Purification of inosine monophosphate by thin layer chromatography and detection of inosine monophosphate within cellular messenger RNA

This dissertation describes the development and use of novel thin layer chromatography (TLC) purification methods that allowed for the specific and sensitive detection of inosine monophosphate (IMP) within cellular mRNA. Radioactive 5' nucleoside monophosphate (5' NMP) mixtures were prepared from mRNA, and then spiked with nonradioactive 5' IMP. This nonradioactive 5' IMP served as a marker for the migration of radioactive 5' IMP and allowed for the conclusive identification of radioactive 5' IMP derived from mRNA. Three consecutive rounds of TLC purified 5' IMP away from the more abundant ribonucleotides found within RNA (adenosine monophosphate, guanosine monophosphate, cytidine monophosphate, and uridine monophosphate), as well as other minor modified nucleotides found within RNA. These purification methods allowed for the detection of 1 to 5 parts IMP per 1 million parts NMPs (1-5 ppm). Radioactive 5' NMPs were prepared from polyadenylated RNA isolated from a number of different rat tissues. TLC analyses of these 5' NMPs showed that 5' IMP was most abundant in brain mRNA, and the tissue specific abundance of 5' IMP directly correlated with the expression of genes encoding adenosine deaminases that act on RNA (ADARs). Quantitation of this endogenous IMP provided the first estimate of the amount of inosine in cellular mRNA. It was also demonstrated that in polyoma virus infected cells, IMP levels in cellular RNA increased with viral titer. Polyoma early (sense) and late (antisense) RNAs are thought to form double-stranded RNAs (dsRNAs) within the cell nucleus and undergo significant modification by ADARs. Cells were infected with polyoma virus and nuclear and cytoplasmic RNA was subsequently isolated at various hours postinfection. TLC analyses showed that 5' IMP was present in nucleotides derived from nuclear RNA isolated at the late stage of polyoma infection but 5' IMP was not present in nucleotides derived from nuclear RNA isolated at the early stage of polyoma infection. These results suggest that stable dsRNA duplexes exist within the nucleus during the late stage of polyoma infection and that these duplexes are modified by ADARs.

PURIFICATION OF INOSINE MONOPHOSPHATE BY THIN LAYER CHROMATOGRAPHY AND DETECTION OF INOSINE MONOPHOSPHATE WITHIN CELL1JLAR MESSENGER RNA by Michael Sean Paul A dissertation submitted to the faculty of The University of Utah in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Biochemistry The University of Utah August 1999 Copyright © Michael Sean Paul 1999 All Rights Reserved THE UNIVERSITY OF UTAH GRADUATE SCHOOL SUPERVISORY COMMITTEE APPROVAL of a dissertation submitted by Michael Sean Paul This dissertation has been read by each member of the following supervisory committee and by majority vote has been found to be satisfactory. Martin C. Rechsteiner ~A.Ut~~ ~es A. McCloskey ~a;!:f~~~ David Grunwal THE UNIVERSITY OF UTAH GRADUATE SCHOOL FIN AL READING APPROV AL To the Graduate Council of the University of Utah: I have read the dissertation of Michael Sean Paul in its final form and have found that (1) its format, citations, and bibliographic style are consistent and acceptable; (2) its illustrative materials including figures, tables, and charts are in place; and (3) the final manuscript is satisfactory to the supervisory committee and is ready for submission to The Graduate School. $!()~/'l~ ,~U~ Date Brenda L. Bass Chair, Supervisory Committee Approved for the Major Department ~c Zc4~(:<- Martin C. Rechsteiner C 0-ChairlDean Approved for the Graduate Council Ann W. Hart Dean of The Graduate School ABSTRACT This dissertation describes the development and use of novel thin layer chromatography (TLC) purification methods that allowed for the specific and sensitive detection of inosine monophosphate (IMP) within cellular mRNA. Radioactive 5' nucleoside monophosphate (5' NMP) mixtures were prepared from mRNA, and then spiked with nonradioactive 5' IMP. This nonradioactive 5' IMP served as a marker for the migration of radioactive 5' IMP and allowed for the conclusive identification of radioactive 5' IMP derived from mRNA. Three consecutive rounds of TLC purified 5' IMP away from the more abundant ribonucleotides found within RNA (adenosine monophosphate, guanosine monophosphate, cytidine monophosphate, and uri dine monophosphate), as well as other minor modified nucleotides found within RNA. These purification methods allowed for the detection of 1 to 5 parts IMP per 1 million parts NMPs (1-5 ppm). Radioactive 5' NMPs were prepared from polyadenylated RNA isolated from a number of different rat tissues. TLC analyses of these 5' NMPs showed that 5' IMP was most abundant in brain mRNA, and the tissue specific abundance of 5' IMP directly correlated with the expression of genes encoding adenosine deaminases that act on RNA (ADARs). Quantitation of this endogenous IMP provided the first estimate of the amount of inosine in cellular mRNA. It was also demonstrated that in polyoma virus infected cells, IMP levels in cellular RNA increased with viral titer. Polyoma early (sense) and late (antisense) RNAs are thought to form double-stranded RNAs (dsRNAs) within the cell nucleus and undergo significant modification by ADARs. Cells were infected with polyoma virus and nuclear and cytoplasmic RNA was subsequently isolated at various hours postinfection. TLC analyses showed that 5' IMP was present in nucleotides derived from nuclear RNA isolated at the late stage of polyoma infection but 5' IMP was not present in nucleotides derived from nuclear RNA isolated at the early stage of polyoma infection. These results suggest that stable dsRNA duplexes exist within the nucleus during the late stage of polyoma infection and that these duplexes are modified by ADARs. v CONTENTS ABSTRACT .......................................... ~ ................................................... iv ACKNOWLEDGMENTS ....................................................................... viii Chapter 1. INTRODUCTION ........................................................................ 1 References .................................................................................. 21 2. PURIFICATION OF 5' IMP BY THIN LAYER CHROMATOGRAPHY .............................................................. 27 Introduction ............................................................................... 28 Results and Discussion ............................................................. 29 C oncl usions ................................................................................ 57 Materials and Methods ............................................................. 58 References .................................................................................. 67 3. INOSINE EXISTS IN mRNA AT TISSUE­SPECIFIC LEVELS AND IS MOST ABUNDANT IN BRAIN mRNA ...................................................................... 69 Introduction ............................................................................... 70 Results ....................................................................................... 71 Discussion .................................................................................. 75 Materials and Methods ............................................................. 76 Acknowledgements .................................................................... 77 References .................................................................................. 77 4. DETECTION OF INOSINE WITHIN NUCLEAR RNA ISOLATED FROM POLYOMA VIRUS-INFECTED CELLS ................................................................... 78 Introduction ............................................................................... 79 Results and Discussion ............................................................. 82 Conclusions ................................................................................ 91 Materials and Methods ............................................................. 92 References .................................................................................. 94 5. PERSPECTIVES ....................................................................... 95 References .................................................................................. 99 VII ACKNOWLEDGMENTS I thank my wife, Amy Kathleen, for not only her love and support during these years but also for the motivation that she provided when all that I saw were doubts. I acknowledge my daughter, Gabrielle Marie, for all the smiles that she provided and wish to thank my parents, H.C. and Cathryn Paul, for their support and belief in my goals. I thank Dan and Judy Regan for their support and encouragement and a special thanks to Lucile Hannifin for preparing the way. I thank all my committee members for their support and guidance and I especially thank Brenda L. Bass for her years of support and her desire to shape me and all of her students into independent and critical thinkers. I thank all my Bass Lab members, especially Ron Hough, Bert Ley, and Jana Brubaker. And as always and in all things, I thank God. Inosine exisits in mRNA at tissue-specific levels and is most abundant in brain mRNA, M. S. Paul, B. L. Bass. EMBO J. 17, 1120-1127, (1998), reprinted with permission, copyright 1998, Oxford University Press. CHAPTER 1 INTRODUCTION 2 In prokaryotes and eukaryotes, the purine ribonucleotide inosinate plays a central role in the de novo biosynthesis of the major purine ribonucleotides adenylate and guanylate (Figure 1.1) (1). Inosine monophosphate (IMP) is converted to adenosine monophosphate (AMP) by amination using adenylosuccinate as the amino donor group. Guanosine monophosphate (GMP) is derived from IMP via a xanothosine monophosphate intermediate. Enzymes involved in the metabolism of adenosine nucleotides and nucleosides also produce inosine as a key intermediate. The deamination of AMP by adenyl ate deaminase to IMP and the deamination of adenosine to inosine by adenosine deaminase are important biochemical regulatory pathways (2). In addition to its role in purine metabolism, InOSIne occurs as a modified nucleotide in certain mature transfer RNA (tRNA) molecules. Most classes of cellular RNA contain posttranscriptionally modified nucleosides and by mid-1994, more than 93 different modified nucleosides were identified in RNA (3,4). Transfer RNA contains the greatest variety and the largest number of modified nucleosides; 79 of the reported 93 modified nucleosides are found in tRNA. It has been estimated that in some tRNAs from higher eukaryotes, the extent of modifications may be as high as 25% of the total tRNA nucleosides (5). The nature of these modifications ranges from so­called simple modifications, such as base and ribose methylations, to much more complex structural modifications termed hypermodifications (6). Nucleotide isolation procedures and sequence analyses of purified yeast tRNA, or soluble RNA (S-RNA), revealed the presence of inosine and 1- methylinosine as two of nine constituent "unusual" nucleotides in yeast alanine tRNA (tRNAAla). o t~ -NH NA) I N Ribose-P04 IMP y ~ e e OOC-CH2-CH-COO I NH adenylosuccinate I NH NAO Ribos1e-P04 H xanthosine monophosphate NH2 (b) ~N I) I N Ribose-P04 -....-... AMP o (d) -- 1 NJ: RibOS~-P04 H NH2 ....... GMP Figure 1.1. De novo biosynthesis of purine ribonucleotides. Adenosine monophosphate (AMP) and guanosine monophosphate (GMP) are synthesized from inosine monophosphate (IMP). Reaction steps indicated by arrows are catalyzed by: (a) adenylosuccinate synthetase, (b) adenylosuccinate lyase, (c) IMP dehydrogenase, and (d) GMP synthetase C..:l 4 Subsequent analyses revealed the presence of inosine within mammalian tRNAs (7-9). Eight isoaccepting tRNA species contain inosine at position 34 (inosine-34) of the anticodon loop, including: tRNAAla, tRNALeu, tRNAIle, tRNAVal, tRNASer, tRNAPro, tRNAThr, and tRNAArg. (10). The presence of inosine at position 34, the "wobblelt position of the anticodon, greatly expands the codon recognition potential of a particular tRNA due to the ability of inosine to base pair with uridine, cytidine, and adenosine (11). Studies of the biosynthesis of inosinic acid in Escherichia coli (E. coli) tRNA determined that inosine constitutes approximately 0.14% of E. coli tRNA nucleotides, corresponding to approximately 1 inosine residue per 8-9 tRNA molecules (12). The enzymatic mechanism(s) that converts the genomically encoded adenosine at position 34 in yeast tRNA to inosine is consistent with a hydrolytic deamination of the C-6 position of the adenine ring, contrary to a previously reported transglycosylase mechanism whereby the inosine base, hypoxanthine, is inserted with free adenine being released (13,14). Substrate specificity studies of the yeast tRNA:adenosine-34 deaminase indicate that beyond the requisite adenosine-34, the deaminase reaction was independent of a specific nucleotide sequence. However, the efficiency of the deamination of adenosine-34 was dependent on the purine/pyrimidine composition of the anticodon loop and the overall 3D structure (13). Substrates that most efficiently promoted adenosine-34 deamination contained a purine base at position 35 (31 of adenosine-34) and retained base-pairing properties of the proximal stem structure. Certain tRNAs also contain N1-methylinosine. The enzymatic mechanism(s) that creates N1-methylinosine at position 37 appears to be a 5 twostep mechanism in which C-6 of adenine is first deaminated to produce inosine, followed by the subsequent S-adenosylmethionine (SAM)-dependent methylation of the N-l position (10). Although ribosomal RNAs contain a number of modified nucleosides, their diversity is much less than that seen in tRNA (28 of the reported 93 modified nucleosides are found in rRNA) , and they fall within three catagories; 21-O-methylations, base methylations, and numerous pseudouridines. Of these, inosine has only been detected as a methylated ribose form (21-O-methylinosine) (3,15,16). Nucleoside modification of mRNA in eukaryotic cells is even less diverse than that found in rRNA; only 12 of the reported 93 modified nucleosides are found in mRNA (3,4). All mRNA nucleoside modifications detected to date consist of base methylations and 21-O-methylations. Characterizations of these modified nucleosides led to the conclusion that eukaryotic mRNAs contain approximately 6-7 methylated nucleosides per mRNA. These are distributed evenly between base methyl- and 21-0- methylnucleosides with approximately 50% of the methyl groups being found within N6-methyladenosine (m6A). The formation of m6A is a sequence specific, SAM-dependent modification that does not alter the base-pairing properties of adenosine (17,18). The biological function of m6A is unknown but studies have suggested a role in mRNA processing such as maturation and/or mRNA transport (18,19). Other modified nucleosides that have been detected in eukaryotic and viral mRNAs include: the terminal mRNA "cap" structure present on virtually all precursor and mature eukaryotic mRNA molecules (17,19,20), 5- methylcytidine (21,22), N6,2'-0-dimethyladenosine (20), N6,N6,21-O- 6 trimethyladenosine (23), N2, 7 -dimethylguanosine and N2 ,N 2,7 - trimethylguanosine (24), and 3,21-O-dimethyluridine (23). RNA editing is also a cellular mechanism that further modifies mRNA nucleosides. RNA editing is a posttranscriptional process that can alter the coding potential of an mRNA molecule by the insertion or deletion of nucleotides and by the modification of encoded nucleotides (26). RNA editing was first observed as the insertion of multiple uridylate residues into the trypanosome mitochondrial cytochrome oxidase subunit II gene (coxI!) transcript (25). RNA editing is now recognized as an important