Alternative trans-splicing of constant and variable exons of a Drosophila axon guidance gene, lola

Alternative splicing is a crucial source of diversity in the proteomes of higher eukaryotes (for reviews, see Black 1998; Graveley 2001; Maniatis and Tasic 2002). Some genes encode an extraordinarily large number of alternatively spliced variants. For example, the mammalian neurexin genes produce over 1000 variants of a presynaptic cell-adhesion molecule linking the pre- and postsynaptic compartments of synapses and essential for Ca2+-triggered neurotransmitter release (Missler et al. 2003). More remarkably, Dscam, a Drosophila axon guidance receptor gene, encodes over 38,000 variants (Schmucker et al. 2000). These genes are implicated in nervous system functions and/or development, suggesting a crucial role for alternative splicing in establishing the highly complex organization of animal tissues. Given that each isoform has its own function, alternative splicing must be strictly controlled to express proper isoforms in proper cells at the proper timing. Splicing can occur either in cis or in trans: cis-splicing joins exons within a pre-mRNA, whereas trans-splicing joins exons of independently transcribed pre-mRNAs. Several cases of trans-splicing have been reported in mammalian cells (Shimizu and Honjo 1993; Fujieda et al. 1996; Caudevilla et al. 1998; Frantz et al. 1999; Chatterjee and Fisher 2000; Takahara et al. 2000; Finta and Zaphiropoulos 2002; Flouriot et al. 2002; Tasic et al. 2002; Wang et al. 2002). However, these occur with very low frequency, and their functional significance has not been demonstrated, leading to the proposal that they may represent “splicing noise” (Tasic et al. 2002; Wang et al. 2002). In contrast, trans-splicing, as an essential mechanism for generation of mRNAs, most likely occurs in the Drosophila modifier of mdg4 [mod(mdg4)] locus, because of its unusual genomic organization: Some of the alternatively spliced exons are encoded by the opposite DNA strand (Dorn et al. 2001; Labrador et al. 2001; Mongelard et al. 2002). mod(mdg4) encodes 26 isoforms, each consisting of identical N termini containing a BTB domain but variable C termini, 20 of which contain the conserved C2H2-type zinc finger motifs (Dorn et al. 2001). Surprisingly, seven isoforms are encoded by the antiparallel DNA strand, and are indeed expressed in vivo (Dorn et al. 2001). Therefore, generation of mRNA by trans-splicing is obvious in the mod-(mdg4) locus. No case of trans-splicing has been reported to date in genes with typical genomic organization, in which all exons are sequentially encoded by the same DNA strand. The Drosophila longitudinals lacking (lola) gene encodes a family of BTB-Zn finger transcription factors required for a number of axon guidance decisions in the developing Drosophila nervous system (Giniger et al. 1994; Madden et al. 1999; Crowner et al. 2002; Goeke et al. 2003). lola is a large and complex locus, spanning over 60 kb and consisting of 32 exons aligned on the same DNA strand, and it produces at least 80 splicing variants through multiple promoter activities and alternative splicing (Fig. 1A; Ohsako et al. 2003). There are four promoters initiating transcription at distinct exons corresponding to different positions within the 5′UTR (exons 1–4). Each splicing variant has a constant sequence (exons 5–8) encoding the N-terminal region containing a BTB dimerization domain (Bardwell and Treisman 1994) and one or two exons alternatively selected from 20 groups of exons (exons 9–32) encoding a C-terminal variable region, 17 of which have unique zinc finger motifs. Almost all of the 17 putative DNA binding domains have unique combinations of predicted DNA contact residues, suggesting that they are likely to have unique DNA target specificities. Functional diversity of lola may be further increased through the BTB domain-mediated dimerization of isoforms. The lola isoforms are expressed in a complex pattern of tissues in the embryo, with some forms present in broad and overlapping domains and others in single tissues or dynamic patterns of isolated cells (Goeke et al. 2003). Furthermore, mutations inactivating single lola isoforms show defects in specific subsets of lola-dependent axon guidance choice points (Goeke et al. 2003). These findings suggest that alternative splicing of lola pre-mRNA plays critical roles in specifying a diverse variety of axon guidance decisions. Figure 1. Interallelic complementation among five lethal mutations in lola. (A) Schematic representation of the genomic organization of the lola locus. The entire locus consists of 32 exons; gray, white, and black boxes indicate 5′ variable, constant, and ... Although the number of isoforms produced in this locus is not of the extent of those in Neurexin or Dscam, it suffers several mechanistic complexities. First, the alternatively spliced exons are the terminal ones in the mRNA, and therefore are associated with as many as 20 potential polyadenylation sites. Thus, proper expression of isoforms requires strict regulation of all these potential polyadenylation sites in coordination with the alternative splicing. Second, it is difficult to see how multiple splice variants are coexpressed within a single cell. Somehow, only a single 3′ end can be linked to the common exons in a single processed transcript, even though other 3′ ends are competent to be joined in the same cell. Trans-splicing may be one possibility to overcome these potential problems. However, despite similarities in molecular features of lola and mod(mdg4), the possibility of trans-splicing is less obvious in this case, because all lola exons are sequentially encoded in the same DNA strand. Therefore, careful experimental analysis is necessary to conclude whether or not trans-splicing is involved in the generation of lola mRNA. In this report, we show that at least some lola isoforms are generated through alternative trans-splicing of the constant and variable exons. Genetic tests demonstrate that mutations in lola common exons can complement mutations in variable exons, and analysis of single nucleotide polymorphisms (SNPs) shows that this reflects the presence of chimeric, wholly wild-type mature transcripts in these mutant animals. This trans-splicing is highly efficient, with about half of the mature transcripts for some isoforms arising from trans-splicing events. Efficient trans-splicing is prevented by chromosome rearrangements, suggesting that the process requires chromosome pairing. Finally, we find that the final lola exon is expressed autonomously from a promoter in the preceding intron. The combination of internal promoters and trans-splicing may simplify the problem of achieving the complex pattern of lola isoform expression during development.