TIBS 17 - F E B R U A R Y 1 9 9 2
MESSENGER RNA EDITING~-6 is the term coined to describe the post-transcriptional alteration of an RNA's informational capacity other than by the established processes of RNA splicing, 5' end formation, and 3' endonucleolytic cleavage and polyadenylation. In its most radical form, the editing of mitochondrial mRNA in kinetoplastid protozoa (Trypanosoma, Leishmania, Crithidia) can create half the mRNA by uridine insertion and deletionT,S; Extra nucleotides can also be inserted into paramyxovirus RNAs by inducing RNA polymerase errors 2. In subtler forms of mRNA editing, specific nucleotides are modified to alter the mRNA sequence, This category includes the interconversion of multiple C and U residues in some plant mitochondrial transcripts 3,6, C to U changes in chloroplast mRNA 9, the extensive deamination of adenosines in double stranded (ds) RNAs,]°, the discrete modification of a glutamine codon (CAG) to an arginine codon (CGG) in mammalian neural glutamategated ion channel mRNAu, and the discrete modification of a glutamine codon (CAA) in apolipoprotein B (apo B) mRNA to create a translation stop codon (UAA)4,12,13. Recent work has expanded our understanding of how the editing of apo B mRNA takes place, how it is controlled, and its physiological role.
Two forms of apolipoprotein (apo) B are found in mammals. The shorter form is translated from an edited mRNA in which a specific cytidine base is deaminated to a uridine, creating a new stop codon. Apo B mRNA editing is mediated by a site-specific cytidine deaminase that recognizes a downstream target sequence in the RNA. The enzyme has no energy or cofactor requirements and no RNA component, and thus bears no obvious relationship to RNA processing events such as splicing or polyadenylation. While apo B mRNA editing activity may have arrived late in evolution to target dietary lipid to the liver in mammals, the discovery of the editing activity in tissues and cells that do not express apo B suggests a more widespread role in the generation of RNA and protein diversity.
in the liver by translation of the fulllength unedited mRNA. It is required for the secretion of endogenously synthesized lipid and transports triglyceride to peripheral tissues, but subsequently delivers cholesterol to all tissues of the body by the low density lipoprotein (LDL) receptor pathway. In contrast to humans, where solely apo
B100 is made in the liver, rodents produce both apo B48 and apo B100 from the liver.
The biology of apo B mRNA editing The apo B gene is expressed primarily in the liver and intestine, with some expression in the kidney proximal tubules (P. J. Cooper and J. Scott,
The structure and function of apo B In mammals two forms of apo B, both products of the same gene, transport
cholesterol and triglyceride in the blood 4. Apo B48 (241 kDa) is made in small intestinal enterocytes, and is generated as a result of apo B mRNA editing (Fig. 1). It is necessary for the transport of dietary fat as triglyceriderich lipoproteins called chylomicrons. The ~hylomicrons deliver triglyceride to peripheral tissues, after which the chylomicron remnant with its residual lipid is rapidly cleared by the liver through the interaction of another apolipoprotein (apo E) with its receptor. Apo B100 (512 kDa) is synthesized
CAA Gene 5'
U~, ~ .(A)n
1 [ ~ - Protein
J ~ COOH Upoprotein LDLreceptor assembly binding Apo-BlO0
Ct~,,.-U~, Uka. v v .(A)n C.~..-Ut~. ~ (A)n 21s2
1 [ ~ [ NH2 COOH Lipoprotein assembly Apo-B4B
Rgure 1 P. Hedges and J. Scott were formerly at the Division of Molecular Medicine, MRC Clinical Research Centre, Harrow, HA1 3UJ, UK. P. Hodges is now at the Institute of Cell and Molecular Biology, University of Edinburgh, Edinburgh, EH9 3JR, UK and J. Scott is at the Department of Medicine, Royal Postgraduate Medical School, London, W12 ONN, UK.
