Future strategies for the systematic identification of GFP-tagged synaptic proteins on a genome-wide scale

As discussed in chapter 5.7 a substantial proportion of genes with no or only a few small introns is unlikely to be identified in an exon-trap screen. One tempting way of identifing these genes is to target exons instead of introns. While the concept of an exon-fusion screen in contrast to an exon-trap screen was never tested in Drosophila, a similar approach has been successfully used in vitro to GFP-tag different molecules in a random fashion (Sheridan et al., 2002).

Fig. 40 Schematic drawing of an exon-trap screen.

The transgenic vector gets integrated into an intron subsequent splicing leads to the production of a fusion protein

As already briefly discussed in chapter 2.4 residual vector (Fig. 40, blue box, central panel) and marker sequences (Fig. 40, pink box, central panel) need to be removed in an exon-trap as well as in an exon-fusion screen. The spliced mRNA (Fig. 40, lower panel) should ideally encode only protein (Fig. 40, black box) and GFP (Fig. 40, green box) sequences. This is accomplished by splicing out all unnecessary sequences 5’

and 3’ of the GFP cassette. In an exon-trap screen this is achieved as follows. The 5’ end splicing occurs between the splice acceptor site (SA) 5’ of GFP and the splice donor (SD) site of the upstream exon (of the protein in which the vector has integrated). At the same time the SD site 3’ of GFP splices with the SA site of the downstream exon (Fig. 40, central panel). Thereby it is possible to remove all vector and marker sequences. This is not possible in an exon-fusion screen. Here the ends of the transposable element will be part of the final protein. In a in vitro system, for example, a Tn5 transposon was used to insert GFP in a glutamate receptor (see chapter 4.1.6). The ends of the transposon form short (7 AA) linkers between the GFP and the protein of interest (Sheridan et al., 2002).

While in this case the marker (kanamycin resistance) was removed by subsequent digestion of the DNA, this is not possible in an in vivo exon-fusion screen. In vivo it is possible to screen without any marker, since positive events can be scored based on GFP expression. Alternatively the marker can be removed in vivo by flanking it with SD and SA sites (Fig. 41, upper panel). The GFP fusion protein is then produced as follows: All of the vector sequence 5’ of the GFP (terminal ends of transposable element) serves a linker (Fig. 41, lower panel, blue box 5’ of GFP) between the tagged protein (Fig. 41, lower panel, black boxes) and the GFP (Fig. 41, lower panel, green box). The marker (Fig. 41, central panel, pink box) itself gets spliced out by the flanking SD and SA sites (Fig. 41, red boxes). Only some basepairs 3’ of the SA site (Fig. 41, lower panel, blue box) remain in the final protein as linker between the GFP and the protein.

Fig. 41 Schematic drawing of an exon-fusion screen. The transgenic vector is integrated into an exon and subsequent splicing leads to the production of a fusion protein

For such a strategy it would be the most advantageous to use a vector which is well characterized and has small terminal repeats, which do not harbor stop codons. The minos (Franz and Savakis, 1991) transposable element, as well as piggyBac and P-elements might be suitable for that purpose. P-P-elements and minos might be particularly useful, since they are known to have very small minimal sequences required for transposition (Spradling and Rubin, 1982; Mullins et al., 1989) and (Graeme Davis, personal communication). Changing the strategy in that way might substantially facilitate reaching a high degree of saturation. Additionally to these forward genetic approaches also reverse genetics of a preselected class of genes might help to identify and GFP-tag synaptic proteins. Biochemical PSD characterization and cDNA isolation (Langnaese et al., 1996; Yoshimura et al., 2004) showed that about 500 proteins are enriched in synaptic

spines. After identifying the corresponding Drosophila homologs 100-200 candidates could be GFP-tagged. Yeast-two-hybrid screens against proteins known to have synaptic localization might provide further candidates. The corresponding constructs can be produced via long-range PCR on cDNA clones. The PCR products can be directly inserted in a precut pUAST transgenic vector, already containing an N- or C-terminal GFP. The localization of these proteins could then be scored in vivo. Transgenic injections are very efficient in Drosophila. About 1000-1500 embryos can be injected per day. On average one transgenic animal is isolated per 30 injected embryos. Injecting 100 embryos for each candidate should allow the injection of 10 candidates per day. The combination of exon-trap screening, exon-fusion screening and direct GFP-tagging of candidates might lead to the identification of the most proteins present at the Drosophila neuromuscular junction within the next few years.

To facilitate the genome-wide exon-trap screen it was decided to perform this screen in a big consortium of laboratories. In the larval screen only 13 out of 322 GFP positive lines were of immediate interest for the two participating laboratories of Dr.

Stephan Sigrist and Dr. Anne Ephrussi. Within the “Deutsche Forschungsgemeinschaft Schwerpunkt Polarity” a consortium of laboratories joined together, which are interested in different tissues and developmental processes in the embryo, the larva or the adult. These laboratories include the laboratories of Dr. Suzanne Eaton, Dr. Anne Ephrussi, Dr.

Christian Dahmann, Dr. Marcos Gonzales-Gaitan, Dr. Christian Klämbt, Dr. Eli Knust, Dr.

Arno Müller, Dr. Andreas Wodarz and Dr. Stephan Sigrist. The aim of this cooperation is to produce 10000 GFP-positive lines within the next two years.

6. Appendix