component of gene expression in eukaryotes. RNA editing mechanisms and the full scope of edited messages have been reviewed (26-28). One form of RNA editing relevant to this dissertation is adenosine deamination, a form of base modification, catalyzed by a class of enzymes that convert adenosine residues to inosine residues within RNA substrates which are largely base-paired (29). In 1987, an activity from Xenopus laevis was discovered that appeared to "unwind" a synthetic RNA molecule that was completely double-stranded (an intermolecular duplex) and this activity was termed an "unwinding" activity (30,31). Experiments aimed at determining the in vitro substrate specificity of the unwinding activity showed that the unwinding activity was inhibited by incubation with excess double-stranded RNA (dsRNA), but not with dsDNA, or a variety of single-stranded RNA or DNA. Subsequent work elucidated that this observed unwinding activity was due to a covalent modification of adenosine residues to inosine residues within the dsRNA substrate and the unwinding activity was renamed the unwinding/modifying activity (32,33). Mechanistic studies using double-stranded RNA substrates and partially purified enzyme from Xenopus extracts showed that the 7 mechanism of adenosine-to-inosine (A-to-I) conversion was consistent with hydrolytic deamination (34). Based on the enzymatic requirement for dsRNA and deamination mechanism, the enzyme was named dsRNA f!denosine ,deaminase, or dsRAD (35). Subsequent work has led to the purification of dsRAD from a variety of organisms and molecular cloning has identified a number of genes encoding dsRAD-related proteins (29). Thus, it has become necessary to standardize the nomenclature and the enzymes are now referred to as f!denosine ,deaminases that act on RNA (ADAR) (36). Based on in vitro enzyme specificity and nucleotide sequence similarity, two distinct enzymes, ADAR1 (dsRAD), and ADAR2, have been identified. Cloned cDNAs that show nucleotide sequence similarity to either ADAR1 or ADAR2 but have yet to be shown to have deaminase activity are referred to as "ADAR-like." ADAR Activity Since the initial detection of ADAR activity in Xenopus laeuis, ADAR activity has been detected ubiquitously throughout the animal kingdom (37). In vitro studies with dsRNA substrates and crude protein extracts identified ADAR activity in a variety of mammalian primary tissues and cell types such as kidney, spleen, brain, and cells of the immune system and indicated that ADAR activity is primarily localized to the nucleus (38). Studies with Xenopus laeuis indicated that ADAR activity is likewise localized to the nucleus, or Germinal Vesicle (GV), in stage VI oocytes. During meiotic maturation and GV breakdown, ADAR activity is released into the cytoplasm where it remains through the eight-cell embryo stage, and subsequently is absent from the cytoplasm of post-mid blastula transition embryos (30,31,39). 8 ADAR activity was shown to be localized to the cytoplasm of follicular cells and unfertilized oocytes of the silkmoth Bombyx mori but localized to the nucleus in a B. mori ovarian-derived cell line (40). These data suggest that cellular localization may playa role in the regulation of ADAR activity. Experiments on short intermolecular RNA duplexes aimed at determining substrate specificity indicated that ADAR1 had no strict sequence specificity but did have a preference for the nucleotide immediately 5' to the modified adenosine (5' nearest neighbor preference) (41). The order of 5' nearest neighbor preference was determined to be A=U>C>G. Furthermore, it was found that the proximity of the duplex termini to the edited adenosine was a determinant in the selection of adenosines for deamination. Purification and Cloning of ADAR1 ADAR1 proteins have been purified from Xenopus laevis, bovine, and chicken. Purified ADAR1 from Xenopus and bovine thymus display an apparent molecular weight of 115 kDa-120 kDa (35,42-44). Purification of ADARl from bovine liver nuclear extracts identified three molecular weight forms of the protein, 93 kDa, 88 kDa, and 83 kDa but evidence from this study suggested that these molecular weight variants may have resulted from partial proteolysis during purification (43). ADAR1 was purified from chicken lung as a 140 kDa protein using, surprisingly, a purification protocol designed to isolate proteins with affinity for Z-DNA (44). In all cases, deamination assays indicated that the purified enzymes alone were capable of converting adenosines to inosines in synthetic dsRNA substrates. 9 ADAR1 cDNAs have been cloned for Xenopus, rat and human, and a partial clone was isolated for the bovine ADAR1 (45-49). The human gene encoding ADAR1 maps to chromosome 1 (50,51). Analyses of full-length ADAR1 clones reveal the presence of putative nuclear localization signals, three conserved but functionally distinct dsRNA binding motifs (dsRBMs), and a C-terminal catalytic domain (deaminase motif). Band shift assays with human ADAR1 and an oligonucleotide assumed to adopt a Z-form DNA structure mapped a DNA binding domain (Za) in the N-terminal region of the protein (52). Interestingly, ADAR1 cDNA was also isolated in a screen for interferon (lFN)-regulated mRNAs in human cell lines. The encoded protein was shown to be constitutively expressed as a 110 kDa protein in the nucleus of human cell lines: a larger, IFN-inducible 150 kDa form, was shown to be expressed both in the cytoplasm and nucleus (48). Southern blot analyses reveal that human ADAR1 is a single-copy gene. Characterization of human ADAR1 genomic clones show that three splice variants encoding active deaminases exist: hADAR1a, whose nucleic acid sequence corresponded to the previously identified human cDNAs, hADAR1b which contains a 26-amino acid deletion, and hADAR1c which contains, in addition to the deletion found in hADAR1b, a 19-amino acid deletion (47). Northern blot analyses of human ADAR1 reveal a single transcript, in all tissues examined, of approximately 6.7-7.5 kb (46,48,49). Human ADAR1 shows consistently high expression in brain and lung tissue, and low expression in skeletal muscle, with discrepancies between studies regarding the level of ADAR1 expression in heart tissue. Although rat ADAR1 cDNA has been cloned, there are no corresponding southern analyses 10 to determine whether rat ADARI is a single- or multicopy gene. In addition, northern analyses were not performed to determine the expression pattern of rat ADARl. Chapter III of this dissertation presents the first northern analyses of rat ADARl, which revealed three transcripts of7.0 kb, 5.5 kb, and 4.3 kb in all tissues examined. Rat ADAR1 expression was observed to be highest in brain tissue, intermediate in lung and heart, and lowest in skeletal muscle and testis. It has yet to be determined whether these multiple transcripts arise from alternative splicing of a single ADAR1 transcript or transcription from different rat ADAR1 genes. Xenopus clones revealed two distinct classes of cDNAs, both of which encode active deaminases (45). Northern analyses indicated that these distinct cDNAs, xADAR1.1 and xADAR1.2 each express alternatively spliced mRNAs. Xenopus ADARl.1 expresses a 5.3 kb and a 4.3 kb transcript, and xADAR2 expresses a 4.3 kb and a 3.8 kb transcript. Purification and Cloning of ADAR2 ADAR2 cDNAs have been cloned from rat, mouse, and human, and the human gene maps to chromosome 21 (53-57). Characterization of ADAR2 cDNAs shows the presence of a conserved C-terminal deaminase domain, two dsRBMs, and a shorter N-terminal as compared to ADAR1. Characterization of rat, mouse, and human ADAR2 cDNA clones all reveal the presence of alternatively spliced variants (36,53,56,58). At least two isoforms exist in mouse and four to eight isoforms in rat: rADAR2a, rADAR2b, rADAR2c, and rADAR2d (8. Rueter and R. Emeson unpublished results) Four isoforms of hADAR2 exist: hADAR2a, hADAR2b, hADAR2c, and hADAR2d (54-56). Two of these isoforms, hADAR2b and hADAR2c, contain a 40-amino acid repeat 11 sequence (Alu cassette) located in the deaminase domain. All four recombinately-expressed hADAR2 isoforms showed deaminase activity when tested on fully double-stranded synthetic RNA substrates, although greater deaminase activity is exhibited by the hADAR2a and hADAR2b isoforms. Deaminase assays with a synthetic RNA that was not fully double-stranded, but that did contain a putative endogenous editing site, indicated that hADAR2c and hADARd were inactive on this substrate while hADAR2a and hADAR2b edited these substrates efficiently. Northern analyses show that rat ADAR2 is expressed in brain and peripheral tissues as one major transcript of approximately 7 kb. Additional tissue specific transcripts were identified in brain (-9 kb) and testis (-3 kb) (53). Relative expression of rADAR2 is highest in brain and lung, intermediate in heart, spleen, liver, testis, and kidney, and lowest in skeletal muscle. Northern analyses of human adult tissue show an hADAR2 expression pattern different to that seen for rADAR2. Although hADAR2 expression was also highest in brain and lowest in skeletal muscle, there is relatively little, to no, detectable expression in lung, liver, or kidney (54). However, northern analyses of human fetal tissue, in contrast to adult tissues, show significant hADAR2 expression in lung, liver, and kidney (55,56). All northern analyses presented were performed on commercial RNA blots and no indication was given as to whether the results of these northern analyses were reproducible. Experience in our laboratory with commercial RNA blots such as these indicated that the integrity of the blotted RNA varies substantially between individual blots. 12 ADAR2 protein purified from a human cell line displays an apparent molecular weight of 90 kDa and like ADAR1 is capable of converting adenosines to inosines in synthetic dsRNA substrates. (59,60). Cloning of ADAR-Like Genes A third member of the ADAR family, RED2, was cloned from rat brain based on sequence similarity to rADAR2 (previously named RED1). Characterization of the RED2 cDNA indicate that RED2 is more similar to rADAR2 than to rADAR1, and like rADAR2, RED2 contains two dsRBMs and a highly conserved C-terminal catalytic domain. Compared to rADAR2, RED2 contains a 54 amino acid extended N-terminus that includes an arginine-rich motif. Northern analyses revealed the brain-specific expression of a major transcript of 8.5 kB and a minor transcript of 4.4 kB (58). In contrast to ADAR1 and ADAR2, recombinantly expressed RED2 displayed no deaminase activity on either a completely base-paired synthetic dsRNA substrate or a synthetic RNA containing putative in vivo editing sites. Human RED2 maps to chromosome 10, highlighting the fact that all three putative human ADAR family members map to different chromosomes (61). Two other cloned cDN As show sequence similarity to the dsRBM and the catalytic domain of ADARs suggesting a relationship to the ADAR family. The first, Tenr, was cloned from a mouse testis expression library based on the ability of the protein to bind to an RNA probe corresponding to a portion of the 3'-UTR of a germ cell-specific mRNA (62). Characterization of the Tenr cDNA sequence revealed a putative -70 kDa gene product that displayed 54% amino acid sequence similarity with the C-terminal portion of xADAR1 (45). The C-terminal sequence of Tenr showed a 51% amino acid similarity to the 13 C-terminus of hADARl (48). Northern analyses indicated that Tenr was expressed exclusively in testis. The second, T20H4.4, was originally identified by the Caenorhabditis elegans genome project as an open reading frame and the T20H4.4 cDNA was subsequently cloned (Hough et aI, manuscript in preparation). The T20H4.4 mRNA encodes a 55 kDa protein which displays approximately 62% amino acid sequence similarity with the C-terminal region of xADARl and hADARl (45,48). Although sequence similarity suggests that Tenr and T20H4.4 are related members of the ADAR family, it has not yet been shown that either protein has deaminase activity. Candidate ADAR-Edited RNAs The ubiquitous expression of ADAR activity and the broad tissue expression of ADAR mRNAs suggest the importance of adenosine deamination in the regulation of gene expression in eukaryotes; however prior to the work described in this thesis, inosine had not been detected within nucleotides derived from cellular mRNA (3). In addition, there are surprisingly few endogenous transcripts thought to be edited by ADARs. As in most cases of RNA editing, the initial identification of these endogenous substrates resulted from a comparative analysis of the genomic DNA sequence with the corresponding cDNA sequence. ADAR editing of cellular transcripts results in a guanosine residue in the cDNA at the position of an adenosine in the genomic sequence. Early studies revealed that inosine, like guanosine, preferentially base-pairs with cytidine (63,64). In the cases of RNA editing by adenosine deamination, inosine in the RNA transcript will preferentially base pair with cytidine during reverse transcription, resulting in cytidine incorporation during first-strand cDNA synthesis, and guanosine 14 incorporation at this site during second-strand cDNA synthesis. Sequence comparison of cDNA and genomic clones then reveals this adenosine-to­guanosine (A-to-G) transition. Comparative DNA analyses of viral RNAs also identified a number of viral transcripts thought to be edited by ADARs based on uri dine to cytidine transitions (U-to-C) as well as A-to-G transitions (29). For example, certain viruses contain an RNA genome that replicates through an RNA intermediate (antigenome) (65). ADAR editing of the antigenome would result in an A-to-G change in the antigenome that, after replication, results in a U-to-C change in the genome. Based on these comparative DNA analyses, 10 cellular RNAs and eight viral RNAs are thought to be edited by adenosine deamination in vivo, but as mentioned, the presence of inosine within these endogenous transcripts has not been directly observed and it was generally believed that inosine is not found within mRNA (3,29). The best evidence for the presence of inosine within an endogenous mRNA came from the elegant development of a method which was designed to specifically cleave RNA after inosine (66). Application of this method to a cellular candidate ADAR substrate results in the specific cleavage at two known editing sites. Although this study does not rule out the possibility of the presence of a modified nucleotide other than inosine at these edited sites, inosine most easily satisfies the criteria required for cleavage. This method provides the first evidence that an inosine-specific cleavage method recognizes specific sites in candidate endogenous ADAR-edited RNAs. A review of the cellular and viral RNAs thought to undergo editing by ADARs reveals two groups of edited RNAs based on the extent of A-to-G (or U-to-C) changes in a defined region of the cDNA sequence. The first group 15 shows editing at <10% of the total adenosines in a given sequence and is referred to as selective deamination. The second group includes sequences that show a substantially greater amount of editing, in some cases -50% of the adenosines, and is termed hypermutation deamination (29). All but one cellular mRNA thought to be edited by ADARs fall within the selective deamination group of substrates. The exception, as discussed below, is found in Drosophilia melanogaster. A mammalian example of a hypermutated cellular mRNA has yet to be identified (67). In contrast, all but one viral RNA thought to be edited by ADARs fall within the hypermutation deamination group of candidate substrates. The exception to this group is found in hepatitis delta virus and is discussed below (68). Putative Inosine-Containing Cellular RNAs The best-characterized examples of selective deamination in mammals occur on transcripts encoding the ionotropic glutamate receptors (glur). Ionotropic glutamate receptors are ligand-gated ion channels which mediate neuronal responses to glutamate, the principal excitatory neurotransmitter in mammalian brain. The ionotropic glutamate receptors have been classified into three main groups based on receptor-gating in response to selective pharmacological agonists: N-methyl-D-aspartate (NMDA), a-amino-3- hydroxy-5-methyl-4-isoxazole propionic acid (AMPA), and kainic acid (KA). Molecular cloning experiments have subdivided these groups into 16 subunits (69). Selective deamination of certain glutamate receptors leads to a variety of important consequences including altered channel gating properties, ion permeability, and agonist-induced desensitization kinetics (70). 