[Genome Editing Tools and their Application in Experimental Ophthalmology]. / Genome Editing Tools und ihr Einsatz in der experimentellen Augenheilkunde.

New genome editing tools in molecular biology are revolutionising precise genome surgery and have greatly influenced experimental ophthalmology too. Aside from the commonly used nuclease-based platforms, such as the zinc-finger nucleases (ZFN) and transcription activator-like effector nucleases (TALEN), CRISPR/Cas systems, clustered regularly interspaced short palindromic repeats (CRISPR) and CRISPR-associated (Cas) genes, perform very efficiently in site-specific DNA cleavage within living cells. DNA double strand breaks (DSB) are repaired through two different conserved repair pathways: NHEJ (non-homologous end joining) and HDR (homology directed repair). By using the correct DNA templates, these repair pathways can be used to knock out defective genes or to repair mutations. Genome editing technology lays the ground for new strategies in basic science, biotechnology, and biomedical science, as well as clinical studies with genome editing. Therapeutic gene editing strategies are now concentrating on diseases in the retina, due to the comparatively easy accessibility of the eye and with local application in vivo.

Type II restriction endonucleases: structure and mechanism.

Type II restriction endonucleases are components of restriction modification systems that protect bacteria and archaea against invading foreign DNA. Most are homodimeric or tetrameric enzymes that cleave DNA at defined sites of 4-8 bp in length and require Mg2+ ions for catalysis. They differ in the details of the recognition process and the mode of cleavage, indicators that these enzymes are more diverse than originally thought. Still, most of them have a similar structural core and seem to share a common mechanism of DNA cleavage, suggesting that they evolved from a common ancestor. Only a few restriction endonucleases discovered thus far do not belong to the PD...D/ExK family of enzymes, but rather have active sites typical of other endonuclease families. The present review deals with new developments in the field of Type II restriction endonucleases. One of the more interesting aspects is the increasing awareness of the diversity of Type II restriction enzymes. Nevertheless, structural studies summarized herein deal with the more common subtypes. A major emphasis of this review will be on target site location and the mechanism of catalysis, two problems currently being addressed in the literature.

Refractive state of tree shrew eyes measured with cortical visual evoked potentials.

PURPOSE: To determine the refractive state of tree shrew eyes using visual evoked potentials (VEP's) recorded from primary visual cortex and compare the values with those obtained with streak retinoscopy and with an autorefractor. METHODS: VEP's were recorded in seven normal tree shrews and three animals in which approximately 5 D of myopia (relative to control eye) was induced by monocular -5 D lens wear. While the animals were awake, refractive correction was measured with an autorefractor before and after cycloplegia (1% atropine and 2.5% phenylepherine). When anesthetized, cycloplegic refractive correction was measured with streak retinoscopy. Then VEP's were produced with square-wave counterphased (1 Hz) high-contrast checkerboard patterns near the animals' high spatial frequency cutoff. Spherical lenses (2 D steps) were placed before the eye, and the VEP (average of 128 sweeps) was measured to determine the lens that produced the largest first positive peak (P1). RESULTS: VEP's were obtained over a broad range of trial lenses. Tuning was narrower when check sizes were small. In normal and control eyes, the P1 amplitude was largest, on average, for a trial lens of (mean +/- SD) -0.6 +/- 1.6 D (corrected for working distance but not vertex distance). The mean streak retinoscopy value (spherical equivalent at the corneal plane) was 7.0 +/- 0.8 D, and mean autorefractor values were 4.0 +/- 1.1 D (cycloplegic) and 3.7 +/- 1.2 D (noncycloplegic). In the eyes that compensated for a -5 D lens, the largest P1 values occurred with lenses with a power of -6.3 +/- 3.2 D. Thus, the VEP measure showed a similar treated vs. control eye difference as did streak retinoscopy (treated eyes, 4.7 +/- 0.4 D myopic) and the autorefractor (treated eyes, 4.8 +/- 0.5 D myopic). CONCLUSIONS: Normal tree shrew eyes are approximately emmetropic. The hyperopic values obtained with streak retinoscopy and the autorefractor are consistent with the presence of a "small-eye artifact" in tree shrews. Eyes that have compensated for a -5 D lens are myopic by approximately the value of the lens.

