the pro-Rp oxygen of the 3′ neighboring phosphate participates in nucleophile formation (Haruki et al. Reference Haruki, Tsunaka, Morikawa, Iwai and Kanaya2000; Nowotny et al. Reference Nowotny, Gaidamakov, Crouch and Yang2005).

Fig. 7.

The RNase H superfamily consists of various endonucleases. (a) E. coli RNase H1 is the prototype of this superfamily. (b) The active site of Bacillus halodurans RNase H complexed with an RNA/DNA hybrid substrate. (c) Tn5 transposase (complexed with a cleaved DNA product) represents the HIV integrase-like recombinases with the DDE motif. (d) RNase H2 has the same topology as RNase H1 but the second and fourth conserved carboxylates are differently located. (e) The active site of Tth Argonaute, essential for RNAi pathways is similar to that of RNase H1. (f) RuvC is a HJ resolvase. (g) The C-terminal nuclease domain of UvrC may use two conserved carboxylates and a His for metal-ion coordination and catalysis. (h) Monomeric Hsp70 ATPases contain two RNase H-like domains. The duplicated domains are both shown in blue- to red-colored gradients. The Asp residues in the first β-strand (D10 and D199) are conserved throughout the Hsp70 family and essential for the ATPase activity. Proteins, nucleic acids and divalent cations are shown in the same scheme as in Fig. 6. The green spheres represent the K⁺ ions.

6.2.2 Transposase and retroviral integrase with the DDE motif

When the structure of HIV integrase and bacterial Mu transposase were determined, it was a surprise to find that these DNA endonucleases bear clear similarity to RNase H1 (Dyda et al. Reference Dyda, Hickman, Jenkins, Engelman, Craigie and Davies1994; Rice & Mizuuchi, Reference Rice and Mizuuchi1995). Sequence alignment predicts that many cellular and retroviral enzymes catalyzing cut-and-paste sequence-specific transposition/recombination have the RNase H fold and a DDE motif, which is approximately equivalent to the 1st, 3rd and 4th carboxylates of RNase H1 (Fig. 7a–c). They include the human RAG1 and RAG2 proteins essential for immunoglobulin gene rearrangement (Kim et al. Reference Kim, Dai, Mundy, Yang and Oettinger1999; Landree et al. Reference Landree, Wibbenmeyer and Roth1999), and eukaryotic Mariner and Hermes transposases (Hickman et al. Reference Hickman, Perez, Zhou, Musingarimi, Ghirlando, Hinshaw, Craig and Dyda2005; Richardson et al. Reference Richardson, Colloms, Finnegan and Walkinshaw2009). Integrases and transposases are at least dimeric when associated with two donor-DNA ends and engaged in recombination. As revealed by the Tn5 transposase–DNA complex structures, the three carboxylates in these integrases and transposases are symmetrically arranged around the two metal ions (Davies et al. Reference Davies, Goryshin, Reznikoff and Rayment2000; Steiniger-White et al. Reference Steiniger-White, Rayment and Reznikoff2004). The first Asp coordinates both and the remaining two coordinate one metal ion each (Fig. 7c). The recently reported structure of a retroviral integrase complexed with two donor DNA ends, which are nearly perpendicular to each other rather than antiparallel as in the case of Tn5, confirms the active site configuration (Hare et al. Reference Hare, Gupta, Valkov, Engelman and Cherepanov2010). The symmetry surrounding the two metal ions has implications for the DNA transposition reaction. These integrase and transposase often catalyze two or more consecutive phosphoryltransfer reactions and the 3′-OH product of the first cleavage reaction can be used as a nucleophile for the second strand transfer reaction (Kennedy et al. Reference Kennedy, Haniford and Mizuuchi2000). It has thus been proposed that the two metal ions switch their roles in coordinating the nucleophile and the leaving group in the two consecutive steps (Nowotny, Reference Nowotny2009; Nowotny et al. Reference Nowotny, Gaidamakov, Crouch and Yang2005; Steitz & Steitz, Reference Steitz and Steitz1993).

6.2.3 RNase H2

RNase H2 is the second cellular RNase H found from bacteria to humans (Cerritelli & Crouch, Reference Cerritelli and Crouch2009). Eukaryotic RNase H2 consists of two auxiliary subunits, and all three subunits of human RNase H2 have been implicated in the Aicardi–Goutieres syndrome, a heritable autoimmune disease caused by either TREX1 or RNase H2 deficiency due to the accumulation of residual RNA and DNA (Crow & Rehwinkel, Reference Crow and Rehwinkel2009; Crow et al. Reference Crow, Leitch, Hayward, Garner, Parmar, Griffith, Ali, Semple, Aicardi, Babul-Hirji, Baumann, Baxter, Bertini, Chandler, Chitayat, Cau, Dery, Fazzi, Goizet, King, Klepper, Lacombe, Lanzi, Lyall, Martinez-Frias, Mathieu, Mckeown, Monier, Oade, Quarrell, Rittey, Rogers, Sanchis, Stephenson, Tacke, Till, Tolmie, Tomlin, Voit, Weschke, Woods, Lebon, Bonthron, Ponting and Jackson2006b). RNase H2 differs from RNase H1 in that it can cleave a single ribonucleotide embedded in a DNA duplex. Crystal structures of bacterial RNase H2 reveal an RNase H1-like structure with altered locations of catalytic carboxylates (Fig. 7d). Only the first and third of the four catalytic carboxylates are superimposable with those of RNase H1. The second carboxylate (E) is immediately located next to the first one (D) instead of 40 residues away. The reaction coordinate is yet to be determined.

6.2.4 Argonaute and PIWI

Argonautes (Ago) and PIWI are the key players in the RNAi pathway and the RNase activity cleaves target mRNA when it forms base pairs with guide RNA (Joshua-Tor, Reference Joshua-Tor2006; Nowotny & Yang, Reference Nowotny and Yang2009). But the catalytic residues are not always conserved, and some Ago and PIWI without catalytic activity function in RNAi pathways by binding complementary guide and target RNA duplexes (Farazi et al. Reference Farazi, Juranek and Tuschl2008; Joshua-Tor, Reference Joshua-Tor2006). For those having RNase activity, the three catalytic residues, either DDD or DDH, are approximately equivalent to the DDE in the integrase family and the 1st, 3rd and 4th carboxylates of RNase H1 (Joshua-Tor, Reference Joshua-Tor2006). The catalytic mechanism is revealed in a series of elegant crystallographic studies of Ago–nucleic acid complexes (Wang et al. Reference Wang, Juranek, Li, Sheng, Tuschl and Patel2008, Reference Wang, Juranek, Li, Sheng, Wardle, Tuschl and Patel2009). Similar to transposases, the first carboxylate coordinates the two metal ions and the remaining two coordinate one metal ion each (Fig. 7e).

6.2.5 RuvC and Ydc2

Bacterial RuvC and mitochondrial Cce1/Ydc2 are the HJ resolvases that have the RNase H fold (Ariyoshi et al. Reference Ariyoshi, Vassylyev, Iwasaki, Nakamura, Shinagawa and Morikawa1994; Ceschini et al. Reference Ceschini, Keeley, Mcalister, Oram, Phelan, Pearl, Tsaneva and Barrett2001; Iwasaki et al. Reference Iwasaki, Takahagi, Shiba, Nakata and Shinagawa1991; West & Connolly, Reference West and Connolly1992; White & Lilley, Reference White and Lilley1997). These proteins are obligatory dimers (Bennett et al. Reference Bennett, Dunderdale and West1993). The active site consists of four carboxylates, and the first two appear to be equivalent to the 1st and 3rd carboxylates of RNase H1, but the carboxylate at the end of β4 is Glu instead of Asp (Fig. 7f). The last two carboxylates are located on the C-terminal α-helix two residues apart. Based on biochemical analyses, a protein–DNA complex model has been proposed (Fogg et al. Reference Fogg, Kvaratskhelia, White and Lilley2001; Yoshikawa et al. Reference Yoshikawa, Iwasaki and Shinagawa2001). How these carboxylates together with the DNA substrate coordinate the metal ions is yet to be elucidated.

6.2.6 UvrC (C-terminal domain)

UvrC is a bacterial endonuclease involved in nucleotide excision repair. It makes two incisions, one on each side of a lesion (Verhoeven et al. Reference Verhoeven, Van Kesteren, Moolenaar, Visse and Goosen2000). UvrC has two nuclease domains in one polypeptide chain, one for each incision. The C-terminal domain of UvrC, which makes the incision 5′ to a lesion, is homologous to RNase H and shares two catalytic carboxylates with RNase H1 (1st and 3rd) (Karakas et al. Reference Karakas, Truglio, Croteau, Rhau, Wang, Van Houten and Kisker2007). But the third catalytic residue in UvrC may be a His that participates in single metal-ion coordination in the absence of DNA substrate (Fig. 7g). Interestingly, replacement of this conserved His with Ala only moderately reduces the enzymatic activity (Karakas et al. Reference Karakas, Truglio, Croteau, Rhau, Wang, Van Houten and Kisker2007). How the active site is formed upon substrate binding and whether the conserved Asp adjacent to the His is involved in catalysis awaits future analysis.

6.2.7 Structural roles and the ATPase catalytic core of Hsp70

The RNase H-like fold has been found in many proteins, where it plays a structural role, e.g. in T7 DNA polymerase and mismatch recognition protein MutS (Obmolova et al. Reference Obmolova, Ban, Hsieh and Yang2000) and RNase E (Callaghan et al. Reference Callaghan, Marcaida, Stead, Mcdowall, Scott and Luisi2005). Intriguingly, it is found in pre-mRNA splicing protein Prp8 (Pena et al. Reference Pena, Rozov, Fabrizio, Luhrmann and Wahl2008; Ritchie et al. Reference Ritchie, Schellenberg, Gesner, Raithatha, Stuart and Macmillan2008), where the first carboxylate is conserved and may play an essential structural rather than a catalytic role (Yang et al. Reference Yang, Zhang, Xu, Heroux and Zhao2008). Finally, two RNase H-like domains are found in actin and Hsp70 proteins and form the ATPase catalytic core (Flaherty et al. Reference Flaherty, Mckay, Kabsch and Holmes1991). Most interestingly, a conserved Asp is found in the first β-strand of each RNase H domain in Hsp70 (D10 and D199) just as in the nucleases of the RNase H superfamily and both are essential for cation binding (K⁺ and Mg²⁺) and ATPase activity (Johnson & McKay, Reference Johnson and Mckay1999) (Fig. 7h).