16 The AMPA class of receptors are comprised of four subunits, glur-a, glur-b, glur-c, and glur-d, and are characterized by low Ca++ permeability. Functional characterization and sequence analyses of AMPA receptors identified a site specific amino acid residue, arginine (R:CGG codon), in the glur-b subunit as responsible for the low Ca++ permeability observed in recombinant heteromeric channels (71,72). Recombinant AMPA channels retained substantial permeability to Ca++ when expressed in the absence of glur-b, and sequence analyses of glur-a and glur-c revealed a glutamine (Q:CAG codon) at this Ca++ "permeability-determining" amino acid position. Genomic DNA analyses revealed that each AMP A subunit is encoded by a single gene and each subunit gene contains a glutamine codon at this position, indicating this QIR codon change (CAG to CGG) results from RNA editing by adenosine deamination of the glur-b transcript (73). Sequence analyses indicate that >99% of glur-b mRNAs are edited. Additional comparative DNA sequence analyses revealed another RNA editing event that generates a glycine codon (G;GGG codon) from a genomically-encoded arginine (R;AGG codon) in the glur-b, glur-c, and glur-d subunits, termed the RiG site (74,75). Comparative DNA analyses also show QIR site editing of certain KA mRNAs (gluR-5 and glur-6) and additional editing sites in these mRNAs as well (76). Molecular cloning and mutagenesis analyses revealed that RNA editing of the AMPA and KA receptor transcripts at the QIR and RIG sites is mediated by, and dependent upon, sequences present in the downstream intron which are complementary to exon sequences surrounding the editing sites. These sequences are termed editing site complementary sequences (ECS) (77-79). Additional A-to-G conversions were identified in the intronic cDNA sequences where the corresponding pre-RNA was proposed 17 to be involved in base-pairing. These results revealed that RNA editing occurred on the glur pre-mRNAs and accounted for the lack of adenosine to inosine conversion seen with glur-b cDNA transcripts when incubated with Xenopus nuclear extracts (M. Paul and B. Bass, unpublished results). In vitro experiments revealed that editing of the QIR and RIG sites proceeded by adenosine to inosine conversion (80-83). The QIR site is edited most efficiently by ADAR2, while the RIG site is edited by ADARl and ADAR2 with equal efficiences (53,59,60). RNA editing by selective deamination has been observed with a second neurotransmitter receptor mRNA, the serotonin, or 5-hydroxytryptamine (5- HT), receptor subtype 2C (5-HT2C) mRNA. Unlike the ionotropic glutamate receptors, serotonin receptors belong to the G protein-coupled receptor family. Comparative sequence analyses identify four editing sites within the second intracellular loop of this receptor that result in potentially two isoleucine-to­valine (IN) changes and an asparagine-to-serine (N/S) amino acid change (84). Quantitative cDNA analyses reveal that seven protein isoforms are expressed in a brain region-specific manner. However, functional studies of edited 5-HT2C isoforms determined that only the fully edited 5-HT2C isoform (VSV) exhibited reduced G protein coupling. As observed with glutamate receptor editing, intron sequences downstream to the editing sites are required for editing. In vitro assays revealed A-to-I conversion with the 5- HT2C mRNA and cotransfection analyses indicated both ADARl and ADAR2 are involved in 5-HT2C editing. The mammalian glycosyltransferase enzyme a2,6-sialyltransferase (ST) transcript is proposed to be edited by selective deamination of a single adenosine. Two ST isoforms that differ at amino acid position 123 were 18 cloned from rat liver eDNA libraries. One form contained a tyrosine at this position (ST tyr; TAG) and the second contained a cysteine residue (ST cys; TGG) (85). Genomic sequence analyses determined the single gene encoding ST in rat contained a tyrosine codon at position 123 (86). The ST isoforms show differences in their catalytic efficiencies, proteolytic processing, and subcellular localization. No deamination assays were performed on ST in vitro transcripts and there was no indication that the ST gene was analyzed for the presence of an ST ECS. The human ST is encoded by a single gene and sequence analyses of the corresponding human genomic clones should facilitate the identification of ECS's involved in ST mRNA editing (87). Two cellular transcripts have been identified in nonmammalian species that are proposed to be edited by ADARs. The first was identified by the molecular cloning of a squid Kv2 potassium (K+) channel. Comparative DNA analyses identified 17 A-to-G transitions within a 360 nucleotide genomic region corresponding to coding sequences (88). Southern analyses confirm the existence of a single gene encoding this K + channel and cDNA analyses of eDNA clones from individual animals reveal the expression of 28 unique edited transcripts with the relative proportion of A-to-G changes at specific positions ranging from 3% to 100%. Functional expression of certain edited constructs corresponding to positions that are edited at high frequencies indicate that selective deamination alters the gating properties of these K+ channels. Like ST, no deamination assays have been performed on the in vitro Kv2 transcripts and no corresponding ECS has been identified, although the genomic sequence was cloned. The second nonmammalian candidate ADAR substrate was identified in Drosophilia and provided the only example to date of a cellular transcript 19 edited by hypermutation deamination. The 4f-rnp gene is a single-copy nuclear gene thought to encode two proteins, each containing one RNA recognition domain (67) Comparative DNA analyses reveal hypermutation deamination of 4f-rnp (-40% A-to-G). Putative Inosine-Containing Viral RNAs The term biased hypermutation has been used to describe the numerous nucleotide conversions that occur in certain viral RNAs. Numerous U-to-C and A-to-G conversions have been observed in certain viral transcripts and these conversions are thought to be the result of hypermutation deamination (29). The potential role of ADARs in biased hypermutation was first recognized with the identification of numerous U-to­C and A-to-G conversions in cDNAs derived from the measles virus matrix (M) gene in viral strains isolated from persistent infections (89). Acute measles infections can develop into persistent viral infections known as subacute sclerosing panencephalitis (SSPE) and measles inclusion body encephalitis (MIBE). Analyses of an M gene cDNA sequence isolated from a MIBE patient indicated 50% U-to-C transitions and 2% A-to-G transitions. It was predicted that these transitions were the result of ADAR editing of double-strand RNA structures formed during viral transcription and replication (90). Subsequent analyses of other SSPE patients revealed substantial hypermutation ofM gene cDNAs in these strains as well (91). Hypermutation deamination editing by ADARs has also been suggested for the numerous conversions seen in vesicular stomatitis virus, human parainfluenza virus, respiratory synctial virus, and in an avian retrovirus (92-96). In these cases the extent ofU-to-C and A-to-G transitions 20 range from -20%-50%. Hypermutation deamination editing is also proposed to produce the extensive A-to-G transitions observed in cDNAs derived from the early-strand transcripts of polyoma virus (97). ADAR editing of polyoma viral RNAs and experiments aimed at determining the presence of inosine in cells infected with polyoma virus is discussed in Chapter IV of this dissertation. Selective deamination editing has been observed in hepatitis delta virus (HDV). The HDV genome is a closed circular RNA that replicates through an RNA antigenome. Two forms of the HDV protein (HDV antigen or HD-Ag) are produced by RNA editing of the antigenome. The sequence change is passed to the genome during replication to allow for two proteins to be synthesized from a single open reading frame. The unedited HDV genome produces the smaller HD-Ag and selective deamination converts an amber stop codon (VAG) to a tryptophan codon (VGG) resulting in the larger form of HD-Ag (68). This site, consistent with editing nomenclature for glutamate receptors, was termed the amberlW site. In vitro assays in which antigenomic RNA transcripts were incubated with xADARl revealed A-to-I conversion. As both forms of HD-Ag are necessary in viral replication and packaging, selective deamination by ADARs seems to playa central role in the lifecycle ofHDV. 21 References 1. Zalkin, H. and Dixon, J. E. (1992) Prog. Nucleic Acid Res. Mol. Biol., 42, 259-287. 2. Blackburn, G. M. and Gait, M. J. (1996) In Blackburn, G. M., and Gait, M. J. (eds.), Nucleic Acids in Chemistry and Biology. Second Edition Oxford University Press, New York, pp. 147-174. 3. Limbach, P., Crain, P. and McCloskey, J. (1994) Nucleic Acids Res., 22(12), 2183-2196. 4. Crain, P. F. and McCloskey, J. A. (1996) Nucleic Acids Res., 24, 98-99. 5. Grosjean, H., Bjork, G. and Maden, B. E. H. (1995) Biochimie, 77, 3-6. 6. Nishimura, S. (1979) In Schimmel, P. R., SolI, D., and Abelson, J. N. (eds.), Transfer RNA: Structure, Properties, and Recognition. Cold Spring Harbor Laboratory, Cold Spring Harbor, pp. 59-79. 7. Hall, R. H. (1963) Biochem Biophys Res Commun, 13,394-398. 8. Holley, R. W., Everett, G. A., Madison, J. T. and Zamir, A. (1965) J. Biol. Chem., 240,2122-2127. 9. Staehelin, M., Rogg, H., Bagulay, B. C., Ginsberg, T. and Wehrli, W. (1968) Nature, 219, 1363-1365. 10. Grosjean, H., Auxilien, S., Constantinesco, F., Simon, C., Corda, Y., Becker, H. F., Foiret, D., Morin, A., Jin, Y. X., Fournier, M. and Fourrey, J. L. (1996) Biochimie, 78, 488-501. 11. Watson, J. D., Hopkins, N. H., Roberts, J. W., Steitz, J. A. and Weiner, A. M. (1987) In Gillen, J. R. (ed.), Molecular Biology of the Gene. Fourth Ed. The Benjamin/Cummings Publishing Company, Inc., Menlo Park, pp. 431-462. 12. Kammon, H. O. and Spengler, S. J. (1970) Biochim. et Biophys. Acta 213,352-364. 13. Auxilien, S., Crain, P. F., Trewyn, R. W. and Grosjean, H. (1996) J. Mol. Biol., 262,437-458. 14. Elliott, M. S. and Trewyn, R. W. (1984) J. Biol. Chem., 259(4), 2407- 2410. 15. Gray, M. W. (1976) Nucleic Acids Research, 3, 977-988. 22 16. Maden, B. E. H. (1990) Progress in Nucleic Acid Research, 39, 241-303. 17. Rottman, F., Shatkin, A. J. and Perry, R. P. (1974) Cell, 3, 197-199. 18. Tuck, M. T. (1992) Int. J. Biochem., 24(3), 379-386. 19. Narayan, P. and Rottman, F. M. (1992) Advances in Enzymology & Related Areas of Molecular Biology, 65, 255-85. 20. Wei, C.-M., Gershowitz, A. and Moss, B. (1975) Cell, 4, 379-386. 21. Sommer, S., Salditt-Georgieff, M., Bachenheimer, S., Darnell, J. E., Furuichi, Y., Morgan, M. and Shatkin, A. J. (1976) Nucleic Acids Research, 3(3), 749-765. 22. Dubin, D. T., Stollar, V., Hsuchen, C.-C., Timko, K. and Guild, G. M. (1977) Virology, 77, 457-470. 23. Bangs, J. D., Crain, P. F., Hashizume, T., McCloskey, J. A. and Boothroyd, J. C. (1992) J. Biol. Chem., 267(14), 9805-9815. 24. HsuChen, C.-C. and Dubin, D. T. (1976) Nature, 264, 190-191. 25. Benne, R., Van den Burg, J., Brakenhoff, J. P., Sloof, P., Van Boom, J. H. and Tromp, M. C. (1986) Cell, 46(6), 819-826. 26. Bass, B. L. (1993) In Gesteland, R., and Atkins, J. (eds.), The RNA World. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, pp. 383-418. 27. Benne, R. (1996) Current Opinion in Genetics and Development, 6,221- 231. 28. Smith, H. C., Gott, J. M. and Hanson, M. R. (1997) RNA, 3, 1105-1123. 29. Bass, B. L. (1997) Trends Biochem. Sci., 22, 157-162. 30. Bass, B. L. and Weintraub, H. (1987) Cell, 48(4),607-613. 31. Rebagliati, M. R. and Melton, D. A. (1987) Cell, 48(4), 599-605. 32. Bass, B. L. and Weintraub, H. (1988) Cell, 55(6), 1089-1098. 33. Wagner, R. W., Smith, J. E., Cooperman, B. S. and Nishikura, K. (1989) Proc Natl Acad Sci USA, 86(8), 2647-2651. 34. Polson, A. G., Crain, P. F., Pomerantz, S. C., McCloskey, J. A. and Bass, B. L. (1991) Biochemistry, 30(49), 11507-11514. 23 35. Hough, R. F. and Bass, B. L. (1994) J Biol Chem, 269(13), 9933-9939. 36. Bass, B. L., Nishikura, K., Keller, W., Seeburg, P. H., Emeson, R. B., O'Connell, M. A., Samuel, C. E. and Herbert, A. (1997) RNA, 8(9),947- 949. 37. Bass, B. L. (1992) Seminars in Developmental Biology, 3,425-433. 38. Wagner, R. W., Yoo, C., Wrabetz, L., Kamholz, J., Buchhalter, J., Hassan, N. F., Khalili, K., Kim, S. D., Perussia, B., McMorris, F. A. and et al. (1990) Molecular & Cellular Biology, 10(10), 5586-5590. 39. Saccomanno, L. and Bass, B. L. (1994) Mol. Cell. Biol., 14(8),5425-32. 40. Skeiky, Y. A. W. and Iatrou, K. (1991) Journal of Molecular Biology, 218,517-527. 41. Polson, A. G. and Bass, B. L. (1994) EMBO J., 18(23),5701-11. 42. O'Connell, M. A. and Keller, W. (1994) Proc Natl Acad Sci USA, 91(22), 10596-10600. 43. Kim, D., Garner, T. L., Sanford, T., Speicher, D., Murray, J. M. and Nishikura, K. (1994) J. Biol. Chem., 269(18), 13480-13489. 44. Herbert, A., Lowenhaupt, K., Spitzner, J. and Rich, A. (1995) Proc Natl Acad Sci USA, 92(16), 7550-4. 45. Hough, R. F. and Bass, B. L. (1997) RNA, 3, 356-370. 46. O'Connell, M. A., Krause, S., Higuchi, M., Hsuan, J. J., Totty, N. F., Jenny, A. and Keller, W. (1995) Mol. Cell. Biol., 15(3), 1389-1397. 47. Liu, Y., George, C. X., Patterson, J. B. and Samuel, C. E. (1997) J Biol Chem, 272(7),4419-4428. 48. Patterson, J. B. and Samuel, C. E. (1995) Mol Cell Biol, 15(10), 5376- 5388. 49. Kim, D., Wang, Y., Sanford, T., Zeng, Y. and Nishikura, K. (1994) Proc. Natl. Acad. Sci. USA, 91(24), 11457-11461. 50. Weier, H. D., George, C. X., Greulich, K. M. and Samuel, C. E. (1995) Genomics, 30(2), 372-5. 51. Wang, Y., Zeng, Y., Murray, J. M. and Nishikura, K. (1995) J. Mol. Biol., 254(2), 184-95. 24 52. Herbert, A., AIfken, J., Kim, Y.-G., Mian, 1. S., Nishikura, K. and Rich, A. (1997) Proc. Natl. Acad. Sci. USA, 94,8421-8426. 53. Melcher, T., Maas, S., Herb, A., Sprengel, R., Seeburg, P. H. and Higuchi, M. (1996) Nature, 379,460-464. 54. Gerber, A., O'Connell, M. A. and Keller, W. (1997) RNA, 3(5),453-63. 55. Lai, F., Chen, C.-X., Carter, K. C. and Nishikura, K. (1997) Mol Cell Biol, 17,2413-2424. 56. Mittaz, L., Scott, H., Rossier, C., Seeburg, P., Higuchi, M. and Antonarakis, S. (1997) Genomics, 41, 210-217. 57. Villard, L., Tassone, F., Haymowicz, M., Welborn, R. and Gardiner, K. (1997) Somatic Cell and Molecular Genetics, 23(2), 135-145. 58. Melcher, T., Maas, S., Herb, A., Sprengel, R., Higuchi, M. and Seeburg, P. H. (1996b) J Biol Chem, 271(50), 31795-31798. 59. O'Connell, M. A., Gerber, A. and Keller, W. (1997) J Biol Chem, 272, 473-478. 60. Yang, J.-H., Sklar, P., Axel, R. and Maniatis, T. (1997) Proc. Natl. Acad. Sci. USA, 94, 4354-4359. 61. Mittaz, L., Antonarakis, S. E., Higuchi, M. and Scott, H. S. (1997) Human Genetics, 100,398-400. 62. Schumacher, J. M., Lee, K., Edelhoff, S. and Braun, R. E. (1995) Biology of Reproduction, 52(6), 1274-83. 63. Speyer, J. F., Lengyel, P., Basilio, C. and Ochoa, S. (1962) Proc. Natl. Acad. Sci. USA, 48,441-448. 