Homology Modeling and Mutational Analysis of Ho Endonuclease of Yeast

Corresponding author: Dina Raveh, Ben Gurion University of the Negev, Beersheva, 84105 Israel., raveh{at}bgumail.bgu.ac.il (E-mail)

Communicating editor: M. D. ROSE

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Ho endonuclease is a LAGLIDADG homing endonuclease that initiates mating-type interconversion in yeast. Ho is encoded by a free-standing gene but shows 50% primary sequence similarity to the intein (protein-intron encoded) PI-SceI. Ho is unique among LAGLIDADG endonucleases in having a 120-residue C-terminal putative zinc finger domain. The crystal structure of PI-SceI revealed a bipartite enzyme with a protein-splicing domain (Hint) and intervening endonuclease domain. We made a homology model for Ho on the basis of the PI-SceI structure and performed mutational analysis of putative critical residues, using a mating-type switch as a bioassay for activity and GFP-fusion proteins to detect nuclear localization. We found that residues of the N-terminal sequence of the Hint domain are important for Ho activity, in particular the DNA recognition region. C-terminal residues of the Hint domain are dispensable for Ho activity; however, the C-terminal putative zinc finger domain is essential. Mutational analysis indicated that residues in Ho that are conserved relative to catalytic, active-site residues in PI-SceI and other related homing endonucleases are essential for Ho activity. Our results indicate that in addition to the conserved catalytic residues, Hint domain residues and the zinc finger domain have evolved a critical role in Ho activity.

HO endonuclease initiates a mating-type switch in the yeast Saccharomyces cerevisiae by making a double-strand break (DSB) in a 24-bp cognate sequence in the mating-type gene MAT (KOSTRIKEN and HEFFRON 1984 ). Repair of the DSB is by gene conversion using one of the silent HM cassettes as a template and results in substitution of the resident MAT allele with a sequence of the opposite mating type (STRATHERN et al. 1982 ; SIMON et al. 2002 ). Ho has homology to LAGLIDADG homing endonucleases (NEFF 1993 ), rare-cutting enzymes that cleave long (>14–40 bp) asymmetrical target sequences in the minor groove, leaving 4-nucleotide (nt) 3' cohesive ends. Ho (F-SceII) is encoded by a freestanding nuclear gene; however, other homing endonucleases are encoded by open reading frames (ORF) embedded in genetically mobile introns or within self-splicing inteins [protein introns (PI); BELFORT and ROBERTS 1997 ; JURICA and STODDARD 1999 ; KOWALSKI and DERBYSHIRE 2002 ]. Repair of the DSB made by the endonuclease promotes mobility of its host intron/intein into its cognate site. The survival of Ho has been attributed to its ability to promote a mating-type switch, thus allowing progeny of a single haploid spore to mate and sporulate (GIMBLE 2000 ). Eight conserved sequence motifs are found in intein-encoded LAGLIDADG endonucleases, and the primary sequence of Ho displays all the intein motifs, including those involved in protein splicing (PIETROKOVSKI 1994 ; PERLER 2000 ).

Structures have been determined using X-ray crystallography for a number of LAGLIDADG endonucleases (CHEVALIER and STODDARD 2001 ). The Chlamydomonas reinhardtii 23S rRNA intron-encoded I-CreI has a single copy of the LAGLIDADG motif and acts as a homodimer, cleaving an almost palindromic homing site (HEATH et al. 1997 ; JURICA et al. 1998 ; CHEVALIER et al. 2001 ). The ßßßß fold of the I-CreI subunit is found in two copies in the monomeric endonucleases, yeast PI-SceI (DUAN et al. 1997 ; HU et al. 2000 ), I-DmoI (SILVA et al. 1999 ), and PI-PfuI (ICHIYANAGI et al. 2000 ). These monomeric forms are thought to have arisen by gene duplication (LYKKE-ANDERSEN et al. 1996 ; SILVA et al. 1999 ). In all the structures the two LAGLIDADG -helices pack against each other with a pair of catalytic aspartic acid residues at the C-terminal ends of each helix. The LAGLIDADG -helix is followed by two anti-parallel ß-strands with a loop of varying length between them. The length of the ß-strands and size of the intervening loop are correlated with cognate site length, whereas individual residues in this region dictate site specificity (JURICA and STODDARD 1999 ; CHEVALIER and STODDARD 2001 ). All LAGLIDADG enzyme structures solved to date show the central two-helix bundle followed by two ß-strands and the ßßßß fold represents the domain topology for the family (JURICA and STODDARD 1999 ; CHEVALIER and STODDARD 2001 ).

The primary sequence of Ho has 50% similarity to PI-SceI (HIRATA et al. 1990 ). The crystal structure of PI-SceI shows two distinct domains: Domain I corresponds to the protein-splicing activity and is composed of residues from both the N and the C terminus of the primary sequence; domain II is an intervening endonuclease domain centered around the two LAGLIDADG -helices with a structure resembling that of the I-CreI homodimer (DUAN et al. 1997 ; HU et al. 2000 ). Within domain I protein-splicing activity is contained in a horseshoe-like structural motif known as the Hedgehog and intein (Hint) domain (HALL et al. 1997 ; HU et al. 2000 ). In PI-SceI an additional DNA recognition region (DRR) is present in domain I where residues important for DNA binding have been identified (HE et al. 1998 ; MOURE et al. 2002 ). PI-SceI domain I expressed by itself in Escherichia coli is able to bind a PI-SceI cognate site oligomer whereas isolated endonuclease domain II does not bind (GRINDL et al. 1998 ). In addition to PI-SceI a DRR can be identified in the primary sequence of Ho and of PI-CtrI of Candida tropicalis. PI-PfuI has a unique stirrup fold inserted between the Hint and endonuclease domains that may be analogous (ICHIYANAGI et al. 2000 ).