The single copy of the human apo B gene spans 43 kilobases and is divided into 29 exons. A 14 kilobase mRNA encodes the 512 kDa apo BIO0 protein. In the middle of the huge (7572 base) exon 26, RNA editing of the mRNA or pre-mRNA changes codon 2153 (CAP,,glutamine) to a stop translation codon (UN~,),Some of the edited RNA is subsequently truncated at cryptic polyedenylation sites to create 7 kilobase mRNAs. Apo B48 protein (241 kDa) consists of the amino-terminal 2152 amino acids of apo BIO0. It is fully cornpetent for lipoprotein assembly and secretion, but lacks the carboxy-terminal domain of apo BIO0 that mediates LDL receptor binding. (D1992,ElsevierSciencePublishers,(UK) 0376-5067/92/$05.00 77
TIBS 17 - FEBRUARY1992 Table I. Percentage of steady-state apo B RNA which has been edited in various mouse tissues (M. S. Davies et al., unpublished) - In the CaCO-2 cell line before and after differentiation to an enterocyte-like state 17, and in various ceil lines transfected with a reporter gene 14
percentageediting Mouse tissues
Intestine Liver Kidney Brain Spleen Testis
Cell lines Human
98 57 52 70 70 44 CaCc-2 CaCc-2 HuTu G-292 MG-63 U-20S HOS KHOS A431
Colonicadenocarcinoma inducedto differentiate uninduced Duodenaladenocarcinoma Osteosarcoma Osteosarcoma Osteosarcoma Osteosarcoma Osteosarcoma Epidermoidcarcinoma Hepatocarcinoma
>50 200 raM) and by vanadyl ribonucleoside unpublished). However, apo B tran- the development of other adult diges- complexes26 (>1 raM), but not by tetrascripts can be detected by PCR in most tive functions, the intestinal mRNA is hydrouridine24, an inhibitor of cytidine tissues (M. S. Davies, D. M. Driscoll, edited to a progressively greater extent deaminase. J. K. Wynne and J. Scott, unpublished), around birth, achieving complete editThe apo B editing enzyme is a proAlthough this probably represents ing in adults. In the rat liver, editing is tein or protein complex, with no aberrant transcription, it does allow also induced, but on an independent detectable RNA component. The activity editing to be assessed in these sites, time-course and reaching a lower final is destroyed by proteases2L24,25, heat The production of apo B48 is under tis- plateau. These changes most probably denaturation24,26,and by sulfhydryl, imisue-specific, developmental and hor- represent true developmental patterns dazole, or guanidino modification 2S,and mona] control. The proportion of apo B rather than responses to changing hor- has the buoyant density of pure transcripts that is edited varies con- mona] and nutritional status, a view protein 2s (1.5 g ml-~). While the editing siderably between tissues (Table I) ~4. supported by the induction of apo B activity of a crude extract can be inThe editing activity of different tissues editing upon differentiation of the CaCo-2 hibited by nucleases25, partially purified can be demonstrated by transfection of small intestine-like cell line 16,]7. editing enzyme is no longer sensia reporter gene with an apo B editing Apo B mRNA editing is also under tive 24,2s,suggesting that nuclease treatsite into cultured cells (Table I). Many hormonal and nutritional regulation, ment generates oligoribonucleotide cell lines show little or no editing ac- For instance, thyroid hormone induces inhibitors of the editing enzyme. Its size tivity, but several osteosarcoma cell editing of apo B mRNA in rat liver, poss- by gel filtration has been estimated as lines and one epidermoid cell line, ibly secondarily to an effect on lipo- 125 kDa24; it sediments through glycerol which do not express endogenous apo genesis~8J9. Fasting induces a modest gradients at llS (240 kDa)2s. B, edit the transgene RNA. Thus editing decrease in editing; refeeding with carboThe interaction of cellular proteins activity can be unrelated to apo B hydrate dramatically increases lipogen- with apo B mRNA has been investigated expression, although the otl3er RNA esis and editing is nearly complete Is,2°. by glycerol gradient sedimentation, targets for editing remain unknown. RNA gel mobility shift and RNA-protein Editing activity is induced during the Editingby cell extracts crosslinking by UV light. Smith et al. 26 development of rat and human enteroA breakthrough in the investigation have demonstrated by gradient sedicytes and rat hepatocytes ~s. Both tissues of apo B mRNA editing was the develop- mentation that the apo B RNA substrate express only unedited apo B mRNA merit of an in vitro reaction that edits assembles with rat liver proteins over early in development. In parallel with exogenously added apo B RNA and a 3 h into llS (240 kDa) and then 27S