64. Basilio, C., Wahba, A. J., Lengyel, P., Speyer, J. F. and Ochoa, S. (1962) Proc. Natl. Acad. Sci. USA, 48, 613-616. 65. Darnell, J., Lodish, H. and Baltimore, D. (1986), Molecular Cell Biology. Scientific American Books, Inc., New York, pp. 208-218. 66. Morse, D. P. and Bass, B. L. (1997) Biochemistry, 36(28), 8429-8434. 67. Petschek, J. P., Mermer, M. J., Scheckelhoff, M. R., Simone, A. A. and Vaughn, J. C. (1996) J. Mol. Biol., 259(5),885-90. 68. Polson, A. G., Bass, B. L. and Casey, J. L. (1996) Nature, 380, 454- 456. 25 69. Seeburg, P. H. (1993) Trends in Neurosciences, 16(9), 359-365. 70. Seeburg, P. H. (1996) J. Neurochem., 66(1), 1-5. 71. Verdoon, T. A., Burnashev, N., Monyer, H., Seeburg, P. H. and Sakmann, B. (1991) Science, 252(1715-1718). 72. Hume, R. 1., Dingledine, R. and Heinemann, S. F. (1991) Science, 253, 1028-1031. 73. Sommer, B., Kohler, M., Sprengel, R. and Seeburg, P. H. (1991) Cell, 67(1), 11-19. 74. Kohler, M., Kornau, H.-C. and Seeburg, P. H. (1994) J. Biol. Chem., 269, 17367-17370. 75. Lomeli, H., Mosbacher, J., Melcher, T., Hoger, T., Geiger, J. R., Kuner, T., Monyer, H., Higuchi, M., Bach, A. and Seeburg, P. H. (1994) Science, 266, 1709-1713. 76. Kohler, M., Burnashev, N., Sakmann, B. and Seeburg, P. H. (1993) Neuron, 10, 491-500. 77. Higuchi, M., Single, F. N., Kohler, M., Sommer, B., Sprengel, R. and Seeburg, P. H. (1993) Cell, 75(7), 1361-70. 78. Egebjerg, J., Kukekov, V. and Heinemann, S. F. (1994) Proc. Natl. Acad. Sci. U. S. A., 91(22), 10270-10274. 79. Herb, A., Higuchi, M., Sprengel, R. and Seeburg, P. H. (1996) Proc. Natl. Acad. Sci. U. S. A., 93(5), 1875-80. 80. Hurst, S. R., Hough, R. F., Aruscavage, P. J. and Bass, B. L. (1995) RNA, 1(10), 1051-60. 81. Melcher, T., Maas, S., Higuchi, M., Keller, W. and Seeburg, P. H. (1995) J. Biol. Chem., 270(15), 8566-70. 82. Rueter, S. M., Burns, C. M., Coode, S. A., Mookherjee, P. and Emeson, R. B. (1995) Science, 267(5203), 1491-4. 83. Yang, J. H., Sklar, P., Axel, R. and Maniatis, T. (1995) Nature, 374(6517),77-81. 84. Burns, C. M., Chu, H., Rueter, S. M., Hutchinson, L. K., Canton, H., Sanders-Bush, E. and Emeson, R. B. (1997) Nature, 387,303-308. 85. Weinstein, J., Lee, E. U., McEntee, K., Lai, P.-H. and Paulson, J. C. (1987)J. Biol. Chem., 262, 17735-17743. 26 86. Ma, J., Qian, R., Rausa, F. M. I. and Colley, K. J. (1997) J. Biol. Chern., 272(1),672-679. 87. Wang, X., Vertino, A., Eddy, R., Byers, M. G., Jani-Sait, S. N., Shows, T. B. and Lau, J. T. Y. (1993) J. Biol. Chern., 268(6), 4355-4361. 88. Patton, D. E., Silva, T. and Bezanilla, F. (1997) Neuron, 19, 711-722. 89. Cattaneo, R., Schmid, A., Eschle, D., Baczko, K., ter-Meulen, V. and Billeter, M. A. (1988) Cell, 55(2), 255-65. 90. Bass, B. L., Weintraub, H., Cattaneo, R. and Billeter, M. A. (1989) Cell, 56(3),331. 91. Cattaneo, R. (1994) Curro Opin. Genet. Deuel., 4(6), 895-900. 92. O'Hara, P. J., Nichol, S. T., Horodyski, F. M. and Holland, J. J. (1984) Cell, 36(4), 915-924. 93. Rueda, P., Garcia-Barreno, B. and Melero, J. A. (1994) Virology, 198, 653-662. 94. Felder, M. P., Laugier, D., Yatsula, B., Dezelee, P., Calothy, G. and Marx, M. (1994) J. Virol., 68(8), 4759-67. 95. Hajjar, A. M. and Linial, M. L. (1995) J. Viral., 69(9), 5878-82. 96. Kim, T., Mudry, R. A. J., Rexrode, C. A. I. and Pathak, V. K. (1996) J. Virology, 70(11), 7594-7602. 97. Kumar, M. and Carmichael, G. G. (1997) Proc. Natl. Acad. Sci. U.S.A., 94, 3542-3547. CHAPTER 2 PURIFICATION OF 51 IMP BY THIN LAYER CHROMATOGRAPHY 28 Introduction It has been known for more than 35 years that inosine exists within mature tRNA, but prior to work presented in this thesis, inosine had not been directly observed within nucleotides derived from cellular messenger RNA (mRNA) (1). With the discovery of a number of adenosine deaminases that act on RNA (ADARs) and the growing list of RNAs thought to be deaminated by ADARs in vivo, the question remained as to why inosine had not been detected in cellular mRNA. The likely answer to this question is that there are no established protocols for the purification of inosine from cellular mRNA. In protein purification, successive rounds of purification are often required before a specific cellular protein of interest is purified away from more abundant cellular proteins and can be visualized by standard methods, for example by Comassie blue staining. By analogy, before inosine within cellular mRNA can be detected, it will be necessary to purify it away from the more abundant cellular ribonucleotides. We have developed a thin layer chromatography (TLC)-based inosine purification protocol that allowed for the sensitive detection of radiolabeled inosine monophosphate (IMP) within the background of the more abundant cellular ribonucleotides. TLC has traditionally been used for the separation and detection of modified nucleotides in tRNA (2-5). In addition, there are a large number of alternative methods for the identification and quantitation of modified nucleosides from RNA such as reversed-phase high-performance liquid chromatography (RP-HPLC) and liquid chromatography-mass spectrometry (LCIMS) (6-8). We chose to purify inosine by TLC for a number of reasons: first, TLC allowed for the simultaneous analysis of multiple samples; second, with TLC there was no contamination of samples between column runs, and 29 third, the use of [32p] greatly increased the sensitivity of detection, and the radioactivity was easily discarded which precluded radioactive contamination of chromatographic apparatuses such as HPLC columns. Results and Discussion TLC Purification of Inosine 5' Monophosphate Most of the TLC systems used to identify modified nucleotides in transfer RNA (tRNA) were developed for the analysis of radiolabeled 5' nucleoside monophosphates (5' NMPs) and thus, we chose to base our novel inosine purification method on the analysis of [32P]-labeled 5' NMPs (5' [32p]NMPs). The general scheme for the labeling and purification of 5' [32PHMP starting with a mixture of 3' NMPs is outlined in Figure 2.1. For the control experiments, we used commercial 3' NMPs as the starting material. The 3' NMPs were 5' end-labeled with T4 polynucleotide kinase (PNK) and ,),_[32P]ATP. The resultant radiolabeled nucleoside 3',5' bisphosphates (*pNp in Figure 2.1) were digested with nuclease PI to yield [32P]-labeled 5' NMPs (*pN in Figure 2.1). Conclusive identification of a modified nucleotide has traditionally been based on the chromatographic comigration with corresponding UV markers (4). Thus, a key feature of our purification procedure was the addition of nonradioactive 5' IMP to the radioactive 5' NMP mixture (see circle in Figure 2.1) which allowed us to monitor the migration of 5' IMP by UV absorbance at all stages of purification. 3' nucleoside monophosphates (3' NMPs or Np) Np I Label 5' end with T4 PNK t and 'Y-[32p]ATP *pNp *pG *pC tI Digest with Nuclease PI & add nonradioactive 5' IMP (pI) *pA *pU pI *pI tI Psuucrcifeiscsaivtieo nT LoCf 5 s' tIeMpsP by *pI + pI 30 Figure 2.1. General scheme for the purification of5' IMP. For control reactions, commercial 3' NMPs (Nps) were 5'-labeled with 'Y-[32p]ATP and T4 PNK The 3',5'-nucleoside bisphosphates (*pNp) were digested to 5' [32p]NMPs (*pN) with Nuclease Pl. Asterisk indicates the 5' [32Pl-label. Nonradioactive 5' IMP (pI) was added to the radiolabeled 5' NMP mixture and 5' [32P1IMP was purified by successive TLC steps. 31 Two-dimensional (2D) TLC has routinely been used for the separation of the numerous tRNA modified nucleotides (3-5). We tried three different methods for the purification of IMP, each of which included a 2D-TLC separation as purification step one. Nomenclature for the various purification methods is described in Materials and Methods. For example, purification method A, step one, is referred to as A.1, purification method A, step two, is referred to as A.2, and purification method A, step three, is referred to as A.3 and likewise for method B and method C. A.1 and B.1, shown in Figure 2.2A and 2.2B, used an anion-exchange cellulose matrix, polyethyleneimine-impregnated cellulose (PEl-Cellulose), for the separation of the nucleotide mixture. Prior to TLC analysis of the radiolabeled 51 NMP mixture, nonradioactive 51 AMP, 51 GMP, 51 CMP, 51 UMP, and 5' IMP UV markers were analyzed in each buffer system to determine the relative migration, or Rf values, of each 51 NMP (data not shown). Control experiments typically started with 125-250 finol each of 31 AMP, 31 GMP, 31 CMP, and 31 UMP (3' NMPs) and positive control experiments typically included 5-10 fmol of 31 IMP. For negative controls, IMP was left out of the reactions. When IMP was present, the ratio between the NMPs and IMP was held at a constant ratio of 100 parts NMPs to 1 part IMP. Both A.1 (Figure 2.2A) and B.1 (Figure 2.2B) efficiently separated 51 AMP, 5' GMP, 51 CMP, and 5' UMP (5' NMPs). Exposure of the plates to UV light allowed for the detection of nonradioactive 5' IMP that was outlined with a dotted circle (see Figure 2.2A-B). Although both methods separated the 5' NMPs, 5' IMP comigrated with 51 GMP in A.1 and with an unknown spot of radioactivity in B.2 (denoted by an asterisk in Figure 2.2B). A pU *, .~ pG/pI _ I '" . ~ 20t 10 B pA pG pC ' pA J. ~ '. - ...... ·1 • • 10 pU pC pI * I . "ATP" Figure 2.2A-B. Autoradiograms of step one ofTLC purification methods A and B. Relative migration ofnucleotides are shown to the right and left of each plate. The directions of each dimension of 2D-TLC development are shown in the bottom left corner of each plate. TLC plates were exposed to film for -30 seconds to 2 minutes. Migration of nonradioactive 5' IMP is outlined with a dotted circle and asterisks denote radioactive species of unknown identity. Only the auto radiograms for samples derived from 500 fmol NMPs (+) 5 finol IMP are shown. (A) Purification Method A: Unincorporated y_[32p]-ATP from the PNK reaction is labeled ATP. (B) Purification Method B: Unincorporated y_[32pJ-ATP from the PNK reaction is bracketed and labeled "ATP." C;..J t\:) 33 This radioactive spot did not comigrate with any of the known standards and most likely was a side-product of the labeling reaction. A significant advantage to A.1 was the fact that the unreacted y_[32P]ATP from the PNK reaction remained at the origin; in B.1, this material migrated substantially in both dimensions. As indicated in Figure 2.2B, the y- [32P]ATP from the PNK labeling reaction was comprised of multiple radioactive spots (bracketed and labeled "ATP" in Figure 2.2B). C.1, shown in Figure 2.2C, used a cellulose matrix for the separation of the nucleotide mixture. Although analysis of nonradioactive 5' NMP standards by UV absorption in C.1 control experiments showed distinct spots corresponding to each 5' NMP marker (Figure 2.2D), the C.1 control plate gave uninterpretable results due to a smearing of radioactivity (Figure 2.2C). C.1 is termed a "Nishi plate" and typically, excess y-[32P]ATP from the PNK labeling reaction is reacted with 2 mM glucose and 0.008 units of yeast hexokinase for 10 minutes at 37° C, and subsequently chased with 5 nmol nonradioactive ATP for 20 minutes at 37° C. This reaction drives the y_[32P] from y_[32P]ATP into radioactive glucose 6-phosphate (G-6-P) which migrates well away from any major or modified nucleotide (4). A representative Nishi plate shown in Figure 2.2D contained G-6-P and 5' NMP markers. G-6-P ran well away from the 5' NMP markers (see top left corner of TLC plate). However, in the experiment presented in Figure 2.2C, excess y-[32P]ATP was not treated as described because it was feared that the hexokinase and/or the glucose could be contaminated with nucleotides that would interfere with our analyses. c D pII 2D 11 ~lIIIII':: ~ 10 pU pC * pA pG/pI 2D 1 10 Figure 2.2C-D. Autoradiograms of step one of TLC purification method C and the 5' NMP control plate for method C. Relative migration of nucleotides are shown to the right of each plate. The directions of each dimension of 2D­TLC development are shown in the bottom left corner of each plate. TLC plate in (C) was exposed to film for ,..,30 seconds to 2 minutes. Migration of nonradioactive 5' IMP is outlined with a dotted circle in (C) and asterisk in (D) denotes radioactive species of unknown identity. The autoradiogram for samples derived from 500 fmol NMPs (+) 5 fmol IMP is shown in (C) and the autoradiogram for the control 5' NMPs is shown in CD). (C) Purification Method C: Smearing of radioactivity was likely a result of the migration of unincorporated y_[32pJ-ATP after the PNK reaction. (D) Pi indicates migration of inorganic phosphate and G-6-P indicates migration of glucose-6-phosphate. 5' IMP comigrated with 5' GMP (pG/pl) as determined by UV absorbance of the nonradioactive 5' NMP markers. UJ ~ 35 Thus, the radioactive smear observed in Figure 2.2C was most likely due to the various radioactive species found in the y_[32P]ATP as observed in B.1 (Figure 2.2B). Even with this smearing, exposure of C.1 to UV light allowed for the detection of 5' IMP which was outlined by a dotted circle as with A.1 and B. Control experiments indicated that in C.1, 5' IMP comigrated with 5' GMP as in A.1 (Figure 2.2A). In addition to the smearing observed with C.1, there were significant disadvantages of this system that included the volatility of isobutyric acid, a very foul-smelling solvent, and the length of TLC development (-20 hours total development time). The second purification step in all three methods was a one­dimensional (lD) TLC system that allowed for the analysis of multiple samples on the same TLC plate, a significant, labor-saving advantage. Since 5' IMP comigrated with 5' GMP on both A.1 and C.1, the same solvent system was used to separate 5' IMP from 5' GMP in A.2 and C.2 (see Materials and Methods). Autoradiography of A.2 and C.2, combined with the UV detection of nonradioactive 5' IMP, showed the efficient separation of 5' IMP from 5' GMP as seen in Lanes 1 and 2 in Figure 2.3A (A.2) and Figure 2.3C (C.2). Control samples indicated that ATP does not mi.grate on this system (data not shown) and autoradiography of C.2 indicated some "carry-over" of the y_[32P]ATP from the PNK reaction which remained at the origin (ATP; Figure 2.3C). This confirmed that the smearing seen in C.1 was due in part to y-[32P]ATP. Autoradiographic and UV analysis of B.2 (Figure 2.3B) indicated that the second purification step used in method B efficiently separated 5' IMP from the comigrating unknown radioactivity seen in B.1 (denoted by an asterisk in Figure 2.2B and Figure 2.3B). A 500 fmol NMPs B 500 fmol NMPs c 500 fmol NMPs 5 5 5 .•. \ pI I I . .. ~\ \ \ 1 • ,. \.,,: • I pI f ' l \-..