Ho is the only member of this large family of >200 homing endonucleases that has a C-terminal zinc-binding domain with a putative role in cognate site recognition (RUSSELL et al. 1986 ). We therefore undertook a functional analysis of Ho to identify residues important for its activity, using a mating-type switch as a bioassay for the ability of the enzyme to make a site-specific DSB at the MAT locus. We created a homology model of Ho on the basis of the coordinates of PI-SceI, using the program MODELLER (Molecular Structure) and tested the importance of residues of the protein-splicing domain for Ho activity. We superposed the model onto four solved LAGLIDADG endonuclease structures, I-CreI, I-DmoI, PI-SceI, and PI-PfuI, to identify putative active site residues that were then tested by site-directed mutagenesis. Furthermore we analyzed the unique zinc finger domain of Ho and determined the borders of this region.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Yeast plasmids and strains:
HO was cloned into pYES2 (Invitrogen, San Diego) as a HindIII fragment starting 171 bp upstream of the first ATG. Single-residue substitutions were made with the Quikchange kit (Stratagene, La Jolla, CA). Each mutation introduced a novel diagnostic restriction site and was confirmed by direct sequencing. Primers are listed in Table 1. The plasmids were transformed into a ho strain of yeast, JKM120: HML, MATa HMRa, ho, ade-1, leu 2-3,112, lys5, trp1::hisG, ura3-52. The ho deletion removes the promoter and 705 bp of ho sequence (J. K. MOORE and J. E. HABER, unpublished results). Mutant HO genes were cloned as green fluorescent protein (GFP) fusions in pHY315-GFP (KAPLUN et al. 2003 ), using primers HoSpeF and HoSacR. The 54-bp C-terminal truncation employed primer pHYTruncR. The fusion proteins were expressed in msn5 mutant cells: W303 background, leu2, ura3-52 his3-200 msn5::HIS3, obtained from J. Hood.

Ho activity assays:
Southern analysis: Cell cultures (50 ml) were grown overnight in minimal medium with 2% raffinose. HO expression was induced by addition of 2% galactose for 3 hr. Five milliliters were taken for a mating-type test, and the rest was used for DNA extraction. The DNA was digested with HindIII and run on a 0.8% agarose gel for blotting. A radioactive probe was prepared by random primer extension using a 4.3-kb MAT fragment as a template (MANIATIS et al. 1982 ). Ho cleavage of MAT gives two bands of 3.5 and 1.5 kb.

Mating-type test: pYES-HO and the mutated versions were transformed into JKM120 that is MATa. Induced and noninduced cells were mated with a MAT tester to determine their mating efficiency. The activity of the HO genes was assayed by measuring the ability of the MATa transformants to mate with a MATa tester strain and give rise to diploids on minimal plates. Mating tester strains were DC14 (MATa, his1) and DC17 (MAT, his1). Standard techniques and media composition were from SHERMAN et al. 1986 .

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Homology model of Ho based on the structure of PI-SceI:
The sequences of Ho and PI-SceI are 50% similar with few gaps in the alignment, which was helpful in our model building. Ho and PI-SceI have 29% identity across the intein domain and 35% identity across the endonuclease domain, the highest similarity being in the LAGLIDADG motifs (Fig 1).

View larger version (56K): In this window In a new window Download PPT slide   Figure 1. Intein motifs of PI-SceI and Ho. Sequence alignment of PI-SceI and Ho showing Hint sequence motifs A, N2, B, F, and G; endonuclease motifs C, D, E, and H; and the unique C-terminal zinc binding domain of Ho. The DNA recognition region (DRR) of PI-SceI is boxed as are the C-X2-C motifs of the three putative zinc fingers.

We constructed a homology model of Ho on the basis of the coordinates of the PI-SceI apoprotein as a template and a primary sequence alignment of Ho with PI-SceI made using GAP of the Wisconsin Genetics Computer Group program (Wisconsin Package Version 10.0). Side chains in the PI-SceI structure were mutated to the Ho sequence on the basis of this alignment using the program QUANTA (Molecular Simulations). Insertions and deletions in the alignment were also incorporated into the mutated structure. The mutated structure together with the parent PI-SceI structure and the sequence alignment were used as input for the program MODELLER (Molecular Structure). This program performs an energy minimization of the mutated structure while employing constraints to the atomic coordinates of aligned residues in the parent structure to generate a refined homology model.

The Ho homology model includes residues 1–465 with the exception of residues 98–108 that correspond to a disordered loop in the PI-SceI structure (residues 93–102). Initially these residues were present in the homology model, but failed to converge during the minimization process due to the lack of structural information in the PI-SceI parent structure and were therefore removed from the final model. Two other disordered loops in the PI-SceI structure (residues 271–279 and 369–374) were included in the Ho model with poorer but acceptable geometry.