Cervical epithelioid carcinoma
TIBS 17 - FEBRUARY 1 9 9 2 (1400 kDa) complexes, and have proposed that the latter is the active 'editosome' - after the term spliceosome. Greeve et al. 2s have also demonstrated that, before binding RNA, the editing activity from rat enterocytes sediments at 11S. However, in this study no lag period was required prior to catalysis and no higher order complex was assembledontheRNAsubstrate. Hence, RNA editing activity showed no similaxity to the spliceosomal or polyadenylation complexes 25. The RNA mobility shift assay shows that proteins bind specifically to apo B RNA. UV light crosslinks one 40 kDa protein to apo B mRNA, with specificity for the editing site 2v. As yet there is no direct connection of the editing activity with either the crosslinking proteins or the RNA-protein assemblies.
Mechanism of editing When [o~-3~p]CTPis incorporated into an apo B mRNA substrate, in vitro editing specifically creates a [32p]uridine monophosphate residue solely at the editing site ~4,zs. Several important conclusions can be drawn from this simple observation. Firstly, the product of RNA editing is a simple uridine base; no complex hypermodified base is created. Secondly, the editing reaction does not involve excision and replacement since the 32p-labelled phosphate of the base is preserved. There is no known polymerase that can remove a nucleotide and add a new nucleotide in its place without replacing the 5'-phosphate with the donor's phosphate. The editing reaction more probably involves sitespecific cytidine deamination of R N A . An alternative is transglycosylation, whereby the phosphoribosyl chain is maintained, and a new base is exchanged onto the ribose. However, no donor base has been identified. Finally, it should be considered that the reaction may be transamination, capturing,4he amine to transfer it to another molecule or perhaps catalysing the forward and reverse reactions, interconverting U and C residues, Definition of the mRNAtarget site The apo B sequence can be transferred into other mRNAs and edited in vivo in transfected cells. The length of apo B sequence that can be successfully edited varies in different RNA contexts ~6,29.The optimal size is 55 bases or more but in vitro, 26 bases of apo B mRNA in a 73 base RNA is edited with about 25% efficiency2L3°. This core
(a) human baboon pig rabbit rat mouse human
site 1 site 1 site 1 site 1 site 1 site1 site2
C->U ****** **** C AAUU UGAUCAGUAUAUUAAAG C AAUU UGAUCAGUAUAUUAAAG C AAUU UGAUCAGUAUAUUAAAG C AAUU UGAUCAGUAUAUUAAAG IGAUA C A A U U UGAUCAGUAUAUUA[~G IGAUA C AAUU UGAUCAGUAUAUUAAAG A--~ C AAU~?,AUGAUCr0"-~U~].~UUUA * ** * * * * * * *
[GAUA IGAUA GAUA IGAUA
( b ) 1 O0 -
50 ~ ~" ~ "°
A +6 C Z_> U
'°°11 i illli,,
0 . . . . . . . . . . . . . . . . . . . • .... • ....... UA U -6 A U G A U A ~__>~,AU U U +6 GA U C A G +12 UA U A U U A+19 A -9 C U
Rgure 2 (a) Nucleotide sequence at apo B mRNA editing site from six mammalian species compared to the second editing site discovered in human apo B mRNA31. The starred nucleotides are those demonstrated by mutation to be most important to editing specificity. (b) Mutational analysis of the human apo B mRNAediting site3°. The nucleotide sequence is numbered relative to the editing site, with negative numbers upstream. Blocks of six nucleotideswere changed to their complementary sequence by site-directed mutagenesis and the RNAs assayedfor editing in vitro (upper). The efficiency of editing is presented as a percentageof the efficiency of a reference unmutated RNA. Mutations at the two blocks downstreamof the editing site show the most severe effects on editing. By random mutagenesis,individualnucleotidepositions spanningthe editing site were changedto one, two or all three of the alternative nucleotides(lower). The efficiencyof editing was averagedfor all mutations at each site and includes the data of Chen et al. 22 Mutations at positions +5 to +15 (5'-UGAUCAGUAUA-3')show the most extreme effects.