} ! ~ ," , I I \ . I I \. ' • t f ',." .,./ ... '' ,." .1' Jl (\ ' ., . /\ pI ' I I • ' ," . \ \j •\ y'J * G1 P • . • ATP 1 2 1 2 1 2 Figure 2.3. Autoradiograms of step two ofTLC purification methods A, B, and C. Relative migration of nucleotides are shown on the left side of each plate and that for nonradioactive 5' IMP with a dotted circle. ID-TLC development was bottom to top as shown. Samples contained 500 fmol NMPs (+) 5 fmol of IMP or 500 fmol NMPs (-) IMP, as indicated at the top of each figure. (A) Purification Method A. (B) Purification Method B: asterisk denotes radioactive species of unknown identity that partially comigrated with 5' IMP on step one. (C) Purification Method C: ATP indicates the "carry-over" unincorporated y_[32P]-ATP from step one. UJ en 37 Longer autoradiographic exposure of Figures 2.3A, 2.3B, and 2.3C revealed significant background radioactivity which migrated throughout the region where 5' IMP migrated (data not shown). This background radioactivity precluded the detection of 5' IMP and did not allow for efficient quantitation of 5' IMP. Due to this background radioactivity, it was necessary to further purify IMP. Purification step three in all methods was a ID TLC purification and all methods used a cellulose TLC plate and the same buffer system (see Materials and Methods). Again, exposure to UV light allowed for the detection of nonradioactive 5' IMP in each case. Autoradiographic analysis of purification step three allowed for the visualization of 5' [32P]IMP in the positive control reactions (Figure 2.4A-C) To further verify the migration of the purified 5' [32P]IMP, each sample was compared to a 5' [32P]IMP marker (see Lane 1 in Figure 2.4A-C). Importantly, the negative control reactions in A.3 and B.3, which did not contain IMP in the PNK reaction, did not show any detectable radioactivity that comigrated with 5' IMP (see Lane 2 in Figure 2.4A and Lane 3 in 2.4B). In contrast, analysis of C.3 revealed significant radioactivity in the negative control reactions that comigrated with 5- IMP (see Lane 3 :n Figure 2.4c). This result verified that C.3 was not an effective purification and since there were other disadvantages of purification method C as mentioned previously, purification method C was not used in further experiments. Since 5' [32P]IMP could not be visualized until purification step three, only purification step three is shown in all the following TLC results. Unless indicated otherwise, purification steps one and two were performed as described (Figure 2.2A and B and Figure 2.3A and B). A pI ~~ :o'V~ ~'<J-<' <; "-~'b--<' 500 fmol NMPs 5 ~ c:,'V~~~Q,-<' <;"-~~ B pI pA 500 fmol NMPs 5 c pI ~ c:,'V~'\.Q,-<' <; "-~'b--<' 500 fmol NMPs 5 1 2 3 1 2 3 1 2 3 Figure 2.4. Autoradiograms of step three ofTLC purification methods A, B, and C. Relative migration of 5' IMP (pI) is shown to the left of each figure as determined by the migration of nonradioactive 5' IMP (not shown) and the 5' [32PHMP marker shown in Lane 1 of each figure. ID-TLC development was bottom to top as shown. Samples contained 500 fmol NMPs (+) 5 fmol of IMP or 500 fmol NMPs (-) IMP as indicated at the top of each figure. (A) Purification Method A. (B) Purification Method B: Migration of 5' [32pJAMP which partially comigrated with 5' IMP on steps 1 and 2 is indicated to the left of the figure. (C) Purification Method C: Lane three shows contaminating radioactivity that comigrated with 5' IMP. UJ 00 39 Sensitivity of 5' [32Pl Detection After the development of two purification procedures (A and B) which allowed for the selective purification and detection of 5' IMP, experiments were performed to determine the level of 5' [32P]-IMP that could be detected. In an effort to increase the efficiency of the PNK reaction, a variety of reagents known to increase the activity of T4 PNK, such as polyamines, sulfhydryl reagents, and glycogen were included in the reaction (9,10). In contrast to previous results, and contrary to what we expected (Figure 2.4A and B), a number of A.3 and B.3 analyses of control experiments using these new reagents revealed the presence of radioactivity that comigrated with the nonradioactive 5' IMP (data not shown). Experiments were performed to determine the nature of this radioactivity. These experiments revealed that two of the reagents we included in our reactions, spermidine and glycogen, were contaminated with 3' IMP that became labeled during our protocol (see Figure 2.1). Figure 2.5A and 2.5B show that when spermidine or glycogen were included in our labeling protocol, without exogenously added 3' NMPs, we detected radioactivity that comigrated with 5' IMP (Lane 5, Figure 2.5A and Lane 4, Figure 2.5B). Purification step one and step two are not shown, but analyses of these plates revealed other radioactive spots that comigrated where the known 5' NMP UV standards migrated on control plates. Thus, the radioactivity detected in Lane 5 of Figure 2.5A and in Lane 4 of Figure 2.5B was 5' [32P]IMP, a result of contamination of spermidine and glycogen with various nucleotides including 3' IMP. Spermidine and glycogen were excluded from further experiments. A * pI 1 fS ~ 2 ~bs­'--~ ~~ ~ • 3 0~' '--~ '--~ ~~ 0~' fS ~ 0~ ~ ~bs- B pI 4 5 IMP spike, fmoles 1 500 fInol NMPs 5 2 3 glycogen 4 Figure 2.5. Contamination of commercial reagents with IMP. T4 PNK reactions containing either spermidine or glycogen, in the absence of NMPs, were analyzed by TLC purification method B for the presence of contaminating IMP (step 3 is shown). The 5' [32p]lMP marker is shown in Lane 1 in each plate and the migration of 5' IMP (pI) is shown to the left of each figure. Abbreviations: ATP = y_[32pJ-ATP, T4 = T4 PNK, and Sp. = spermidine. (A) Contamination of spermidine with IMP. All reactions contained 5 mM DTT and 2x OPA+ buffer. Asterisk denotes contaminating radioactivity from the film cassette. (B) Contamination of glycogen with IMP. T4 PNK reactions were carried out as described. ~ o 41 Figure 2.6 shows the PhosphorImage results of decreasing the amount of IMP from 1 femtomole (fmol) to 0.1 fmol in the presence of 100-fold excess NMPs. In the negative control, IMP was left out of the PNK labeling reaction which contained 100 fmol NMPs. Analysis of Lanes 2-4 showed that as expected, decreasing the amount of 3' IMP in the PNK reaction resulted in the decrease of the 5' [32P]IMP signal. UV analysis confirmed that this radioactivity comigrated with nonradioactive 5' IMP and also comigrated with the 5' [32P]IMP marker (Lane 1). Importantly, the negative control (Lane 5) that did not contain IMP but that did include the same concentration of the NMPs as in Lane 2, showed no detectable radioactivity. Thus, it is clear from the results shown in Figure 2.6, that the IMP purification methods developed here not only allowed for the purification of 5' IMP but also for the very sensitive (sub-fmoD detection of 5' IMP as well. We wished to extend the purification protocol to the identification of inosine within a specific endogenous messenger RNA. The glutamate receptor subunit b (glur-b) cDNA in rat brain had been shown to contain an adenosine to guanosine transition that resulted in a glutamine (Q) to arginine (R) codon change (QIR site) (11). This change did not occur in the other members of the AMPA class of glutamate receptors, glur-a, glur-c, and glur-d. Since inosine base-pairs like guanosine, this A-to-G change was indicative of editing by ADARs, and in fact, in vitro studies showed that ADARs could deaminate glur-b mRNA at the QIR site (12-15). ECCLES HEALTH sCln~ .E. v pI 1 100 1 2 50 0.5 3 10 0.1 4 100 NMPs IMP 5 Figure 2.6. Subfemtomoles of5' [32pJIMP can be detected after TLC purification. T4 PNK reactions containing 1 fmol IMP (Lane 2),0.5 fmol IMP (Lane 3), 0.1 fmol IMP (Lane 4), or (-) fmol IMP (Lane 5) were analyzed by TLC purification method B (step 3 is shown). Amount of NMPs (fmol) is indicated at top of figure and migration of 5' IMP (pI) is shown at the left of the PhosphorImage. 42 Northern Analyses and Quantitation of Glur-b mRNA in Rat Brain Poly A± RNA 43 To determine how sensitive the inosine purification method had to be In order to detect inosine within endogenous glur-b RNA, we performed northern blot analyses and RNA slot blot analyses on rat brain polyA + RNA using synthetic glur-a and glur-b RNA as controls. Hybridization of 10 mg rat brain poly A+ RNA with a glur-b 100-mer DNA oligonucleotide which flanked the glur-b mRNA QIR site revealed two transcripts, approximately 6 kB and 4 kB, of similar abundance (Figure 2.7). This result was consistent with published northern results which identified two glur-b transcripts in rat brain poly A + RNA (5.9 kB and 3.9 kB) also of similar abundance (16). A single band was identified in the lane containing synthetic glur-b RNA and there was little cross hybridization seen with the glur-a synthetic RNA that confirmed the specificity of the glur-b 100-mer DNA probe. Prior to our analyses, no quantitative data were available as to how much glur-b was present in rat brain poly A + RNA. Quantitation of a RNA slot blot which compared rat brain poly A+ RNA to various amounts of synthetic glur-b RNA indicated that there was approximately 0.5 fmol glur-b mRNA in 10 mg rat brain polyA+ RNA (data not shown). Hybridization and Isolation of the QIR Site-Containing Glur-b Fragment From Rat Brain Poly A± RNA Sl Nuclease (Sl) and Mung Bean Nuclease (MBN) has been used routinely in the mapping ofmRNA fragments (17-19). glur-a syn. RNA (-5 kb) glur-b syn. RNA (-3.7 kb) 1 2 -- 6 kb -- 4 kb 3 4 Figure 2.7. Northern blot analyses of:fjur-b mRNAin rat brain poly A+ RNA. Northern blot was probed with the [ PJ-labeled glur-b 100-mer DNA oligo. Ten Ilg rat brain poly A + is shown in Lane 4 and the relative sizes of glur-b splice variants are shown to the left of the figure and were estimated from RNA markers (not shown). Lane 1, [poly A- (+) glurJ, contains 20 ~ rat brain poly A- RNA spiked with 1 fmol glur-a ( .... 5 kb) and 1 fmol glur-b ( .... 3.7 kB) synthetic RNAs. Relative sizes of synthetic RNAs are shown to the left of the figure. Twenty Ilg rat brain total RNA (Lane 2) and 20 Ilg rat brain poly A­RNA (Lane 3) were also analyzed. Although Lanes 1-3 were taken from a shorter exposure than was Lane 4, there was mimimal cross-hybridization between the glur-b probe and the glur-a synthetic RNA on the longer exposure. 44 45 Hybridization of RNA transcripts to antisense DNA probes followed by digestion with these single-stranded nucleases results in the formation of RNA:DNA heteroduplexes, or hybrids, which are typically analyzed by polyacrylamide gel electrophoresis (PAGE). A hybridization and isolation protocol method was developed that allowed for the specific hybridization of glur-b mRNA to the glur-b 100-mer DNA probe. As mentioned, the A-to-G change identified in the glur-b cDNA was not identified in the glur-a cDNA. Thus, for an endogenous negative control, a DNA probe that hybridized to the region of the glur-a mRNA which corresponded to the same region in the glur-b mRNA (sequences flanking the QIR site) was synthesized. The general scheme for the hybridization and isolation of the glur-b and glur-a hybrids is shown in Figure 2.8. Briefly, control hybridization reactions typically contained synthetic glur-a or glur-b RNA spiked into 10 mg of rat poly A- RNA. This poly A- RNA was the flow-through fraction of oligo-dT column chromatography of total rat brain RNA and contained mainly rRNA and tRNA. The synthetic glur RNA/poly A- RNA mixture was hybridized with radiolabeled glur-a or glur-b DNA probe, digested with MBN, and analyzed by native PAGE. Specific RNA:DNA hybrids were confirmed by incubation of synthetic glur-b RNA with the glur-a DNA probe and conversely, incubation of glur-a synthetic RNA with the glur-b DNA probe. These hybridizations were subsequently digested with MBN and analyzed by 12% native PAGE and no hybrids were detected with either synthetic glur RNA incubated with the noncognate DNA probe. ,I \ /' ;I' / RNA + -- I, -- _-"'"' ~ DNA + Digest with MEN and gel purify RNA:DNA hybrid I ~ RNA:DNA hybrid ,-- - .... *pG Digest to Np with RNase T2 Label 5' end with PNK & y_[32p]ATP Digest with Nuclease PI & add nonradioactive 5' IMP (pI) ........ *pA ..... , '\ *pU *pC *pI / " , , ........ -- pI ;I' /' .... -"' I Purificaticn of 5' IMP by t successive TLC steps *pI + pI 46 Figure 2.8 Overall scheme for the hybridization, isolation, and purification of IMP from RNA:DNA hybrids. Rat brain poly A+ RNA or synthetic glur RNA transcripts were hybridized with either [32P]-glur-a or [32P]-glur-b specific DNA oligos, digested with MBN, and RNA:DNA hybrids were purified by 12% native PAGE. Gel purified hybrids were heat denatured, RNA digested, and 5' [32P]-labeled as described in Materials and methods (and as in Figure 2.1). 5' [32p]IMP was purified from the radiolabeled 5' NMP mixture by TLC purification method A or method B. 47 However, hybridization of either glur synthetic RNA with its cognate DNA probe revealed the presence of a slower migrating species indicating RNA:DNA hybrid formation. RNA:DNA hybrid formation was verified by RNase H digestion followed by 120/0 native PAGE (data not shown). Figure 2.9A shows a representative 12% native PAGE analysis of 10 mg rat brain poly A+ RNA, or 0.5 fmol synthetic glur b RNA, hybridized with the radiolabeled glur-b DNA probe. The relative migration of the free DNA probe is shown in Lane 1 and the addition of MBN to the DNA probe in the absence of either synthetic glur-b RNA or poly A+ RNA resulted in the complete digestion of the DNA (Lane 2). Hybridization with either 0.5 fmol synthetic glur-b RNA (Lane 3) or 10 mg poly A+ RNA (Lane 4) resulted in the formation of a glur-b RNA:DNA hybrid as indicated by the slower migrating species (compare Lanes 3 and 4 to Lane 1). Comparison of the relative intensities of the glur-b synthetic hybrid (Lane 3) and the glur-b endogenous hybrid (Lane 4) indicated that the hybridization reactions were fairly quantitative based on the estimation of 0.5 fmol glur-b mRNA per 10 mg rat brain poly A + RNA (discussed above). In order to increase the amount of endogenous hybrids, thereby increasing the 5' [32P]IMP signal in the subsequent TLC analysis of the endogenous glur-b hybrid, three separate hybridization reactions were performed. Each hybridization reaction contained 10 mg rat brain poly A+ RNA and 0.1 picomole (pmol) of either the glur-a or the glur-b DNA probe. Hybridizations were digested with MBN and glur-a or glur-b hybridizations were combined prior to analysis by 12% native PAGE (Figure 2.9B). Different amounts of radiolabeled glur DNA probe were included on the gel in order to estimate the amount of endogenous glur-a hybrid. A B glur-b DNA glur-a DNA glur-b DNA rgalut rp-obl MyR ABN+NA + ++ ++ glurr aDtN pAol, yf mA+o l 2 1 0.5 ++ 2 1 0.5 ++ 1 234 1 2 3 4 5 6 7 8 Figure 2.9. Hybridization of glur-b synthetic RNA and rat brain poly A + RNA with [32PJ-labeled glur-specific DNA probes. Hybridizations were performed in Formamide Binding Buffer and incubated at 30°C for 8-12 hours. Hybridizations were digested with 60 U/mL MBN and RNA: DNA hybrids were analyzed by 12% native PAGE (19:1). (A) Hybridization of 10 Jlg rat poly A + RNA (Lane 4) or 0.5 fmol synthetic glur-b RNA (Lane 3) with 0.1 lmol [32pJ­glur- b DNA. Free [32PJ-glur-b DNA migration is shown in Lane 1. (B) Hybridization of 10 Jlg rat poly A RNA with 0.1 pmol [32PJ-glur-a DNA (Lane 4) or with 0.1 pmol [32PJ-glur-b DNA (Lane 8). The hybridization reactions initially contained 0.1 pmol [32PJ-glur DNA prior to the MBN digestion as indicated by the (+) symbols. Migration of free [32PJ-glur DNA is shown in Lanes 1-3 and Lanes 5-7. Amount of free [32P]-glur DNA is shown at the top Lanes 1- 3 and Lanes 5-7. ~ 00 49 Lanes 1-3 show a titration of 2 fmol (Lane 1), 1 fmol (Lane 2), and 0.5 fmol (Lane 3) of glur-a DNA probe and Lanes 5-7 show the similar titration of the glur-b DNA probe. Hybridization of rat poly A + RNA with glur-a DNA followed by MBN digestion resulted in the formation of an endogenous glur-a hybrid (Lane 4). The efficiency of the PNK labeling of the glur-a DNA probe was consistently less than labeling of the glur-b DNA probe (compare Lanes 1-3 to Lanes 5-7). Based on this and on the intensity of the glur-a endogenous hybrid compared to the intensity of the glur-b endogenous hybrid (compare Lane 4 to Lane 8), it seems that glur-a mRNA is expressed to a greater degree in 10 mg rat brain poly A + RNA than is glur-b mRNA. Since glur-a was not expected to be edited by ADARs, and thus should not contain inosine, this greater amount of glur-a should still provide an appropriate endogenous negative control for subsequent TLC analyses. TLC Purification of IMP Derived from Glur Hybrids To analyze the nucleotide composition of the endogenous glur-a and glur-b hybrids, the hybrids were first excised from the gel and eluted from the gel slices (Materials and Methods). The hybrids were heat denatured, digested to 3' NMPs with RNase T2. Radiolabeled 5' [32P]NMPs were then produced according to our standard protocol (see Figure 2.8). To assay whether RNase T2 digestion products derived from the DNA probes would interfere with TLC analysis of 5' [32P]IMP, we also eluted and processed the glur-a and glur-b DNA samples from the gel shown in Figure 2.9B (Lane 1 and Lane 5). The endogenous hybrids and the DNA samples were analyzed by TLC purification method B and the results are shown in Figure 2.10. pI 1 2 3 4 Figure 2.10. Detection of IMP from endogenous glur-a and glur-b hybrids and from glur-a and glur-b DNA samples. Hybrids and DNA samples were gel purified by native 12% PAGE (see Figure 2.9), processed (see Figure 2.8), and IMP was purified by TLC purification method B (step 3 is shown). Migration of 5' IMP (pI) is shown to the left of the autoradiogram. 