Engineering Homing Enzymes with Novel Functions

4.1 Changing the Recognition Site Specificity of Homing Endonucleases

Homing endonucleases have become invaluable tools to study complex genomes. By virtue of its ability to introduce unique breaks into complex genomes that mimic lesions made by DNA-damaging agents, I-Scel is commonly used to study DNA repair pathways in organisms ranging from yeast to humans (see, for example, Johnson and Jasin 2001). Moreover, this enzyme stimulates gene targeting over 1000-fold in mammalian systems through homologous recombination (Rouet et al. 1994). However, the current repertoire of homing endonuclease specificities is inadequate to effect gene targeting at desired loci. Gene targeting using the available homing enzymes is limited by the necessity of having to first insert the recognition sites at a locus by conventional targeting methods. The ability to tailor homing endonucleases to effect site-specific cleavage at pre-existing sites would permit repair of defective genes at any given locus. Homing endonucleases could also be engineered to act as transcriptional repressors of genes that are not normally regulated. Uncoupling the DNA-binding and nuclease activities could be accomplished by mutating the acidic residues that bind the metal ion cofactors, leaving intact the site-specific DNA-binding activity (Gimble and Stephens 1995). In addition, site-specific proteins could be developed as therapeutic agents that deliver tethered small molecules to defined regions of the genome.

The approaches used to alter homing endonuclease specificity range from those that target a limited number of contacts to subtly modify DNA recognition to others that generate gross changes in the protein-DNA interaction surface. What also distinguishes these strategies is that some focus on rational design based on high-resolution X-ray structures while others rely more on combinatorial methods coupled with genetic screens and selections. As enumerated below, there are strengths and weaknesses to each approach.

4.1.1 Altering Homing Endonuclease Specificity by Domain Shuffling

One strategy to alter homing endonuclease specificity has been to exchange an entire intein domain or portions of it with domains from other inteins or unrelated DNA-binding proteins. An advantage of working with inteins is that part of their DNA-binding specificity originates in the protein-splicing domain, which, in some cases, can be uncoupled from the endonuclease domain that cleaves the DNA. Some of these engineered proteins have been modeled after chimeric endonucleases created by fusing the non-specific catalytic domain of Fokl, a class IIS restriction enzyme, with DNA-binding domains or

DNA-binding proteins (Kim et al. 1996,1998). For example, the Pl-Scel protein splicing domain, which makes independent base-specific contacts to a subset of the total Pl-Scel recognition sequence (Grindl et al. 1998; Moure et al. 2002), was fused to the Fokl catalytic domain to generate a functional endonuclease (Hu and Gimble, unpubl. data). As expected, this enzyme exhibits less specificity than full-length Pl-Scel, which establishes base-specific contacts to the entire recogntion sequence using both protein domains. Furthermore, rather than cleaving at unique sites on each DNA strand like Pl-Scel, the Pl-Scel/ Fokl chimera cleaves at multiple neighboring sites (Hu and Gimble, unpubl. data). Regardless, this work demonstrates that the Pl-Scel protein-splicing domain can be recruited as a site-specific DNA-binding module to target other protein domains to specific sequences.

Domain shuffling between the Pl-Scel and its Candida tropicalis analog, Pl-CtrIP, was used to design altered specificity proteins (Steuer et al. 2004). Replacement of the Pl-Scel protein-splicing domain with that of Pl-CtrIP resulted in inactive proteins. Thus, proper register between domains is likely to be critical for domain shuffling to work. However, an active chimera results when the DNA recognition region (DRR) of Pl-Scel, which is a subdomain of the protein-splicing domain that makes some of the base-specific DNA contacts, was replaced with its analog from Pl-CtrIP. Interestingly, this protein displayed a small preference for the Candida tropicalis site, which differs from the Saccharomyces cerevisiae site at six nucleotide positions, suggesting that at least a portion of the specificity has been transferred to Pl-Scel (Steuer et al. 2004).