6.3.1 Restriction endonucleases (BamHI, HincII, MutH, EcoR124I, etc.)

Nearly 70% REases belong to this superfamily (Orlowski & Bujnicki, Reference Orlowski and Bujnicki2008). Many REases in this family have a Pro immediately before the conserved Asp making the PD-EXK motif and a conserved N-terminal Glu, which contributes to the metal-ion coordination as observed in EcoRV, BglI, BamHI and MutH of type-II REase family (Horton & Perona, Reference Horton and Perona2004; Lee et al. Reference Lee, Chang, Joseph, Ghirlando, Rao and Yang2005; Newman et al. Reference Newman, Lunnen, Wilson, Greci, Schildkraut and Phillips1998; Viadiu & Aggarwal, Reference Viadiu and Aggarwal1998; Winkler et al. Reference Winkler, Banner, Oefner, Tsernoglou, Brown, Heathman, Bryan, Martin, Petratos and Wilson1993). These REases are most often dimeric (Bitinaite et al. Reference Bitinaite, Wah, Aggarwal and Schildkraut1998), but monomeric REase also exist (Lee et al. Reference Lee, Chang, Joseph, Ghirlando, Rao and Yang2005; Sokolowska et al. Reference Sokolowska, Kaus-Drobek, Czapinska, Tamulaitis, Szczepanowski, Urbanke, Siksnys and Bochtler2007; Xu et al. Reference Xu, Roberts and Guo2005). By varying dimeric interfaces and thus the relative positions of the two catalytic centers, dimeric REases can cleave DNA to generate blunt ends or staggered ends with various 5′ or 3′ overhangs (Fig. 8b and c) (Ban & Yang, Reference Ban and Yang1998b; Newman et al. Reference Newman, Lunnen, Wilson, Greci, Schildkraut and Phillips1998). The catalytic module invariably approaches DNA from the minor groove side, and the sequence-specific binding is conducted by a separate module/subdomain in the major groove (Fig. 8d). The first two carboxylates of the DEK motif coordinate the metal ions. The third, which usually is hydrogen bonded with both the nucleophilic water and the DNA-binding module in the major groove (Fig. 8d), has been attributed to couple DNA sequence recognition with the cleavage reaction (Lee et al. Reference Lee, Chang, Joseph, Ghirlando, Rao and Yang2005).

Type-I and type-III restriction/modification systems differ from type-II in the requirement of nucleotide triphosphate and DNA translocation (Dryden et al. Reference Dryden, Murray and Rao2001; Murray, Reference Murray2000), but their endonucleases have a similar active site composed of D and EXK motifs (Niv et al. Reference Niv, Ripoll, Vila, Liwo, Vanamee, Aggarwal, Weinstein and Scheraga2007). The crystal structure of EcoR124I of type-I motor-endonuclease subunit shows that the conserved DEK motif indeed forms the catalytic center although the Pro before the first conserved D is absent (Lapkouski et al. Reference Lapkouski, Panjikar, Janscak, Smatanova, Carey, Ettrich and Csefalvay2009).

6.3.2 Phage λ and RecE 5′–3′ exonuclease

The active site of the PD-(D/E)XK motif is found among bacterial and phage 5′–3′ exonucleases, which act on dsDNA and produce a protruding 3′ ssDNA for strand annealing or invasion in homologous recombination (Kolodner et al. Reference Kolodner, Hall and Luisi-Deluca1994; Stahl et al. Reference Stahl, Thomason, Poteete, Tarkowski, Kuzminov and Stahl1997). RecE together with the ssDNA annealing protein RecT has been exploited for enhanced homologous recombination for in vivo gene manipulation (Muyrers et al. Reference Muyrers, Zhang and Stewart2000; Zhang et al. Reference Zhang, Buchholz, Muyrers and Stewart1998). Phage λ exonuclease (aka Red recombinase) and E. coli RecE are trimeric and tetrameric, respectively, and both form a toroidal structure, whose central hole is proposed to bind DNA for processive degradation (Kovall & Matthews, Reference Kovall and Matthews1997; Zhang et al. Reference Zhang, Xing, Herr and Bell2009) (Fig. 8e). A conserved Glu precedes the DEK motif in Phage λ exonuclease (Fig. 8f). In RecE, the DEK motif becomes DDK and it is preceeded by a conserved His (Fig. 8g). Whether the multiple active sites in these highly processive exonucleases work sequentially or stochastically is not known yet.

6.3.3 RecB and AdnAB ssDNA nuclease

The nuclease domains of the bacterial motor-nuclease machines RecBCD and AdnAB are closely related to RecE and REases (Singleton et al. Reference Singleton, Dillingham, Gaudier, Kowalczykowski and Wigley2004; Sinha et al. Reference Sinha, Unciuleac, Glickman and Shuman2009). They have both exonuclease and endonuclease activities on ssDNA (Yu et al. Reference Yu, Souaya and Julin1998). Although RecB contains a single nuclease domain, it is one of the three subunits in the helicase/nuclease RecBCD complex. RecBCD forms a cylindrical molecule and DNA threads through the central hole similar to the cases of RecE and phage λ exonuclease. But the DNA becomes unwound by the helicase activity carried by RecB and RecD of RecBCD or by AdnA and AdnB (Unciuleac & Shuman, Reference Unciuleac and Shuman2009). The nucleases then nick the ssDNA and ultimately generate 3′ ssDNA overhangs for homologous recombination (Dillingham & Kowalczykowski, Reference Dillingham and Kowalczykowski2008; Sinha et al. Reference Sinha, Unciuleac, Glickman and Shuman2009). Deviating from the canonical DEK motif, the Glu is replaced by Asp in RecB as in RecE. RecB does not have the Pro preceding the DDK motif. A conserved His N-terminal to the DDK is likely to participatein catalysis (Fig. 8h).

Dna2 ubiquitous in all eukaryotes has a helicase domain and a RecB-like nuclease domain containing the DEK motif (Budd et al. Reference Budd, Choe and Campbell2000). Dna2 works with ssDNA-binding protein during DNA replication and removes excessively long primers (Kim et al. Reference Kim, Kim, Ryu, Kim, Hurwitz and Seo2006; Masuda-Sasa et al. Reference Masuda-Sasa, Imamura and Campbell2006).

6.3.4 HJ resolvase: T7 endonuclease I

Phage T7 endonuclease I and archaeal Hjc are HJ resolvases, whose active sites contain the REase-like (E)-PD-EXK motif (Lilley & White, Reference Lilley and White2001). These HJ resolvases are homodimeric and recognize four-way junctions in a distorted stacked-X form as shown in the crystal structures of enzyme–DNA complexes (PDB: 2PFJ and 2WJ0) (Hadden et al. Reference Hadden, Déclais, Carr, Lilley and Phillips2007). Recognition and resolution of HJ are recently reviewed (Declais & Lilley, Reference Declais and Lilley2008). T7 endonuclease I cleaves DNA by the two-metal-ion mechanism with the two most conserved carboxylates coordinating two metal ions (Hadden et al. Reference Hadden, Déclais, Carr, Lilley and Phillips2007) (Fig. 8i).

6.3.5 Structure-specific endonuclease: XPF, Hef, Rad1 and Mus81

The archaeal and eukaryotic structure-specific nucleases, Hef, Rad1, XPF and Mus81 all share the conserved nuclease domain, known as ERCC4 domain. The catalytic motif of D-ERK is homologous to that of REases (Chang et al. Reference Chang, Kim, Choi, Lee and Cho2008; Newman et al. Reference Newman, Murray-Rust, Lally, Rudolf, Fadden, Knowles, White and Mcdonald2005; Nishino et al. Reference Nishino, Komori, Ishino and Morikawa2003) (Fig. 8j). These nucleases are required for DNA repair, processing of a stalled replication fork, and repair-associated recombination (for an in-depth review, read Ciccia et al. Reference Ciccia, Mcdonald and West2008). In addition to the DEK motif, two conserved carboxylates, Asp and Glu, N-terminal to it are located in the active site (Fig. 8j). DNA substrates can vary from a 3′ flap, replication fork, to a HJ, all of which have the common feature of a 3′ flap, and cleavage occurs at the single- and double-strand junction. Although archaeal Hef nuclease is homodimeric, all eukaryotic nucleases in this family, for example, Rad1, XPF and Mus81, form heterodimers (Rad1/Rad10, XPF/ERCC1 and Mus81/Eme1, Mus81/Eme2, or Mus81/Mms4) with only one active subunit. Interestingly, even with homodimeric nucleases it is likely that only one subunit is engaged in catalysis at a given time (Newman et al. Reference Newman, Murray-Rust, Lally, Rudolf, Fadden, Knowles, White and Mcdonald2005). Based on the conservation of the active site, two-metal-ion catalysis is predicted.

6.3.6 DNA mismatch-repair endonuclease very short-patch repair (Vsr)

E. coli Vsr endonuclease specifically cleave 5′ to the T of mismatched T:G base pair, which results from the deamination of 5-methyl-cytosine in a specific sequence (Hennecke et al. Reference Hennecke, Kolmar, Brundl and Fritz1991). The crystal structures of Vsr and Vsr complexed with DNA cleavage products reveal the REase fold in spite of the absence of the last two conserved catalytic residues of the DEK motif (Bunting et al. Reference Bunting, Roe, Headley, Brown, Savva and Pearl2003; Tsutakawa et al. Reference Tsutakawa, Jingami and Morikawa1999). A conserved Glu preceding the central uneven β-hairpin and two His residues C-terminal to it are in the catalytic center and may be involved in metal-ion coordination and nucleophile activation (Fig. 8k). In addition to the absolutely conserved Asp at the beginning of the uneven β-hairpin, a backbone carbonyl oxygen in the second β-strand also contributes to the metal-ion coordination in the Vsr–DNA cleavage product complex (Fig. 8k). The presence of two Mg²⁺ ions in the enzyme–product complex structure is highly suggestive of the two-metal-ion mechanism as observed for the other members of this superfamily.

6.3.7 Rai1/Dom3Z for mRNA degradation and processing

Human Dom3Z and yeast Rai1 are the newest enzymes found to have the conserved REase fold and the (E)-D-EXK motif (Xiang et al. Reference Xiang, Cooper-Morgan, Jiao, Kiledjian, Manley and Tong2009) (Fig. 8l). Before the crystal structure of Rat1–Rai1 complex was determined, Rai1 was thought to activate the 5′ to 3′ exoribonuclease Rat1 in the processing of RNA (Xue et al. Reference Xue, Bai, Lee, Kallstrom, Ho, Brown, Stevens and Johnson2000). But the structure reveals a divalent cation-binding site in Rai1. The authors proceed to show that the Rai1 homologue Dom3Z binds GDP, possesses a pyrophosphatase activity and can remove pyrophosphate from a 5′ triphosphorylated RNA. However, the GDP in the crystal structure is too far from the (E)-D-EXK motif to be either a substrate or product (Fig. 8l). Whether Rai1 has nuclease activity is yet to be determined.

6.4.1 5′ FENs: T4 RNase H, yeast Rad27 and human FEN1

The 5′–3′ exonuclease intrinsic to E. coli DNA Pol I is the founding member of the superfamily (Klett et al. Reference Klett, Cerami and Reich1968; Setlow & Kornberg, Reference Setlow and Kornberg1972), which includes Phage T4 RNase H, yeast Rad27 and human FEN1. The exonuclease is active on either RNA or DNA and its main function is to remove RNA primers in lagging strand synthesis (Bhagwat et al. Reference Bhagwat, Hobbs and Nossal1997; Liu et al. Reference Liu, Kao and Bambara2004). These nucleases also contain the structure-specific endonuclease activity and can cleave the 5′ displaced ssDNA flap, such as the displaced primers in lagging strand DNA synthesis. The latter activity gives rise to the name of flap endonuclease, FEN1 (Harrington & Lieber, Reference Harrington and Lieber1994). In bacteriophages, archaea and mammals, the 5′–3′ exo and FENs can exist as an independent protein, for example, T4 RNase H (a mis-nomenclature; Bhagwat et al. Reference Bhagwat, Hobbs and Nossal1997). In fact, T4 RNase H has the same activities and structure as the 5′–3′ exonuclease of E. coli Pol I (Fig. 9a, b). To avoid confusion, phage T4 RNase H has been renamed as T4FEN (Tomlinson et al. Reference Tomlinson, Atack, Chapados, Tainer and Grasby2010). FEN activity is not only involved in DNA replication but also in long-patch base excision and DSB repair (Shen et al. Reference Shen, Singh, Liu, Qiu, Zheng, Finger and Alas2005).

