The polyadenylation signal of rice tungro bacilliform virus (RTBV) was char
acterized by mutational and deletion analysis. The cis-acting signals requi
red to direct polyadenylation conformed to what is known for plant poly(A)
signals in general and were very similar to those of the related cauliflowe
r mosaic virus. Processing was directed by a canonical AAUAAA poly(A) signa
l, an upstream UG-rich region considerably enhanced processing efficiency,
and sequences downstream of the cleavage site were not required. When prese
nt at the end of a transcription unit, the cis-acting signals for 3'-end pr
ocessing were highly efficient in both monocot (rice) and dicot (Nicotiana
plumbaginifolia) protoplasts. In a promoter-proximal position, as in the vi
ral genome, the signal was also efficiently processed in rice protoplasts,
giving rise to an abundant "short-stop" (SS-) RNA. The proportion of SS-RNA
was considerably lower in N. plumbaginifolia protoplasts. In infected plan
ts, SS-RNA was hardly detectable, suggesting either that SS-RNA is unstable
in infected plants or that read-through of the promoter-proximal poly(A) s
ite is very efficient. SS-RNA is readily detectable in transgenic rice plan
ts (A, Kloti, C. Henrich, S, Bieri, X. He, G. Chen, P. K. Burkhardt, J. Wun
n, P. Lucca, T. Hohn, I. Potrylus, and J, Futterer, 1999, Plant Mol. Biol,
40:249-266), thus the absence of SS-RNA in infected plants can be attribute
d to poly(A) site bypass in the viral context to ensure production of the f
ull-length pregenomic viral RNA. RTBV poly(A) site suppression thus depends
both on context and the expression system; our results suggest that the ci
rcular viral minichromosome directs assembly of a transcription-processing
complex with specific properties to effect read-through of the promoter-pro
ximal poly(A) signal.