Other strategies to alter specificity have focused on shuffling the domains within the catalytic region of intron-encoded homing endonucleases. Relative to engineering inteins, tinkering with these endonucleases is more ambitious because the DNA-binding determinants are tightly linked to the catalytic center due to their close proximity, making it more difficult to alter binding without also affecting the catalytic machinery (Epinat et al. 2003). Another reason is that the complex hydration shell and shared metal ion cofactor that are key components of the active sites are positioned by residues at the interface of the two domains and can be easily disrupted by alteration of either domain.

Two groups fused the N-terminal domain of I-Dmol to the C-terminal domain of I-Crel to create chimeric endonucleases that cleave hybrid substrates derived from each parent, but not the parental sites (Chevalier et al. 2002; Epinat et al. 2003). Here, the underlying assumption is that each domain would continue to bind its respective native half-site within the composite substrate. Initial structural modeling of the chimeric protein revealed several steric clashes between the domain interface residues. Simply substituting these residues with alanine resulted in insoluble proteins (Chevalier et

-Dmol l-Crel

-Dmol l-Crel

Fig. 1. Engineering of a chimeric homing endonuclease. The X-ray crystal structures of the parent enzymes I-Dmol (1B24 [Silva et al. 1999]) and I-Crel (1G9Z [Chevalier et al.

2001]) are depicted in green and blue, respectively. Their DNA substrates are shown below each protein, with each half-site labeled separately (I-Dmol substrate half-sites D1 and D2, orange; I-Crel substrate half-sites CI and CI', magenta). I-Dmol, an asymmetric enzyme, binds the substrate only in the depicted orientation, but I-Crel, a symmetric homodim-er, binds to its substrate in either orientation. The I-Dmol N-terminal domain (green) and the I-Crel C-terminal domain (blue) were fused to yield H-Drel (1M0W [Chevalier et al.

2002]). H-Drel only cleaves the two chimeric substrates that are shown of the four that are possible from different combinations of I-Dmol and I-Crel half-sites, and it cleaves neither of the parent enzyme substrates

Fig. 1. Engineering of a chimeric homing endonuclease. The X-ray crystal structures of the parent enzymes I-Dmol (1B24 [Silva et al. 1999]) and I-Crel (1G9Z [Chevalier et al.

2001]) are depicted in green and blue, respectively. Their DNA substrates are shown below each protein, with each half-site labeled separately (I-Dmol substrate half-sites D1 and D2, orange; I-Crel substrate half-sites CI and CI', magenta). I-Dmol, an asymmetric enzyme, binds the substrate only in the depicted orientation, but I-Crel, a symmetric homodim-er, binds to its substrate in either orientation. The I-Dmol N-terminal domain (green) and the I-Crel C-terminal domain (blue) were fused to yield H-Drel (1M0W [Chevalier et al.

2002]). H-Drel only cleaves the two chimeric substrates that are shown of the four that are possible from different combinations of I-Dmol and I-Crel half-sites, and it cleaves neither of the parent enzyme substrates al. 2002). In order to reduce the number of different combinations of interface residues to be tested of the total sequence space, one group applied computational algorithms to search through 8xl017 sequence combinations to identify those with low local free energy amino acid sequences at the interface (Chevalier et al. 2002). Sixteen of these were further screened for solubility using a LacZ-based screen in vivo. One chimeric variant, E-Drel, now termed H-Drel according to convention (Roberts et al. 2003), exhibits a specificity that is distinct from either of the two parent enzymes (Fig. 1). It does not cleave either of the parental substrates and, of the four chimeric substrates with different combinations of I-Crel and I-Drel half-sites, H-Drel only cleaves two, those with one specific dmo half-site fused to either of the ere half-sites. Thus, the I-Dmol N-terminal domain recognizes a specific half-site of its substrate, as expected for an asymmetric enzyme, while the I-Crel domain, which originates from a symmetric homodimer, interacts with either ere half-site. The DNA cleavage geometry is identical to all other characterized LAGLIDADG enzymes, yielding a four base, 3'-overhang, and the cleavage activity is similar to I-Crel. The 2.4 A X-ray structure of H-Drel closely resembles the computationally redesigned model, confirming the accuracy of the method (Chevalier et al. 2002). A second group engineered a chimera termed DmoCre having the same domain architecture as H-Drel (Epinat et al. 2003). The cleavage properties of DmoCre and H-Drel were similar even though different mutations were made at the domain interface, indicating that there are multiple solutions to the problem of engineering the LAGLIDADG a-helical interface. A constraint of domain shuffling is that the specificity of the resulting chimera is always directly related to the specificities of the parent enzymes. These studies illustrate that the partial independence of LAGLIDADG domains permits them to be recombined to yield novel reagents.

