In mature chloroplasts, the DNA (cpDNA) is surrounded by a potentially genotoxic environment that would make the mitochondrial DNA milieu look like a “nadree” (picnic). And yet, the slower evolution of cpDNA compared to other cellular genomes suggests that this organelle must have efficient mechanisms for repairing DNA. Unfortunately, those mechanisms have been barely noted, much less studied. This dissertation describes a novel approach that was developed to study how chloroplasts of Arabidopsis repair the most severe form of DNA damage, a double-strand break (described in Chapter 2). The success with this approach also prompted the development of a new method for site-specific modification of tobacco cpDNA that is described in Chapter 3. To study the consequences and repair of a break in the circular plastid genome, we developed an inducible system based on a psbA-intron endonuclease from Chlamydomonas (I-CreII) that specifically cleaves the psbA gene of Arabidopsis. The protein was targeted to the chloroplast using the rbcS1 transit peptide, and activation of the nuclear gene was made dependent on an exogenous inducer (β-estradiol). In Chlamydomonas, I-CreII cleavage at psbA was repaired, in the absence of the intron, by homologous recombination between repeated sequences (20-60 bp) that are abundant in that genome. By comparison, Arabidopsis cpDNA is very repeat-poor. Nonetheless, phenotypically strong and weak transgenic lines were obtained, and shown to correlate with I-CreII expression levels. Southern blot hybridizations indicated a substantial loss of psbA, but not cpDNA as a whole, in the strongly-expressing line. PCR analysis identified deletions nested around the I-CreII cleavage site that were indicative of repair using microhomology (6-12 bp perfect repeats, or 10-16 bp with mismatches) or no homology. The results provide evidence of alternative repair pathways in the Arabidopsis chloroplast that resemble the nuclear microhomology-mediated and nonhomologous end-joining pathways, in terms of the homology requirement. Moreover, when taken together with the results from Chlamydomonas, plus other considerations, the data suggests that an evolutionary relationship may exist between the repeat structure of cpDNA and the organelle’s ability to repair broken chromosomes. Taking advantage of the inducible I-CreII system, I developed a method to delete defined regions of cpDNA in tobacco, which was named DREEM (for direct repeat and endonuclease mediated). Chloroplast transformation was used to introduce an I-CreII cleavage site adjacent to an aadA:gfp marker and flanked by a direct repeat of 84 bp. When chloroplast-targeted I-CreII was induced with β-estradiol during germination, complete loss of the aadA:gfp marker occurred by SSA-type repair involving the 84-bp direct repeat. I obtained additional evidence for DREEM effectiveness by deleting 3.5 kb of native cpDNA that included part of the large ycf1 gene. DREEM can be used for other modifications besides gene deletions, partly because it is seamless and leaves no trace of introduced DNA. Since expression of the endonuclease is controlled by steroid application (and concentration), and the deleted cpDNA is probably destroyed during the SSA process, this inducible gene-ablation technique could enable the study of essential chloroplast genes in vivo. / text

Homing endonucleases (HEs) typically have one of four types of catalytic domains (LAGLIDADG, GIY-YIG, H-N-H, His-Cys), and a DNA-binding region(s) that provides specificity. I-CreII, which is encoded by the psbA4 intron from Chlamydomonas reinhardtii, is unusual in containing two catalytic motifs: H-N-H and GIY-YIG. A previous study showed that I-CreII cleavage leaves 2-nt 3' OH overhangs similar to GIYYIG endonucleases, but that it also has a flexible metal requirement like H-N-H enzymes. Also, alanine substitution of several conserved residues in the GIY-YIG motif and two in the H-N-H motif did not produce a clear catalytic mutant, although some variants had strongly reduced DNA binding. Thus, in order to identify the catalytic motif, I substituted additional amino acids in both domains with alanine, and identified three histidines in the H-N-H motif that are likely to be involved in catalysis. To gain insight into how I-CreII interacts with its ~30-bp homing-site DNA, three types of DNA protection analysis were performed. Hydroxyl-radical footprinting, which reveals regions of tight DNA binding, indicated that I-CreII binds strongly to a region downstream of the cleavage and intron-insertion sites, corresponding to bp 2-10 of exon 5. There was also partial protection around the cleavage site, but only on the top strand, which is consistent with the enzyme's tendency to cleave this strand first. DNase I protection, which can reveal less closely-bound regions of target DNA, gave a larger footprint than hydroxyl-radical protection, with the additional region lying upstream of the cleavage site. These results also suggest that DNA backbone-binding downstream of the cleavage site involves sugars and phosphates, whereas upstream it is mainly with phosphates. DMS protection, which probes guanines on the N-7 position in the major groove, did not provide any evidence of major groove binding (at least not through guanines). DNase I protection could also be performed on the I-CreII variants that had reduced DNA affinity. The N161A variant was instructive in that it showed reduced protection of a T-A bp very close to the cleavage site, providing support for a catalytic role for the H-N-H motif and a possible constraint for modeling. Of the GIY-YIG motif variants, the footprint of the G231E/K245A variant was distinctly useful in that it was preferentially effected downstream of the cleavage site. This result suggested the H-N-H and GIY-YIG motifs are co-linear with their targets in the homing site. Structural modeling of the GIY-YIG domain of I-CreII using the solved I-TevI domain as template provided evidence for a unique insertion in the I-CreII structure that disrupted a catalytic α-helix; the insertion is predicted to be a positively charged, hairpinlike loop anchored by two antiparallel β-strands. I propose that this insertion can explain the evolutionary conversion of this catalytic endonuclease domain into a DNA-binding domain. These findings should also help to understand other dual-motif H-N-H/GIY-YIG endonucleases, such as I-CmoeI.