sequence of 26 ribonucleotides is highly conserved among mammalian apo B sequences (Fig. 2), but not in lower vertebrates (chick and frog) that do not edit apo B mRNA (N. O. Davidson, pers. commun.). It is able to confer editing specificity, but probably requires a structure or sequence context that allows access of the editing enzyme. For instance, an AU-rich context appears preferable to a CC-rich sequence 29. Site-directed mutagenesis of the editing site has been used to define the sequence requirements for editing. In the nine nucleotides surrounding the edited C (nucleotides -3 to +5), 20 out
of 22 mutations allowed efficient editing in vitro, suggesting that the sequence specificity of apo B mRNA is lax22. In a more extensive mutagenesis, Shah et aL examined the 55-nucleotide sequence that undergoes efficient editing in vitro (Fig. 2) 3°. In agreement with work of Chert et aL 22, mutation in positions immediately surrounding the editing site led to little change or even to increased editing. However, they identified an l l-nucleotide sequence downstream of the editing site (5'UGAUCAGUAUA-3';positions +5 to +15) where most mutations abolished or greatly reduced editing in vitro 3°.
TIBS 17 - FEBRUARY 1992
No RNA secondary structure prediction is consistent with the mutation results, but spacing or higher order structure must be involved, as deletions and insertions are poorly tolerated. A preliminary hypothesis is that the editing enzyme recognizes a downstream binding site and then searches for a cytidine at a fixed distance upstream
U + NH3 No cofactor or energy requirement I=igu• 3 A representation of the apo B mRNA editing enzyme, emphasizing the recognition of a target RNA sequence
(Fig. 3)26,3°.Some flexibility exists, as other cytidines and the deamination of a cytidine at a fixed distance introduced in the vicinity upstream, of the edited C can be edited, either independently or in concert with the natural ness or localization on an abnormally site. For instance, only 2% of RNA con- long exon could also contribute, lmtaining the mutation CAA to CCA is portantly, this form of RNA editing can edited, but most of the edited RNA that now be dissociated from creation of a has been modified contains UUA rather termination codon. This RNA editing than UCA or CUA22. This implies that, may be an important force for generonce properly bound to a substrate ating other forms of protein and RNA RNA, the editing enzyme can catalyse diversity. Finally, the finding of the secmultiple deaminations. However, the ond edited site is a dramatic validation activity appears to be site-specific and for the ability of the in vitro editing not processive. Cytidines only two activity to reproduce its in vivo specinucleotides from the editing site (in ficity, CAC mutants) are poorly edited 22, and the next upstream cytidine in natural Comparisonwith other forms of RNAediting mRNA (position -11) is not edited in RNA editing has been divided into vivo or in vitro2~,22,3°,3L two classes: 'insertional editing' versus Recently, a second RNA editing site 'substitutional editing '32. In the first was discovered in human apo B mRNA3L class, editing of kinetoplastid mitochonCytidine 6802 is edited in vitro with drial mRNA appears to involve base about 15% of the efficiency of the pri- pairing to specific guide RNAs with mary editing site cytidine 6666. From poly(U) tails, identification of editing intestinal cDNA libraries, two of 11 apo sites in the mRNA by mismatches with B cDNAs had a T at site 6802 (all con- the guide RNA, cleavage of the mRNA taining T at the primary editing site), by phosphate transesterification to the demonstrating that editing of this sec- poly(Lr)tail, and re-ligation of the mRNA ond site occurs in vivo. The second site halves with added or deleted U has a sequence similar to the primary residues by reverse transesterification 7. editing site (Fig. 2), including seven of A similar mechanism may underlie the the 11 nucleotides of the proposed tar- insertion