4.1.2 Altering Homing Endonuclease Specificity Using Genetic Screens and Selections

Complementary strategies to obtain altered specificity enzymes have used genetic selections or screens in bacteria to isolate variants from complex libraries. In theory, if the entire sequence space of a protein could be screened, variants with any desired specificity could be obtained without prior structural or biochemical knowledge of the system. In reality, however, the complexity of the library that can be probed, which increases exponentially with the linear increase in the number of randomized residue positions, is limited by the number of library members that can be transformed into E. coli and assayed. Since the upper limit on the number of independent transformants for a given library is ~107-109, the size of the sequence that can be completely randomized (e.g. 20 different codons at n different positions) with 99% confidence that all variants will be obtained is limited to n=6-8 (Lowman and Wells 1991). Thus, mutagenizing an entire protein domain has not been possible, and it has been necessary, instead, to use existing structural information to target specific interactions for analysis.

The genetic methods that have been used to alter homing endonucle-ase specificity can be broadly divided into those that select/screen for DNA-binding activity and those that select/screen for catalytic activity. Selections for DNA-cleavage activity are inherently preferable because variants are obtained with the desired property. Variations that contribute to cleavage activity by affecting any of the steps of the reaction pathway, including ligand binding, metal ion cofactor binding, coupling between DNA binding and catalysis, the catalytic reaction, or product release, will be selected. By contrast, variants that are selected for DNA binding are not guaranteed to be catalytically active. If the DNA-binding site is tightly coupled to the catalytic site, it may be impossible to perturb the DNA-binding determinants without also decreasing the DNA-cleavage activity. Moreover, since these methods select for binding of the protein in the ground state, it is expected that isolated variants will have low turnover numbers. This latter constraint may not be an issue for homing endonucleases, however, since these evolved to have extreme specificity, rather than high turnover numbers, in order to cleave at single genomic target sites. In general, genetic selections are preferred to genetic screens since they generally permit a larger part of the total sequence space to be sampled in a single experiment.

An adaptation of a bacterial two-hybrid strategy (Joung et al. 2000) selected altered specificity variants of the intein-encoded Pl-Scel homing endonu-clease (Gimble et al. 2003). This DNA-binding selection consists of three components (Fig. 2A); a protein fusion between the yeast GalllP protein and PI-Scel, a second protein fusion between the E. coli RNA polymerase a subunit (RNAPa) and the yeast Gal4 protein, and a DNA target that includes a mutant Pl-Scel recognition sequence located upstream of a weak lac promoter that controls the selectable HIS3 and aadA genes (Gimble et al. 2003). GalllP/PI-Scel variants expressed from a plasmid library that bind to the mutant target site can be selected because they recruit the RNAPa/Gal4 fusion protein to the promoter through the GalllP-Gal4 interaction, thereby conferring bacterial growth on histidine- and spectinomycin-selective media. The study targeted a protein/DNA interaction within the Pl-Scel protein-splicing domain because its distance from the endonuclease active sites minimizes the deleterious effects of the mutations on catalysis. Whether this DNA-binding selection can be successfully applied to intron-encoded homing endonucleases, where the DNA-binding determinants and the catalytic residues are immediately adjacent and presumably tightly coupled, is unclear. A residue that makes a critical