Identification of Asp218 and Asp326 as the principal Mg2+ binding ligands of the homing endonuclease PI-SceI.

The monomeric homing endonuclease PI-SceI harbors two catalytic centers which cooperate in the cleavage of the two strands of its extended recognition sequence. Structural and biochemical data suggest that catalytic center I contains Asp218, Asp229, and Lys403, while catalytic center II contains Asp326, Thr341, and Lys301. The analogy with I-CreI, for which the cocrystal structure with the DNA substrate has been determined, suggests that Asp218 and Asp229 in catalytic center I and Asp326 and Thr341 in catalytic center II serve as ligands for Mg(2+), the essential divalent metal ion cofactor which can be replaced by Mn(2+) in vitro. We have carried out a mutational analysis of these presumptive Mg(2+) ligands. The variants carrying an alanine or asparagine substitution bind DNA, but (with the exception of the D229N variant) are inactive in DNA cleavage in the presence of Mg(2+), demonstrating that these residues are important for cleavage. Our finding that the PI-SceI variants carrying single cysteine substitutions at these positions are inactive in the presence of the oxophilic Mg(2+) but active in the presence of the thiophilic Mn(2+) suggests that the amino acid residues at these positions are involved in cofactor binding. From the fact that in the presence of Mn(2+) the D218C and D326C variants are even more active than the wild-type enzyme, it is concluded that Asp218 and Asp326 are the principal Mg(2+) ligands of PI-SceI. On the basis of these findings and the available structural information, a model for the composition of the two Mg(2+) binding sites of PI-SceI is proposed.

A model for the PI-SceIxDNA complex based on multiple base and phosphate backbone-specific photocross-links.

We have synthesized different oligodeoxynucleotides carrying, in single positions of the >36 bp recognition site of PI-SceI, photoreactive base analogues (5-iododeoxypyrimidines) or phosphate modifications (p-azidophenacylphosphorothioates) and used them in photocross-linking experiments with PI-SceI to probe the protein-DNA interface of the specific complex between the homing endonuclease PI-SceI and its DNA substrate. One base-specific and several backbone-specific cross-links were analyzed in detail: the cross-linking positions were identified by Edman degradation of isolated cross-linked peptidexoligodeoxynucleotide adducts and confirmed by site-directed mutagenesis. Based on these results and the crystal structure of PI-SceI, a model for the structure of the PI-SceIxDNA complex is proposed.

The monomeric homing endonuclease PI-SceI has two catalytic centres for cleavage of the two strands of its DNA substrate.

The monomeric homing endonuclease PI-SceI cleaves the two strands of its DNA substrate in a concerted manner, which raises the question of whether this enzyme harbours one or two catalytic centres. If PI-SceI has only one catalytic centre, one would expect that cross-linking enzyme and substrate should prevent reorientation of the enzyme required to perform the second cut after having made the first cut: PI-SceI, however, when cross-linked to its substrate, is able to cleave both DNA strands. If PI-SceI has two catalytic centres, one would expect that it should be possible to inactivate one catalytic centre by mutation and obtain a variant with preference for a substrate nicked in one strand; such variants have been found. The structural homology between the catalytic domain of PI-SceI having a pseudo 2-fold symmetry, and I-CreI, a homodimeric homing endonuclease, suggests that in PI-SceI active site I, which attacks the top strand, comprises Asp218, Asp229 and Lys403, while Asp326, Thr341 and Lys301 make up active site II, which cleaves the bottom strand. Cleavage experiments with modified oligodeoxynucleotides and metal ion mapping experiments demonstrate that PI-SceI interacts differently with the two strands at the cleavage position, supporting a model of two catalytic centres.

Photocross-linking of the homing endonuclease PI-SceI to its recognition sequence.

