3 Materials and Methods
4.4 Summary of Results
5.1.2 Ectopic Gene Targeting
Quantitative PCR results demonstrated that hPARP-1 was present in one copy in heterozygous hPARP-1 ES cells and mice, and in two copies in homozygous hPARP-1 mice (Figure 4.8).
Importantly, despite the apparent site-specific homologous recombination of the targeting vector in all three hPARP-1 ES cell clones, subsequent mouse breeding and qPCR analyses revealed that the intended replacement of the mParp-1 with the hPARP-1 did not occur, since the mPARP-1 was still present in two gene copies in ES cells of all three clones as well as in heterozygous and homozygous hPARP-1 mice (Figure 4.8 and Table 4.2). [N.B. it is therefore more accurate to designate mice with one targeted chromosome “hemizygous” rather than
“heterozygous” (Schenkel 2006)]. These findings fit in with the lower signal intensities of the hPARP-1 fragments in Southern blot analyses as mentioned above. Consistent with the situation of two gene copies of mParp-1 plus two copies of hPARP-1 in homozygous mutant mice, Southern blot analysis using DNA of hPARP-1 homozygotes displayed equal signal intensities for the mParp-1 and hPARP-1 fragments (Figure 4.11 B).
Fluorescence in situ hybridization (FISH) analyses confirmed the presence of a remaining copy of the mParp-1 on the wild-type locus on chromosome 1 (Figure 4.6). Furthermore and consistent with the preceding data, FISH analyses revealed the integration of an ectopic hPARP-1 copy in close proximity to endogenous mParp-1 locus on chromosome 1 without showing a clear tendency for a distal or proximal location. Since the technical resolution of FISH analysis is limited to 2-3 Mb (Trask 1991), this indicates that the hPARP-1-targeted site lies within kb up to several Mb from the endogenous mParp-1 locus.
The initial hypothesis that mParp-1 pseudogenes might have been targeted can be ruled out:
While hPARP-1 pseudogenes are present in the human genome on chromosomes 13 and 14 (Cherney et al. 1987; Baumgartner et al. 1992), BLAST searches did not yield any evidence for the presence of an mParp-1 pseudogene or extended homologies in the mouse genome. A further potential explanation was based on the fact that large DNA fragments are prone to breakage by hydrodynamic shearing (Yang and Seed 2003). Yet, homologous recombination of multiple or truncated fragments of the targeting vector used in this study can be excluded, because only full-length PARP-1 proteins, of human as well as murine origin, were detected in mutant ES clones.
The most likely explanation for the above results, however is similar to „ectopic gene targeting‟ a phenomenon described with „ends-in‟ insertion-type vectors in non-ES-cell mammalian gene targeting. (Adair et al. 1989; Adair et al. 1990; Itzhaki and Porter 1991; Dorin et al. 1992; Scheerer and Adair 1994; Dellaire et al. 1997; Adair et al. 1998; McCulloch et al.
2003).
Based on the data of this thesis and in line with reports of ectopic gene targeting, the following model is an explanation the outcome of the gene targeting approach (Figure 5.1):
Upon electroporation, the homology arms of the targeting vector were rendered single-stranded by a 5‟-to-3‟ nuclease activity and the resulting single-single-stranded 3‟ ends invaded the endogenous wild-type locus (Sun et al. 1991; Barzel and Kupiec 2008). DNA polymerases then extended the arms bidirectionally using endogenous sequences as templates, as shown by DNA sequencing of flanking PCR amplicons and Southern blot analyses. Subsequently, the whole vector resolved and integrated at an adjacent position.
Figure 5.1: Proposed model for ectopic gene targeting in hPARP-1 ES cells.
Upon the transfection of murine ES cells with the hPARP-1 targeting vector, the vector homology arms were rendered single-stranded by a 5‟-to-3‟ nuclease activity and the resulting single-stranded 3‟ ends invaded the endogenous mParp-1 wild-type locus. DNA polymerases then extended the arms bidirectionally using endogenous sequences as templates. Subsequently, the whole vector resolved and integrated at an adjacent position on chromosome 1.
The phenomenon of ectopic gene targeting was first described in Chinese hamster ovary (CHO) cells in targeting of the adenine phoshoribosyl transferase (APRT) locus and after retroviral transduction of rat cells (Adair et al. 1989; Ellis and Bernstein 1989). Several reports followed, demonstrating ectopic gene targeting in CHO, murine fibroblasts, and hybridoma cells, respectively (Adair et al. 1990; Itzhaki and Porter 1991; Aratani et al. 1992; Scheerer and Adair 1994; Adair et al. 1998). All these experiments used „ends-in‟ insertion-type vectors and an elongation of the targeting vector occurred in a unidirectional manner from a single resected 3‟ end of the vector (Belmaaza and Chartrand 1994). The first evidence of bidirectional elongation of an „ends-in‟ integration vector in mammalian cells was found by McCulloch et al.
in a hybridoma cell line. They also showed that extension of the vector ends can exceed 16 kb in length (McCulloch et al. 2003).
