Mark P. Sowden 1,2, Matthew J. Eagleton 3 and Harold C. Smith 1,2,4, * Nucleic Acids Research, 1998, Vol. 26, No. 7

1644 1652 Nucleic Acids Research, 1998, Vol. 26, No. 7 1998 Oxford University Press Apolipoprotein B RNA sequence 3 of the mooring sequence and cellular sources of auxiliary factors determine the location and extent of promiscuous editing Mark P. Sowden 1,2, Matthew J. Eagleton 3 and Harold C. Smith 1,2,4, * 1 Department of Biochemistry and Biophysics, 2 Department of Pathology, 3 Department of Surgery and 4 The Cancer Center, University of Rochester School of Medicine and Dentistry, Box 712, 601 Elmwood Avenue, Rochester, NY 14642, USA Received December 15, 1997; Revised and Accepted February 9, 1998 ABSTRACT Apolipoprotein B (apob) RNA editing involves a cytidine to uridine transition at nucleotide 6666 (C6666) 5 of an essential cis-acting 11 nucleotide motif known as the mooring sequence. APOBEC-1 (apob editing catalytic sub-unit 1) serves as the site-specific cytidine deaminase in the context of a multiprotein assembly, the editosome. Experimental over-expression of APOBEC-1 resulted in an increased proportion of apob mrnas edited at C6666, as well as editing of sites that would otherwise not be recognized (promiscuous editing). In the rat hepatoma McArdle cell line, these sites occurred predominantly 5 of the mooring sequence on either rat or human apob mrna expressed from transfected cdna. In comparison, over-expression of APOBEC-1 in HepG2 (HepG2-APOBEC) human hepatoma cells, induced promiscuous editing primarily 5 of the mooring sequence, but sites 3 of the C6666 were also used more efficiently. The capacity for promiscuous editing was common to rat, rabbit and human sources of APOBEC-1. The data suggested that differences in the distribution of promiscuous editing sites and in the efficiency of their utilization may reflect cell-type-specific differences in auxiliary proteins. Deletion of the mooring sequence abolished editing at the wild type site and markedly reduced, but did not eliminate, promiscuous editing. In contrast, deletion of a pair of tandem UGAU motifs 3 of the mooring sequence in human apob mrna selectively reduced promiscuous editing, leaving the efficiency of editing at the wild type site essentially unaffected. ApoB RNA constructs and naturally occurring mrnas such as NAT-1 (novel APOBEC-1 target-1) that lack this downstream element were not promiscuously edited in McArdle or HepG2 cells. These findings underscore the importance of RNA sequences and the cellular context of auxiliary factors in regulating editing site utilization. INTRODUCTION Apolipoprotein (ApoB) mrna editing alters a genomically encoded glutamine codon (CAA) at position 2153 to a STOP codon (UAA) in RNA through site-specific cytidine deamination (1,2). The process is mediated by the 27 kda cytidine deaminase, APOBEC-1 (apob editing catalytic sub-unit 1), which has been cloned from human, rat, rabbit and mouse (3 6). The enzyme functions as a homodimer (6,7), that binds to RNA weakly and non-specifically (8,9). The sequence elements that determine site specific editing have been identified by deletion and site-directed mutagenesis studies (10,11). An 11 nucleotide (nt) mooring sequence is both necessary and sufficient to promote editing of an appropriately located upstream cytidine. The mooring sequence also functions to promote editing, albeit at a reduced efficiency, when inserted into a heterologous gene (12,13) or alternative sites in apob mrna (14). An enhancer sequence located immediately 5 of the cytidine to be edited is also critical for efficient levels of editing. Length and AU content of RNA sequences more distally located from the editing site also contribute a bulk RNA context that enhances editing in vitro (12). To edit apob mrna, APOBEC-1 requires auxiliary proteins for editosome assembly and efficient site-specific editing activity (4,15 20). The auxiliary proteins have a widespread distribution and can be demonstrated in extracts from cells that do not synthesize apob mrna (5,21). These factors can interact specifically with APOBEC-1 and some may bind apob mrna (18 20). The proportion of total cellular apob mrna that is edited is subject to developmental, hormonal and dietary regulation. Increased apobec-1 mrna abundance correlated with increased apob mrna editing upon dietary modulation in rats and developmentally in the human small intestine (22 24). In contrast, a proposed increase in auxiliary factor abundance led to the increase of hepatic apob mrna editing during embryonic and early postnatal development in the rat or following hormone administration whilst apobec-1 mrna levels remained unaltered (25). *To whom correspondence should be addressed. Tel: +1 716 275 4267; Fax: +1 716 273 1027; Email: hsmith@bphvax.biophysics.rochester.edu