PI-SceI is an intein-encoded protein that belongs to the LAGLIDADG family of homing endonucleases. According to the crystal structure and mutational studies, this endonuclease consists of two domains, one responsible for protein splicing, the other for DNA cleavage, and both presumably for DNA binding. To define the DNA binding site of PI-SceI, photocross-linking was used to identify amino acid residues in contact with DNA. Sixty-three double-stranded oligodeoxynucleotides comprising the minimal recognition sequence and containing single 5-iodopyrimidine substitutions in almost all positions of the recognition sequence were synthesized and irradiated in the presence of PI-SceI with a helium/cadmium laser (325 nm). The best cross-linking yield (approximately 30%) was obtained with an oligodeoxynucleotide with a 5-iododeoxyuridine at position +9 in the bottom strand. The subsequent analysis showed that cross-linking had occurred with amino acid His-333, 6 amino acids after the second LAGLIDADG motif. With the H333A variant of PI-SceI or in the presence of excess unmodified oligodeoxynucleotide, no cross-linking was observed, indicating the specificity of the cross-linking reaction. Chemical modification of His residues in PI-SceI by diethylpyrocarbonate leads to a substantial reduction in the binding and cleavage activity of PI-SceI. This inactivation can be suppressed by substrate binding. This result further supports the finding that at least one His residue is in close contact to the DNA. Based on these and published results, conclusions are drawn regarding the DNA binding site of PI-SceI.

The protein splicing domain of the homing endonuclease PI-sceI is responsible for specific DNA binding.

The homing endonuclease PI- Sce I consists of a protein splicing domain (I) and an endonucleolytic domain (II). To characterize the two domains with respect to their contribution to DNA recognition we cloned, purified and characterized the isolated domains. Both domains have no detectable endonucleolytic activity. Domain I binds specifically to the PI- Sce I recognition sequence, whereas domain II displays only weak non-specific DNA binding. In the specific complex with domain I the DNA is bent to a similar extent as observed with the initial complex formed between PI- Sce I and DNA. Our results indicate that protein splicing domain I is also involved in recognition of the DNA substrate.

Structural and functional analysis of the homing endonuclease PI-sceI by limited proteolytic cleavage and molecular cloning of partial digestion products.

PI-SceI is a member of an unusual class of rare cutting homing endonucleases produced by an autocatalytic protein splicing from a precursor. To analyze the structural and functional domain organization of the endonuclease PI-SceI and to examine whether the DNA binding activity can be structurally separated from the catalytic activity, we performed limited proteolytic digestion experiments with various proteases. Two protease-resistant fragments spanning the N- and C-terminal halves of the nuclease were identified using different proteases which cleave the protein in the same region. Each fragment contains one of the two conserved LAGLIDADG motifs. The products of the limited proteolytic digests were shown to remain associated and to exhibit specific DNA binding but to be inactive in DNA cleavage. Different from what is observed with native PI-SceI, only one complex is formed as shown in an electrophoretic mobility shift assay. Expression clones for the N- and C-terminal protein fragments obtained by tryptic digestion were constructed, and the proteins PI-SceI-N and PI-SceI-C were purified. Only PI-SceI-N exhibits DNA binding activity. Bending experiments with PI-SceI-N, a mixture of PI-SceI-N and PI-SceI-C, as well as the products of the limited tryptic digest show that a DNA substrate with the full length recognition sequence is bent by 45;. This degree of bending is also observed with a DNA containing only the right side of the recognition sequence, corresponding to one of the DNA cleavage products of PI-SceI. Our results demonstrate that the N-terminal half of PI-SceI which lacks one of the two LAGLDADG motifs is able to bind to DNA specifically and to induce one of the distortions observed to occur in the process of DNA binding by PI-SceI. These results are discussed in light of the recently solved crystal structure of PI-SceI and used to refine a model for the mechanism of DNA binding and cleavage by PI-SceI.

