The polyadenylation site of Mimivirus transcripts obeys a stringent 'hairpin rule'

Abstract

Mimivirus, a giant DNA virus infecting Acanthamoeba, is revealing an increasing list of unique features such as a 1.2-Mb genome with numerous genes not found in other viruses, a uniquely conserved promoter signal, and a particle of unmatched complexity using two distinct portals for genome delivery and packaging. Herein, we contribute a further Mimivirus distinctive feature discovered by sequencing a panel of viral cDNAs produced for probing the structure of Mimivirus transcripts. All Mimivirus mRNAs are polyadenylated at a site coinciding exactly with unrelated, but strongly palindromic, genomic sequences. The analysis of 454 Life Sciences (Roche) FLX cDNA tags (150,651) confirmed this finding for all Mimivirus genes independent of their transcription timings and expression levels. The absence of a suitable palindromic signal between adjacent genes results in transcripts encompassing multiple ORFs in the same or even in opposite orientations. Surprisingly, Mimivirus tRNAs are expressed as polyadenylated messengers, including an ORF/tRNA composite mRNA. To our knowledge, both the nature and the stringency of the "hairpin rule" defining the location of polyadenylation sites are unique, raising once more the question of Mimivirus's evolutionary origin. The precise molecular mechanisms implementing the hairpin rule into the 3'-end processing of Mimivirus pre-mRNAs remain to be elucidated.

Figure 1.

Histogram of the initial positions of the palindromes scoring above threshold with respect to the stop codons. X = 0 is the position of the downstream stop codon of each Mimivirus gene. Palindrome positions refer to their 5′ extremities. A bin width of 10 nt was used. The presence of the palindromes strongly correlates with the beginning of the 3′-UTR, just after the stop codon. Forty palindromes overlap the stop codon, leading to extremely short 3′-UTRs (Table 1).

Figure 2.

Experimentally validated Mimivirus polyadenylated transcript structures. Palindromic sequences are represented by hairpins; ORFs are indicated by open arrows; and transcripts by red arrows. (A) Single ORF followed by a palindromic polyadenylation signal; ∼70% of transcripts might fall in this category; 27 have been experimentally validated through a gene-by-gene analysis. (B) Convergent ORFs sharing the same palindromic polyadenylation signal; four such pairs have been validated (R257/ L258, R453/ L454, R497/L498, R528/L529). (C) “Polycistronic” transcripts, wherein polyadenylation occurs within the first palindromic signal encountered. Two such transcripts have been validated (L416-417, R882-883). (D) “Pseudo-polycistronic” messenger, wherein the 3′-UTR encompasses a neighboring ORF in the reverse orientation. Two such cases have been validated (L356-R355, L778-R777). (E) Infrequent cases of polyadenylation occurring within a low-scoring palindrome (e.g., L164) or following an AATAAA motif (e.g., R502, L532, polycistronic R463-465). (F) A unique case of a polyadenylated bi-cistronic transcript mixing the R901 ORF and a downstream neighboring tRNA^Leu. (G) Polyadenylated transcripts of individual tRNA genes (tRNA^Trp, tRNA^His). (H) Polyadenylated transcripts encompassing two adjacent tRNA genes (tRNA^His + tRNA^Cys) coexisting with a single gene polyadenylated transcript (tRNA^His).