of C residues into slime mould get sequence. Since this second editing mitochondrial mRNAs~'32. The insertion site is within the 3' untranslated region of G residues in paramyxovirus mRNA of apo B48 mRNA, it is unclear what is apparently co-transcriptional, due to effect editing would have. If this editing RNA polymerase stuttering 2. occurs independent of the primary editThe modification of apo B mRNA is ing event, it is again unclear what effect different from insertional mRNA editing, a threonine (ACA) to isoleucine (AUA) No guide RNA is implicated, the phoschange would have on the apo B protein, phodiester backbone of the RNA is not Testing other nearby sites with similar cut, and the new base is synthesized in sequences found none that were edited situ. It falls into a second class of RNA in vitro or in vivo. editing, substitutional editing, that In summary, a limited stretch of pri- blurs the distinction between editing mary RNA sequence downstream of and RNA modification. Apo B editing the edited nucleotides is an important appears to bear more mechanistic simideterminant of editing. Other elements larity to tRNA, snRNA and rRNA base such as secondary structure, AU-rich- modifications, ds RNA adenosine deam-
ination, and the C to U changes reported in plant mitochondrial and chloroplast mRNA, and the editing of brain glutamate receptor channel mRNA. The myriad of tRNA, snRNA and r R N A modifications can modulate the stability of certain base pairing interactions, but only a few of these changes alter the fundamental identity of the base. The most striking of these is the creation of lysidine, whereby the attachment of a lysine amino acid to the 2 position of the cytidine ring causes the base to be read as uridine and
changes the anticodon specificity33. Another U-like base, 5-carbamoylmethyi uridine, is found in the anticodon of tRNAP'°(U*GG) while a CGG anticodon is predicted by the gene 34,35. Here, the cytidine may be deaminated as well as hypermodified. The unusual selenocysteine tRNA[ser]s~c undergoes two different patterns of C to U and U to C modifications3E The gene encoding this tRNA has been sequenced and the tRNAs sequenced by physiochemical separation, so it is certain in this example that uridine, and not a U-like base, is produced. It is intriguing to speculate on connections between the modification of a natural suppressor tRNA and mRNA editing that can create termination codons. An adenosine deaminase selectively modifies a large number of adenosine residues to inosine (I) in ds RNA, without obvious sequence specificity and only a modest preference for dinucleotides 5,~°. The activity is conserved between insects and vertebrates, widespread among cell types, regulated in development and in the cell cycle, and may have a function in vivo either to promote the degradation of specific endogenous or viral RNAs or perhaps to alter the'coding capacity of mRNAs. This adenosine deaminase, like the apo B mRNA cytidine deaminase, recognizes a target structure in RNA and performs base-specific deamination, but it appears to act processively and extensively modifies the target. Intriguingly, this adenosine deaminase can cooperate with an RNA binding protein (the HIV virus Tat protein) to perform a single site-specific modification37. Site-specific adenosine deamination may also underlie the editing of the mRNA encoding synaptic glutamategated ion channels u. Permeability to calcium is prevented by the introduction of a specific arginine residue into the channel. While glutamine (CAG) is encoded in the genome, post-transcrip-
TIBS 17 -
tional editing creates the arginine codon (CGG or perhaps CIG). The six genes encoding this class of receptors are virtually identical in nucleotide sequence in this region and are coexpressed in the same cells, yet the mRNAs are edited to very different extents: not edited at all, completely edited or edited at an intermediate level. As for ape B mRNA editing, the infermation directing the editing could lie at a distance downstream of the edited base, perhaps in the following intron, The editing activity could be similar to the ape B RNA editing enzyme or, since both activities are expressed in the brain, they may share protein subunits, An equally intriguing possibility is that a distant mRNA sequence is comp-