The mechanism of DNA cleavage by the type II restriction enzyme EcoRV: Asp36 is not directly involved in DNA cleavage but serves to couple indirect readout to catalysis.

Three different mechanisms have been proposed to describe DNA cleavage by the type II restriction endonuclease EcoRV, which differ in the number and function of metal ions directly involved in catalysis and the different roles assigned to amino acid residues in the active sites and a phosphate group of the substrate. There are only four acidic amino acid residues close to the scissile bond: the essential Asp74 and Asp90, the non-essential Glu45, and Asp36. We show here that Asp36 can be exchanged for alanine, with only minor effects on the cleavage rate of the nearby phosphodiester bond, excluding that Asp36 could be directly involved in catalysis. Hence, the two versions of the two-metal-ion mechanism are not compatible with the experimental data, because too few ligands for two metal ions are present near the active site of EcoRV. Our result, thus, supports the one-metal-ion mechanism for EcoRV. We suggest that Asp36 has an allosteric effect by which specific contacts between one strand of the DNA and one subunit of the enzyme trigger the activation of one catalytic center. Given the similar structures of the active sites of EcoRV, EcoRI, BamHI, PvuII and FokI, as well as the occurrence of a characteristic catalytic motif in several other restriction enzymes, we conclude that these enzymes most likely share a similar mechanism of DNA cleavage, whose characteristic feature is the involvement of only one Mg2+ ion in catalysis.

Intra- vs intersubunit communication in the homodimeric restriction enzyme EcoRV: Thr 37 and Lys 38 involved in indirect readout are only important for the catalytic activity of their own subunit.

EcoRV is a dimer of two identical subunits which together form one binding site for the double-stranded DNA substrate. Concerted cleavage of both strands of the duplex requires intersubunit communication to synchronize the two catalytic centers of EcoRV. Here we address the question of how contacts to the DNA backbone trigger conformational changes which lead to the activation of both catalytic centers. The structure of the specific EcoRV-DNA complex shows that a region including amino acids Thr 37 and Lys 38 is involved in interactions with the DNA backbone and is a candidate for intersubunit communication. Homodimeric EcoRV T37A and K38A variants have a 1000-fold reduced catalytic activity. To examine whether Thr 37 and Lys 38 of one subunit affect the catalytic center in the same subunit and/or in the other subunit, we have produced heterodimeric variants containing a Thr 37 --> Ala or Lys 38 --> Ala substitution in one subunit combined with a wild type (wt) subunit (wt/T37A and wt/K38A) or with a subunit which contains an amino acid substitution (Asp 90 --> Ala) in the active site (D90A/T37A and D90A/K38A). Cleavage experiments with supercoiled pAT153 show that wt/T37A and wt/K38A preferentially nick the DNA. A steady-state kinetic analysis of the cleavage of an oligodeoxynucleotide substrate shows that the activity of wt/T37A and wt/K38A is half of that of wild type EcoRV, whereas D90A/T37A and D90A/K38A are almost inactive. These results demonstrate that Thr 37 and Lys 38 affect primarily the catalytic center in their own subunit and that both subunits of EcoRV can be activated independently of each other. We suggest that Thr 37 and Lys 38 control the catalytic activity of the active site in their own subunit by positioning alpha-helix B.

The Flavobacterium okeanokoites adenine-N6-specific DNA-methyltransferase M.FokI is a tandem enzyme of two independent domains with very different kinetic properties.

