The carboxy-terminal domain (CTD) of the large subunit of RNA polymerase II (pol II) is essential for several co-transcriptional pre-messenger RNA processing events, including capping, 3 0 -end processing and splicing. We investigated the role of the CTD of RNA pol II in the coordination of A to I editing and splicing of the ADAR2 (ADAR: adenosine deaminases that act on RNA) pre-mRNA. The auto-editing of Adar2 intron 4 by the ADAR2 adenosine deaminase is tightly coupled to splicing, as the modification of the dinucleotide AA to AI creates a new 3 0 splice site. Unlike other introns, the CTD is not required for efficient splicing of intron 4 at either the normal 3 0 splice site or the alternative site created by editing. However, the CTD is required for efficient co-transcriptional auto-editing of ADAR2 intron 4. Our results implicate the CTD in site-selective RNA editing by ADAR2 and in coordination of editing with alternative splicing.

ADAR enzymes, adenosine deaminases that act on RNA, form a family of RNA editing enzymes that convert adenosine to inosine within RNA that is completely or largely double-stranded. Site-selective A-->I editing has been detected at specific sites within a few structured pre-mRNAs of metazoans. We have analyzed the editing selectivity of ADAR enzymes and have chosen to study the naturally edited R/G site in the pre-mRNA of the glutamate receptor subunit B (GluR-B). A comparison of editing by ADAR1 and ADAR2 revealed differences in the specificity of editing. Our results show that ADAR2 selectively edits the R/G site, while ADAR1 edits more promiscuously at several other adenosines in the double-stranded stem. To further understand the mechanism of selective ADAR2 editing we have investigated the importance of internal loops in the RNA substrate. We have found that the immediate structure surrounding the editing site is important. A purine opposite to the editing site has a negative effect on both selectivity and efficiency of editing. More distant internal loops in the substrate were found to have minor effects on site selectivity, while efficiency of editing was found to be influenced. Finally, changes in the RNA structure that affected editing did not alter the binding abilities of ADAR2. Overall these findings suggest that binding and catalysis are independent events.

The ADAR family of RNA-editing enzymes deaminates adenosines within RNA that is completely or largely double stranded. In mammals, most of the characterized substrates encode receptors involved in neurotransmission, and these substrates are thought to be targeted by the mammalian enzymes ADAR1 and ADAR2. Although some ADAR substrates are deaminated very promiscuously, mammalian glutamate receptor B (gluR-B) pre-mRNA is deaminated at a few specific adenosines. Like most double-stranded RNA (dsRNA) binding proteins, ADARs bind to many different sequences, but few studies have directly measured and compared binding affinities. We have attempted to determine if ADAR deamination specificity occurs because the enzymes bind to targeted regions with higher affinities. To explore this question we studied binding of rat ADAR2 to a region of rat gluR-B pre-mRNA that contains the R/G editing site, and compared a wild-type molecule with one containing mutations that decreased R/G site editing. Although binding affinity to the two sequences was almost identical, footprinting studies indicate ADAR2 binds to the wild-type RNA at a discrete region surrounding the editing site, whereas binding to the mutant appeared nonspecific.

Scanning force microscopy (SFM) can be used to image biomolecules at high resolution. Here we demonstrate that singlemolecule analysis by SFM complements biochemical data on RNA protein binding and can provide information that cannot be obtained by the usual biochemical methods. We have used this method to study the interaction between the RNA editing enzyme ADAR2 and RNA transcripts containing selective and nonselective editing sites. The natural selectively edited R/G site from glutamate receptor subunit B (GluR-B) was inserted into an RNA backbone molecule consisting of a completely doublestranded (ds) central part and incompletely paired ends derived from potato spindle tuber viroid (PSTVd). This molecule was efficiently edited at the R/G site, but promiscuous editing occurred at nonselective sites in the completely double-stranded region. The construct was also used to analyze binding of ADAR2 to wild-type and modified R/G editing sites in relation to binding at other nonselectively edited sites. Editing analysis together with SFM allow us to differentiate between binding and enzymatic activity. ADAR2 has been reported to have a general affinity to dsRNA. However, we show that there is a prominent bias for stable binding at sites selectively edited over other edited sites. On the other hand, promiscuous editing at nonselective sites apparently results from transient binding of the enzyme to the substrate. Furthermore, we find distinct sites with nonproductive binding of the enzyme.