To analyze the extent of vector homology arm elongations and to examine the impact on genes flanking the endogenous mPARP-1 locus, gene copy numbers of mPARP-1-flanking genes, i.e. Gm821 (proximal) and Lin9 (distal), were determined by qPCR (Figure 4.9).
According to these results, Gm821 was duplicated in lines #113 and #225, and Lin9 in line
#225. These findings demonstrate that elongation of vector homology arms was quite extensive, since both sequences tested are located more than 40 kb apart from the mParp-1 locus.
Strikingly, the mechanism of elongation of the homology arms is fully consistent with the synthesis-dependent strand annealing (SDSA) pathway, which is seen as the predominant mechanism for cellular double strand break (DSB) repair during HR (section 1.2.1) (Helleday et al. 2007). The above data therefore demonstrate that the DSB repair machinery in the form of the SDSA pathway is involved in the processes that occur during mammalian gene targeting.
Since ectopic homologous recombination was observed in all three independently generated ES cell clones, the nature of the targeted locus seems to favor SDSA over the resolution of the Holliday junction by two recombination events as is the case in classical gene targeting (Li and Baker 2000; Li et al. 2001).
Interestingly, the PARP-1 protein itself could play a role in such a scenario, since PARP-1 might exert an anti-recombinogenic activity (section 1.1.2.5.2.3). In particular, it was shown that gene targeting frequencies in Parp-1 knock-out ES cells is three times higher than in wild-type ES cells (Dominguez-Bendala et al. 2006). Furthermore, the present study revealed an apparent feedback regulation of PARP-1 expression levels (Figure 4.17), which is consistent with previous reports showing that PARP-1 can bind to its promoter sequences and act as a negative modulator of the activity of its own promoter (Oei et al. 1994; Soldatenkov et al. 2002;
Lonskaya et al. 2005; Potaman et al. 2005). Therefore, it is tempting to speculate that high expression of PARP-1 in ES cells (Figure 4.13) caused PARP-1 to accumulate at the site of recombination, i.e. the mParp-1 promoter, during gene targeting, which might have counteracted gene replacement and directed the targeting event to a nearby position on chromosome 1.
The question of random or non-random ectopic integration of the elongated targeting vector has by now not been fully clarified. However, Dellaire and colleagues showed that unidirectional ectopic gene targeting in murine fibroblasts followed a bimodal distribution: the targeting vector either integrated non-randomly within 3 Mb of the target gene or in a random fashion on another chromosome (Dellaire et al. 1997). Although it was shown that targeting efficacy increases with the length of the homology arms (Moens et al. 1992), as little as 14 to 25 bp of homology is sufficient to trigger integration of the targeting vector by homologous recombination (Rubnitz and Subramani 1984; Ayares et al. 1986). Therefore, it is possible that
in all three clones the elongated targeting vector integrated at the same site of a potential microhomology in a non-random manner. Interestingly, 2.8 Mb distal of the mParp-1 locus a microhomology region of the mParp-1 promoter of 62 bp in length was identified, which could be the ectopic integration site.
Ectopic gene targeting events are believed to occur as frequently as classical gene targeting (Dorin et al. 1992; Dellaire et al. 1997). Therefore, it is rather surprising no such event has ever been described, when an „ends-out‟ gene replacement vector was used (McCulloch et al. 2003).
This thesis appears to be the first report of an ectopic gene targeting event with bidirectional copying of endogenous sequences exceeding 40 kb onto the ends of an gene replacement vector in murine ES cells. This finding is of particular interest, since it represents an important caveat for the gene targeting technology in murine ES cells in general, and during the generation of knock-in mice in particular. In the case of knock-in mice positive gene targeting cannot simply be validated by screening for loss of protein expression as in knock-out approaches. As evaluated in this thesis, qPCR and FISH analyses represent two reliable techniques to detect ectopic gene targeting in ES cell clones that were identified to be positive for site-specific homologous recombination by PCR or Southern blot. A future routine employment of either qPCR or FISH analysis in knock-in approaches before blastocyst injection and subsequent generation of chimeric mice is advisable in order to validate the replacement of the gene of interest.
In summary, a novel hPARP-1 mouse model was generated via ectopic gene targeting in ES cells. In addition to the endogenous wild-type mParp-1 locus, homozygous mutant mice carry two copies of the full-length hPARP-1 coding sequence, which is under the transcriptional control of the murine regulatory sequences (i.e. promoter, terminator).