The Flavobacterium okeanokoites adenine-N6-specific DNA-methyltransferase, M.FokI, modifies both adenine residues within its asymmetric recognition sequence 5'-GGATG/CATCC-3'. It is a fusion protein comprising two independent enzymes. We have cloned, overexpressed and purified full-length M.FokI as well as both individual domains and analyzed their kinetics of DNA methylation using unmethylated and hemimethylated oligodeoxynucleotide substrates. Our data show that both domains of M.FokI methylate DNA independently of each other but cooperate in DNA binding. In agreement to former studies, the N-terminal domain of M.FokI modifies the upper strand of the recognition sequence (GGATG). It strongly prefers hemimethylated (5'-GGATG/CmATCC-3'; mA = N6-methyladenosine) over unmethylated substrates. In contrast, the C-terminal domain prefers unmethylated DNA substrates. Surprisingly, in addition to methylating the lower strand of the recognition sequence (CATCC), M.FokI-(335-647) also modifies the upper strand (GGATG), albeit with a lower activity. In addition, methylation was detected at CACCC sites, but not at sites in which a central AT dinucleotide is flanked by at least one A x T or T x A base pair. These results suggests that M.FokI-(335-647) either has a very degenerate specificity for (G/C)AT(G/C) and sites similar to CATCC or GGATG or, alternatively, that it has a dual specificity for CATCC and GGATG.

EcoRV-T94V: a mutant restriction endonuclease with an altered substrate specificity towards modified oligodeoxynucleotides.

Synthetic oligodeoxynucleotides with single methyl phosphonate (mp) substitutions were used for an analysis of the contribution of phosphate contacts to the recognition of the cleavage site by the restriction endonuclease EcoRV. Only in the last position within the recognition sequence, is the methyl phosphonate substitution tolerated by the enzyme. The wild-type enzyme cleaves the Sp diastereomer of the oligodeoxynucleotide GACGATATmpCGTC and the unmodified sequence with equal rates, whereas the Rp diastereomer is cleaved much more slowly. Inspection of the crystal structure of an EcoRV-DNA complex revealed that the non-bridging oxygen atoms of the phosphodiester bond between the T and C bases are in hydrogen bonding distance of the hydroxyl group of the amino acid Thr94. We therefore tried to engineer a variant of EcoRV that would prefer a methyl phosphonate linkage over a normal phosphodiester bond and produced mutants with amino acid exchanges at position 94. One of them, Thr94Val, shows a dramatically reduced activity towards the unmodified DNA and does not accept the Rp diastereomer, but cleaves the Sp diastereomer with the same rate as wild-type EcoRV. Its selectivity, i.e. the ratio of cleavage rates determined for the unmodified and modified substrates, differs by three orders of magnitude from that of the wild-type enzyme.

Binding, bending and cleavage of DNA substrates by the homing endonuclease Pl-SceI.

To characterize the interaction between the homing endonuclease PI-SceI and DNA, we prepared different DNA substrates containing the natural recognition sequence or parts thereof. Depending on the nature of the substrates, efficient cleavage is observed with a DNA containing approximatel 30 bp of the natural recognition sequence using supercoiled plasmids, approximately 40-50 bp using linearized plasmids and > 50 bp using synthetic double-stranded oligodeoxynucleotides. Cleavage of supercoiled plasmids occurs without accumulation of the nicked intermediate. In the presence of Mn2+, DNA cleavage by PI-SceI is more efficient than with Mg2+ and already occurs with substrates containing a shorter part of the recognition sequence. The requirements for strong binding are less stringent: a 35 bp oligodeoxynucleotide which is not cleaved is bound as firmly as other longer oligodeoxynucleotides. PI-SceI binds with high affinity to one of its cleavage products, a finding which may explain why PI-SceI hardly shows enzymatic turnover in vitro. Upon binding, two complexes are formed, which differ in the degree of bending (45 degrees versus 75 degrees). According to a phasing analysis bending is directed into the major groove. Strong binding, not, however, cleavage is also observed with the genetically engineered enzymatically inactive variant comprising amino acids 1-277. Models for binding and cleavage of DNA by PI-SceI are discussed based on these results.

The production and characterization of artificial heterodimers of the restriction endonuclease EcoRV.