To date, crystal structures of several apo-nucleases in this family and phage T4 and archaeal FEN in complex with DNA have been determined (Ceska et al. Reference Ceska, Sayers, Stier and Suck1996; Chapados et al. Reference Chapados, Hosfield, Han, Qiu, Yelent, Shen and Tainer2004; Devos et al. Reference Devos, Tomanicek, Jones, Nossal and Mueser2007; Hwang et al. Reference Hwang, Baek, Kim and Cho1998; Kim et al. Reference Kim, Eom, Wang, Lee, Suh and Steitz1995). The catalytic center often contains 6–7 highly conserved Asp and Glu residues (Fig. 9a–c). The four most conserved carboxylates located after β1, β3 and β4 are reminiscent of the catalytic carboxylates in RNase H (Fig. 7a). The remaining negatively charged residues in the catalytic center is likely to be involved in binding of a third metal ion, which may facilitate substrate binding and thus indirectly contribute to the catalysis (Syson et al. Reference Syson, Tomlinson, Chapados, Sayers, Tainer, Williams and Grasby2008). However, in the absence of knowledge of the active site complete with scissile phosphate and metal ions, the reaction coordinates and chemistry for bond breakage remain uncertain.

6.4.2 XPG and HJ resolvases Gen1 and Yen1

XPG is required for nucleotide excision repair to incise at the 3′ side of a lesion and shares sequence homology with FEN1 (Habraken et al. Reference Habraken, Sung, Prakash and Prakash1993; O'Donovan et al. Reference O'DONOVAN, Davies, Moggs, West and Wood1994; Scherly et al. Reference Scherly, Nouspikel, Corlet, Ucla, Bairoch and Clarkson1993). Recently isolated eukaryotic HJ resolvases Gen1 and Yen1 are also found to be homologous to FEN1 (Ip et al. Reference Ip, Rass, Blanco, Flynn, Skehel and West2008). Interestingly, like FEN1, human XPG and its yeast homolog Rad2 both exhibit 5′-exonuclease activity (Habraken et al. Reference Habraken, Sung, Prakash and Prakash1994). Although no crystal structure has been reported for these endonucleases, the conserved active site residues and functional assay indicate that they all recognize a 5′ flap or branch structure and endonucleolytically cleave DNA at or adjacent to a branch point (including a ss–ds junction), likely by the two-metal-ion mechanism.

6.4.3 Human and yeast 5′ exonuclease 1 (Exo1)

It is perhaps not surprising to find that human and yeast 5′ exonuclease I, which are implicated in mismatch and DSB repair (Genschel et al. Reference Genschel, Bazemore and Modrich2002; Szankasi & Smith, Reference Szankasi and Smith1995), are homologous to FEN1 and XPG and share the same conserved active site. Like its homologues, Exo1 also has the FEN activity (Lee & Wilson, Reference Lee and Wilson1999; Tran et al. Reference Tran, Erdeniz, Dudley and Liskay2002). This FEN activity of ExoI is reported to be stimulated by Wrn protein (Sharma et al. Reference Sharma, Sommers, Driscoll, Uzdilla, Wilson and Brosh2003). Exo1 also uses the two-metal-ion mechanism.

6.4.4 5′ exoribonuclease Rat1 (XRN2) and XRN1

Recently, the structure of mouse Rat1 or XRN2, the 5′ exoribonuclease for RNA processing, has been reported (Xiang et al. Reference Xiang, Cooper-Morgan, Jiao, Kiledjian, Manley and Tong2009). Rat1 is highly similar to the structure of archaeal FEN1 with a large N-terminal loop wrapping around the protein and a helical tower (Fig. 9c, and d). The highly conserved carboxylate-rich active site implicates a metal-ion-dependent mechanism of catalysis by these XRNs. It is unclear what makes them specific for RNA degradation and whether XRNs may have endonuclease activities if a proper substrate-binding partner is present. Eukaryotic XRN1, the major 5′–3′ exonuclease of RNAs, shares the conserved catalytic residues of XRN2 (Szankasi & Smith, Reference Szankasi and Smith1996) and is predicted to have a FEN1-like structure and active site.

6.4.5 PIN domain RNA exonuclease

PIN (PilT N-terminal domain) nuclease is a recent addition to the FEN1 superfamily. It is a component of the exosome for RNA turnover and is also involved in nonsense mediated mRNA decay (Arcus et al. Reference Arcus, Backbro, Roos, Daniel and Baker2004; Clissold & Ponting, Reference Clissold and Ponting2000; Eberle et al. Reference Eberle, Lykke-Andersen, Muhlemann and Jensen2009). Crystal structures of PIN nucleases reveal a stripped down minimal core of FEN1 (Glavan et al. Reference Glavan, Behm-Ansmant, Izaurralde and Conti2006) (Fig. 9e). Like many members of this superfamily, PIN has both 5′-exo and endonucleolytic activities (Eberle et al. Reference Eberle, Lykke-Andersen, Muhlemann and Jensen2009; Schneider et al. Reference Schneider, Leung, Brown and Tollervey2009). Three Asp residues after β1, β3 and β4 are highly conserved among PIN nucleases as well as all FEN1 and XPG superfamily members. The fourth (two residues away from the third) can be Asp, Glu, Asn, or His (Arcus et al. Reference Arcus, Backbro, Roos, Daniel and Baker2004; Glavan et al. Reference Glavan, Behm-Ansmant, Izaurralde and Conti2006). These four are superimposable with the core catalytic residues of all nucleases in the FEN1 superfamily (Fig. 9).

6.7.1 DNase I

DNase I is most prevalent in mammals and is also found in bacteria, but not in low eukaryotes or plants. DNase I functions in apoptosis and has been implicated in the autoimmune disease SLE (systemic lupus erythematosus) (Martinez Valle et al. Reference Martinez Valle, Balada, Ordi-Ros and Vilardell-Tarres2008). The crystal structure of bovine pancreatic DNase I – DNA complex was determined nearly 20 years ago (Weston et al. Reference Weston, Lahm and Suck1992). In the crystal structures of DNase I without DNA, two Ca²⁺ ions were observed at a distance from the active site and thought to play structural and substrate-binding roles (Chen & Liao, Reference Chen and Liao2006). DNase I is most active in the presence of mixed Ca²⁺ and Mg²⁺ (Pan & Lazarus, Reference Pan and Lazarus1999). All the catalytic residues (four Asp and Glu, two His and one Asn) are located at the end of the central 4 β-strands (↓↑↑↓) of both β-sheets (Fig. 11a–c). Two conserved His residues surrounding the scissile phosphate have been shown to be important for the general acid-and-base catalysis, and the His mutations can be rescued by imidazole (Chen et al. Reference Chen, Lai, Lai, Huang, Lin and Liao2007a). The scissile phosphate is distorted and both pro-Rp and pro-Sp oxygens interact with DNase I (Figs 11c, 5c). Although divalent cation is chelated away by 20 mM EDTA and thus absent in the crystal structure of enzyme–substrate complexes (Weston et al. Reference Weston, Lahm and Suck1992), the requirement of two Asp residues (D168 and D212) and Mg²⁺ for DNA cleavage suggests that metal ions are likely to be essential for catalysis (Jones et al. Reference Jones, Worrall and Connolly1996).

6.7.2 AP endonuclease Exo III and APE1

Exonuclease III (Exo III), which is the major apurinic/apyrimidinic endonuclease in E. coli and its eukaryotic homologue APE1 share the conserved tertiary structure and active site with DNase I (Gorman et al. Reference Gorman, Morera, Rothwell, De La Fortelle, Mol, Tainer, Hickson and Freemont1997; Mol et al. Reference Mol, Kuo, Thayer, Cunningham and Tainer1995). These nucleases cleave the phosphodiester bond 5′ to an abasic (apurinic/apyrimidinic) site endonucleolytically and also cleave exonucleolytically to remove a 3′-dRP (dRP=phosphoglycolate, phosphate, or α,β-unsaturated aldehyde) (Demple & Harrison, Reference Demple and Harrison1994). Crystal structures of Exo III and APE1 alone (Gorman et al. Reference Gorman, Morera, Rothwell, De La Fortelle, Mol, Tainer, Hickson and Freemont1997; Mol et al. Reference Mol, Kuo, Thayer, Cunningham and Tainer1995) and APE1 complexed with an abasic DNA substrate without divalent cation and complexed with metal ion and cleavage product have been determined (Mol et al. Reference Mol, Izumi, Mitra and Tainer2000). The active site is highly similar to that of DNase I, except for the replacement of the H134 and E78 by Y171 on the 3′ leaving group side (Fig. 11c and d). Two Asp and the C-terminal His surrounding the scissile phosphate are superimposable between DNase I and APE1.

In the APE1–DNA complex structure, the abasic nucleotide is flipped out of the DNA double helix. Concomitantly, the backbone phosphate turns as well and the pro-Sp non-bridging oxygen points inward instead of outward (Fig. 11d). The distortion of the scissile phosphate is identical to that in the DNase I complex, where the pro-Sp oxygen rotates toward the minor groove (Fig. 11c and d). In RNase H and many other two-metal-ion-dependent nucleases, the pro-Sp oxygen coordinates both metal ions (Fig. 5). Here the pro-Sp oxygen is stabilized by the C-terminal His in APE1 and DNase I. Instead, it is likely that the pro-Rp oxygen coordinates two metal ions for the nucleophilic attack (see additional discussion in section 6.16). Consistent with this prediction, sulfur replacement of pro-Sp or pro-Rp oxygen of the scissile phosphate eliminates endonucleolytic cleavage by Exo III (Takeuchi et al. Reference Takeuchi, Lillis, Demple and Takeshita1994).

6.7.3 RNase E and G

RNase E is an endoribonuclease participating in the degradation of messenger and regulatory RNA and the processing of tRNA, rRNA and tmRNA precursors (Worrall & Luisi, Reference Worrall and Luisi2007). In bacteria, it often functions in a multienzyme assembly known as the RNA degradosome, which contains RNA helicase, enolase and exoribonuclease PNPase. RNase G and RNase E are paralogs and share the conserved active site. RNase E forms a homotetramer with D2 symmetry (Callaghan et al. Reference Callaghan, Grossmann, Redko, Ilag, Moncrieffe, Symmons, Robinson, Mcdowall and Luisi2003). Each active site is composite and composed of two polypeptide chains, each of which contributes a two-layered αβ structure (Fig. 11e) (Callaghan et al. Reference Callaghan, Marcaida, Stead, Mcdowall, Scott and Luisi2005). The catalytic domain of a single RNase E subunit (one αβ structure) is similar to half of DNase I (Fig. 11f). The dyad axis that relates the two RNase E subunits that form a composite active site, however, is approximately perpendicular to the β-sheet instead of parallel to it as in the DNase I structure (Fig. 11a and e). Dimerization of RNase E actually results in a cluster of conserved carboxylates on either end of the β-sheets and thus two active sites instead of one as in DNase I (Callaghan et al. Reference Callaghan, Marcaida, Stead, Mcdowall, Scott and Luisi2005). The catalytic residues of the RNase E and their arrangement bear similarity to DnaQ and RNase H (Figs. 6 and 7). The mechanism for RNase E catalysis is likely two-metal-ion dependent with details yet to be revealed.