Despite being encoded by a free-standing gene, the primary sequence of Ho contains the Hint sequence motifs, with inactivating mutations in critical functional residues (PERLER et al. 1997 ), in addition to the endonuclease domain II (Fig 2A). Domain I of Ho aligns well with the three-dimensional structure of PI-SceI and is composed of 185 N-terminal residues encompassing intein sequence motifs A, N2, and B. The C-terminal part extends from Glu428 to Ser465 and includes intein sequence motifs F and G. The endonuclease domain, characterized by two repeats of the ßßßß fold of the I-CreI subunit, begins at approximately Pro186 and ends at Arg427; it includes intein sequence motifs C, D, E, and H (Fig 1 and Fig 2). The sequence downstream of domain II is composed of intein motifs F and G that form the C-terminal portion of the Hint fold. Downstream of this sequence is the unique 120-residue zinc finger domain. This region of Ho has no homology to PI-SceI and was not included in the modeling.

View larger version (52K): In this window In a new window Download PPT slide   Figure 2. (A) Superposition of Ho homology model onto PI-SceI structure; the Hint and DRR that comprise the protein splicing domain are indicated, as is the endonuclease domain. The backbone of PI-SceI is in orange, the N-terminal domain I residues of Ho are in pink, the endonuclease residues are in blue, and the C-terminal domain I residues are in green. Residues of Ho that were mutated are indicated with red side chains and are labeled; numbers of parallel residues of PI-SceI appear after the slash. The zinc finger domain of Ho is not shown. (B) Hint domain and part of DRR extension of Ho with residues M1 and S465 that delineate the borders of the model indicated. P186 and R427 are the borders of the intervening endonuclease domain. Also marked are residues that served for truncations and deletions described in the text. (C) Slab showing the side chains of the active site residues of PI-SceI (red) and of Ho (blue) that were mutated. Numbering of residues is as for Ho.

The root mean square deviation between the PI-SceI structure and Ho model is 0.365 Ų. The largest differences correlate to insertions in the Ho sequence at loops in the endonuclease domain (Fig 2A). There is a two-residue insertion preceding block D of the conserved intein motifs, which contains a proposed catalytic residue (Lys308). The conformation of these inserted residues appears perturbed, and it is likely that in the native structure a change in the loop structure preceding this insertion would be observed.

Dissection of the role of the domain I residues in Ho activity:
We deleted the N-terminal 112 residues of Ho and found that this led to loss of activity. This contradicts an earlier finding of activity for the truncated gene (NAHON and RAVEH 1998 ). In PI-SceI residues Arg90 and Arg94 of the DRR have a major role in cognate site binding (HE et al. 1998 ). The model places Ho Lys99 in the vicinity of these residues of PI-SceI (Fig 2A) and we therefore mutated Lys99 to alanine. We found that this abrogated the ability of Ho to initiate a mating-type switch. The truncation and the mutation indicate that in Ho as in PI-SceI, domain I residues are important for site-specific DNA cleavage (Fig 3 and Fig 4).

View larger version (75K): In this window In a new window Download PPT slide   Figure 3. Southern analysis to test for activity of Ho mutants. (Top) wt, cleavage of MAT after induction of pYES-HO into two bands of 3.5 and 1.5 kb (arrows); V, pYES control; ZFs, C-terminal truncation of Ho—remaining protein consisting of residues 1–446 has no endonucleolytic activity; K308A substitution; K417A substitution; R286A substitution; and Ala286 restored to arg, the restored allele shows cleavage of MAT. (Bottom) wt; FG is an internal deletion of the C-terminal F and G intein motifs, residues 446–465, and extends to residue 586; 15 is a C-terminal truncation of 15 residues and extends to residue 571; 54, of 54 C-terminal residues to residue 532; 112N, of 112 N-terminal residues; and ZF1, ZF2, and ZF3 are substitutions of the upstream two cysteine residues of each putative zinc finger motif with alanine residues.

View larger version (58K): In this window In a new window Download PPT slide   Figure 4. Mating-type tests of Ho and mutant versions. MATa cells transformed with a specific HO plasmid were mated with the MAT tester control and then tested for their ability to mate with the MATa tester strain. (A) Ho; N-113, N-terminal truncation of 113 residues; K99A in the DRR; FG, internal deletion of intein motifs F and G (residues 446–465); FG and ZFs, C-terminal truncation of intein motifs F and G and the zinc finger domain (remaining endonuclease extends from 1 to 446). Ho and FG show mating with the MATa tester. (B) Ho; -15C, C-terminal deletion of 15 residues; -54C, C-terminal deletion of 54 residues; ZF1, mutations of C466A and C469A; ZF2, C508A and C511A mutations; ZF3, C558A and C561A substitutions. Ho and the 15-residue deletion show mating with the MATa tester. (C) Ho, D333A, K308A, K417A, R286A, and A286R restored. Ho and the restored R286 show mating, and K308A shows residual mating ability. (Plate) Serial 10-fold dilutions of MATa cells transformed with column 1, Ho-K308A, and column 2, Ho, were mated with the control MAT tester. Both cells show the same mating efficiency. In columns 3 and 4 Ho-K308A and Ho, respectively, were mated with the MATa tester. The K308A mutation reduces the mating-type switching efficiency of Ho 10-fold.

In the Ho model the C-terminal part of domain I starts at Arg427, 56 residues downstream of domain II intein motif H, and ends at Ser465 before the first Cys of the putative zinc finger domain (Fig 2B). It includes intein sequence motifs F and G. We truncated both these sequence motifs and the zinc finger domain to make a protein that extends from residue 1 to Gly446 (Fig 2B). This C-terminal domain-truncated Ho displays no activity. However, when we restored the zinc finger domain by making an internal deletion of motifs F and G (residues 446–465) we found that this construct could induce a mating-type switch (Fig 3 Fig 4 Fig 5).

