A Drosophila full-length cDNA resource

Abstract

Background: A collection of sequenced full-length cDNAs is an important resource both for functional genomics studies and for the determination of the intron-exon structure of genes. Providing this resource to the Drosophila melanogaster research community has been a long-term goal of the Berkeley Drosophila Genome Project. We have previously described the Drosophila Gene Collection (DGC), a set of putative full-length cDNAs that was produced by generating and analyzing over 250,000 expressed sequence tags (ESTs) derived from a variety of tissues and developmental stages.

Results: We have generated high-quality full-insert sequence for 8,921 clones in the DGC. We compared the sequence of these clones to the annotated Release 3 genomic sequence, and identified more than 5,300 cDNAs that contain a complete and accurate protein-coding sequence. This corresponds to at least one splice form for 40% of the predicted D. melanogaster genes. We also identified potential new cases of RNA editing.

Conclusions: We show that comparison of cDNA sequences to a high-quality annotated genomic sequence is an effective approach to identifying and eliminating defective clones from a cDNA collection and ensure its utility for experimentation. Clones were eliminated either because they carry single nucleotide discrepancies, which most probably result from reverse transcriptase errors, or because they are truncated and contain only part of the protein-coding sequence.

Figure 1

A putative example of RNA editing as revealed by comparison of cDNA and genomic DNA sequences. (a) Gene models for CG18314 based on sequence of two DGCr1 full-length cDNA clones (GH15292.c, GH08370.c) that differ at their 5' and 3' termini. Although the cDNAs have alternative 5' and 3' UTRs and are alternatively spliced, they share the same protein-coding potential (shown in blue). CG18314 encodes a G-protein-coupled receptor of the rhodopsin family, containing a seven-transmembrane protein domain (7tm_1; the red bar shows the extent of the domain) with similarity to β₂-adrenergic receptors of mouse (X15643, E value = 9e-23) and human (M15169, E = 8e-22). Shown hatched is a 310-bp portion of cDNA sequences with A-to-G nucleotide variation. (b) Sequence alignments of this 310-bp portion of genomic sequence, two cDNA and three EST sequences (GH14918, GH14553, HL02270). Shown in yellow are codons with A-to-G nucleotide variation. Above the genomic nucleotide sequence is its translated amino-acid sequence starting at amino acid 224 of the protein. Comparing the cDNA nucleotide sequence to the genomic sequence identifies 10 A-to-G nucleotide variations. Two are silent, seven result in amino-acid changes, and one alters the stop codon, allowing two additional amino acids to be encoded. The amino acids that are affected are shown below the nucleotide sequence (red letters in a gray circle). Two of the amino-acid changes (N224S and S229G) map to the conserved seven-transmembrane protein domain. The Anopheles gambiae genomic draft contains sequence encoding this protein (gi|21299606|gb|EAA11751.1| (AAAB01008960) agCP5433) which is highly conserved at the amino-acid sequence level (E = e-168) and also encodes N and S at these sites. To sample additional transcripts of this gene, we performed gene-specific RT-PCR to amplify the region shown in (b). From a total of 64 independent transcripts we confirmed the 10 cases of editing diagrammed above, and identified 15 new sites of A-to-G nucleotide variations. A list of these putative editing sites showing the resulting amino-acid change and the number of times this change was observed, given in parentheses, is as follows: N224D (2), N224S (12), L225L (9), N227S (1), S229G (9), H230R (1), M231V (1), L236L (16), A239A (1), P246P (2) E254G (1), I272I (1), I275M (1), I281V (1), S286G (1), K306R (16), K308R (5), K308G (8), Q312Q (1), A313A (1), L315L (31), I316V (52), *323W (44) and S324G (4).