50 51 Purification of IMP revealed a substantial amount of 5' [32Pl IMP derived from the glur-b hybrid (Lane 1) and unexpectedly, revealed a substantial amount of 5' [32P1IMP derived from the glur-a hybrid (Lane 2). Purification of IMP also revealed a small amount of 5' [32P1IMP present in the glur-a and glur-b DNA samples (Lane 3 and Lane 4). Since control hybridization experiments confirmed that there was no cross-hybridization between the glur-a DNA probe and synthetic glur-b RNA (data not shown), the 5' [32P]IMP observed in the glur-a hybrid sample (Lane 2) was not due to cross-hybridization of the glur-a DNA probe to the glur-b mRNA. Likewise, the small amount of 5' [32P1IMP observed with the DNA samples (Lane 3 and Lane 4) could not account for the substantially more, and essentially identical, amounts of 5' [32P1IMP detected in both endogenous hybrids. It seemed possible that the 5' [32P1IMP we detected in both endogenous hybrids derived from inosine-containing RNAs within endogenous rat brain poly A+ RNA (other than glur-b) which were not completely digested by MBN. Such partially digested RNAs could have comigrated with the endogenous glur-b and glur-a hybrids on the 12% native PAGE and if so, then these RNAs would have been co-purified and would have contributed to the observed 5' [32P]IMP. To address the possibility that partially digested endogenous RNA may have comigrated with the endogenous glur-b and glur­a hybrids and contributed to the observed 5' [32P1IMP (Figure 2.10, Lane 1 and Lane 2), a second MBN digestion step was included after hybrid purification by 12% native PAGE. The hybrids were then repurified after the second MBN digestion step by 15% native PAGE. 52 In addition to analyzing the glur-a and glur-b endogenous samples by this new protocol, we included in our analyses two additional hybrids. The glur-b DNA probe was hybridized with two synthetic RNA molecules, both corresponding to the glur-b sequence, one containing a single inosine (WI RNA), and the other containing an adenosine in place of the inosine (WA RNA). Hybridization reactions contained 1 fmol of the WI RNA or WA RNA mixed with 10 mg rat poly A- RNA (the flow-through fraction of oligo-dT column chromatography of total rat brain RNA). These RNA mixtures were hybridized with 0.1 pmol of the glur-b DNA probe, the WI and WA hybrids were gel purified, eluted, and processed according to the standard protocol. TLC analyses revealed 5' [32P]IMP derived from the WI hybrid and unexpectedly, revealed 5' [32p] IMP derived from the W A hybrid as well (data not shown). Experiments aimed at addressing why both the WI and WA hybrids contained IMP revealed that RNase T2, like spermidine and glycogen, was contaminated with 3' IMP (data not shown). However, unlike spermidine and glycogen, we were not able to exclude this reagent from our protocols and accordingly, the amount of IMP found in RNase T2 defined our background (-0.5 fmols). Contamination of RNase T2 was the most likely source of the 5' [32P]IMP we detected upon analysis of the glur-a and glur-b DNA samples shown in Figure 2.10 (Lane 3 and Lane 4) Figure 2.11A shows the B.3 analysis of 5' [32P]IMP derived from the synthetic WI and WA hybrids. TLC purification revealed a significant 5' [32p]IMP signal derived from the WI hybrid and a lower 5' [32P]IMP signal in the WA hybrid which derived from the IMP contamination of RNase T2 (compare Lane 1 to Lane 3). ~~ ~ ~ :<:"0 p.~ ~v -;;if! ~~ ~"Y ~ :<:<ir p.~ ~v -;;~'V A ~~ ~ ~~~ ~"'~~ ~ <;'" .0 B ~~'> ,# ~~~ ~",~'l <;'" pI pI pA 1 2 3 4 1 2 3 4 Figure 2.11. TLC purification of IMP from WI and WA synthetic hybrids and from endogenous glur-a and glur­b hybrids after 2x MBN digestion and 2x gel purification. 5' [32p]IMP was purified by TLC purification method B (step 3 is shown). Migration of 5' IMP (pI) and 5' AMP (pA) is shown to the left of each autoradiogram and the lanes containing 5' [32pJAMP are labeled at the top. (A) Analysis of 5' [32pJIMP from WI (Lane 1) and WA (Lane 3) hybrids. (B) Analysis of 5' [32P]IMP from glur-b (Lane 1) and glur-a (Lane 3) hybrids. Cl1 CJJ 54 To ensure the fidelity of the indirect transfer procedure used to transfer 5' [32P]IMP from one TLC purification step to the next (see Material and Methods), 5' [32P]AMP was likewise transferred. Comparison of the relative intensities of 5' [32P]AMP (Lane 2 to Lane 4) revealed that the indirect transfer procedure was quantitative. Figure 2.11B shows the B.3 analysis of 5' [32p]IMP derived from the glur-a and the glur-b endogenous hybrids, both isolated after two rounds of MBN digestion and purification from two different polyacrylamide gels, is shown in Figure 2.11B. We observed essentially equal amounts of 5' [32P]IMP in both the glur-b and glur-a hybrids (compare Lane 1 to Lane 3). These results were similar to the results observed with the endogenous glur-a and glur-b hybrids after only one round of MBN digestion and one gel purification (Lane 1 and Lane 2, Figure 2.10). As in Figure 2.11A, 5' [32P]AMP was transferred in the same manner as 5' [32P]IMP (Lane 2 and Lane 4). Although we expected the glur-b endogenous hybrid to contain inosine, the results presented in Figure 2.11B implied that there was an equal amount of inosine in the glur-a hybrid. Com parison of both the endogenous sam pIes (Lane 1 and Lane 3 in Figure 2.11B) to the synthetic WA sample (Lane 3, Figure 2.11A) confirmed that the 5' [32P]IMP observed in the endogenous samples was not due entirely to the IMP contamination of RNase T2. These results stengthened the hypothesis that the source of the observed 5' [32P]IMP was inosine-containing RNAs within endogenous rat brain poly A+ RNA that were not completely digested by MBN, or these inosine-containing RNAs may be double-stranded and therefore would not be digested with MBN. 55 If endogenous inosine-containing RNAs were the source of the 5' [32P]IMP observed, then they must be of similar length (-100 base pairs) to the glur-a and glur-b hybrids. To test whether these putative endogenous inosine-containing RNAs comigrated only with the glur-a and glur-b hybrids or throughout the gel, we performed additional experiments. Figure 2.9B indicated that the endogenous glur-a hybrid ran approximately 1 "hybrid width" above the relative location of the endogenous glur-b hybrid (compare Lane 4 to Lane 8). Accordingly, material from the gel region directly below the glur-a hybrid and material from the gel region directly above the glur-b hybrid were analyzed for 5' [32P]IMP by the standard purification protocol. Figure 2.12 shows the A.3 purification of 5' [32P]IMP derived from these endogenous samples and from the endogenous glur-a and glur-b hybrids. The 5' [32P]IMP spots appear different from previous TLC purifications due to the fact that the direct transfer procedure was used instead of the indirect transfer procedure (Material and Methods). Control experiments revealed that the direct transfer procedure was quantitative and reduced the time of 5' [32P1IMP purification substantially. From time to time, samples transferred by the direct transfer method showed aberrant migration relative to normal migration (compare Lane 2 to Lanes 3- 5). However, as indicated by the dotted circles in Figure 2.12, the 5' [32P1IMP invariably comigrated with the nonradioactive 51 IMP marker as determined by UV visualization. Figure 2.12 includes: glur-a hybrid ("hybll ; Lane 2), gel purified material directly below the endogenous glur-a hybrid (lhyb-1"; Lane 3), endogenous glur-b hybrid ("hyb"; Lane 4), and gel purified material directly above the endogenous glur-b hybrid ("hyb+l; Lane 5). glur-a glur-b hyb hyb - 1 hyb hyb + 1 pI 1 2 3 4 5 Figure 2.12. Detection of IMP in the glur-a hybrid, glur-b hybrid, and in rat poly A + RNA that did not comigrate with the glur-a hybrid or the glur-b hybrid. Migration of 5' IMP (pI) is shown to the left of the figure, the migration of nonradioactive 5' IMP is shown as a dotted circle, and 5' [32PJIMP marker is shown in Lane 1. "Hyb-1" indicates the region of the gel 1 hybrid-width below the glur-a hybrid (hyb) was excised. "Hyb+l" indicates the region of the gel 1 hybrid-width above the glur-b hybrid (hyb) was excised. Material from these re~ions were analyzed in the same fashion as in the endogenous hybrids. 5' [3 P]IMP was purified by TLC purification method B (step 3 is shown). 56 57 Consistent with previous results, analysis of Figure 2.12 revealed 5' [32PHMP in both the endogenous glur-a and glur-b hybrids (Lane 2 and Lane 4). In addition, there was significant 5' [32P]IMP in the material analyzed from regions of the gel which did not comigrate with the glur hybrids (Lane 3 and Lane 5). These results suggested that the 5' [32P]IMP consistently detected in both the glur-a hybrid and glur-b hybrid, and now detected in regions of the gel distinct from the endogenous hybrids, was likely due to the presence of inosine-containing RNAs from the rat brain poly A + starting material. Interestingly, the amount of 5' [32P]IMP found in the material from different regions of the gel indicated that there were substantial amounts of inosine-containing RNAs in rat brain poly A + RNA. The hybridization and MBN digestion protocols would select for RNAs which formed structures resistant to MBN digestion, such as double-stranded RNA structures, a hypothesis that is consistent with the double-stranded RNA structure requirement for the adenosine-to-inosine modifications catalyzed by ADARs. Conclusions This chapter describes the development of novel TLC purification methods that allowed for the purification and sensitive detection of 5' [32PHMP. Application of these methods in experiments aimed at detecting inosine within a cellular mRNA thought to be edited by ADARs in vivo (glur-b mRNA) revealed the possibility that inosine-containing poly A+ RNAs may be more prevalent than previously recognized. Although it was not definitely proven that glur-b mRNA contains inosine, nor that the 5' [32P]IMP we 58 observed was derived from poly A+ RNA, the development of these methods enabled the experiments presented in Chapter III that provided the first conclusive identification of inosine within cellular mRNA. Materials and Methods RNA Isolation Synthetic RNA was transcribed from plasmids (provided by P. Seeburg) according to standard protocols using T3 or T7 RNA polymerase (Ambion). Inosine-containing synthetic RNA (WI) and the corresponding adenosine-containing synthetic RNA (WA) were provided by D. Morse in Brenda Bass's lab at the University of Utah. Rat brains were provided by S. Rogers at the University of Utah or alternatively, obtained from Pel-Freez. Poly A+ RNA was isolated directly from rat brains using a FastTrack 2.0 mRNA isolation kit (In Vitrogen). Alternatively, rat brains were homogenized in, and RNA extracted with, 4 M guanidinium isothiocyanate, 20 mM NaOAC (pH 5.5), 0.1 mM DTT and 0.5% n-Iauryl sarcosine. The guanidinium isothiocyanate homogenate was successively extracted with phenol:chloroform and chloroform, followed by ethanol precipitation with 0.1 volume (voL) 3.0 M NaOAc (pH 5.5) and 2 vol. 100% ethanol. Poly A+ RNA was then selected from the pellet using the FastTrack 2.0 mRNA isolation kit. Hybridization Reactions and Nuclease Digestions DNA oligonucleotides were obtained from the University of Utah DNA Core Facility. Glur-a 100-mer sequence: 5'- CCT GGG GGA AAT GTC ACA TCC TTG CTG CAT GAA GGC CCC CAG GGA GAA CCA CAG GCT GTT 59 GAA TAT GCC AAA CTC ATT TGA CTG GTC ACT GGT TGT CTG T -3'. Glur-b 100-mer sequence: 5'- TCT TGG CGA AAT ATC GCA TCC TTG CCG CAT AAA GGC ACC CAA GGA AAA CCA GAG ACT ATT AAA AAT CCC AAA TTC ATT AGT TGA TTC ACT ACT TTG TGT T -3'. DNA oligonucleotide labeling reactions typically contained 10-20 pmol DNA oligonucleotide, 1 mM y-[32P]ATP (Amersham, 6000Cilmmol, 150mCilml), 5 mM DTT, 2x One-Phor-All Buffer Plus (2x OPA+; Pharmacia), and 6 units (U) T4 polynucleotide kinase (T4 PNK; Pharmacia) in a final volume of 10 J.LL. Labeling reactions were incubated 37° C for 20 minutes followed by the addition of a second aliqout of T4 PNK and incubated 37° C for 20 minutes. Radiolabeled oligonucleotides were successively extracted with phenol:chloroform and chloroform, and precipitated with 2 vol. ethanol and 0.1 vol. NaOAc (pH 5.5) at -20° C. Incorporated counts per minute (cpm) were determined by DEAE filter paper binding (10). Hybridization reactions typically contained 0.1 pmol radiolabeled glur­a or glur-b DNA oligonucleotide incubated with 0.5-10 fmol synthetic RNA mixed with 10 mg rat brain polyA- RNA or incubated with 10 mg rat brain polyA+ RNA. Oligonucleotide, RNA, and H20 were combined in a final volume of 3-4 J.LL and mixed with 30 J.LL formamide binding buffer (FBB): 80% deionized formamide, 0.4 M NaCI, 40 mM Pipes buffer (pH 6.7), and 1 mM EDTA (pH 8.0). Hybridization mixtures were vortexed extensively and incubated at 85° C for 10 minutes followed by immediate incubation for 8-12 hours at 30° C. After incubation, 300 ilL Mung Bean Nuclease (MBN) digestion buffer containing 10 mM NaOAc (pH 4.5),50 mM NaCI, 0.1 mM ZnCI, 1 mM DTT, and 60 U per mL MBN (Gibco-BRL) was added to each hybridization reaction and incubated 1 hour at 42° C. Digestion reaction 60 were stopped with 0.1% SDS and nucleic acids were precipitated with 0.1 vol. NaOAc (pH 5.5), 0.5 ilL 20 mg/mL glycogen (BMB) and 2 vol. 100% ethanol. RNA:DNA hybrids were recovered by centrifugation and resuspended in 10 ilL H20. Hybrids were purified by 12% native polyacrylamide:bis acrylamide (19:1) gel electrophoresis (PAGE) for 8-12 hours at 125-200 V. For preparative purposes, the hybrids were visualized by autoradiography and excised from the gel. Where indicated, the hybrids were re-purified by 15% PAGE (19:1). For analytically purposes, the native gel was soaked in 5% glycerol, dried under vacuum, and visualized by autoradiography. Preparation of 5' Radiolabeled Nucleoside Monophosphates Excised RNA:DNA hybrids were crushed in a 1.5 mL eppendorf tube with a sterile glass rod. The hybrids were subsequently eluted from the polyacrylamide gel pieces by rocking incubation at room temperature in 0.5 M NH40Ac, 0.1 mM EDTA, and 0.1% SDS for 12-16 hours. The solution was separated from the polyacrylamide gel pieces by centrifugation through a Spin-X column (Costar) and precipitated with 0.1 vol. NaOAc (pH 5.5) and 2 vol. 100% ethanol. Hybrids were recovered by centrifugation and resuspended in 10 ilL 50 mM NaOAc (pH 4.5), denatured at 65° C for 10 minutes, and digested to 3' nucleoside monophosphates (3' NMPs) with 5 U Ribonuclease T2 (RNase T2 , GIBCO-BRL) for 30 minutes at 37° C. A second 5 U aliquot of RN ase T2 was added and digestions were incubated for an additional 30 minutes at 42° C. Samples were successively extracted with phenol:chloroform, chloroform, and ether, followed by lyophilization. 