View larger version (47K): In this window In a new window Download PPT slide   Figure 5. Diagram showing the results of mutations and deletions in the C terminus of Ho. (A) Internal deletion of intein sequence motifs F and G does not affect Ho activity. (B) A 15-residue C-terminal truncation does not affect activity. (C) A 54-residue C-terminal truncation abolishes activity. (D) First zinc finger C466A and C469A mutations inactivate Ho. (E) Second zinc finger C508A and C511A mutations inactivate. (F) Third zinc finger C558A and C561A substitutions inactivate.

Dissection of the zinc finger domain of Ho:
The zinc finger structure is stabilized by a zinc ion bound to four cysteine moieties and three linearly ordered zinc fingers can be predicted from the primary sequence of the zinc-binding domain of Ho (RUSSELL et al. 1986 ). The first two putative zinc fingers of Ho correspond to the C2C2C2C2 zinc finger type and have a sequence Cys-X2-Cys-X16-Cys-X2-Cys-X18-Cys-X2-Cys-X10-Cys-X2-Cys (KLUG and SCHWABE 1995 ). They are followed by 32 residues that precede a putative third zinc finger of sequence C-X2-C-X12-H-X2-C that terminates 9 residues before the C terminus of Ho. We constructed a 15-residue C-terminal truncation of Ho that eliminates half the putative third zinc finger. This truncated form of Ho retains endonucleolytic activity. The zinc finger domain is, however, essential for Ho activity as a 54-residue truncation of Ho that eliminates the entire third zinc finger and 26 upstream residues caused Ho to lose its activity (Fig 3 Fig 4 Fig 5). We therefore substituted two alanine residues for the upstream cysteines of each putative zinc finger. Mutation of the cysteine residues of each of the three putative zinc fingers resulted in loss of Ho activity (Fig 4 and Fig 5).

We have shown previously that Ho is a very unstable protein with a half-life of 8 min (KAPLUN et al. 2000 ). To ensure that mutant versions of Ho that had lost the capability to induce a mating-type switch were stable and were imported into the nucleus, we fused the inactive 54-residue C-terminal truncation and the three zinc finger mutants to GFP to determine their nuclear accumulation. GFP-fusion proteins were expressed in cells deleted for the Msn5 nuclear exportin as in these mutants Ho remains within the nucleus (KAPLUN et al. 2003 ). We found that in all cases Ho was imported into the nucleus, although each of the zinc finger mutants showed a certain reduction in nuclear accumulation of Ho compared with native protein. This was somewhat more pronounced for the mutations in the upstream zinc fingers (Fig 6). Nuclear accumulation of these mutant Ho proteins resembles the less efficient karyopherin-mediated import of PI-SceI in meiotic cells (NAGAI et al. 2003 ).

View larger version (150K): In this window In a new window Download PPT slide   Figure 6. Localization of C-terminal zinc finger mutations and deletions of Ho using GFP-fusion proteins expressed in cells deleted for the Msn5 nuclear exportin. Ho54 is the C-terminal deletion to residue 532; the zinc fingers are numbered from the N terminus as in Fig 5.

Superposition of the Ho model onto the aligned structures of I-CreI, PI-SceI, and I-DmoI and identification of catalytic site residues:
The crystal structures of the four LAGLIDADG endonucleases I-CreI, PI-SceI, I-DmoI, and PI-PfuI can be superposed on the basis of the central helices formed by the LAGLIDADG sequence. The I-CreI dimer corresponds to the pseudosymmetric endonucleolytic domain of monomeric PI-SceI, I-DmoI, and PI-PfuI. The enzymes thus contain two symmetric catalytic sites with each site cleaving one DNA strand of the cognate sequence across the minor groove (JURICA et al. 1998 ; CHEVALIER et al. 2001 ).

The superposition of structures shows that diverse residues are found in the periphery of the active sites and that only the acidic side chain of the LAGLIDADG sequence that coordinates the divalent cation in the I-CreI/DNA structure (Asp20) is conserved in all four structures (CHEVALIER and STODDARD 2001 ). Cleavage of the phosphodiester bond requires a general base to activate the nucleophile, a Lewis acid to stabilize the pentacoordinate transition state, and a general acid to protonate the 3' leaving group. I-CreI uses a three-metal mechanism, the central metal being shared by two separate active sites (CHEVALIER et al. 2001 ). The general base and acid appear to be within a well-ordered solvent network positioned by peripheral residues and the first metal ion. The protein makes no direct contact to the nucleophile, the scissile phosphate, or the 3' leaving group.

Guided by the superpositions and homology model we carried out mutagenesis studies on putative critical residues in the active site of Ho (Fig 2C). Asp218 and Asp326 are the metal-coordinating residues of the PI-SceI active site; they are Asp222 and Asp333 in Ho. We mutated Asp333 of Ho to alanine by site-directed mutagenesis and this resulted in a loss of activity (Fig 3 and Fig 4).