deaminase and adenosine deaminase 38, for which a crystal structure is known and a reaction mechanism proposed 39. The active site residues of adenosine deaminase are found in regions of limited primary sequence homology with AMP deaminase 4°, and dCMP deaminase 41,42. The different deaminases vary considerably in their subunit molecular weight (20 to 93 kDa) and multimer composition (from monomers to hexamers). Since the deaminase enzymatic activity can be carried by a relatively small protein domain, the ape B editing activity may have a related subunit or domain joined to a targeting protein responsible for RNA recognition. The separation of deaminase and recognition domains may explain the spacing between the
lementary to this region, and the intramolecular ds RNA is deaminated by the unwinding enzyme, The RNA editing that appears most similar to the apo B mRNA activity is the plant mitochondrial and chloroplast mRNA editing 3,6,u. Each of these activities change single C residues to be read as U, although an interesting difference in the plant mitochondrial system is the back-editing reaction, i.e. amination of U to form C. Multiple C residues are edited in most mRNAs, and rRNA and introns can be edited, too. The plant mitochondrial enzyme may also be a cytidine deaminase, perhaps being reversible or with a transaminase activity. However, plant mRNA target sequences are not edited by mammalian apo B mRNA editing extracts 3~, and one group has proposed similarity to the insertional editing mechanism, based on loose similarity between edited sites and their homology to hypothetical guide RNA sequences 3. Most telling for the difference between ape B editing and the plant activity is the report 6 that a mutant wheat strain edits aberrantly, creating G from A, A from G, and A from U. If these modifications are due to altered editing activity, they cannot be explained by a mutant cytidine de-
RNA recognition site and the editing site. The glutamate receptor mRNA editing activity, ds RNA deaminase, tRNA modification enzymes, plant mitechondrial and chloroplast RNA editing activity may have similar structure, or even share subunits.
Comparisonto other deaminases The conversion of cytosine and cytidine nucleoside derivatives to their corresponding uracil derivatives is accomplished by a number of enzymes specific for different roles in the biosynthesis and degradation of nucleic acids: cytosine deaminase, cytidine deaminase, dCMP deaminase, and dCTP deaminase. Further, mechanistic similarities have been demonstrated between cytidine
Cattaneo, Nicholas Davidson, Dolph Hatfieid and Kazuko Nishikura for providing access to unpublished results.
References 1 Bass, B. L. (1991) Nature349, 370-371 2 cattaneo, R. et al. (1990) Experientia 46, 1142-1148 3 Mulligan, R. M. (1991) Plant Cell 3, 327-330 4 Scott, J. (1989)Curr. opin. Cell. Biol. 1, 1141-1147 5 wagner, R. W. et al. (1989) Prec. NatlAcad. Sci. USA 86, 2647-2651 6 Walbot, V. (1991) Trends Genet. 7, 37-39 7 Blum, B. et al. (1991) Cell65, 543-550 8 Stuart, K. (1991) Trends Biochem. Sci. 16, 68-72 9 Hoch, B. et al. (1991) Nature 353, 178-180 l O Bass, B. L. and Weintraub, H. (1988) Cell55, 1089-1o98 11 Sommen,B. et al. (1991) Cell67, 11-19 12 Chen,S-H.et al. (1987) Science 238, 363-366 13 Powell, I.. M. et al. (1987) Cell 50, 831-840 14 Bostr6m, K. et al. (1990) J. Biol. Chem. 265,
22446-22452 15 Wu, J. H. et al. (1990) J. Biol. Chem. 265,
16 BostrOm,K. et al. (1989) J. Biol. Chem. 264, 15701-15708 17Jiao,s., Moberly,J. B. and Schonfeld, G. (1990) J. Lipid Res. 31, 695-700 18 Baum, C. L., Teng, B-B. and Davidson, N. 0. (1990) J. Biol. Chem. 265, 19263-19270
1 9 Davidson, N. O. et al. (1988) J. Biol. Chem.