A novel approach to studying the inter- and intrasubunit communication required for the activity of homodimeric proteins is described. It was developed for the restriction endonuclease EcoRV, but should also be useful for other homodimeric enzymes. Two ecorV genes encoding different EcoRV mutants are coexpressed in the same Escherichia coli cell leading to homo- and heterodimeric variants of the enzyme. The two ecorV genes carry either a 5' extension coding for the glutathione-S-transferase or a His6-tag. The EcoRV heterodimer produced in vivo is separated from the two EcoRV homodimers and purified to homogeneity by affinity chromatography. Purified EcoRV heterodimers are stable and are not subject to reassortment of the subunits. To investigate the interdependence of the two catalytic centers, EcoRV heterodimers consisting of one subunit with wild type sequence and one subunit with amino acid substitutions in the PD...(D/E)XK motif, characteristic for the active sites of many restriction endonucleases, were produced. While the homodimeric EcoRV active site mutants are catalytically inactive, the heterodimeric EcoRV variants with one active and one inactive catalytic center display a twofold reduced activity toward oligodeoxynucleotide substrates compared to the wild type, and preferentially nick supercoiled plasmid DNA. From these results we conclude that in the wild type enzyme both catalytic centers function independently of each other.

Analysis of the mechanism of the Serratia nuclease using site-directed mutagenesis.

Based on crystal structure analysis of the Serratia nuclease and a sequence alignment of six related nucleases, conserved amino acid residues that are located in proximity to the previously identified catalytic site residue His89 were selected for a mutagenesis study. Five out of 12 amino acid residues analyzed turned out to be of particular importance for the catalytic activity of the enzyme: Arg57, Arg87, His89, Asn119 and Glu127. Their replacement by alanine, for example, resulted in mutant proteins of very low activity, < 1% of the activity of the wild-type enzyme. Steady-state kinetic analysis of the mutant proteins demonstrates that some of these mutants are predominantly affected in their kcat, others in their Km. These results and the determination of the pH and metal ion dependence of selected mutant proteins were used for a tentative assignment for the function of these amino acid residues in the mechanism of phosphodiester bond cleavage by the Serratia nuclease.

Engineering novel restriction endonucleases: principles and applications.

Restriction endonucleases cleave DNA with remarkable sequence specificity. In this review, we summarize the status of, and prospects for, engineering restriction endonucleases with new specificities. Such variants could be of considerable commercial value because restriction enzymes are among the most frequently used enzymes in molecular biology, and not all the desirable specificities are available. While it has not yet been possible to effect specificity changes, mutant have been described that (1) exhibit relaxed specificity, (2) favour modified substrates over their natural substrates, (3) discriminate between cleavage sites located in different sequences, (4) prefer metal ions other than Mg2+ as cofactors for cleavage, or (5) possess site-specific DNA-nicking activity.

Introduction of asymmetry in the naturally symmetric restriction endonuclease EcoRV to investigate intersubunit communication in the homodimeric protein.

Type II restriction endonucleases are dimers of two identical subunits that together form one binding site for the double-stranded DNA substrate. Cleavage within the palindromic recognition site occurs in the two strands of the duplex in a concerted manner, due to the action of two catalytic centers, one per subunit. To investigate how the two identical subunits of the restriction endonuclease EcoRV cooperate in binding and cleaving their substrate, heterodimeric versions of EcoRV with different amino acid substitutions in the two subunits were constructed. For this purpose, the ecorV gene was fused to the coding region for the glutathione-binding domain of the glutathione S-transferase and a His6-tag, respectively. Upon cotransformation of Escherichia coli cells with both gene fusions stable homo- and heterodimers of the EcoRV variants are produced, which can be separated and purified to homogeneity by affinity chromatography over Ni-nitrilotriacetic acid and glutathione columns. A steady-state kinetic analysis shows that the activity of a heterodimeric variant with one inactive catalytic center is decreased by 2-fold, demonstrating that the two catalytic centers operate independently from each other. In contrast, heterodimeric variants with a defect in one DNA-binding site have a 30- to 50-fold lower activity, indicating that the two subunits of EcoRV cooperate in the recognition of the palindromic DNA sequence. By combining a subunit with an inactive catalytic center with a subunit with a defect in the DNA-binding site, EcoRV heterodimers were produced that only nick DNA specifically within the EcoRV recognition sequence.