Pairwise comparisons between structures show that a glutamine residue (Gln47) is conserved between I-CreI and both I-DmoI active sites (Gln42, the amine group in the side chain of Asn129). Mutation of this residue in I-CreI and the homologous residue in another LAGLIDADG enzyme, I-CeuI (Gln93), abolishes catalytic activity (SELIGMAN et al. 1997 ; TURMEL et al. 1997 ). No homologous residue is present in either of the two PI-SceI active sites. Conversely, a lysine residue is conserved between I-CreI (Lys98 and 98') and both PI-SceI active sites (Lys301, Lys403), but is found at only one active site in I-DmoI (Lys120; SILVA et al. 1999 ) and of PI-PfuI (ICHIYANAGI et al. 2000 ; CHEVALIER and STODDARD 2001 ). In the I-CreI/DNA structure this lysine is within 6.5 Å of the scissile phosphate. Mutation of I-CreI Lys98 abolishes catalytic activity (SELIGMAN et al. 1997 ). In PI-SceI mutation of Lys301 abolishes cleavage whereas mutation of Lys403 reduces cleavage 50-fold. This difference may be related to the asymmetry of the cognate sequence (GIMBLE et al. 1998 ) or to the order of strand cleavage (ICHIYANAGI et al. 2000 ). Ho has both symmetrically situated lysine residues (Lys308 and Lys417) and we mutated each lysine residue separately. Each residue change created a novel diagnostic restriction site (Table 1) and positive clones were confirmed by direct sequencing. Cells expressing Ho with alanine instead of either Lys308 or Lys417 did not show cleavage of MAT by Southern analysis (Fig 3). K417A did not induce a mating-type switch; however, the K308A mutant showed residual mating-type interconversion activity. Quantitation by serial dilutions indicated that the K308A mutation reduced mating-type switching 10-fold (Fig 4). In PI-SceI it is the K403A (Ho K417) substitution that retains residual activity.

Another basic residue observed in the catalytic site of the I-CreI/DNA structure is Arg51, a side chain also identified in mutation screens. Arg51 may have its counterparts in PI-SceI with Arg231 and His343, where mutations to Ala cause a small reduction in activity (GIMBLE et al. 1998 ). These residues are not conserved in the Ho model; they align with Leu235 and Tyr354 of Ho, respectively. However, Arg286 aligns with Asn281 of PI-SceI, a residue that contacts the DNA in the cocrystal (MOURE et al. 2002 ) and is shown to be critical for activity (HU et al. 1999 ). Indeed we found that substitution of Arg286 with alanine abolished Ho activity in vivo. Restoration of R286A to Arg286 with a small fragment of Ho gave a switching efficiency equal to the Ho controls (Fig 3 and Fig 4).

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Homing endonucleases recognize extremely long cognate sites despite their small size and this ensures high specificity and protects the host genome from spurious cleavage. Inteins from which the endonuclease moiety has been deleted or rendered nonfunctional are capable of protein splicing and this has led to a model for intein evolution in which an ancestral protein-splicing coding sequence was colonized by an invasive endonuclease (DERBYSHIRE et al. 1997 ). Recently an artificial bifunctional intein was reconstructed by inserting the gene for I-CreI into the ORF of the GyrA mini-intein of Mycobacterium xenopi (FITZSIMONS HALL et al. 2002 ). Intron-encoded LAGLIDADG endonucleases such as I-CreI have both DNA-binding and cleavage functions (JURICA et al. 1998 ). However, in PI-SceI most of the binding energy comes from interactions with domain I (GRINDL et al. 1998 ). Footprinting and DNA cross-linking experiments of PI-SceI followed by the crystal structure of PI-SceI bound to its cognate DNA have identified the residues that contact the target DNA. Domain II residues contact base pairs -10 to +5 whereas domain I residues contact base pairs +6 to +21 upstream of the cleavage site. Within this latter region contacts between PI-SceI Arg94 of the DRR and base pairs +18 and +19 stabilize a distortion of the DNA (MOURE et al. 2002 ). The cognate sites of PI-SceI and of Ho can be aligned, and optimal alignment is achieved by introduction of a 2-bp gap into the HO sequence (GIMBLE and WANG 1996 ). Critical residues identified in the HO cognate sequence (NICKOLOFF et al. 1986 ) and in the PI-SceI recognition sequence coincide in the region recognized by endonuclease domain II; those recognized by domain I and the DRR are not conserved (GIMBLE and WANG 1996 ). The DRR provides some of the uniformly distributed phosphate contacts and this may be the function of Ho DRR residue Lys99 that we find to be critical for Ho activity. We find that the domain I residues contributed by the C-terminal sequence of the Hint domain (intein motifs F and G) are totally dispensable. These residues assemble in the center of the Hint horseshoe structure in functional inteins (ICHIYANAGI et al. 2000 ) and are not involved in contacts with DNA in the PI-SceI-DNA cocrystal (MOURE et al. 2002 ).

Furthermore, our data show that the zinc finger domain has an essential role consistent with in vitro studies demonstrating that zinc ions are essential for Ho activity (JIN et al. 1997 ). Our finding that the downstream His-X2-Cys pair of the third putative zinc finger can be truncated without any effect on Ho activity implies that the successive pairs of Cys-X2-Cys sequences do not form linearly ordered classical zinc fingers, but that this region probably forms a more complex zinc finger fold with the ligands holding the zinc interspersed in the sequence (SCHWABE and KLUG 1994 ). The cognate sites of PI-SceI and Ho endonuclease are very similar and both the VMA1 gene that hosts PI-SceI and HO are 80,000 bp apart on chromosome IV (GIMBLE and WANG 1996 ). The zinc finger domain may have been acquired during invasion of an ectopic site on this chromosome by an ancestral variant with a relaxed site specificity. Single point mutations of the MAT target sequence (MATinc) protect the cells from Ho cleavage (WEIFFENBACH et al. 1983 ). The extra domains of the intein-encoded homing endonucleases, the DRR of PI-SceI, the stirrup of PI-PfuI, and the DRR and unique zinc finger domain of Ho may enhance the site specificity of binding, although we cannot exclude other roles. For example, the single zinc finger of intron endonuclease, I-Tev1, serves as a distance determinant in the flexible linker and determines the spacing between the site of DNA binding and the site of cleavage by the catalytic moiety (DEAN et al. 2002 ).

