Optimizing a vector for a genome-wide screen for proteins localized at synapses

After the successful larval screen (Heidelberg screen) it was decided to perform a genome-wide screen in a big consortium of laboratories. The sorting of the embryos for the genome-wide screen is performed in collaboration with Dr. Christian Klämbt at the University of Münster. Therefore, it will be referred to as the Münster screen. The aim of this co-operation is to produce 10000 GFP positive lines within the next two years.

Extrapolating the discovery rate of NMJ localizing lines predicts that more than 100 lines with expression at the NMJ should be obtained.

Part of this thesis was the construction of a new set of vectors to be used for the screen of this consortium was. It is not feasible to balance and to keep 10000 lines.

Therefore, it is planned to keep heterozygous stocks over several generations, just by flipping the vials. Heterozygous white expressing flies have an advantage in propagation compared to flies, which lost the insert (and the white marker, and are therefore blind again). All laboratories will evaluate the stocks within 4 weeks of their production, interesting ones will be sequenced. After removing clones of identical insertions all interesting, unique stocks will be balanced and kept. This workflow requires a selection marker, which gives the flies carying the insert an advantage compared to flies, which have lost the insert. Therefore, white was used as marker and not 3xP3 as used for the Heidelberg screen. Moreover, the piggyBac backbone itself was modified. In the Heidelberg Screen (vector p1 and p2) the “marker (3xP3-dsred) GFP-cassette” replaced a 0,8 kb deletion in the piggyBac transposase open reading frame (maps of all constructs are shown in Fig. 10). By serial deletion Fraser and others determined the minimal piggyBac sequences required for transposition in vivo (Malcom Fraser, personal communication). Based on their results it was decided to place the GFP-marker cassette in piggyBac vectors that contain only these minimal piggyBac ends. This should increase the transposition efficacy, since it is known that the transposition efficacy is higher, when the insert is smaller (Malcom Fraser, personal communication).

One main advantage of piggyBac is that it excises only precisely, thus leaving no potential “second site hit” back (Cary et al., 1989; Fraser et al., 1995; Elick et al., 1996;

Fraser et al., 1996). Nonetheless, it would be very valuable to be able to do imprecise excisions (as usually done with P-elements) to create mutants in the genes of interest.

Combining the advantages of P-elements and piggyBac, P-ends were placed in the piggyBac vector (compare Fig. 10). Thus the screen is a piggyBac screen, which has many advantages over P-element based screens (compare chapter 2.5). Once a functional GFP-fusion is identified the stock can be crossed to P-transposase. Thereby mutants can be easily obtained. The P-ends were either placed flanking the entire GFP-marker cassette (p3: piggyM P white SA-GFP-SD P) or only flanking the selection GFP-marker (p4: piggyM P white P SA-GFP-SD P see Fig. 10). Placing the P-ends to only flank the selection marker allows sorting for imprecise excisions.

The fact that transgenics lines of these two constructs were easily obtained (Christian Klämbt, personal communication) proved that the piggyBac ends must be functional in vivo. Nonetheless, the remobilization of these constructs using transgenically expressed piggyBac transposase worked only very poorly (Christian Klämbt, personal communication). One possibility was that the P-ends and the piggyBac ends are too close together, causing problems with piggyBac remobilization. To rule out this problem the GFP-marker cassette used to construct p4 was put into the piggyBac backbone called piggyL (piggy large). Vectors based on this backbone (p1 and p2) were successfully used in the Heidelberg screen, and remobilized well. The new construct was named piggyL P white P SA-GFP-SD (p5) (Fig. 10). Transgenic animals were easily obtained. These constructs remobilized well in the eye. Nonetheless, germ line transposition was often ineffective. (Christian Klämbt, personal communication). Finally one strain could be isolated, which showed an even better mobilization rate than p2, the construct used for the Heidelberg screen (Christian Klämbt, personal communication). The Münster screen focused on mainly late embryo stages (12-26h, 25°C) (Christian Klämbt, personal communication), while the Heidelberg screen sorted mainly 22-36 h (25°C) old first instar larvae. In the Heidelberg screen both the handling of first instar larvae, as well as the number of positively sorted animals, were significantly better (five times more positive larvae) compared to sorting embryos. Partially this can also been attributed to the fact that

in Heidelberg a lot of time was spent on optimizing conditions for larvae, which were of prime interest. In Münster mainly embryos are sorted for the consortium. Therefore an embryo sorter is used in Münster (Copas Express, Union Biometrica, Somerville, MA, USA), while in Heidelberg the more flexible COPAS Select (Union Biometrica, Somerville, MA, USA) sorter was used. The sort rate of 1:2000 (Christian Klämbt, personal communication) for embryos (using p5 compared to a ratio of 1:5000 in p2 animals) is good. It remains to be clarified to what degree these numbers reflect true positive events.

In Heidelberg roughly 1:10000 embryos were positive upon manual resorting (using p2), while in first instar sorts 1:2000 sorted larvae were true positives. If most embryos sorted in Münster prove of to be positive upon resorting, this would indicate that the embryo sort protocol was substantially improved and could be a suitable alternative to sorting first instar larvae. If the number of true positives is less than 50% of the sorted embryos, sorting of first instar larvae should be considered as an alternative. Collectively this data shows that the established construct should allow meeting the goal of establishing 10000 GFP positive lines, which will contribute substantially to the characterization of the both pre- and postsynaptic compartment.

5. Discussion

5.1 Establishing a new assay to study molecular dynamics during