61 3' NMPs from the RNase T2 reaction were resuspended in 4 J..LL H20 and 5' radiolabeled with 1 mM y-[32P1ATP, 6 U T4. PNK in 2x OPA+ supplemented with 5 mM DTT in a final volume of 10 J..LL. Where indicated, T4 PNK reactions were supplemented with 1 J..LL 20 mg/mL glycogen (BMB) or 2 mM spermidine (USB). Alternatively, commercial 3' IMP, 3' AMP, 31 GMP, 3' CMP, and 31 UMP (Sigma) were used as the 3' NMP starting material in the PNK reactions. Reactions were incubated at 37° C for 30 minutes, followed by heat-inactivation of T4 PNK at 75° C, 10 minutes. Volume was adjusted with 36.4 J..LL H20 and the 5' [32Pl, 3' -nucleoside bisphosphates were digested to 5' [32P1NMPs with 1.5 U Nuclease PI (BMB) at 37° C for 30 minutes. Reactions were successively extracted with phenol:chloroform, chloroform, and ether, followed by lyophilization. Thin Layer Chromatography Samples were resuspended In 10 pI H20 and 1 pI 100 mM nonradioactive 5'-IMP (Sigma). 51 AMP, 5' GMP' 5' UMP, and 5' CMP were obtained from Sigma. 5' [32p]IMP was purified by one of the following purification methods. Purification Method A, Step One (A.1) Samples were spotted in 2.5 pI aliquots onto a polyethyleneimine­impregnated cellulose (PEl; Bodman) thin layer, 2.5 cm from the lower right corner, and 2.5 cm from the bottom of the plate. After loading, plates were soaked in anhydrous methanol (J.T. Baker) for 10 minutes and dried in fume­hood. Plates were successively developed in the first dimension with H20 to the origin, 0.25 N acetic acid (Mallinckrodt) to 9 cm from the origin, and 0.8 N 62 formic acid (Sigma) to 4 cm on a Whatman #1 paper wick stapled to the top of the PEl plate. Plates were dried in a stream of air in a fume hood, soaked in anhydrous methanol for 10 minutes, and dried in a fume hood. Plates were rotated 90" with respect to the first dimension and developed in the second dimension with H20 to the origin and 0.22 M Tris Buffer (pH 8.0) to 4 cm on a Whatman #1 paper wick stapled to the top of the PEl plate. N onradiolabeled 5' IMP was visualized under short-wave ultraviolet (UV) light and the region corresponding to 5' IMP migration was marked with a pencil. The thin layer was exposed for 30 seconds to 2 minutes on Kodak X­AR film. The region corresponding to 5' IMP migration was cut-out from each 2D-TLC thin layer with a pair of scissors, transferred to a second thin layer by either a direct transfer procedure or an indirect transfer procedure and rechromatographed. Indirect transfer. The PEl matrix was scraped from the plastic backing into a 1.5 mL eppendorf tube. The nucleotidic material was eluted from the PEl matrix by rocking incubation in 0.5 mL 0.25 M ammonium carbonate solution (pH 8.6) for 12-16 hours at room temperature (3). After incubation, the PEl/ammonium carbonate mixture was placed in a Spin-X column and the eluted material was recovered by centrifugation. The volume of the ammonium carbonate solution was adjusted to 1.5 mL with H20 and lyophilized. The lyophilized samples were resuspended in 1.5 mL H20 and re-Iyophilized. This lyophilization procedure was repeated 3-4x until the majority of ammonium carbonate had evaporated. Samples were resuspended in 10 JlL H20 and loaded (as in A.1) on a fresh TLC plate. Direct transfer. The matrix side of the PEl cut-outs were brought into direct contact with a fresh PEl thin layer presoaked in anhydrous methanol 63 and the PEl cut-outs were placed 2.5 cm from the bottom of the thin layer, at least 2.0 cm from one another, and at least 2.0 cm from the side of the thin layer. The PEl cut-outs were held in place by two metal rulers, -25.5 cm long and -1 cm wide. In order to ensure a good seal, it was important the length of the rulers were longer than the width of the TLC plate (20 cm in these experiments). An adhesive-backed magnetic strip (OfficeMax) was applied to one of the metal rulers ensuring a firm seal along the length of the rulers. This enabled the direct transfer of the nucleotides from the PEl cut-outs to a fresh TLC plate for re-chromatography. The magnet-containing ruler was used on the plastic side of the TLC plate. Purification Method A, Step Two (A.2) Samples were transferred to a presoaked PEl plate (identical to A.l) and successively developed with H20 to the origin and IN sodium formate buffer (pH 3.4) to 4 cm on a Whatman 3mm paper wick stapled to the top of the thin layer. Sodium formate buffer (pH 3.4) was prepared by diluting a stock concentration of formic acid (26.5 N) to 1 N and titrating with concentrated NaOH (Sigma). Plates were dried in a fume hood and soaked for 10 minutes in anhydrous methanol and exposed to film for -30 minutes. Nonradioactive 5' IMP was again visualized under shortwave UV light, marked with pencil and this region was excised from the PEl plate. The material contained in the region corresponding to nonradioactive 5'-IMP was transferred directly or indirectly to the next purification step. 64 Purification Method A, Step Three (A.3) Samples from A.2 were transferred to a Cellulose (Kodak) thin layer and rechromatographed in saturated NH4S04:isopropanol:0.lM NaOAc, pH 6.0 (79:2:19). Plates were dried and nonradiolabeled 5' IMP was again visualized under UV light, marked with pencil, and the cellulose thin layer was analyzed by autoradiography. Purification Method B, Step One (B.l) Samples were applied (as in A.l) to a PEl plate presoaked in methanol and developed successively in the first dimension with H20 to the origin, 0.2 M LiCI for 2 minutes, 1.0 M LiCI for 6 minutes, and 1.6 M LiCI to 4 cm on a Whatman #1 paper wick stapled to the top of the PEl plate. Plates were dried in a stream of air in a fume hood, soaked in anhydrous methanol for 10 minutes, and dried in a fume hood. Plates were rotated 900 with respect to the first dimension and successively developed in the second dimension with 0.5 M sodium formate buffer (pH 3.4) for 30 seconds, 2.0 M sodium formate buffer (pH 3.4) for 2 minutes, and 4.0 M sodium formate buffer (pH 3.4) to 4 cm on a Whatman #1 paper wick stapled to the top of the PEl plate. Nonradiolabeled 5' IMP was visualized under short-wave UV light and the region corresponding to 5' IMP migration was marked with a pencil. The PEl plate was exposed for 30 seconds to 2 minutes on Kodak X-AR film. The material contained in the region corresponding to nonradioactive 5'-IMP was transferred indirectly or directly to the next purification step as described. 65 Purification Method B, Step Two (B.2) PEl plates were developed with H20 to the origin followed by with 1N acetic acid to 3 cm on the PEl and 1 N acetic acid:3 M LiCI (9:1) to 4 cm on a Whatman 3mm paper wick stapled to the top of the thin layer. Plates were dried in a fume hood and soaked for 10 minutes in anhydrous methanol and exposed to film. Nonradioactive 5' IMP was again visualized under shortwave UV light, marked with pencil and this region was excised from thin layer. The material contained in the region corresponding to nonradioactive 5'-IMP was transferred indirectly or directly to the next purification step as described. Purification Method B, Step Three (B.3) Step three was identical to purification method A, step three. Purification Method C, Step One (C.1) Samples were applied to a cellulose thin layer and developed in the first dimension with isobutyric acid:NH40H:H20 (66:1:33) until the solvent front reached the top of the cellulose thin layer (-8hours). Plates were allowed to dry in a stream of air in a fume hood, rotated 90° with respect to the first dimension and developed in the second dimension with isopropanol:HCI:H20 (70:15:15) until the solvent front reached the top of the cellulose thin layer (-12 hours). Plates were dried in a stream of air in a fume hood. Nonradiolabeled 5' IMP was visualized under short-wave UV light marked with a pencil. The thin layer was exposed for a 30 seconds to 2 minutes on Kodak X-AR film. The material comigrating with nonradioactive 5' IMP was eluted from the cellulose matrix by rocking incubation for 12-16 66 hours in H20. After incubation, the celluloseIH20 mixture was placed in a Spin-X column and the eluted material was recovered by centrifugation and lyophilized. Purification Method C, Step Two (C.2) and Purification Method C, Step Three (C.3) Samples from C.l were resuspended in 10 f.1L H20 and loaded in 2.5 f.1L aliqouts, 2.5 cm from the bottom of a methanol-soaked PEl plate and at least 2.0 cm from one another and 2.0 cm from the side of the PEl plate. Purification method C, step two was identical to A.2. Material comigrating with nonradioactive 51 IMP was transferred indirectly to purification method C, step three. Step three was identical to step three in both purification method A and method B. 67 References 1. Limbach, P. A., Crain, P. F., and McCloskey, J. A. (1994) Nucleic Acids Res., 22(12), 2183-2196. 2. Randerath, K. and Randerath, E. (1964) Journal of Chromatography, 16, 111-125. 3. Randerath, K. and Randerath, E. (1980) Meth. Enzymol., XII, Part A, 323-334. 4. Silberklang, M., Gillum, A. M. and RajBhandary, U. (1979) Methods in Enzymology, IIX, 58-109. 5. Gupta, R. C., E. Randerath, and K. Randerath. (1976) Nucleic Acids Res., 3(11), 2915-2921. 6. Buck, M., Connick, M. and Ames, B. A. (1983) Analytical Biochemistry, 129, 1-13. 7. Gehrke, C. W. and Kuo, K. C. (1989) Journal of Chromatography, 471, 3-36. 8. Kowalak, J. A. (1994) . Ph.D. Thesis, University of Utah. 9. Lillehaug, J. R. and Kleppe, K. (1975) Biochemistry, 14(6), 1225-1229. 10. Sambrook, J., Fritsch, E.F., Maniatis, T. (ed.) (1989) Molecular Cloning: A Laboratory Manual. Second Ed. Edited by Nolan, C. Cold Spring Harbor Press, Plainview. 11. Sommer, B., Kohler, M., Sprengel, R. and Seeburg, P. H. (1991) Cell, 67(1), 11-19. 12. Hurst, S. R., Hough, R. F., Aruscavage, P. J. and Bass, B. L. (1995) RNA, 1(10), 1051-60. 13. Melcher, T., Maas, S., Higuchi, M., Keller, W. and Seeburg, P. H. (1995) J. Biol. Chem., 270(15), 8566-70. 14. Rueter, S. M., Burns, C. M., Coode, S. A., Mookhetjee, P. and Emeson, R. B. (1995) Science, 267(5203), 1491-4. 15. Yang, J. H., Sklar, P., Axel, R. and Maniatis, T. (1995) Nature, 374(6517), 77-81. 16. Durand, G. M. and Zukin, R. S. (1993) Journal of Neurochemistry, 61(6), 2239-2256. 68 17. Benne, R., Van den Burg, J., Brakenhoff, J. P., Sioof, P., Van Boom, J. H. and Tromp, M. C. (1986) Cell, 46(6), 819-826. 18. Berk, A. J. and Sharp, P. A. (1977) Cell, 12, 721-732. 19. Malo, M. S. (1990) Nucleic Acids Research, 18:20),6159. CHAPTER 3 INOSINE EXISTS IN mRNA AT TISSUE-SPECIFIC LEVELS AND IS MOST ABUNDANT IN BRAIN mRNA 70 The EMBO Journal Vol.17 No.4 pp.1120-1127, 1998 Inosine exists in mRNA at tissue-specific levels and is most abundant in brain mRNA Michael S.Paul and Brenda L.Bass1 Department of Biochemistry and Howard Hughes Medical Institute, 50 North Medical Drive, liniversity of litah. Salt Lake City. liT 84132. USA 'Corresponding author e-mail: [e-mail] The general view that mRNA does not contain inosine has been challenged by the discovery of adenosine deaminases that act on RNA (ADARs). Although inos­ine monophosphate (IMP) cannot be detected in crude preparations of nucleotides derived from poly(A)+ RNA, here we show it is readily detectable and quanti­fiable once it is purified away from the Watson-Crick nucIeotides. We report that IMP is present in mRNA at tissue-specific levels that correlate with the levels of ADAR mRNA expression. The amount of L'\1P present in poly(A)+ RNA isolated from various mammalian tissues suggests adenosine deamination may play an important role in regulating gene expression, particu­larly in brain, where we estimate one IMP is present for every 17 000 ribonucIeotides. Keywords: ADARideaminaseiinosinelRNA editing Introduction During protein synthesis, amino acids are specified by mRNA codons, which consist of triplet combinations of the Watson-Crick nucleotides. Covalent modifications of the Watson-Crick nucleotides, like those that occur within tRKA (Limbach etal., 1994), could potentially increase the number of ways in which a particular amino acid is specified, but are not thought to be prevalent in mRNA. In fact, although eukaryotic mRNAs and many viral mRNAs are 'capped' at their 5' tenninus with a 7- methylguanosine and proximal ribose methylations, only a single modified nucleoside has been detected within coding sequences (reviewed in Narayan and Rottman, 1992). However, this modified nucleoside, 6-methyladen­osine, which is created by the methylation of adenosine (A), does not alter codon meaning since it also pairs with uridine (Engel and von Hippe), 1978). The discovery of .!l,denosine ge.!),ffiinases that act on RNA (ADARs; Bass and Weintraub, 1988; Melcher et aI., 1996; Bass et aI., 1997) has led to speculations that inosine (I), the product of adenosine deamination, may exist within mR.t~A (reviewed in Simpson and Emeson, 1996; Bass, 1997). Inosine prefers to base-pair with cytidine, and accordingly, is read by the translational machinery as guanosine (Basilio eta!., 1962). Except at certain wobble positions, an A to 1 modification within a codon would result in the specification of an alternate amino acid, and the synthesis of a protein that was not genomically 1120 encoded. Such post-transcriptional changes to the genom­ically encoded sequence of an mRNA are not unpreceden­ted, and fall into a group of reactions known as RNA editing reactions (reviewed in Bass, 1997; Kable et al., 1997; Smith etal., 1997). However, all previously described RNA editing reactions that occur on mRNAs involve the conversion of one Watson-Crick nucleotide into another, or the insertion or deletion of Watson-Crick nucleotides, rather than the creation of a non-Watson­Crick nucleotide such as inosine. Since the first observation of RNA editing in 1986 (Benne et ai., 1986), many examples and types of RNA editing have been discovered, and the process is now recognized as an important way to regulate gene expression in eukaryotes. Based on discrepancies between genomic and cDNA sequences that are consistent with adenosine dearnination (e.g. A to G transitions), several mammalian glutamate receptor (gluR) mRNAs (Sommer eta!., 1991; Lomeli et aI., 1994), mammalian 5HT 2C serotonin receptor mR~A (Burns et aI., 1997), hepatitis delta virus (HDV) anti· genomic RNA (Polson et ai., 1996) and mammalian a2,6- sialyltransferase mRNA (Ma et al., 1997) have been pro· posed to be edited by ADARs in vivo. For each of these RNAs, adenosine deamination is proposed to cause a functionally important codon change. For example, an A to I change within gluR-B pre-mRNA is thought to be responsible for changing a glutamine codon to an arginine codon (QIR editing site), as well as an arginine codon to a glycine codon (RiG editing site). Both of these editing events have been correlated with functionally important changes in the ion channels formed from the altered gluR subunits (reviewed in Seeburg, 1996). In support of the idea that the editing sites predicted by A to G transitions correspond to in vivo deamination sites, incubation of in vitro transcribed gluR·B, 5HT2c and HDV antigenomic RNAs with various ADARs leads to dearnination of adenosines at the proposed editing sites (Melcher et al., 1995; Rueter et ai., 1995; Yang et aI., 1995; Polson et al., 1996; Burns et aI., 1997). Also, a recently developed method for specifically cleaving RNA at inosine residues shows cleavage of endogenous rat brain gluR-B mRNA at both the QIR and the RiG editing sites (Morse and Bass, 1997). Despite the existence of a great deal of circumstantial evidence for the presence of inosine within mRNA, inosine monophosphate (IMP) has never been detected among ribonucleotides derived from mRNA. In fact, although inosine has long been known to exist in tRNA (reviewed in Grosjean et aI., 1996), the general view is that mRNA does not contain inosine (Limbach et aI., 1994). Indeed, analyses of nucleotides derived from ribonuclease-digested mRNA in our own laboratory have repeatedly emphasized this general view, and led to the detection of only the four Watson-Crick nucleotides. © Oxford University Press We reasoned that our inability to detect inosine within cellular mRNA might be analogous to the unfeasibility of detecting certain proteins within a crude cellular extract. Obviously, in such cases, the specific activity of the protein must be increased by purification in order for the protein to be detected above the background established by other, more abundant proteins. In this light, we designed a scheme by which IMP could be purified from the ribonucIeotides of cellular poly(A)+ RNA, using successive thin-layer chromatography (TLC) steps. Here we report that IMP is indeed a component of poly(A)+ R..