Fig. 2. Bacterial selections and screens for altered specificity homing endonucleases. a Bacterial two-hybrid binding selection (Gimble et al. 2003). GalllP/PI-Scel fusion variants expressed from a plasmid library associate with a RNAPa/Gal4 fusion protein encoded by a separate plasmid. If the Pl-Scel variant binds to a homing site located on an F', it recruits the RNAPa subunit proximal to the weak Plac promoter, leading to an increase in HIS3 and aadA expression and growth on histidine-selective and spectinomycin-selective media, b Screen for I-Crel cleavage activity (Seligman et al. 2002). Plasmid-encoded I-Crel derivatives that bind and cleave an I-Crel homing site located on an F' lead to its elimination. Concomitant loss of an adjacent kanamycin antibiotic marker yields kanamycin-sen-sitive cells, c Selection for homing endonuclease activity (Gruen et al. 2002). Co-existence of two plasmids kills bacterial cells because one expresses an amber nonsense allele of a toxic gene product, barnase, and the other expresses an amber tRNA suppressor. However, when the tRNA supressor plasmid also expresses a homing enzyme that cleaves the homing site on the barnase expression plasmid, the cells survive due to elimination of the barnase gene

Fig. 2. Bacterial selections and screens for altered specificity homing endonucleases. a Bacterial two-hybrid binding selection (Gimble et al. 2003). GalllP/PI-Scel fusion variants expressed from a plasmid library associate with a RNAPa/Gal4 fusion protein encoded by a separate plasmid. If the Pl-Scel variant binds to a homing site located on an F', it recruits the RNAPa subunit proximal to the weak Plac promoter, leading to an increase in HIS3 and aadA expression and growth on histidine-selective and spectinomycin-selective media, b Screen for I-Crel cleavage activity (Seligman et al. 2002). Plasmid-encoded I-Crel derivatives that bind and cleave an I-Crel homing site located on an F' lead to its elimination. Concomitant loss of an adjacent kanamycin antibiotic marker yields kanamycin-sen-sitive cells, c Selection for homing endonuclease activity (Gruen et al. 2002). Co-existence of two plasmids kills bacterial cells because one expresses an amber nonsense allele of a toxic gene product, barnase, and the other expresses an amber tRNA suppressor. However, when the tRNA supressor plasmid also expresses a homing enzyme that cleaves the homing site on the barnase expression plasmid, the cells survive due to elimination of the barnase gene contact, Arg-94, and four neighboring residues were randomized to generate a plasmid library for the selection. Altered specificity proteins were selected that gain the ability to bind to substrates containing mutations at two nucleotides contacted by Arg-94 in wild-type Pl-Scel. A powerful feature of the method is that the binding stringency of the isolates could be increased by including 3-aminotriazole, a competitive inhibitor of HIS3, during the selection. The DNA-binding specificities of the selected variants ranged from being relaxed (i.e. able to cleave the wild-type and mutant targets equally) to being dramatically shifted to preferring the selection targets. None of the variants displayed the same degree of specificity as wild-type Pl-Scel (Gimble et al. 2003). To achieve this level of specificity, a negative selection against binding to the wild-type recognition sequence may need to be applied concomitant with the positive selection for binding to the new target site.