6.8.1 RNase Z for tRNA maturation

RNase Z is an endonuclease that processes the 3′ end of tRNA precursors (pre-tRNA) and is also known as the 3′ tRNase. It is widely spread from bacteria, archaea to eukaryotes (Schiffer et al. Reference Schiffer, Rosch and Marchfelder2002). RNase Z cleaves 3′ to the discriminator base, which is the last in the acceptor stem and primes the CCA addition. If CCA is already present, RNase Z can cleave after CCA, but cleavage before CCA is severely inhibited thereby avoiding futile cycles of CCA addition (Minagawa et al. Reference Minagawa, Takaku, Takagi and Nashimoto2004). The crystal structure of RNase Z (de la Sierra-Gallay et al. Reference De La Sierra-Gallay, Pellegrini and Condon2005; Ishii et al. Reference Ishii, Minagawa, Takaku, Takagi, Nashimoto and Yokoyama2005; Kostelecky et al. Reference Kostelecky, Pohl, Vogel, Schilling and Meyer-Klaucke2006) confirms the sequence-based prediction of its resemblance to β lactamase (Schiffer et al. Reference Schiffer, Rosch and Marchfelder2002) and the two Zn²⁺-based catalytic mechanism (Minagawa et al. Reference Minagawa, Takaku, Takagi and Nashimoto2004). Five conserved His and two conserved Asp residues are involved in metal-ion coordination (Fig. 12b).

6.8.2 CPSF-73 and RNase J1

Mammalian CPSF-73 was predicted to have the metallo-β-lactamase fold and endonuclease activity (Aravind, Reference Aravind1999). Mutagenesis of the yeast CPSF-73 homolog, Ysh1p, confirms the predicted catalytic nuclease activity and its in vivo function (Ryan et al. Reference Ryan, Calvo and Manley2004). CPSF-73 has both an endo and 5′-exo nuclease activity and is essential for processing the 3′ end of mRNA precursors (Chanfreau et al. Reference Chanfreau, Noble and Guthrie1996; Jenny et al. Reference Jenny, Minvielle-Sebastia, Preker and Keller1996; Marzluff et al. Reference Marzluff, Wagner and Duronio2008; Yang et al. Reference Yang, Sullivan, Marzluff and Dominski2009). Crystal structures of the human CPSF-73 (Mandel et al. Reference Mandel, Kaneko, Zhang, Gebauer, Vethantham, Manley and Tong2006) show that the catalytic domain is essentially identical to RNase Z. Like RNase Z, two Zn²⁺ ions are coordinated by five His and two Asp residues.

Bacillus subtilis RNase J1 is a 5′ exonuclease and functions in pre-rRNA maturation and mRNA decay (Bechhofer, Reference Bechhofer2009; Mathy et al. Reference Mathy, Benard, Pellegrini, Daou, Wen and Condon2007). RNase J1 also has the metallo-β-lactamase fold and shares the conserved catalytic residues of CPSF-73.

6.8.3 Artemis, Snm1 and Pso2

Artemis, which is required to open DNA hairpins in V(D)J recombination, also belongs to the metallo-β-lactamase nuclease family (Moshous et al. Reference Moshous, Callebaut, De Chasseval, Corneo, Cavazzana-Calvo, Le Deist, Tezcan, Sanal, Bertrand, Philippe, Fischer and De Villartay2001). Mutations in Artemis are linked to defective V(D)J recombination and immune deficiency SCID. Like CPSF-73, Artemis has both endonuclease and 5′ exonuclease activities. In addition, the DNA cross-link repair nucleases Snm1 and Pso2 have similar nuclease domains (Callebaut et al. Reference Callebaut, Moshous, Mornon and De Villartay2002; Dominski, Reference Dominski2007). These related RNase and DNase share an additional conserved structural domain, called the β-CASP domain, which is inserted near the C-terminus of the nuclease domain (Mandel et al. Reference Mandel, Kaneko, Zhang, Gebauer, Vethantham, Manley and Tong2006). Although crystal structures of Artemis, Snm1 and Pso2 are not yet determined, based on the sequence homology these nucleases are likely to bind two Zn²⁺ ions in the active site and catalyze DNA cleavage by the two-metal-ion mechanism.

6.15.1 Group I intron

Self-splicing group I introns catalyze two consecutive phosphoryl transfer reactions (Fig. 18a) and are the best characterized example of two Mg²⁺-dependent ribozymes (Stahley & Strobel, Reference Stahley and Strobel2006). The nucleophile in the first step (cleavage reaction of the 5′ exon–intron juction) is the 3′-OH of a free guanosine (or GMP, GTP), which becomes covalently linked to the 5′-end of the intron. In the second step, the 3′-OH of the cleaved exon (product of the first cleavage reaction) serves as the nucleophile and attacks the 3′ intron–exon junction to produce the ligated exons and a free intron (Vicens & Cech, Reference Vicens and Cech2006) (Fig. 18a). Cleavage substrates, the intron–exon junctions, are structured and the bases surrounding each scissile phosphate are paired and stacked in a double helix (Fig. 18b) (Adams et al. Reference Adams, Stahley, Kosek, Wang and Strobel2004). Two divalent cations are coordinated between a non-bridging oxygen (pro-Sp) of the scissile phosphate and a backbone phosphate of the ribozyme, which acts exactly like the most conserved Asp in the active site of many two-metal-ion-dependent protein enzymes (Yang et al. Reference Yang, Lee and Nowotny2006). The two metal ions are likely to switch their roles of activating nucleophiles in the consecutive reaction steps, similar to DNA transposition by retroviral integrase and bacterial transposases (section 6.2.2, Fig. 7c) (Nowotny et al. Reference Nowotny, Gaidamakov, Crouch and Yang2005; Steitz & Steitz, Reference Steitz and Steitz1993).

Fig. 18.

Self-splicing introns. (a) A schematic drawing of phosphoryl transfer reactions catalyzed by group I introns. G–O: represents a free guanosine with its 2′-O serving as the nucleophile. (b) The structure of the catalytic center of group I introns. The exons to be ligated are shown in yellow and the nucleotides of the intron involved in catalyses are highlighted in pink and hot pink. Two Mg²⁺ (purple spheres) are jointly coordinated by the scissile phosphate and a backbone phosphate that mimics the side chain of Asp in protein enzymes. (c) A schematic drawing of phosphoryl transfer reactions catalyzed by group II introns. A–O: represents an internal adenosine with its 2′-O serving as the nucleophile. (d) The structure of the catalytic center of a group II intron in complex with ligated exons as the product of splicing. Two metal ions (purple spheres) again are found in the active site.

6.15.2 Group II intron

Group II introns also catalyze their own excision and function in retro-transposition in addition (Aizawa et al. Reference Aizawa, Xiang, Lambowitz and Pyle2003; Peebles et al. Reference Peebles, Perlman, Mecklenburg, Petrillo, Tabor, Jarrell and Cheng1986). The splicing reaction consists of two steps like group I introns, but group II introns use the 2′-OH of an internal Ade to initiate self-splicing (Doudna & Cech, Reference Doudna and Cech2002). The intron is excised in a lariat form of a 2′–5′ cyclic diester bond (Fig. 18c), which is similar to mRNA splicing by spliceosomes in eukaryotes (Michel et al. Reference Michel, Costa and Westhof2009). The crystal structure of the product form of a group II intron was recently determined, and two metal ions are found in the catalytic center (Toor et al. Reference Toor, Rajashankar, Keating and Pyle2008) (Fig. 18d). A two-metal-ion-dependent catalysis mechanism similar to that of group I introns has been postulated for self-splicing group II introns and spliceosomes, in general, for mRNA splicing (Toor et al. Reference Toor, Keating and Pyle2009).

6.15.3 RNase P

RNase P is one of the first ribozymes identified and the universally conserved nuclease that endonucleolytically cleaves pre-tRNA to generate the mature 5′ end (Guerrier-Takada et al. Reference Guerrier-Takada, Gardiner, Marsh, Pace and Altman1983). In addition to the catalytic RNA subunit, RNase P also includes a variable number of protein subunits (Kirsebom, Reference Kirsebom2007). Crystal structures of two types of bacterial RNase P have been determined (Torres-Larios et al. Reference Torres-Larios, Swinger, Pan and Mondragon2006). These structures reveal a conserved architecture of the catalytic core. But, in the absence of an RNase P-tRNA ribozyme-substrate complex, details of the active site are uncertain. The catalytic mechanism is likely to be similar to that of group I introns in spite of the uncertainty of which ribonucleotide coordinates the Mg²⁺ ions.

6.16.1 E. coli Endo IV (AP endonuclease)

Endo IV is the class II and minor AP endonuclease in E. coli (second to Exo III, see section 6.7.2) and is known for its EDTA-resistant activity (Demple & Harrison, Reference Demple and Harrison1994). Its yeast homologue is known as Apn1. Endo IV has the TIM-barrel fold, and the active site is located at one end of the β barrel, where the three Zn²⁺ ions bind (Fig. 19a). The crystal structures of the enzyme complexed with substrate and product were also determined at high resolutions (Garcin et al. Reference Garcin, Hosfield, Desai, Haas, Bjoras, Cunningham and Tainer2008) (Fig. 19b). Two out of three metal ions, Zn2 and Zn3, are reminiscent of the metal ions A and B, respectively, in the two-metal-ion mechanism (Fig. 19c). However, instead of the pro-Sp non-bridging oxygen of the scissile phosphate jointly coordinating these two metal ions, the scissile phosphate in the Endo IV complex turns toward the minor groove and the pro-Rp oxygen coordinates Zn2 and Zn3. This is very similar to the cases of DNase I and APE1 (Fig. 11c, d). In Endo IV, Zn1 (instead of a His as in APE1 and DNase I) stabilizes the pro-Sp oxygen and facilitates the distortion of the scissile phosphate (Fig. 19c).

In Endo IV, the Zn3 metal ion is coordinated by both the pro-Rp and the 3′ leaving oxygens just like the B site ion in the two-metal-ion mechanism (Fig. 5a and c). Mutational studies also confirm that coordination of Zn3 is more sensitive to amino acid substitution than the other two-metal-ion-binding sites and that the Zn3 metal ion plays a more important role in catalysis (Garcin et al. Reference Garcin, Hosfield, Desai, Haas, Bjoras, Cunningham and Tainer2008). Moreover, for Endo IV to be active, the Zn3 site also has a strong preference for Mn²⁺ over Zn²⁺ (Demple & Harrison, Reference Demple and Harrison1994; Garcin et al. Reference Garcin, Hosfield, Desai, Haas, Bjoras, Cunningham and Tainer2008). Supporting the crystallographic observation, sulfur replacement of the pro-Sp oxygens at the scissile phosphate reduces the Endo IV cleavage activity, but sulfur substitution of the pro-Rp oxygen eliminates the nuclease activity (Takeuchi et al. Reference Takeuchi, Lillis, Demple and Takeshita1994). Five His and four carboxylates contribute to tight binding of the three metal ions (Fig. 19c), which may explain why Endo IV is EDTA resistant. E261, which coordinates the Zn2 ion (equivalent to metal ion A), is suggested to be essential for nucleophile activation and cleavage (Garcin et al. Reference Garcin, Hosfield, Desai, Haas, Bjoras, Cunningham and Tainer2008).