~A, and is present at tissue-specific levels that correlate with the tissue-specific expression of AD AR mRN As. The amount of IMP present in poly(A)+ R:~A isolated from various tissues suggests adenosine deamination may play an important role in regulating gene expression, particularly in brain, where we estimate one IMP is present for every 17 000 nucleo­tides. From our data, we estimate that mammalian brain contains -1800-fold more inosine than can be accounted for by the editing sites within mammalian gluR-B mRNA, suggesting that there are many additional inosine-con­taining R..l\l'As yet to be identified. Results Purification of IMP from rat brain poly(Aj+ RNA Our general scheme for the purification of IMP from cellular RNA is outlined in Figure I. We chose to analyze the RNA in the form of radiolabeled 5' nucleoside monophosphates (5' NMPs; pNs) since existing TLC systems, for the most part, were developed for the analysis of 5' NMPs. RNA samples were first digested to 3' NMPs (Nps) using RNase T2, to generate a substrate for T4 polynucleotide kinase (PNK), which requires a 3' phos­phate. The resulting 3' NMPs were 5' end-labeled with T4 PNK and [y_32p]ATP. The [32P]Nps were then digested with nuclease PI to yield 5' [32P]NMPs. At this pOint, after extracting the samples to remove all protein, the samples were spiked with non-radioactive 5' IMP. This spike was an essential and key feature of our protocol since it allowed us to monitor IMP by UV absorbance, during initial purification steps, when radiolabeled IMP was undetectable. Since several putative ADAR substrates were identified in mammalian brain tissue (Sommer eta!., 1991; Lomeli etal., 1994; Bums etal., 1997), we first attempted to purify IMP from rat brain poly(A)+ RNA. We wanted to be able to quantify any observed radioactive IMP derived from the rat brain RNA, in a manner that accounted for the recovery and labeling efficiencies of each step of the protocol shown in Figure 1. Thus, we ran a number of control samples in parallel. These control samples con­tained RNA that was synthesized in vitro using T3 RNA polymerase and were spiked with various amounts of non­radioactive 3' IMP. Since 3' NMPs become labeled during our protocol (Np ___ *pNp, Figure I), we anticipated that we would purify radioactive IMP in an amount proportional to the amount added, and that the radioactivity values we determined would take into account the losses and effi­ciencies intrinsic to our protocol. As a control for back­ground radioactivity that might co-migrate with IMP, we also processed a T3 transcript that was not spiked with 3' IMP. I I \ 71 Inosine in mammalian mRNA ~RNA /' "- ",. + Digest with RNase T2 Np I Label 5' end with T4 PNK t and y_32P-ATP *pNp I Digest with Nuclease PI t & add non-radioactive 5'-IMP (pI) "- *pG *pA "- \ *pU \ *pC I *pI I "- pI /' ",. I Purification of 5'-IMP by t successive TLC steps *pI + pI Fig. 1. Overall scheme for the purification of IMP. Asterisks indicate nP-Iabeled phosphates. Each step was optimized in control experiments. and RNase T2 digestion was determined to be >95% efficient. By ~onitoring the non-radioactive 5' IMP spike (PI) with UV light, 5' [,2PIIMP (*pI) was purified away from the crude nucleotide mixture (encircled with a dotted line) during successive TLC separations. Ten Jlg of rat brain poly(A)+ RNA, and 10 Jlg of each of the control synthetic RNAs, were processed as in Figure 1 and subjected to the first purification step, a two­dimensional (2D) TLC separation. The autoradiogram for each sample looked essentially identical, and that for the rat brain poly(A)+ RNA is shown in Figure 2A. As anticipated, most of the radioactivity was distributed between radiolabeled 5' AMP, 5' CMP, 5' GMP and 5' UMP, and the [y_32p]ATP remaining from the PNK reaction. Although radioactive IMP was not detected, exposure of the TLC plate to short-wave UV light allowed the detection of the non-radioactive IMP spike, which we outlined with a dotted line (see Figure 2A). Reasoning that 5' [32P1IMP was present in the outlined spot, which overlapped with the 5' [32p]GMP, it was excised from each TLC plate with a pair of scissors. Utilizing a direct transfer method (Randerath and Randerath, 1980), the matrix side of each cut-out was brought into direct contact with a fresh TLC plate, and subjected to the second purification step. The second purification step was a one-dimensional (lD) TLC system, and thus, multiple samples could be processed on a single plate (Figure 2B). Autoradiography 1121 M.S.Paul and S.L.Sass A pC pA pU pG/pi ATP 10 B ___S.. .:;yn.;...;..th..;;eti.;;.,·.;;.,c.;;.,RN.;...;..A_ __ Po~R NA n.:,~i!e? • 10 5 0.5 c __ .....;S...:;yn.;...;..th..;;e.;;.,ti.;;.,c.;;.RN=A;.... __ Poly A+ RNA n.:,~~f:" - 1 0 0.5 pI La"", 1 II 4 5 Fig. 2. Autoradiograms of the three successive TLC separations used to purify [32PllMP. Nucleotide migrations are shown to the right of each plale. and that of the non-radioactive 5' IMP spike. with a dotted outline. For ID separations. chromatography was bottom to top as shown. (A) Purification step 1: the crude nucleotide mixtures for each sample were separated by 2D TLC; only the autoradiogram of the rat brain poly(A)+ sample is shown. (B) Purification step 2: the nucleotide composition is shown for samples derived from rat brain poly(A)+ RNA (lane 6). and synthetic R.!"IAs spiked with 0 (-J, 10. 5, 1 and 0.5 pillol of IMP (lanes 1-5). (C) Purification step 3: the nucleotide composition is shown for samples derived from synthetic RNA or rat brain poly(A)+ RNA; lanes are as in (B). The identity of radioactive spots that are not labeled to the right of the plate is unknown. 1122 72 combined with the UV detection of the non-radioactive IMP indicated that purification step 2 effectively separated 5' GMP from 5' IMP. However, again, the level of background radioactivity was too high to allow detection of radioactive IMP in any of the samples; longer exposure showed substantial background radioactivity that co­migrated throughout the region of 5' IMP migration (data not shown). As in purification step I. non-radioactive 5' IMP was visualized with UV light. outlined with a pencil. cut out. and taken to purification step 3 by the direct transfer method. Purification step 3 was also a ID TLC system, and the autoradiogram of the TLC plate is shown in Figure 2C. In contrast to the analyses performed after purification steps I and 2, after step 3 we observed radioactive IMP in the rat brain poly(A)+ RNA sample, as well as each of the synthetic RNAs spiked with IMP. We verified that all material that co-migrated with IMP actually corresponded to IMP by excising each radioactive IMP spot and re­chromatographing it on a fourth TLC plate with a different solvent; no additional spots were observed, and the calcu­lated amounts ofIMP remained the same (data not shown). A visual comparison of the rat brain poly(A)- sample in Figure 2C with the various spiked synthetic RNAs indicated there was -I pmol of IMP in 10 jlg of rat brain poly(A)+ RNA (compare lane 4 with lane 6). In order tu obtain a more quantitative value, and to minimize inaccuracies due to loading differences, we normalized the radioactivity in the IMP spots using values determined for the amount of radioactivity in the AMP and CMP spots of our starting material (purification step I; see Materials and methods). Although we had no convenient way to correct for errors that occurred during transfer to successive purification steps, control experiments showed that the direct transfer method gave quantitative recovery (data not shown). After normalization, we determined that the 10 jlg ofrat brain poly(A)+ RNA analyzed in Figure 2 contained 1.4 pmol of IMP. Since 10 jlg of poly(A)+ RNA corresponds to -30 nmol of NMP (assuming the molecular weight of each nucleotide is -330 g/mol), this initial experiment indicated that, within rat brain poly(A)+ RNA, there was approximately one IMP for every 21 000 nucleotides. Importantly, the in vitro synthesized control RNA that was not spiked with IMP showed no detectable radioactiv­ity that co-migrated with the UV-visualized IMP (Figure 2C, lane I). This low background in the absence of exogenously added IMP was not by chance but took a great deal of trouble-shooting. In particular, we found that many commercially available reagents were contaminated with material. presumably 3' NMPs, that became labeled during our protocol and resulted in substantial amounts of radioactivity that co-migrated with IMP. In our final protocol. we were able to exclude all but one of these reagents (see Materials and methods). The reagent we could not exclude was RNase T2, which was necessary for the first step of our labeling protocol (see Figure 1) and which appears to bind many nucleotides tenaciously at its active site, inclUding 3' IMP. The contaminating 3' IMP becomes labeled during the protocol of Figure I and can be visualized in control experiments after very long exposure times (data not shown). Thus, the intrinsic contaminating IMP from this reagent defined our back- ground (-0.5 fmol). Nevertheless, even in the presence of this background, our protocol was extremely sensitive and allowed the detection of 0.03 pmol of IMP in the background of 30 nmol of NMPs (I part per million; data not shown). . IMP detected in mRNA was not due to contaminating tRNA The oligo(dT) selection method we used to isolate the rat brain poly(A)+ RNA is considered adequate for removing tRNA and, accordingly, we could not detect tRNA in our samples using various protocols to stain electrophoretically separated RNA (data not shown; see Materials and methods). However, we were worried that even an unde­tectable amount of contamination could yield the amounts of IMP we observed. Thus, we conducted additional experiments designed to rule out the possibility that the IMP observed in rat brain poly(A)'" RNA was due to contaminating tRNA. We first assayed whether the inosine-containing fraction of the poly(A)+ RNA could be precipitated by high salt. The LiCI protocols we used were designed to precipitate large RNAs, such as ruRNA and rRNA, while maintaining the solubility of small RNAs, such as tRNA and 5S RNA (Wallace, 1987). We reasoned that, if the IMP in our poly(A)+ RNA derived from tRNA, it would not be recovered in the precipitated RNA, and we would not observe IMP in our final analysis. We tested the efficacy of our LiC! protocol by its ability to remove tRNA, and thus IMP, from the poly(At RNA fraction, i.e. the RNA that did not bind the oligo(dT) column. Figure 3A shows the amount of IMP purified from 10 )lg of the poly(A)­RNA, with (lane 2) or without (lane 1) the LiCI precipita­tion steps. Clearly, the LiCl precipitation steps were very effective at excluding tRNA from the poly(At RNA. In contrast to its effect on the amount of IMP observed in the poly(A)- sample, essentially no change in the amount of IMP was observed in the poly(A)+ sample after LiCl precipitation (compare lane 3 with lane 4; note that the small difference apparent by eye became insignificant when IMP amounts were normalized). Since ruRNA, but not tRNA, is precipitated by high salt, this result was consistent with the idea that the IMP observed in the poly(A)+ RNA derived from ruRNA. Another effective method for removing tRNA from an . RNA sample is to centrifuge the RNA through high molarity CsC! (Sambrook et ai., 1989). Thus, we per­formed another analysis in which the total RNA sample was subjected to a CsCI centrifugation step prior to oligo(dT) selection. Figure 3B shows the amount of IMP observed in 10 Ilg of the total RNA starting material (lane I), 10 )lg of the RNA fraction that pelleted through a 5.7 M CsCI cushion (lane 2) and 10 Ilg of the RNA fraction that was too small to centrifuge through the CsCI cushion and remained in the superuatant (lane 3). Clearly, the CsCI cushion effectively excluded the IMP-containing RNA from the pellet (lane 2), which contained the rRNA and ruRNA. Consistent with the idea that the IMP observed in lane I derived from tRNA, the amount of IMP per 10 Ilg was greater in the superuatant fraction (lane 3), where tRNA had been concentrated. To determine if the CsCI centrifugation affected the amount of IMP observed in the poly(A)+ RNA, poly(A)+ 73 Inosine in mammalian mRNA B C.CJ CHlb"wptiOJJ T()fo) P",lh.\ SlIp..-n".tal1t pI Lane: 1 c Fig. 3. Effects of high salt precipitation and CsCl centrifugation on the amount of IMP observed in poly(A)+ RNA. RNAs were treated as outlined in Figure I and subjected to the three TLC purification steps of Figure 2. Only the purification step 3 analyses are shown; migration of 5' IMP (pI) is indicated. (A) The autoradiogram shows nucleotides derived from 10 j.l.g of rat brain poly(A)- or poly(A)+ RNA, before (-) and after ( +) precipitation with LiCL The identity of radioactive spots that are not labeled to the right of the plate is unknown, but the spot of slowest migration in lane 1 is removed by LiCI precipitation. suggesting that it may derive from IR."IA. (B) Instead of our normal procedure. which involves selecting poly(At R."IA directly from total RNA, total RNA was first subjected to CsCI centrifugation. The autoradiogram shows IMP deri ved from 10 j.l.g of the total rat brain RNA starting material (lane I), from 10 j.l.g of the CsC! pellet (!ane 2) and from 10 j.l.g of the CsC! supernatant (lane 3). (C) Oligo(dT) was used to purify poly(A) + RNA from the CsCl-pelleted RNA analyzed above (E, lane 2). The autoradiogram shows IMP derived from 10 f,lg of this poly(A) + RNA (lane 5), as well as that from 10 f,lg of synthetic RNA spiked with IMP as indicated (lanes 1-4). RNA was isolated from the CsCI pellet by oligo(dT) selection, and subjected to the labeling protocol of Figure I and the three TLC purification steps used for the experi­ment of Figure 2. Figure 3C shows an autoradiogram of purification step 3. Ten Ilg of the poly(A)- fraction that had been purified by CsCI centrifugation and oligo(dT) selection contained l.0 pmol of IMP (normalized value of lane 5, Figure 3C), a value that was almost identical to that found in previous samples not subjected to the CsCI purification step (e.g. Figure 2C, lane 6). Thus, beginning with two different RNA popUlations, that contained very different amounts of inosine-containing tRNA (compare lanes I and 2 of Figure 3B), our oligo(dT) method selected the same amount of inosine-containing poly(A)'" RNA. Again, this meant that the IMP we observed in the poly(A)+ RNA was not due to the non-specific contamina- 1123 M.S.Paul and B.L.Bass tion by tRNA, and that our normal oligo(dT) selection protocol was very effective in removing tRNA. Although IMP was not readily visible in the RNA found in the CsCI pellet (lane 2. Figure 3B), a faint IMP spot became apparent after long exposure. Comparison of this faint spot with the spiked control RNAs run in parallel (data not shown) indicated that this RNA fraction (10 Ilg) contained -0.1 pmol of IMP. At present we do not know if this IMP derives entirely from the mRNA in the CsCI pellet, or if some of it derives from the rRNA in the pellet; quantification of multiple experiments will be required to obtain a number accurate enough to make this determination. Nevertheless, the low amount of IMP observed in the CsCI pellet means that it is essentially impossible to attribute the IMP observed in the poly(A)­RNA to non-specific contaminati