Superposition of the structural model of Ho onto the solved structures of four LAGLIDADG endonucleases enabled us to pinpoint residues with a potential role in catalysis, to mutate them individually, and to check their role using a simple mating-type switch bioassay. The mutagenesis study shows that the active site residues in Ho are similar to those of I-CreI and PI-SceI. The finding that K308 rather than K417 shows residual activity, rather than the opposite as observed for PI-SceI, may be due to differences between them in the architecture of the hydration sphere that coordinates the metal ions. However, despite these similarities there may be striking differences in the protein/DNA contacts as shown for the isoschizomers I-CreI and I-MsoI (CHEVALIER et al. 2003 ).

The I-CreI cognate site can be cleaved by four different subfamilies of the 23S rRNA intron-encoded I-CreI endonuclease. These proteins share a related structure although only one-third of the residues that interact with DNA are conserved. This high sequence variability may permit the spread of homing endonucleases to new genomic locations by enabling cleavage of variant DNA sequences (LUCAS et al. 2001 ). In contrast, the HO genes of S. cerevisiae and S. bayanus show 93% identity with 100% conservation of the catalytic residues and those we identified by mutation in domain I and in the zinc finger domain. Both genes are also extremely highly conserved (>99% identity) in the allopolyploid S. pastorianus, a hybrid between S. cerevisiae and S.bayanus. Each of the alleles isolated from S. pastorianus recognizes the MAT cognate site and induces a mating-type switch in S. cerevisiae (TAMAI et al. 2000 ). The difference in sequence diversity between subfamilies of I-CreI and Ho may explain why HO is restricted to the genus Saccharomyces.

	FOOTNOTES

¹ Present address: Howard Hughes Medical Institute, Department of Biochemistry, Brandeis University, Waltham, MA 02454.

	ACKNOWLEDGMENTS

We thank the International Union Against Cancer (UICC) for supporting a visit of D.R. to the Stoddard laboratory and the Israel Cancer Association and the Israel Cancer Research Fund (New York) for research support to D.R.

Manuscript received August 7, 2003; Accepted for publication October 31, 2003.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

BELFORT, M. and R. J. ROBERTS, 1997  Homing endonucleases: keeping the house in order. Nucleic Acids Res. 25:3379-3388.[Abstract/Free Full Text]

CHEVALIER, B. S. and B. L. STODDARD, 2001  Homing endonucleases: structural and functional insight into the catalysts of intron/intein mobility. Nucleic Acids Res. 29:3757-3774.[Abstract/Free Full Text]

CHEVALIER, B. S., R. J. MONNAT, JR., and B. L. STODDARD, 2001  The homing endonuclease I-CreI uses three metals, one of which is shared between the two active sites. Nat. Struct. Biol. 8:312-316.[CrossRef][Medline]

CHEVALIER, B., M. TURMEL, C. LEMIEUX, R. J. MONNAT, and B. L. STODDARD, 2003  Flexible DNA target site recognition by divergent homing endonuclease isoschizomers I-CreI and I-MsoI. J. Mol. Biol. 329:253-269.[CrossRef][Medline]

DEAN, A. B., M. J. STANGER, J. T. DANSEREAU, P. VAN ROEY, and V. DERBYSHIRE et al., 2002  Zinc finger as distance determinant in the flexible linker of intron endonuclease I-TevI. Proc. Natl. Acad. Sci. USA 99:8554-8561.[Abstract/Free Full Text]

DERBYSHIRE, V., D. W. WOOD, W. WU, J. T. DANSEREAU, and J. Z. DALGAARD et al., 1997  Genetic definition of a protein-splicing domain: functional mini-inteins support structure predictions and a model for intein evolution. Proc. Natl. Acad. Sci. USA 94:11466-11471.[Abstract/Free Full Text]

DUAN, X., F. S. GIMBLE, and F. A. QUIOCHO, 1997  Crystal structure of PI-SceI, a homing endonuclease with protein splicing activity. Cell 89:555-564.[CrossRef][Medline]

FITZSIMONS HALL, M., C. J. NOREN, F. B. PERLER, and I. SCHILDKRAUT, 2002  Creation of an artificial bifunctional intein by grafting a homing endonuclease into a mini-intein. J. Mol. Biol. 323:173-179.[CrossRef][Medline]

GIMBLE, F. S., 2000  Invasion of a multitude of genetic niches by mobile endonuclease genes. FEMS Microbiol. Lett. 185:99-107.[CrossRef][Medline]

GIMBLE, F. S. and J. WANG, 1996  Substrate recognition and induced DNA distortion by the PI-SceI endonuclease, an enzyme generated by protein splicing. J. Mol. Biol. 263:163-180.[CrossRef][Medline]

GIMBLE, F. S., X. DUAN, D. HU, and F. A. QUIOCHO, 1998  Identification of Lys-403 in the PI-SceI homing endonuclease as part of a symmetric catalytic center. J. Biol. Chem. 273:30524-30529.[Abstract/Free Full Text]

