A rabbit beta-globin polyadenylation signal directs efficient termination of transcription of polyomavirus DNA

A 94-nucleotide SfaNI–AhaIII fragment of rabbit β-globin DNA, spanning nucleotides 1688–1781 of the gene, contains a complete functional polyadenylation signal: the AATAAA hexanucleotide 6 nucleotides downstream of the SfaNI end, together with 82 nucleotides of downstream sequence carrying the additional elements required for efficient cleavage and polyadenylation of β-globin RNA.

The rabbit β-globin polyadenylation signal is recognized efficiently by the cleavage/polyadenylation machinery of mouse 3T6 cells, including during the late phase of polyomavirus infection: S1 nuclease mapping of RNA from 3T6 cells transfected with a plasmid carrying a tandem duplication of the signal shows transcripts polyadenylated almost exclusively at the upstream copy, whether or not the cells are co-infected with polyomavirus, indicating that polyomavirus infection does not deplete or limit the host factors needed to recognize this heterologous signal.

Inserting a single copy of the 94-nucleotide rabbit β-globin polyadenylation signal upstream of, and in the same orientation as, the native L-strand polyadenylation site of the polyomavirus genome (mutant ins 5) directs approximately 92% of L-strand polyadenylation events to the inserted β-globin site; the residual downstream viral site produces only a minor S1-protected band, roughly 7% as intense as the major band from the β-globin site. This contrasts with a polyomavirus mutant carrying a tandem duplication of its own native, weak L-strand polyadenylation signal (mutant ins 4, a 144-nucleotide duplicated fragment), in which polyadenylation events split roughly evenly, about 50% at each of the duplicated upstream and downstream copies — showing that an intrinsically strong, heterologous polyadenylation signal, not sequence duplication alone, is what can redirect processing away from a weak native site.

Inserting the same 94-nucleotide rabbit β-globin fragment at the identical genomic position but in the reverse orientation, opposite to the direction of L-strand transcription (mutant ins 6), abolishes its function as an L-strand polyadenylation signal: no L-strand RNA is polyadenylated within the insert, and L-strand transcripts are instead polyadenylated at the native downstream viral site as in wild-type virus. The rabbit β-globin polyadenylation signal therefore acts only in an orientation-dependent, transcription-direction-specific manner.

Inserting the rabbit β-globin polyadenylation signal into the L strand of the polyomavirus genome (mutant ins 5) increases the overall proportion of virus-specific nuclear RNA that is polyadenylated during infection of mouse 3T6 cells, from 19% to 45% (a 2.5-fold increase) in one experiment and from 22% to 31% (a 1.4-fold increase) in a second, independent experiment, relative to wild-type polyomavirus infection.

Inserting the rabbit β-globin polyadenylation signal into the L strand of the polyomavirus genome increases the efficiency of transcription termination by RNA polymerase II on that strand, from 78–81% per genome traverse in wild-type virus to 94–98% in the ins 5 mutant, reducing the fraction of polymerases that read through the entire viral genome without terminating from 19–22% to 2–6%. A polyadenylation signal that efficiently directs RNA 3′-end cleavage and polyadenylation therefore also functions, at least in part, as a signal for transcription termination by RNA polymerase II.

Duplicating the polyomavirus genome's own native L-strand polyadenylation signal at the same genomic position (mutant ins 4, a 144-nucleotide duplicated fragment) does not increase either the overall efficiency of L-strand RNA polyadenylation or the efficiency of L-strand transcription termination, unlike insertion of the heterologous rabbit β-globin signal at that position. The native polyomavirus L-strand polyadenylation signal is therefore inherently weak, and this weakness is not corrected by increasing its copy number.

Despite carrying 174 extra nucleotides of inserted sequence (yielding a 5484-nucleotide genome) and directing near-complete (94–98%) termination of L-strand transcription — which should largely eliminate the multigenome-length read-through transcripts that are normally spliced into L-strand mRNAs bearing multiple tandem copies of a 57-nucleotide leader sequence — the ins 5 polyomavirus mutant replicates as well as wild-type virus in primary baby mouse kidney cells, producing equivalent viral titers and equivalent amounts of virus-specific RNA. Efficient transcription termination, and the resulting loss of multiply spliced leader mRNAs, is therefore not necessary for productive polyomavirus replication in tissue culture.