Messenger RNA processing sites in Trypanosoma brucei

Abstract

In Kinetoplastids, protein-coding genes are transcribed polycistronically by RNA polymerase II. Individual mature mRNAs are generated from polycistronic precursors by 5′ trans splicing of a 39-nt capped leader RNA and 3′ polyadenylation. It was previously known that trans splicing generally occurs at an AG dinucleotide downstream of a polypyrimidine tract, and that polyadenylation is coupled to downstream trans splicing. The few polyadenylation sites that had been examined were 100–400 nt upstream of the polypyrimidine tract which marked the adjacent trans splice site. We wished to define the sequence requirements for trypanosome mRNA processing more tightly and to generate a predictive algorithm.

By scanning all available Trypanosoma brucei cDNAs for splicing and polyadenylation sites, we found that trans splicing generally occurs at the first AG following a polypyrimidine tract of 8–25 nt, giving rise to 5′-UTRs of a median length of 68 nt. We also found that in general, polyadenylation occurs at a position with one or more A residues located between 80 and 140 nt from the downstream polypyrimidine tract. These data were used to calibrate free parameters in a grammar model with distance constraints, enabling prediction of polyadenylation and trans splice sites for most protein-coding genes in the trypanosome genome. The data from the genome analysis and the program are available from: http://web.cgb.ki.se/daniel/splicemodel.php.

Introduction

The Kinetoplastid protists and their close relatives are the only eukaryotes known in which trans splicing is an obligatory step in maturation of all mRNAs. In Kinetoplastids, transcription of protein-coding genes by RNA polymerase II is polycistronic [1], [2], [3] and polycistrons can be many tens of kb long [3], [4], [5]. Although RNA polymerase II appears to initiate transcription at the boundaries between divergent polycistrons, or between polycistrons and regions transcribed by other RNA polymerases, attempts to identify consensus promoter sequences have failed [6], [7] and multiple initiation sites can be used for transcription of a single locus [5]. In addition, some protein-coding gene arrays, such as the VSG and procyclin expression sites, are transcribed by RNA polymerase I ([8] and references therein).

Mature mRNAs must be monocistronic, and require a cap and a poly(A) tail, in order to be exported from the nucleus and translated efficiently. In Kinetoplastids monocistronic mRNAs are generated by processing of the polycistronic precursors. Capped 5′-ends are formed by a trans splicing reaction in which a capped spliced leader (SL) sequence [9], [10], [11] of about 40 nt is added to the 5′-end of each mRNA [12]. As in cis splicing, the acceptor site is an AG dinucleotide, which is preceded by a polypyrimidine tract of variable length [12], [13], [14]. Addition of a poly(A) tail is required to complete mRNA processing. There is no consensus polyadenylation signal in the 3′-untranslated region (3′-UTR). Instead, evidence obtained from a very small number of loci suggests that polyadenylation occurs within a short region (rather than at a single defined site) 100–400 nt upstream of the next polypyrimidine trans splice signal [13], [15], [16], [17], [18]. In these experiments, plasmids bearing reporter genes were introduced into trypanosomes, and the exact sites of polyadenylation of the reporter transcripts were determined. Changes in the sequences downstream of the reporter gene consistently showed dependence of polyadenylation on the position of a downstream trans splicing acceptor site.

Experiments with permeabilised trypanosomes indicate that trans splicing of a transcript precedes the polyadenylation of its upstream partner and that inhibition of trans splicing leads to concomitant loss of polyadenylation [19]. Abrogation of either one of these mutually dependent RNA processing reactions, either through heat shock [20], RNA interference directed against processing factors [21], or sinefungin treatment (which inhibits trans splicing by preventing methylation of the SL RNA cap [22]) results in the accumulation of dicistronic or longer transcripts [23].

Although very broad outlines of the requirements for mRNA processing have been recognised, details have so far not been examined and the number of loci studied in detail has been very small and restricted to highly abundant transcripts. In addition, it is known that features in addition to polypyrimidine tracts are required for trans splicing and polyadenylation to occur. The intergenic regions of trypanosome transcripts very often contain multiple polypyrimidine tracts, yet a preference for the tract immediately preceding an open reading frame (ORF) has been observed [24]. It is not clear whether the ORF, the 5′-UTR [25], or both are important in determining this preference. In contrast, trans splicing to AG sites lacking any obvious downstream ORF can also occur and direct polyadenylation of the upstream transcript [15]. This can result in the generation of ORF-less processed RNAs and provides a possible explanation for the means by which the most 3′ mRNA in a polycistron is polyadenylated.

Conventional gene-recognition programmes identify many potential trypanosome ORFs which will never be translated into a functional protein because they lack the requisite mRNA processing signals. Identification of these signals will enable genome-wide recognition of the splice sites that serve as indicators of the functionality of potential ORFs [26]. A good algorithm for the prediction of trans splicing signals would therefore be a very useful tool for annotating kinetoplastid genomes as well as determining untranslated regions of mature transcripts.

In the work presented here, we have analysed all available splicing and polyadenylation sites from T. brucei for common sequence signals and used the information to generate a prediction algorithm for trypanosome mRNA processing sites.