Nucleotide sequence of bean golden mosaic virus and a model for gene regulation in geminiviruses

Abstract

We have sequenced the genome of bean golden mosaic virus, which comprises two circular single-stranded DNA molecules (2646 and 2587 nucleotides long) of mostly unique sequence. Comparison of the sequences of bean golden mosaic virus and of cassava latent virus, which share serological relationship but are very different in host range and geographical origin, shows that each virus has a unique 200-nucleotide sequence (common region) on each 2.6-kilobase molecule of its genome. The common regions of the two viruses have no sequence homology except for a short inverted repeat near the 3' end. Six open reading frames were identified that possess considerable sequence homology between the two viruses and, in bean golden mosaic virus, may encode proteins of 15.6, 19.6, 27.7, 29.7, 33.1, and 40.2 kDa. Conserved open reading frames are found in both the viral strand and the complementary strand, are approximately the same size, and are in the same orientation with respect to the common region in both viruses. We propose that temporal regulation in geminiviruses depends on the polarity of transcription and that the common region represents a replication origin and contains elements that serve to modulate gene expression.