GRINDL, W., W. WENDE, V. PINGOUD, and A. PINGOUD, 1998  The protein splicing domain of the homing endonuclease PI-SceI is responsible for specific DNA binding. Nucleic Acids Res. 26:1857-1862.[Abstract/Free Full Text]

HALL, T. M., J. A. PORTER, K. E. YOUNG, E. V. KOONIN, and P. A. BEACHY et al., 1997  Crystal structure of a Hedgehog autoprocessing domain: homology between Hedgehog and self-splicing proteins. Cell 91:85-97.[CrossRef][Medline]

HE, Z., M. CRIST, H. YEN, X. DUAN, and F. A. QUIOCHO et al., 1998  Amino acid residues in both the protein splicing and endonuclease domains of the PI-SceI intein mediate DNA binding. J. Biol. Chem. 273:4607-4615.[Abstract/Free Full Text]

HEATH, P. J., K. M. STEPHENS, R. J. MONNAT, JR., and B. L. STODDARD, 1997  The structure of I-Crel, a group I intron-encoded homing endonuclease. Nat. Struct. Biol. 4:468-476.[CrossRef][Medline]

HIRATA, R., Y. OHSUMK, A. NAKANO, H. KAWASAKI, and K. SUZUKI et al., 1990  Molecular structure of a gene, VMA1, encoding the catalytic subunit of H(+)-translocating adenosine triphosphatase from vacuolar membranes of Saccharomyces cerevisiae.. J. Biol. Chem. 265:6726-6733.[Abstract/Free Full Text]

HU, D., M. CRIST, X. DUAN, and F. S. GIMBLE, 1999  Mapping of a DNA binding region of the PI-sceI homing endonuclease by affinity cleavage and alanine-scanning mutagenesis. Biochemistry 38:12621-12628.[CrossRef][Medline]

HU, D., M. CRIST, X. DUAN, F. A. QUIOCHO, and F. S. GIMBLE, 2000  Probing the structure of the PI-SceI-DNA complex by affinity cleavage and affinity photocross-linking. J. Biol. Chem. 275:2705-2712.[Abstract/Free Full Text]

ICHIYANAGI, K., Y. ISHINO, M. ARIYOSHI, K. KOMORI, and K. MORIKAWA, 2000  Crystal structure of an archaeal intein-encoded homing endonuclease PI-PfuI. J. Mol. Biol. 300:889-901.[CrossRef][Medline]

JIN, Y., G. BINKOWSKI, L. D. SIMON, and D. NORRIS, 1997  Ho endonuclease cleaves MAT DNA in vitro by an inefficient stoichiometric reaction mechanism. J. Biol. Chem. 272:7352-7359.[Abstract/Free Full Text]

JURICA, M. S. and B. L. STODDARD, 1999  Homing endonucleases: structure, function and evolution. Cell Mol. Life Sci. 55:1304-1326.[CrossRef][Medline]

JURICA, M. S., R. J. MONNAT, JR., and B. L. STODDARD, 1998  DNA recognition and cleavage by the LAGLIDADG homing endonuclease I-CreI. Mol. Cell 2:469-476.[CrossRef][Medline]

KAPLUN, L., Y. IVANTSIV, D. KORNITZER, and D. RAVEH, 2000  Functions of the DNA damage response pathway target Ho endonuclease of yeast for degradation via the ubiquitin-26S proteasome system. Proc. Natl. Acad. Sci. USA 97:10077-10082.[Abstract/Free Full Text]

KAPLUN, L., Y. IVANTSIV, A. BAKHRAT, and D. RAVEH, 2003  DNA damage response—mediated degradation of Ho endonuclease via the ubiquitin system involves its nuclear export. J. Biol. Chem. 278:48727-48734.[Abstract/Free Full Text]

KLUG, A. and J. W. SCHWABE, 1995  Protein motifs 5. Zinc fingers. FASEB J. 9:597-604.[Abstract]

KOSTRIKEN, R. and F. HEFFRON, 1984  The product of the HO gene is a nuclease: purification and characterization of the enzyme. Cold Spring Harbor Symp. Quant. Biol. 49:89-97.[Medline]

KOWALSKI, J. C. and V. DERBYSHIRE, 2002  Characterization of homing endonucleases. Methods 28:365-373.[CrossRef][Medline]

LUCAS, P., C. OTIS, J. P. MERCIER, M. TURMEL, and C. LEMIEUX, 2001  Rapid evolution of the DNA-binding site in LAGLIDADG homing endonucleases. Nucleic Acids Res. 29:960-969.[Abstract/Free Full Text]

LYKKE-ANDERSEN, J., R. A. GARRETT, and J. KJEMS, 1996  Protein footprinting approach to mapping DNA binding sites of two archaeal homing enzymes: evidence for a two-domain protein structure. Nucleic Acids Res. 24:3982-3989.[Abstract/Free Full Text]

MANIATIS, T., E. FRITSCH and J. SAMBROOK, 1982 Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

MOURE, C. M., F. S. GIMBLE, and F. A. QUIOCHO, 2002  Crystal structure of the intein homing endonuclease PI-SceI bound to its recognition sequence. Nat. Struct. Biol. 9:764-770.[CrossRef][Medline]