The C to U editing of apo B mRNA by a sequence-specific cytidine deaminase is currently a phenomenon without context. For apo B this switch is regulated by developmental induction and hormonal and metabolic factors. While it remains possible that apo B mRNA editing activity appeared in mammals late in evolution, exclusively to target dietary lipid to the liver, the finding of the editing activity in a wide variety of tissues and cell lines that do not normally express apo B suggests a wider role for this process. Editing would allow the co-ordinated shift in production of alternative versions of other RNAs or proteins. However, apart from the second editing site in ape B mRNA there are no further examples of mRNAs edited by this activity. Identification of the editing activity's target sequence should allow prediction of editing sites in other sequenced genes. As yet,
263, 13482-13485 20 Leighton, J. K. et al. (1990) J. LipidRes. 31, 1663-1668 21 Driscoll, D. M. et al. (1989) Cell 58, 519-525 22Chen, S-H.et al. (1990) J. Biol. Chem. 265, 6811-6816 23 Lau, P. P. et al. (1991) J. Biol. Chem. 266, 20650-20654 24 Driscoll, D. M. and Casanova, E. (1990) J. Biol. Chem. 265, 21401-21403 25Greeve, J., Navaratnam,N. and Scott, J. (1991) Nucleic Acids Res. 19, 3569-3576 26Smith, H. C. et al. (1991) Prec. NatlAcad. Sci. USA 88, 1489-1493 271_au, P. p. et al. (1990) Nucleic Acids Res. 19, 5817-5821 28 Hedges, P. E. et al. (1991)NucleicAcids Res. 19, 1197-1201 29Davies,M. S. etal. (1989)1 Biol. Chem. 264, 13395-13398 30 Shah, R. R. et al. (1991) J. Biol. Chem. 25, 16301-16304 31 Navaratnam,N. et al. (1991) Nucleic Acids Res. 19, 1741-1744 32 Mahendran, R., Spottswood, M. R. and Miller, D. L. (1991) Nature 349, 434-438 33Muramatsu,T. etal. (1988)J. Biol. Chem. 263, 9261-9267 34Keith, G. etal. (1990)Biochim. Biophys. Acta
screening of cDNA libraries or DNA sequence data banks for sequences with similarity to the ape B editing site has not been productive. For the time being we may need to rely on an army of cDNA cloners who are unwilling to ignore a C to T or A to G 'cloning artifact'.
35weill, D. and Heyman,T. (1990) Nucleic Acids Res. 18, 6134 36 Diamond, A. M. et al. (1990) Nucleic Acids Res.
Acknowledgements The a u t h o r s thank Lesley Sargeant for typing the manuscript, and Brenda Bass, Axel Brennicke, Roberto
37 Sharmeen, L. et al. (1991) Prec. Natl Acad. Sci. USA 88, 8096-8100 38 Frick, L. et al. (1989) Biochemistry 28,
9423-9430 (1991) science 252, 1278-1284 40 Chang, Z. et al. (1991) Biochemistry 30, 2273-2280 41 Maley,F. and Maley,G. F. (1990) Prog. Nucleic Acid Res. Mol. Biol. 39, 49-80 42Mclntosh,E. M. and Haynes,R. H. (1986) Mol. Cell. Biol. 6, 1711-1721
39Wilson, D. K., Rudolph, F. B. and Quiocho, F. A.