6.16.2 Nuclease P1 and S1

Nuclease P1 and S1 are single-strand-specific nucleases that degrade DNA or RNA either endonucleolytically or as a 3′ exonuclease (Desai & Shankar, Reference Desai and Shankar2003). P1 nuclease is all α-helical and resembles bacterial phospholipase C (Hough et al. Reference Hough, Hansen, Birknes, Jynge, Hansen, Hordvik, Little, Dodson and Derewenda1989; Romier et al. Reference Romier, Dominguez, Lahm, Dahl and Suck1998) (Fig. 19d). Although P1 nuclease and Endo IV are unrelated in amino acid sequence or tertiary structure, they both use carboxylates and histidines to tightly bind three Zn²⁺ ions even in the absence of substrate (Fig. 19). In P1 nuclease, two Zn²⁺ ions are jointly coordinated by two conserved Asp (Fig. 19e). Similar to Endo IV, the pro-Rp rather than the pro-Sp non-bridging oxygen of the scissile phosphate is intolerant of thio-replacement (Romier et al. Reference Romier, Dominguez, Lahm, Dahl and Suck1998). Although a P1-substrate complex structure is not available, the sulfur replacement experiment is in agreement with the metal ion coordination observed in Endo IV. It is likely that the pro-Rp oxygen of the scissile phosphate together with the two conserved Asp residues coordinates the two catalytic metal ions.

6.17.1 Staphylococcal nuclease

Staphylococcal nuclease (SNase) is a Ca²⁺ dependent extracellular phosphodiesterase that cleaves ssRNA and ssDNA to 3′-phospho mono- and di-nucleotides (Tucker et al. Reference Tucker, Hazen and Cotton1978). The enzyme contains two antiparallel β-sheets and three α-helices (Fig. 20a). The active site is composed of four carboxylates at the end of the large β-sheet. High-resolution crystal structures of its complex with nucleotide analog/inhibitor pdTp and metal ions Ca²⁺ or Co²⁺ (inhibitor) reveal a single but mobile metal ion (Loll et al. Reference Loll, Quirk, Lattman and Garavito1995). Because of the lack of a confirmed general base, the metal ion is postulated to deprotonate and to activate the nucleophilic water (Hale et al. Reference Hale, Poole and Gerlt1993; Loll et al. Reference Loll, Quirk, Lattman and Garavito1995). In the absence of an enzyme–substrate complex, the catalytic mechanism remains undefined. It is possible that SNase uses a single metal ion as observed in the crystal structure, or the two-metal-ion mechanism given the carboxylate-rich active site, or even three metal ions similar to Endo IV and DNase P1/S1 because SNase prefers ssDNA and ssRNA. Since the cleavage product of SNase is 3′-phosphate, which is radically different from all the confirmed two-metal-ion-dependent nucleases, it is likely that it has a different catalytic mechanism.

Fig. 20.

Nucleases that may use the two-metal-ion mechanism. (a) Staphylococcal nuclease (SNase) is shown in a rainbow-colored ribbon diagram. Thymidine diphosphate (pdTp, THP) co-crystallized and the catalytic residues are shown as sticks. A Ca²⁺ ion captured in the structure is shown as a purple sphere. (b) The MutL endonuclease activity resides in its C-terminal domain (CTD). The crystal structure of E. coli MutL CTD (known as LC20) is shown in a rainbow-colored ribbon diagram. Conserved sequence motifs for the metal-ion-binding site are indicated in matching colors of the ribbon diagram and modeled onto the E. coli structure. The potential metal ion ligands are shown as sticks. (c) Cas1 is a CRISPR nuclease. One catalytic domain is shown in rainbow-colored ribbon diagrams with helices shown as rods. The active site is composed of three carboxylates and one His. A bound Mn²⁺ is shown as a purple sphere. The dimerization domain is not shown here. (d) The structure of Cas2. Cas2 is a CRISPR RNase with the ferredoxin fold. Both subunits of the homodimer Cas2 are shown in rainbow color from the blue N- to the red C-terminus. The active site is located at the dimer interface with the conserved and catalytically essential Asp shown as sticks and labeled.

6.17.2 MutL

MutL is an essential enzyme for DNA mismatch repair in bacteria and eukaryotes, and inactivation of MutL in humans leads to dramatically increased susceptibility to hereditary colon cancers (Li, Reference Li2008). All MutL homologues contain a highly conserved N-terminal ATPase domain (Ban & Yang, Reference Ban and Yang1998a). Four years ago, it was discovered that some MutL homologues also contain metal-ion-dependent weak endonuclease activity in the C-terminal dimerization domain (Kadyrov et al., Reference Kadyrov, Dzantiev, Constantin and Modrich2006, Reference Kadyrov, Holmes, Arana, Lukianova, O'Donnell, Kunkel and Modrich2007). A conserved metal-binding site in the C-terminal domain, which is essential for the MutL endonuclease activity (Duppatla et al., Reference Duppatla, Bodda, Urbanke, Friedhoff and Rao2009), is modeled based on the crystal structure of E. coli MutL (Guarne et al., Reference Guarne, Ramon-Maiques, Wolff, Ghirlando, Hu, Miller and Yang2004; Kosinski et al., Reference Kosinski, Plotz, Guarne, Bujnicki and Friedhoff2008; Yang, Reference Yang2007) (Fig. 20b). Mutations of the metal ion-coodinating residues lead to the loss of the MutL endonuclease activity in vitro and mismatch repair capacity in vivo (Fukui et al., Reference Fukui, Nishida, Nakagawa, Masui and Kuramitsu2008).

6.17.3 Cas1

CRISPR (clustered regularly interspaced short palindromic repeats) is a bacterial and archaeal adaptive immune system to resist phage infection (Sorek et al. Reference Sorek, Kunin and Hugenholtz2008). Phage DNA is cleaved to short oligonucleotides and incorporated into host CRISPR between repetitive DNA sequences and the transcription products are used for target degradation if the same phage infection is encountered again. Cas1 and Cas2 are CRISPR-associated proteins essential for CRISPR (Brouns et al. Reference Brouns, Jore, Lundgren, Westra, Slijkhuis, Snijders, Dickman, Makarova, Koonin and Van Der Oost2008). Cas1 is the universally conserved protein in CRISPRs and is a dimeric metal-dependent DNA endonuclease. The crystal structure reveals a novel fold and a bound Mn²⁺ in the absence of DNA substrate (Wiedenheft et al. Reference Wiedenheft, Zhou, Jinek, Coyle, Ma and Doudna2009). The structure contains a helical catalytic domain (Fig. 20b) and an αβ dimerization domain. The helical structure of the Cas1 catalytic domain is reminiscent of DNase P1/S1. With the active site composed of EHDD residues and at least one conserved Asp indispensable for the nuclease activity (Wiedenheft et al. Reference Wiedenheft, Zhou, Jinek, Coyle, Ma and Doudna2009), it is possible that DNA cleavage by Cas1 is by the two-metal-ion mechanism.

6.17.4 Cas2

Cas2 is ferredoxin-like and metal-dependent single-strand endoribonuclease (Beloglazova et al. Reference Beloglazova, Brown, Zimmerman, Proudfoot, Makarova, Kudritska, Kochinyan, Wang, Chruszcz, Minor, Koonin, Edwards, Savchenko and Yakunin2008). It is dimeric, prefers U-rich regions and produces 3′ -OH and 5′-phosphate oligonucleotides. Given its metal-ion dependence and requirement for a conserved Asp residue, a two-metal-ion-dependent catalysis mechanism has been proposed. Interestingly, in the crystal structure of Cas2 protein, the catalytically essential Asp (D10) is situated at the dimeric interface and two symmetry-related Asp residues are 6·5 Å apart (Beloglazova et al. Reference Beloglazova, Brown, Zimmerman, Proudfoot, Makarova, Kudritska, Kochinyan, Wang, Chruszcz, Minor, Koonin, Edwards, Savchenko and Yakunin2008). These two Asp side chains may belong to a single active site (Fig. 20c).

7.1.1 HNH and His-Cys homing endonucleases

Although HNH and His-Cys were initially classified into two separate families of homing endonucleases based on conserved sequence motifs, close scrutiny of amino acid sequences has led to the proposal that they derived from a common ancestor and share the same ββα-Me catalytic core (Friedhoff et al. Reference Friedhoff, Franke, Meiss, Wende, Krause and Pingoud1999; Kuhlmann et al. Reference Kuhlmann, Moore, James, Kleanthous and Hemmings1999). The HNH motif is involved in the catalytic metal coordination, whereas the His-Cys motif coordinates additional metal ions to stabilize the tertiary structure (Fig. 21a) (Stoddard, Reference Stoddard2005). The catalytic metal ion can be Mg²⁺, Zn²⁺, Ca²⁺ or other divalent cations, which are often coordinated by His, Asn and Asp (Fig. 21b). Most noticeably, the ββα-Me motif is nearly a stand-alone domain and separate from the DNA sequence-specific binding domains (Fig. 21a). In the His-Cys homing endonuclease I-PpoI, two structural metal ions (Zn) tie the DNA-binding and ββα-Me catalytic domains together in each subunit (Flick et al. Reference Flick, Jurica, Monnat and Stoddard1998).

7.1.2 HJ resolvase: phage T4 endonuclease VII

T4 endo VII is the only HJ resolvase that contains the ββα-Me motif for catalysis. It forms a homodimer and recognizes an H-shaped four-way junction (Biertümpfel et al. Reference Biertümpfel, Yang and Suck2007) (Fig. 21c and d). The two strands that make a U-turn (aka exchanged strands) in the four-way junction are cleaved. Mg²⁺ is the preferred metal ion for DNA cleavage. Like His-Cys homing endonuclease, a Zn²⁺ is also tightly bound by each T4 endo VII subunit and stabilizes the tertiary structure. Relative to T7 endonuclease I, which has the REase-type active site and uses the two-metal-ion mechanism, T4 endo VII is not as specific for four-way junctions and can cleave 3-way junctions and mismatched DNA heteroduplexes, all of which are characterized by distorted double-helix structures (Jensch & Kemper, Reference Jensch and Kemper1986; Solaro et al. Reference Solaro, Birkenkamp, Pfeiffer and Kemper1993).

7.1.3 Type-II restriction endonucleases

Several type-II restriction endonucleases also contain the ββα-Me motif for DNA cleavage. They include HphI (Cymerman et al. Reference Cymerman, Obarska, Skowronek, Lubys and Bujnicki2006), Eco311 (Jakubauskas et al. Reference Jakubauskas, Giedriene, Bujnicki and Janulaitis2007), KpnI (Chatterjee et al. Reference Chatterjee, Hammond, Blakesley, Adams and Gerard1991; Vasu et al. Reference Vasu, Saravanan, Bujnicki and Nagaraja2008) and Hpy99I (Sokolowska et al. Reference Sokolowska, Czapinska and Bochtler2009). The active site of these sequence-specific endonucleases contains the conserved DH and N residues just like T4 endo VII (Fig. 21d and e). Additional DNA-binding domains in these restriction enzymes target cleavage to a specific sequence. The crystal structure of Hpy99I complexed with its cognate DNA substrate reveals an interlocked homodimer, which bears remarkable similarity to T4 endo VII (Fig. 21c and e). In addition to the catalytic motifs approaching the cognate DNA from the minor groove side, Hpy99I embraces the DNA with sequence-specific binding in the major groove. Different from I-PpoI (Fig. 21a), the catalytic motifs, DNA-binding modules and the dimeric interface in Hpy99I are rolled into one. Concommitantly, the recognition sequence of Hpy99I is much shorter (5 bp) than I-PpoI (14 bp) and the cleavage sites border the major-groove recognition rather than being at a distance as with I-PpoI (Fig. 21a and e).