NAGAI, Y., S. NOGAMI, F. KUMAGAI-SANO, and Y. OHYA, 2003  Karyopherin-mediated nuclear import of the homing endonuclease VMA1-derived endonuclease is required for self-propagation of the coding region. Mol. Cell. Biol. 23:1726-1736.[Abstract/Free Full Text]

NAHON, E. and D. RAVEH, 1998  Targeting a truncated Ho-endonuclease of yeast to novel DNA sites with foreign zinc fingers. Nucleic Acids Res. 26:1233-1239.[Abstract/Free Full Text]

NEFF, N., 1993  Protein splicing: selfish genes invade cellular proteins. Curr. Opin. Cell Biol. 5:971-976.[CrossRef][Medline]

NICKOLOFF, J. A., E. Y. CHEN, and F. HEFFRON, 1986  A 24-bp sequence from MAT locus stimulates intergenic recombination in yeast. Proc. Natl. Acad. Sci. USA 83:7831-7835.[Abstract/Free Full Text]

PERLER, F. B., 2000  InBase, the intein database. Nucleic Acids Res. 28:344-345.[Abstract/Free Full Text]

PERLER, F. B., M. Q. XU, and H. PAULUS, 1997  Protein splicing and autoproteolysis mechanisms. Curr. Opin. Chem. Biol. 1:292-299.[CrossRef][Medline]

PIETROKOVSKI, S., 1994  Conserved sequence features of inteins (protein introns) and their use in identifying new inteins and related proteins. Protein Sci. 3:2340-2350.[Abstract]

RUSSELL, D., R. JENSEN, M. ZOLLER, J. BURKE, and B. ERREDE et al., 1986  Structure of the Saccharomyces cerevisiae HO gene and analysis of its upstream regulatory region. Mol. Cell. Biol. 6:4281-4294.[Abstract/Free Full Text]

SCHWABE, J. W. and A. KLUG, 1994  Zinc mining for protein domains. Nat. Struct. Biol. 1:345-349.[CrossRef][Medline]

SELIGMAN, L. M., K. M. STEPHENS, J. H. SAVAGE, and R. J. MONNAT, JR., 1997  Genetic analysis of the Chlamydomonas reinhardtii I-CreI mobile intron homing system in E. coli.. Genetics 147:1653-1664.[Abstract]

SHERMAN, F., G. FINK and J. HICKS, 1986 Methods in Yeast Genetics. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

SILVA, G. H., J. Z. DALGAARD, M. BELFORT, and P. VAN ROEY, 1999  Crystal structure of the thermostable archaeal intron-encoded endonuclease I-DmoI. J. Mol. Biol. 286:1123-1136.[CrossRef][Medline]

SIMON, P., P. HOUSTON, and J. BROACH, 2002  Directional bias during mating type switching in Saccharomyces is independent of chromosomal architecture. EMBO J. 21:2282-2291.[CrossRef][Medline]

STRATHERN, J. N., A. J. KLAR, J. B. HICKS, J. A. ABRAHAM, and J. M. IVY et al., 1982  Homothallic switching of yeast mating type cassettes is initiated by a double-stranded cut in the MAT locus. Cell 31:183-192.[CrossRef][Medline]

TAMAI, Y., K. TANAKA, N. UMEMOTO, K. TOMIZUKA, and Y. KANEKO, 2000  Diversity of the HO gene encoding an endonuclease for mating-type conversion in the bottom fermenting yeast Saccharomyces pastorianus.. Yeast 16:1335-1343.[CrossRef][Medline]

TURMEL, M., C. OTIS, V. COTE, and C. LEMIEUX, 1997  Evolutionarily conserved and functionally important residues in the I-CeuI homing endonuclease. Nucleic Acids Res. 25:2610-2619.[Abstract/Free Full Text]

WEIFFENBACH, B., D. T. ROGERS, J. E. HABER, M. ZOLLER, and D. W. RUSSELL et al., 1983  Deletions and single base pair changes in the yeast mating type locus that prevent homothallic mating type conversions. Proc. Natl. Acad. Sci. USA 80:3401-3405.[Abstract/Free Full Text]

This article has been cited by other articles:

				J. B. Robbins, M. Stapleton, M. J. Stanger, D. Smith, J. T. Dansereau, V. Derbyshire, and M. Belfort Homing endonuclease I-TevIII: dimerization as a means to a double-strand break Nucleic Acids Res., March 12, 2007; 35(5): 1589 - 1600. [Abstract] [Full Text] [PDF]

				A. Bakhrat, K. Baranes, O. Krichevsky, I. Rom, G. Schlenstedt, S. Pietrokovski, and D. Raveh Nuclear Import of Ho Endonuclease Utilizes Two Nuclear Localization Signals and Four Importins of the Ribosomal Import System J. Biol. Chem., May 5, 2006; 281(18): 12218 - 12226. [Abstract] [Full Text] [PDF]

				L. Kaplun, R. Tzirkin, A. Bakhrat, N. Shabek, Y. Ivantsiv, and D. Raveh The DNA Damage-Inducible UbL-UbA Protein Ddi1 Participates in Mec1-Mediated Degradation of Ho Endonuclease Mol. Cell. Biol., July 1, 2005; 25(13): 5355 - 5362. [Abstract] [Full Text] [PDF]

				V. Koufopanou and A. Burt Degeneration and Domestication of a Selfish Gene in Yeast: Molecular Evolution Versus Site-Directed Mutagenesis Mol. Biol. Evol., July 1, 2005; 22(7): 1535 - 1538. [Abstract] [Full Text] [PDF]