In this work we have investigated a discrepancy between the Drosophila
synapsin cDNA sequence published earlier [26
] and the BDGP sequence at the junction of exons 4 and 5 of the synapsin
gene. In all wild-type lines examined we verified the genomic sequence of the genome project (AE003686) including the normal GT-AG splice consensus. Thus it seems unlikely that this region is polymorphic. The genomic sequence encodes the canonical PKA recognition motif RRxS in the A-domain of the Drosophila
synapsin. This motif is also found in all other known synapsins. In vertebrates phosphorylation at this site apparently is involved in the redistribution of the protein during synaptic activity [25
]. However, in Drosophila
the genomic sequence reveals a single base pair difference to the published cDNA which encodes RGFS at the kinase recognition motif. Since the first canonical start codon (ATG) of the open reading frame of the Drosophila synapsin
gene is located downstream of the conserved A domain and the kinase recognition site, we had earlier identified the amino acid sequence of the N-terminus of the 70 kDa synapsin isoform by Edman degradation of the immuno-affinity purified protein [27
]. This independent data identified an unconventional leucine encoded by CTG as the first amino acid and clearly supported the cDNA encoded motif RGFS, strongly suggesting mRNA editing at this site.
To verify the cDNA sequence and to obtain a semi-quantitative measure of the efficiency of this editing we isolated mRNA of embryos/1st instar larvae, 3rd instar larvae, pupae, and heads and bodies of adults of different wild-type lines and the eye colour mutant w1118 which is frequently used to generate transgenic lines, and directly sequenced RT-PCR products in the region of interest. Surprisingly, in all samples except embryos/1st instar larvae we found no trace of the genomic sequence, indicating that more than 90% of the pre-mRNA was edited, as estimated from the signal to noise ratio of the sequencing trace. (The presence of more than 10% unedited mRNA would have been detected as a double peak at one position, cf. asterisk in Fig. ). Thus, the only major pool of primary transcripts that escapes editing is found very early in development. In two wild-type laboratory strains (WT Oregon-R and Canton-S) a certain fraction of the mRNA apparently was edited in addition at the first arginine codon of the kinase target motif. The resulting sequence GGFS presumably cannot be recognized by kinases.
All A to G discrepancies between genomic and cDNA in Drosophila
investigated so far are due to the activity of the ADAR enzyme which catalyzes an adenosine-to-inosine conversion [8
]. Expression of the Drosophila
ADAR appears to be prominent in the nervous system. Interestingly, pre-mRNAs of several other proteins of the synaptic release machinery were also identified as A-to-I editing targets, such as synaptotagmin, dunc-13, stoned-B, complexin, and lap [8
]. For hydrolytic deamination the enzyme needs a partial double-stranded RNA to form at the editing region. Normally this dsRNA is formed between the editing site and a complementary sequence in a neighbouring intron. Upon searching for a potential editing site complementary sequence (ECS) in the pre-mRNA of the synapsin
gene of Drosophila
we analysed 1 kb surrounding the kinase target site by a computer program which predicts secondary structures of an RNA molecule (MFOLD, [35
]). In this analysis we detected a potential ECS region lying only 90 bp downstream of the edited arginine codon (Fig. ). The ECS has a length of 15 bp similar to the size found e.g. in the ECS of the mammalian glutamate receptor GluR-B mRNA [3
]. Thus the pre-mRNA of Drosophila
synapsin can form a secondary structure containing a double helical stem that could make it a target for ADAR in the nervous system of Drosophila
Figure 6 Potential editing site complementary sequence (ECS) in intron 4 of the synapsin gene and predicted folding of the pre-mRNA. (A) Genomic sequence of synapsin, exon 4 is underlined. Bases belonging to the codons for the two arginines in the protein kinase (more ...)
The new Adar mutant allele isolated here suffered a deletion of the entire first exon and 210 bp upstream sequences which presumably contain essential regulatory sequences, but the coding region remains intact. The fact that flies homozygous for this allele show a very similar phenotype to null mutants and are unable to edit the synapsin pre-mRNA suggests that this new allele is a severe hypomorph or a null allele.
With the methods used here we were unable to detect unedited mRNA in 3rd
instar larvae, pupae, and adults. Such high RNA editing efficiency in adult Drosophila
has also been observed at four different editing sites of the L-type voltage gated calcium channel Ca-alpha 1D
] and in substrates of the ADAR2 enzyme of mammals [38
]. Possibly, unedited versions of these proteins are required earlier during development. This may also be true for Drosophila
synapsin as we find unedited mRNA in embryos/first instar larvae. Like differential splicing, RNA editing is extensively used in Drosophila
to generate protein diversity far beyond what is expected from the number of protein coding genes. Another evolutionary advantage of RNA editing may be the possibility to adjust the ratio of the abundance of two isoforms to any value between 0 and 1, rather than only to the 0, 1/2 or 1 possible by allelic encoding. So at first sight 100% editing would not seem to make much sense. However, if editing was reduced or absent in only a relatively small subset of neurons, we would not be able to detect this in our experiments. On the other hand, editing of the first arginine codon of the RRFS motif occurred only in two out of nine strains investigated and here only with about 50% efficiency. To show that the modification of the RNA sequence is restricted to the conserved N-terminal kinase target motif in the A-domain we also investigated the only other RRxS motif in Drosophila
synapsin. Here no discrepancy between genomic DNA and cDNA sequences was observed. This result also represents an additional control against possible artefacts. Editing at the N-terminus was found in all laboratory strains and also in newly collected flies from different parts of central Germany. We conclude that the RNA editing described here is not an adaptation of inbred stocks to a laboratory environment that leads to degeneration of many adaptations that develop or are maintained under natural selection pressure.
We finally investigated likely functional consequences of the editing of synapsin described here. We measured the in vitro phosphorylation by bovine PKA of N-terminal undeca-peptides containing the edited, the unedited, and, as a negative control, a mutated amino acid sequence. These experiments clearly demonstrate that the peptide containing the genomic RRFS sequence represents an excellent substrate in this assay, while the edited version (RGFS) is not efficiently phosphorylated. Quantitative measurements of Michaelis-Menten constants should eventually be obtained for intact synapsin isoforms rather than on peptides, but such experiments are beyond the scope of this paper. However, the present in-vitro peptide phosphorylation data strongly suggest that editing also influences phosphorylation of synapsin by PKA in vivo. Since synapsin knock-out flies are impaired in learning and/or memory [27
] we speculate that phosphorylation of Drosophila
synapsin in the A domain is subject to cell-specific fine regulation by RNA editing. This now needs to be tested by appropriate phosphorylation assays using wild-type and transgenic flies with targeted mutations that prevent or simulate synapsin phosphorylation in conjunction with studies on the behavioural impairments of these flies as described for synapsin knock-out animals [27