7.1.4 Sugar-non-specific endonucleases

Sugar-non-specific nucleases degrade ssDNA and ssRNA or dsDNA and dsRNA in a sequence-independent manner (Friedhoff et al. Reference Friedhoff, Kolmes, Gimadutdinow, Wende, Krause and Pingoud1996). Members of this family have very diverse functions and can take part in bacterial defense, invasion, nutrient scavenging and eukaryotic programmed cell death. Periplasmic Vvn (Hsia et al. Reference Hsia, Li and Yuan2005) defends host cells by degrading foreign nucleic acids. The secreted serratia and cyanobacterium NucA (Muro-Pastor et al. Reference Muro-Pastor, Flores, Herrero and Wolk1992) nucleases may be regulated depending on growth phases (Chen et al. Reference Chen, Shipley, Ball and Benedik1992). The other secreted nucleases, colicins E7 and E9 (Hsia et al. Reference Hsia, Li and Yuan2005), kill non-self-target cells and thereby enhance the chance of host cell survival under stress. Eukaryotic mitochondrial endonuclease G also belongs to this family and is a key enzyme in cell death (Buttner et al. Reference Buttner, Eisenberg, Carmona-Gutierrez, Ruli, Knauer, Ruckenstuhl, Sigrist, Wissing, Kollroser, Frohlich, Sigrist and Madeo2007). These non-specific nucleases are very efficient and non-discriminative in nucleic acid degradation. Host cells often have specific inhibitors, known as immunity proteins, to avoid self-destruction.

7.3.1 Rep nucleases

Members of the rep family are found in ssDNA and dsDNA bacterial phages, plasmids, plant geminiviruses and animal parvoviruses, such as Adeno-associated virus (Aav) (Hickman et al. Reference Hickman, Ronning, Kotin and Dyda2002). By forming 5′-phosphotyrosyl linkage, these nucleases generate a free 3′-OH to initiate DNA synthesis. In addition to the HUH motif for metal-ion coordination, this superfamily also contains the YxxxY motif with two conserved Tyr residues. The crystal structure of Aav5 Rep reveals a nuclease core consisting of a five-stranded antiparallel β-sheet and a number of surrounding α-helices (Hickman et al. Reference Hickman, Ronning, Perez, Kotin and Dyda2004) (Fig. 23a). Most catalytic residues including HUH are located on a pair of adjacent central β-strands, and the nucleophile Tyr is on a nearby α-helix (Fig. 23a). As predicted, an Mg²⁺ is coordinated by the two absolutely conserved His and a conserved Glu (Fig. 23a). The catalytic mechanism has not been verified by a crystal structure of enzyme-substrate complexes, but the nuclease alone structure resembles that of mob nucleases (see below) and hence the catalysis is likely by the one-metal-ion mechanism.

Fig. 23.

The HuH nucleases. (a) The nuclease domain of Adeno-associated virus 5 (Aav5) Rep shown in rainbow-colored ribbon diagrams. The associated metal ion is shown in purple. (b) The structure of TraI relaxase (Y16F mutant) representing the mob family. Only two nucleotides surrounding the scissile phosphate of the ssDNA substrate are shown as sticks. (c) A close-up view of the TraI active site. Three His side chains coordinate the Mg²⁺, which in turn aligns the scissile phosphate for nucleophilic attach by Y16. (d) Ribbon diagrams of the DNA transposase (TnpA) of IS608 complexed with a 3′-end cleavage product. TnpA is homodimeric. The two domain-swapped protein subunits are shown in yellow and cyan and the two DNAs in hot pink. (e) A close-up view of the composite TnpA active site. The last nucleotide at the 3′ end is also shown. The metal-chelating His residues (of the HUH motif) are from one subunit. The third metal ligand (Q131) and the nucleophile Y127 are from the other subunit.

7.3.2 Mob nucleases (TraI and TrwC)

Mobility of bacterial integrative conjugative elements (ICEs, including conjugative F plasmids and integrative transposons) and ssDNA phage (ϕx174) depends on DNA nicking by mob nucleases, which are also known as relaxases (Khan, Reference Khan2005; Llosa et al. Reference Llosa, Gomis-Ruth, Coll and De La Cruz Fd2002). Even though some mob nucleases recognize ssDNA and make a single-stranded nick, the DNA substrate is actually folded into a base-paired hairpin structure and the scissile phosphate bond is embedded in the double helix. TraI encoded by E. coli F plasmid and TrwC by R388 conjugation plasmid are characterized by two consecutive Tyr residues near the very N-terminus in addition to the HUH motif (Gomis-Ruth & Coll, Reference Gomis-Ruth and Coll2006; Larkin et al. Reference Larkin, Haft, Harley, Traxler and Schildbach2007). The first Tyr serves as the nucleophile. Because the cleavage–ligation reactions catalyzed by these nucleases relax supercoiled DNA, they are called relaxases, very much like topoisomerases (Fekete & Frost, Reference Fekete and Frost2000). In the crystal structure of TraI, the nucleophile Tyr is mutated to Phe and the relaxase is co-crystallized with its specific DNA substrate (Fig. 23b) (Datta et al. Reference Datta, Larkin and Schildbach2003; Larkin et al. Reference Larkin, Datta, Harley, Anderson, Ebie, Hargreaves and Schildbach2005). Several positively charged Lys and Arg side chains facilitate substrate DNA binding. The catalytically essential Mg²⁺ is coordinated by the scissile phosphate, the two absolutely conserved His, and a third His only conserved within the mob family (Fig. 23c). The active site is superimposable with the metal-ion center in colicin E7 (Yang, Reference Yang2008), indicating the conserved metal-ion function and catalytic mechanism.

7.3.3 TnpA DNA transposase of IS200/IS605 family

Nucleases with the HUH motif are also involved in DNA transposition of IS elements (Ton-Hoang et al. Reference Ton-Hoang, Guynet, Ronning, Cointin-Marty, Dyda and Chandler2005). They differ from rep and mob nucleases by possessing only one instead of two conserved Tyr residues besides the HUH motif. These small and dimeric nucleases (<200 aa) are represented by the IS200/IS605 family of DNA transposases (Tnp). The ssDNA transposase (TnpA) from transposon ISHp608 is the best-studied example. TnpA recognizes specific ssDNA sequence and utilizes it for substrate recognition by forming base-paired duplex structures for cleavage (Barabas et al. Reference Barabas, Ronning, Guynet, Hickman, Ton-Hoang, Chandler and Dyda2008). The tertiary structure and catalytic mechanism of TnpA is reminiscent of TraI and TrwC relaxases (Fig. 23), but the nucleophile Tyr is permutated to be near the C-terminus. Each active site is composite with two metal-chelating His residues from one subunit and the nucleophile Tyr from the second subunit by domain swapping (Fig. 23d and e). A third metal-ion ligand, Gln, is located near the Tyr and supplied in trans as well. Interestingly, the domain swapping and subsequent active site formation are entirely dependent on the DNA substrate binding, which induces a gross conformational change (Barabas et al. Reference Barabas, Ronning, Guynet, Hickman, Ton-Hoang, Chandler and Dyda2008).

8.1.1 Type-II topoisomerases

Type-II topoisomerases are divided into distinct IIA and IIB families, both of which cleave dsDNA (Wang, Reference Wang2002). Type-II topoisomerases are also Mg²⁺-dependent ATPases, and Mg²⁺ cannot be omitted from the reaction. These enzymes are dimeric or hetero–tetrameric. Each active site is composite and consists of the nucleophile Tyr on the WHD domain of one subunit and carboxylates on the TOPRIM domain of another subunit (Schoeffler & Berger, Reference Schoeffler and Berger2008). A single Mg²⁺ coordinated by the conserved E and DxD motifs of TOPRIM was first found in the crystal structure of an archaeal type IIB topoisomerase in the absence of DNA substrate (Nichols et al. Reference Nichols, Deangelis, Keck and Berger1999). A Mg²⁺ ion was later found in the active site of yeast Topo II co-crystallized with a nicked DNA substrate (Dong & Berger, Reference Dong and Berger2007) (Fig. 24a).

Fig. 24.

Topoisomerases containing the TOPRIM motif. (a) The active site of S. cerevisiae Topo II in complex with dsDNA substrate. The TOPRIM domain is shown in rainbow-colored ribbon diagram. The nucleophile Y782 and its neighbor R781 are supplied in trans by the other subunit of the dimeric Topo II and shown as yellow sticks. The scissile phosphate and surrounding nucleotides are shown as pink sticks. Nitrogen and oxygen atoms are colored blue and red, respectively. A Mg²⁺ ion (purple sphere) is coordinated by the conserved carboxylates in the TOPRIM domain. (b) Ribbon diagrams of E. coli Topo III (Type IA) in complex with ssDNA substrate. The α-helices are represented by cylinders. The TOPRIM domain (aka domain I) is shown in rainbow colors and the catalytic domain (domain III) carrying the Tyr nucleophile is shown in yellow. The ssDNA substrate in shown as sticks. (c) A close-up view of the active site of Topo III. The scissile phosphate, the conserved carboxylates and the nucleophile Y328 are nearly superimposable with that of Topo II shown in (a). (d) A model of the reaction coordinates of DNA cleavage by type IA and type II topoisomerases. The active sites of the two types of TOPRIM topoisomerase are superimposed. The metal ion observed in Topo II is incorporated in the composite active site and shown as a lilac sphere. A Lys (K8) of Topo III approximately occupies the metal ion A position. The Arg next to the nucleophile (R781 of Topo II or R330 of Topo III) in the catalytic domain further stabilizes the scissile phosphate and neutralizes the developing negative charges.

The location of this metal ion is close to the leaving group and is reminiscent of the B metal ion in RNase H1 and the single metal ion in TraI (Figs 7 and 23). The absence of the second metal ion (equivalent to metal ion A on the nucleophile side) is not a proof of a one-metal-ion mechanism, because the second metal ion may have escaped detection due to a crystallographic artifact. The arrangement of the Tyr and the metal ion relative to the DNA substrate in Topo II is rather similar to the mob nucleases, suggesting that the one-metal-ion mechanism may be applicable here. The absence of the fourth carboxylate in the active site and the presence of the Arg adjacent to the nucleophile Tyr to stabilize the scissile phosphate (Fig. 24a) also lend support to the one-metal-ion mechanism.