Genetic screens and selections for homing endonucleases that cleave altered recognition sequences have been developed. In a study of I-Crel, proteins containing each of the 19 possible non-wild-type amino acid substitutions for a residue that makes a critical DNA contact were expressed and screened in vivo for their ability to cleave and eliminate a reporter plasmid containing a mutant I-Crel recognition sequence (Seligman et al. 2002). Plasmid loss following DNA cleavage at the mutant site results in loss of kanamycin resistance or P-galactosidase activity, depending on whether the plasmid contains the kanR or lacZ genes (Fig. 2B). Shifted, rather than completely altered, specificity proteins were obtained, perhaps due to the lack of a negative selection/screen against cleavage of other recognition sequences. Determination of the crystal structures of these constructs revealed that the new contacts do not significantly perturb the local protein conformation or surrounding DNA contacts (Sussman et al. 2004). The observed independence of the contacts will simplify the engineering of homing endonucleases. A similar method was set up as a genetic selection (Fig. 2C) in which a toxic gene product, barnase, expressed by a reporter plasmid, kills all of the bacteria except those that also encode a homing endonuclease variant that cleaves a homing site within the barnase plasmid (Gruen et al. 2002). Whether the background level of false positive isolates is sufficiently low to obtain altered specificity variants by this method is still unclear. An ingenious approach using phage display, which has been applied to staphylococcal nuclease but not to homing enzymes, permits selection for catalytic activity as well as rapid, iterative rounds of directed evolution to refine endonuclease specificity. Here, M13 phage that express library candidates that are catalytically active cleave a neighboring DNA substrate on the phage surface, thereby causing the selected phage particle, but no others, to be released from a solid support (Pedersen et al. 1998).

4.2 Introducing Molecular Switches into Homing Endonucleases

The ability to temporally or spatially control homing endonuclease activity would increase their utility in vitro and in vivo. Two different approaches can be used to modulate activity, by incorporating a trigger that activates or inactivates the enzyme irreversibly or by inserting a switch that allows the endonuclease to be reversibly turned on and off. One approach to tightly regulate homing enzymes that requires no engineering is to simply use chelating agents to alter the concentration of the divalent metal ion, which is essential to the reaction. However, this strategy is impractical in vivo since it is difficult to modulate metal concentrations within the cell, and because numerous other metabolic activities would be affected by changes in the metal concentration. Another approach, which has been applied to restriction endonucleases (Muir et al. 1997), is to isolate temperature-sensitive variants that become inactivated at high temperature. Altering the temperature in vivo, however, is likely to adversely affect other cellular functions.

Inserting switches or triggers into homing endonucleases that regulate only these proteins would be the most effective means to control their activity. Progress has been made in this respect by showing that PI-SceI can be reversibly regulated using a redox switch (Posey and Gimble 2002). To create the switch, one of two cysteines was inserted into a flexible loop that undergoes a hinge-flap motion in the presence of DNA to make minor-groove contacts, while the second was inserted into an adjacent P-hairpin loop that contacts the DNA in the major groove. In the reduced state, the loops freely undergo the conformational changes that permit them to contact DNA, and the protein is nearly as active as wild-type PI-SceI. However, under oxidizing conditions, formation of a disulfide bond between the cysteines constrains the loops, effectively trapping the protein in a non-productive conformation that binds DNA poorly and is catalytically inactive. Successive cycles of reducing and oxidizing treatments can be used to switch PI-SceI activity on and off (Posey and Gimble 2002). The major drawback of the switch is that it is currently limited to in vitro applications since changing the intracellular redox potential is untenable. The next step will be to develop switches in homing endonucleases that are activated using small molecules that are easily transported into cells and that do not negatively affect cellular metabolism. An alternative strategy may involve "caged" homing enzymes that have been chemically modified at inserted cysteine residues with different ortho-nitrobenzyl moieties that inactive the proteins (Bayley et al. 1998). After exposure to UV-visible radiation, the release of the covalent modification triggers the activation of the protein.

Cure Your Yeast Infection For Good

Cure Your Yeast Infection For Good

The term vaginitis is one that is applied to any inflammation or infection of the vagina, and there are many different conditions that are categorized together under this ‘broad’ heading, including bacterial vaginosis, trichomoniasis and non-infectious vaginitis.

Get My Free Ebook


Post a comment