8.1.2 Type-IA topoisomerases

Type IA topoisomerases are monomeric and nick one strand of DNA (Schoeffler & Berger, Reference Schoeffler and Berger2008). In addition, type-IA enzymes are ATP-independent and relax negative supercoils (with the exception of reverse gyrase). Interestingly, the ssDNA co-crystallized with E. coli Topo III (type IA) adopts the double-helix-like conformation (Changela et al. Reference Changela, Digate and Mondragon2001) (Fig. 24b). Similar to Topo II, the active site of monomeric type IA is assembled between two domains and composed of ExxK and DxD from the TOPRIM domain (aka domain I) and the nucleophile Tyr from domain III (equivalent to the WHD of Topo II) (Fig. 24b and c). It was controversial as to whether metal ion is required for Topo IA. It is now clear that efficient DNA cleavage in addition to the religation by type-IA topoisomerases requires Mg²⁺ (Domanico & Tse-Dinh, Reference Domanico and Tse-Dinh1991). In support of the metal-ion requirement, amino-acid substitutions of the conserved carboxylates in TOPRIM render the mutant topoisomerases defective in DNA supercoil relaxation (Bhat et al. Reference Bhat, Leelaram, Hegde and Nagaraja2009; Zhu & Tse-Dinh, Reference Zhu and Tse-Dinh2000).

The crystal structure of a type-IA topoisomerase (Topo III) with the nucleophile Tyr replaced by Phe and in complex with ssDNA substrate provides the highest-resolution view of the active site of TOPRIM topoisomerases (Changela et al. Reference Changela, Digate and Mondragon2001). Relative to the scissile phosphate, the carboxylates and the Tyr of Topo III occupy positions similar to those in Topo II. In addition to the Arg next to the Tyr, a Lys conserved in type IA stabilizes the scissile phosphate (Fig. 24c). But due to the high salt concentration in the crystallization buffer (1·3 M (NH₄)₂SO₄), soaking in 10 mM MgCl₂ was not successful in detecting metal ion binding (Changela et al. Reference Changela, Digate and Mondragon2001). The superposition of the catalytic residues of E. coli Topo III and yeast Topo II show that the metal ion observed in the Topo II falls right between the conserved carboxylates and scissile phosphate in Topo III (Fig. 24d).

The striking similarities of metal ion coordination and nucleophilic attack of the scissile phosphate between TOPRIM topoisomerases and HUH nucleases TraI and TnpA imply a similar catalytic mechanism. The nuclease activity of Topo IA in the absence of Mg²⁺ may be explained by the conserved Lys and Arg in the active site, which can neutralize the developing negative charge on the scissile phosphate (Fig. 24c). A divalent cation on the leaving group side may not be necessary, but, in the strand ligation step, the metal ion is on the 3′-OH nucleophile side and may become essential for its deprotonation. It is most likely that type II and type IA topoisomerases with so much in common use the same metal-ion-dependent cleavage mechanisms.

The M5 and OLD family nucleases and type-IA and type-II topoisomerases of the TOPRIM superfamily share the conserved TOPRIM domain yet may use a two- versus one-metal-ion-dependent mechanism for DNA cleavage. The divergence appears to originate from different nucleophiles and different environments surrounding the respective scissile phosphate in their cleavage reactions. They may represent an early divergence from a primordial sequence motif and structural domain.

9.1.1 Barnase

Barnase, an extracellular ribonuclease of Bacillus amyloliquefaciens, has been extensively studied for its folding properties, interaction with its intracellular protein inhibitor barstar and its substrate recognition (Fersht & Daggett, Reference Fersht and Daggett2002; Urakubo et al. Reference Urakubo, Ikura and Ito2008). Homologous RNases are found in other bacillus species, e.g. Binase from Bacillus intermedius (Schulga et al. Reference Schulga, Nurkiyanova, Zakharyev, Kirpichnikov and Skryabin1992). Barnase has a characteristically twisted five-stranded contiguous antiparallel β-sheet with long loops surrounding it and short helices covering one side (Fig. 26a). The exposed face of the β-sheet binds nucleic acid substrate. Most of the catalytically essential residues are located on the loops except for the central Glu. The crystal structure of Barnase bound to a tetradeoxynucleotide (instead of RNA to prevent cleavage) reveals the mode of substrate recognition and a possible catalytic mechanism (Buckle & Fersht, Reference Buckle and Fersht1994). Barnase cleaves 3′ to a Gua, which is sandwiched between an Arg and Phe and forms double hydrogen bonds with a Glu (Fig. 26a). The scissile phosphate is neutralized by two Arg and one Lys. The catalytic residues Glu (E73) on the central β-strand and His (H102) on a long loop are on either side of the scissile phosphate and function as the general base and acid, respectively. The conformation of the deoxyribose linked to the scissile phosphate, however, is not in position for the in-line attack in this crystal structure, probably because of the absence of 2′-OH (Buckle & Fersht, Reference Buckle and Fersht1994). Y103, which stacks against the (deoxy)ribose, together with the Arg and Lys residues, may align and distort the substrate for the nucleophilic attack.

9.1.2 RNase T1

RNase T1, RNase Sa and fungal RNase U2, F1 and Ms belong to RNase T1 family and are homologous to Barnase in sequence and tertiary structures (Yoshida, Reference Yoshida2001). They too cleave ssRNA after Gua. The general base and acid pair, Glu and His, are conserved in these RNases. Several crystal structures of RNase T1 bound to various substrate analogs have been reported (Arni et al. Reference Arni, Watanabe, Ward, Kreitman, Kumar and Walz1999), and the complex with a cyclic thiophosphate substrate is particularly revealing (Zegers et al. Reference Zegers, Loris, Dehollander, Fattah Haikal, Poortmans, Steyaert and Wyns1998). RNase T1 can cleave the cyclic phosphodiester to 3′ phosphate, which is the second half of the normal hydrolysis reaction. The sulfur replacement (thiophosphate) greatly reduces the reaction rate. In the crystal structure, the cyclic phosphate is placed between the conserved Glu (E58) and His (H92) and hydrogen bonded to both (Fig. 26b). The ribose is flipped compared to that in Barnase (Fig. 26a) and well placed for the general acid–base catalysis. In addition, positively charged H40 and R77 interact and orient the scissile phosphate. The guanine base preceding the scissile phosphate is sandwiched between two Tyr residues and forms bifurcated hydrogen bonds with a glutamate (E46) in addition.

9.1.3 Restrictocin

The homologous restrictocin and sarcin are ribotoxins and specifically cleave 28S ribosome RNA (Olmo et al. Reference Olmo, Turnay, Gonzalez De Buitrago, Lopez De Silanes, Gavilanes and Lizarbe2001). Structurally these ribotoxins are related to Barnase and RNase T1 (Yang et al. Reference Yang, Gerczei, Glover and Correll2001), yet they specifically recognize a well-folded hairpin loop (sarcin/ricin loop) in the 28S rRNA. Because of the modified 2′-groups used in co-crystallization of restrictocin-RNA to prevent cleavage (Yang et al. Reference Yang, Gerczei, Glover and Correll2001), the RNA substrate is out of register by one nucleotide and the scissile-phosphate mimic is displaced from the catalytic residues. Nevertheless, the structure reveals a dramatically deformed hairpin RNA substrate and the unpaired and flipped-out base 5′ to the scissile phosphate (Correll et al. Reference Correll, Yang, Gerczei, Beneken and Plantinga2004) (Fig. 26c). Such distortions are necessary to align the 2′-OH nucleophile for the in-line attack. The bases surrounding the scissile phosphate are thus splayed much like in the cases of Barnase and RNase T1. The catalytic Glu-His pair is E95 and H136 in Ristrictocin. Like RNase T1, restrictocin also depends on aromatic and positively charged residues to bind and orient the scissile phosphate for the cleavage reaction (Fig. 26c).

9.1.4 Colicin E3, E4 and E6

Colicin E3, E4 and E6 are homologous and specifically cleave 16S rRNA. Bacteriocin Cloacin DF13 is a ribonuclease homologous to E3 and E6 colicins and it again cleaves 16S rRNA (Akutsu et al. Reference Akutsu, Masaki and Ohta1989). The crystal structure of E3 complexed with its inhibitor reveals the similarity to that of Barnase and RNase T1 (Carr et al. Reference Carr, Walker, James, Kleanthous and Hemmings2000) (Fig. 26d). The conserved E62 and H71 assume comparable positions as the catalytic Glu-His pair found in the T1 superfamily. But mutagenic studies show that H58 instead of H71 is critical for catalysis and that E62 and H58 are likely to be the general base and acid pair (Walker et al. Reference Walker, Lancaster, James and Kleanthous2004). Additionally charged residues, Asp, Glu, Lys and Arg, align the active site. Mutagenesis suggests that R40, D55, E60, Y64 and R90 are essential for catalysis (Walker et al. Reference Walker, Lancaster, James and Kleanthous2004). How these RNases recognize their substrate remains unsettled in the absence of a co-crystal structure and additional biochemical studies.

10.1.1 Nuc and DNase II

Nuc of Salmonella typhimurium provided the first glimpse of a PLD family enzyme. It is a homodimer. The conserved His residues from two subunits (H94) align a narrow groove forming one active site and most probably act as the nucleophile and general acid in the trans-esterification reaction (Stuckey & Dixon, Reference Stuckey and Dixon1999) (Fig. 30c). The reaction intermediate may be stabilized by the conserved Lys residues (K96) much as in cleavage by RNase A. The structure and metal-ion-independent catalysis indicate that the DNA substrate is recognized in the single-stranded form. Like all PLD-like DNases, Nuc generates 5′-phosphate and 3′-OH. Metazoa DNase II are homologs of bacterial Nuc and belong to the PLD family. DNase II in lysosomes of macrophages plays an essential role in degrading DNA downstream from CAD/DFF40 after engulfment of apoptotic cells (Nagata, Reference Nagata2007). Without DNase II, residual DNA fragments lead to autoimmune diseases. The Nuc and DNase II are non-sequence specific.

10.1.2 BfiI restriction endonuclease

BfiI restriction endonuclease presents the first crystal structure of Mg²⁺ independent type-II restriction endonucleases (Grazulis et al. Reference Grazulis, Manakova, Roessle, Bochtler, Tamulaitiene, Huber and Siksnys2005; Vitkute et al. Reference Vitkute, Maneliene, Petrusyte and Janulaitis1998). It is a homodimer in solution and the two PLD-like domains form one active site (Lagunavicius et al. Reference Lagunavicius, Sasnauskas, Halford and Siksnys2003) (Fig. 30d). Like all PLD-family enzymes, BfiI cleaves DNA via a two-step trans-esterification mechanism (Sasnauskas et al. Reference Sasnauskas, Connolly, Halford and Siksnys2007). In the first step, it uses the conserved His as a nucleophile to cleave DNA forming a 5′ phosphoenzyme. In the second step, it transfers the phosphodiester bond to a water or alcohol molecule to regenerate the enzyme. The transition intermediate is likely to be stabilized by the conserved Lys residues as in TDP1 (Davies et al. Reference Davies, Interthal, Champoux and Hol2004). Although BfiI cleaves dsDNA, it cleaves one strand first and then the second (Sasnauskas et al. Reference Sasnauskas, Zakrys, Zaremba, Cosstick, Gaynor, Halford and Siksnys2010). Sequence-specific recognition is achieved by the DNA-binding domains that are separate from the catalytic domains (Fig. 30d) and occurs several base pairs away from the cleavage site (Grazulis et al. Reference Grazulis, Manakova, Roessle, Bochtler, Tamulaitiene, Huber and Siksnys2005; Vitkute et al. Reference Vitkute, Maneliene, Petrusyte and Janulaitis1998).

10.2.1 TopIBs

TopIBs relax supercoiled DNA. They can be sequence-specific like viral topoisomerases (Cheng et al. Reference Cheng, Kussie, Pavletich and Shuman1998) or have weak preference for certain DNA sequences like eukaryotic cellular TopIB (Redinbo et al. Reference Redinbo, Stewart, Kuhn, Champoux and Hol1998). Many crystal structures of Topo 1B, from protein alone (Cheng et al. Reference Cheng, Kussie, Pavletich and Shuman1998), protein–DNA complexes (Perry et al. Reference Perry, Hwang, Bushman and Van Duyne2006b; Redinbo et al. Reference Redinbo, Stewart, Kuhn, Champoux and Hol1998) to a cleavage transition mimic (Davies et al. Reference Davies, Mushtaq, Interthal, Champoux and Hol2006), have been determined. The active site contains two Arg, one Lys, one His and the nucleophile Tyr (Fig. 31a). The basic residues orient and neutralize the DNA backbone. One conserved Arg that is hydrogen bonded to both scissile phosphate and the nucleophile Tyr (Fig. 31b) is postulated to be the general base to deprotonate the Tyr (Davies et al. Reference Davies, Mushtaq, Interthal, Champoux and Hol2006; Yakovleva et al. Reference Yakovleva, Chen, Hecht and Shuman2008). The remaining Arg is proposed to work together with the Lys to act as the general acid and protonate the 5′ leaving group (Hwang et al. Reference Hwang, Baek, Kim and Cho1998; Yakovleva et al. Reference Yakovleva, Chen, Hecht and Shuman2008).

Fig. 31.

Metal-independent topoisomerases and Tyr recombinases. (a) The overall structure of Human Topo IB. The protein is shown in rainbow-colored ribbon diagrams. The DNA duplex is shown in pink/blue ribbons. The active site is highlighted with the catalytic residues shown in sticks. The nucleophile Y723 (mutated to Phe) is labeled. (b) A close-up view of the active site of Topo IB. The reaction is stopped because of the Y723F (shown in pink) mutation. The scissile phosphate is neutralized by Arg and Lys side chains. (c) The overall structure of the Cre recombinase complexed with substrate LoxP site. The tetrameric Cre is shown in four colors and the two LoxP in pink and yellow. The DNA-binding domains are above the DNA and the catalytic domains are below. (d) A close-up view of the active site of Cre-LoxP. The scissile phosphate is already pre-nicked. (e) Superposition of the catalytic residues of Topo IB and Cre.

10.2.2 Tyr recombinases: Cre, Flp, Xer and Phage integrase

The representative members of the Tyr recombinase superfamily include phage integrases (λ and HP1) (Biswas et al. Reference Biswas, Aihara, Radman-Livaja, Filman, Landy and Ellenberger2005; Hickman et al. Reference Hickman, Waninger, Scocca and Dyda1997), the Cre–loxP recombination system (Baldwin et al. Reference Baldwin, Martin, Abel, Gelato, Kim, Schultz and Santoro2003; Guo et al. Reference Guo, Gopaul and Van Duyne1997) widely used in the gene knockout techniques, XerC/D for bacterial DNA segregation (Subramanya et al. Reference Subramanya, Arciszewska, Baker, Bird, Sherratt and Wigley1997), and yeast Flp recombinase (Chen et al. Reference Chen, Narendra, Iype, Cox and Rice2000). They are all site-specific recombinases and share the conserved catalytic motif RHR in addition to the Tyr nucleophile. These recombinases are most often bi-lobal with a sequence-specific DNA-binding domain at the N-terminus and catalytic domain at the C-terminus (Fig. 31c). Many crystal structures have been determined along the reaction pathway (Grindley et al. Reference Grindley, Whiteson and Rice2006), and the active site and the surrounding structure of these recombinases can be superimposed with that of Topo IB (Fig. 31d and e). The Tyr nucleophile is located on the third helix in a three-helix bundle, which resembles a classic HTH domain (Yang & Steitz, Reference Yang and Steitz1995a) (Fig. 31d). Positioning the Tyr for nucleophilic attack requires the synapsis of four protein subunits and two palindromic DNA recombination sites (Ghosh et al. Reference Ghosh, Guo and Van Duyne2007a) (Fig. 31c). A HJ is a necessary intermediate in recombination (Voziyanov et al. Reference Voziyanov, Pathania and Jayaram1999). A catalytically competent active site may consist of residues from two adjacent subunits (so-called trans-configuration) (Chen et al. Reference Chen, Narendra, Iype, Cox and Rice2000). The Tyr nucleophile is likely to be deprotonated by one of the two conserved Arg residues as with human and vaccinia Topo IB (Fig. 31e) (Redinbo et al. Reference Redinbo, Champoux and Hol2000; Yakovleva et al. Reference Yakovleva, Chen, Hecht and Shuman2008). The other conserved Arg and Lys residues in the active site orient the scissile phosphate, stabilize the transition state, and may also protonate the 5′-O leaving group.

11.2.1 Diversity of nucleases in a similar cleavage reaction

In the first category, the divergence of restriction endonucleases, HJ resolvases, site-specific recombinases, homing endonuclease and AP endonucleases stand out. As noted by Darwin, greater variation is found closer to the evolutionary source of a species. This principal is exemplified in the recent genomic sequencing of African Bushmen (Schuster et al. Reference Schuster, Miller, Ratan, Tomsho, Giardine, Kasson, Harris, Petersen, Zhao, Qi, Alkan, Kidd, Sun, Drautz, Bouffard, Muzny, Reid, Nazareth, Wang, Burhans, Riemer, Wittekindt, Moorjani, Tindall, Danko, Teo, Buboltz, Zhang, Ma, Oosthuysen, Steenkamp, Oostuisen, Venter, Gajewski, Zhang, Pugh, Makova, Nekrutenko, Mardis, Patterson, Pringle, Chiaromonte, Mullikin, Eichler, Hardison, Gibbs, Harkins and Hayes2010), which reveals that the genetic differences between any two Bushmen is greater than that between a European and an Asian. The diversity of nucleases may indicate that repair of an abasic lesion, homologous and site-specific DNA recombination and restrictive digestion of foreign DNA are ancient processes. The different nucleases that result in a shared outcome may also provide hints of a convergent relationship. For example, two AP endonucleases, Endo IV and Exo III (APE1) (sections 6.7.2 and 6.1.6.1), recognize the abasic site in a very similar way and use the pro-Rp oxygen to coordinate the catalytic two metal ions despite their unrelated tertiary structure and catalytic motifs.

Two cases of diverse nucleases for one biological process are worth noting. HJ or four-way DNA junctions are resolved to two duplexes by nucleases with different structure and mechanisms. Even among bacterial and single-cell eukaryotes there are more than four different HJ resolvases (Table 1). In higher eukaryotes, HJ resolvases exhibit additional diversity both in catalytic motifs and nuclease organization (Fekairi et al. Reference Fekairi, Scaglione, Chahwan, Taylor, Tissier, Coulon, Dong, Ruse, Yates, Russell, Fuchs, Mcgowan and Gaillard2009; Ip et al. Reference Ip, Rass, Blanco, Flynn, Skehel and West2008). HJ resolvases in lower organisms are homodimeric regardless of differences in catalytic mechanisms and symmetrically cleave fully base-paired four-way junctions in antiparallel or square–planar configurations (Biertümpfel et al. Reference Biertümpfel, Yang and Suck2007; Hadden et al. Reference Hadden, Déclais, Carr, Lilley and Phillips2007; Lilley & White, Reference Lilley and White2001). Eukaryotic HJ resolvase SLX1–SLX4 and Mus81-Eme1, however, are heterodimeric, and only one subunit contains the active site (Blanco et al. Reference Blanco, Matos, Rass, Ip and West2010; Ciccia et al. Reference Ciccia, Mcdonald and West2008). Instead of symmetric and fully base-paired junctions, eukaryotic HJ resolvases can recognize and cleave asymmetric ssDNA flaps. In bacteria, HJ often occur due to localized short palindromic sequences, which can be easily arranged in symmetric antiparallel or open-square conformation to be cleaved by homodimeric HJ resolvases. But in eukaryotes, four-way junctions often arise during homologous recombination between two distal sites or sister chromatids, where a symmetric HJ in antiparallel conformation is not easily achievable. The increasing diversity in HJ resolvases may correlate with the increased genomic complexity from bacteria to eukaryotes.

At least four different types of endonucleases are involved in site-specific DNA recombination: two kinds of Tyr recombinases (sections 7.3 and 10.2), Ser recombinases (section 8.2) and many two-metal-ion-dependent transposases (section 6.2.2). The catalytic mechanisms of these nucleases range from no metal, one-metal to two-metal dependent. Ser and Tyr recombinases are rarely found in eukaryotes and are absent in plants and metazoa. In contrast, two-metal-ion-dependent transposases/recombinases are widespread in bacteria, retroviruses and lower and higher eukaryotes. Usage of two-metal-ion catalysis is also associated with the cut-and-paste mechanism and a widened selection of target DNA for recombination.

11.2.2 Adaptation of a catalytic motif in various nucleases

Several catalytic modules of conserved structure and mechanism are found in many DNases and RNases (see sections 6.1–6.4 and 7.1). The wide distribution of a particular catalytic module is likely to reflect its catalytic efficiency. For example, the DEDD motif is associated with a large number of 3′ exonucleases, regardless of whether the substrate is DNA or RNA, and FEN1-like nucleases containing both a 5′ exo and endonuclease activity are found to remove primers in DNA replication and to cleave nucleic acid in HJ resolution, DNA repair and RNA processing.

Two catalytic motifs have the widest adaptation and are associated with the most diverse functions. They are the REase fold with the DEK motif (section 6.3) and the ββα-Me motif (section 7.1). These two catalytic motifs share the following features, which may allow them to be incorporated into different nucleases and different pathways. Firstly, the conserved catalytic motif is modular and small in size. Both the DEK and ββα-Me motifs are often contained within a stretch of 20–30 residues. The DEK motif occurs in a β-hairpin structure (Fig. 8), and the ββα-Me motif consists of a β-hairpin followed by an α-helix (Fig. 21). Thus, the DEK and ββα-Me motifs are readily incorporated into diverse surrounding tertiary structures and adapted to different nuclease activities and pathways. Secondly, both the DEK and ββα-Me motif are extremely tolerant of amino acid substitutions. Except for the first D in the DEK motif and a His serving as the general base in ββα-Me, other catalytic residues can vary greatly (see sections 6.3 and 7.1). Thirdly, the REase-like nucleases are further aided by the two-metal-ion mechanism in substrate specificity and catalytic efficiency.

It is worth noting that metal-ion-dependent mechanisms appear most adaptable. This observation may be explained by the advantages of specificity, efficiency and diversity endowed by metal ions (Yang et al. Reference Yang, Lee and Nowotny2006). Multiple metal-independent mechanisms and varied catalytic motifs are found among RNases, and this is likely due to efficient catalysis using a 2′-OH as the nucleophile. Adaptation of metal-independent mechanisms among DNase is rather limited and with few alterations, e.g. DNase II versus BfiI and TopIBs versus Tyr recombinases. The small number of metal-independent DNase may be a result of the limited types of reactions catalyzed and additional requirements for regulation and inhibition of cleavage activity.