1645 Nucleic Acids Research, 1994, 1998, Vol. Vol. 22, 26, No. No. 17 1645 To evaluate the role of enzyme abundance in determining editing efficiency, APOBEC-1 was experimentally over-expressed in McArdle rat hepatoma cells. These cells and extracts thereof had previously been used to study the cis- and trans-acting factors involved in site-specific apob RNA editing (13,18,20,21,26 28). Over-expression of APOBEC-1 in these cells increased editing efficiency from 14 to 85% (29), comparable with the enhancement seen during development or following metabolic stimulation. In contrast with physiological up-regulation of editing, experimental over-expression of APOBEC-1 also induced editing of cytidines 5 of C6666 that in normal cells or tissues would not have been affected (promiscuous editing) (29). Editing at these additional sites was dependent upon the mooring sequence but did not demonstrate the constraints of proximity to the mooring sequence (spacer element) and the need for the enhancer sequence typical of the wild type editing site (29). ApoB mrna hyper-editing involves conversion of cytidines predominantly 3 of C6666 (but also 5 of C6666) and was described when rabbit APOBEC -1 was targeted for over-expression by liver specific promoters in transgenic mice and rabbits (30). In addition to apob RNA, the mrna encoding the translational repressor, NAT-1 (novel APOBEC-1 target-1) was hyper-edited (31). A low level of editing was also observed at a single cytidine in the mrna encoding a tyrosine kinase TEC-1 RNA (32). Neither NAT-1 nor TEC-1 mrnas were edited in normal mouse liver. In contrast with promiscuous editing in McArdle cells, the significance of the mooring sequence in hyper-editing of apob mrna could not be demonstrated (30,32). Hyper-editing of TEC-1 and NAT-1 mrnas occurred in the proximity of mooring sequence homologs but the generalized distribution of sites precluded a conclusion as to the orientation or distance constraints of these sites relative to the mooring sequence. In light of the hyper-editing data, we have re-evaluated the characteristics of promiscuous editing in the context of different sources of auxiliary factors, APOBEC-1 from different species and the cis-acting RNA elements that might be involved. The results demonstrated that promiscuous editing was generic to all species sources of APOBEC-1. However, different cell type contexts of auxiliary factors could affect which cytidines were promiscuously edited and the proportion of total cellular apob RNAs edited at these sites. Finally, promiscuous editing required the mooring sequence as well as sequence elements downstream of the mooring sequence coincident with a pair of tandem UGAU motifs. MATERIALS AND METHODS Plasmid constructions Expression vectors for wild type and mutant apob RNA substrates were generated in prc/cmv (Invitrogen, CA) as described previously (27). Rat APOBEC-1 cdna (4) was subcloned from prc/cmv-apobec-1 (29) into a modified pcdnaiii vector (Invitrogen, CA) that encodes a hexa-histidine sequence followed by a nine amino acid hemagglutinin (HA) epitope tag (20). Human and rabbit APOBEC-1 were produced from oligo(dt)-primed, small intestine total RNA using AMV reverse transcriptase (Promega), and amplified with HiFi Taq DNA Polymerase (Boehringer) according to the manufacturer s recommendations, in a PCR using APOBEC-1 5 and APO- BEC-1 3 as amplimers. Thermal cycling conditions were: one cycle at 94 C for 3 min, five cycles at 94 C for 45 s, 52 C for 1.5 min, 72 C for 1.5 min and 30 cycles at 94 C for 45 s, 55 C for 1 min, 72 C for 1 min. The PCR products were subcloned into the modified pcdnaiii vector through the EcoRV and XbaI restriction sites in the amplimers and verified by dideoxy DNA sequencing (ABI Prism cycle sequencing system). Cell culture The rat liver hepatoma cell line McArdle RH7777 was maintained, transfected and stable clonal cell lines generated as described previously(29). The human liver hepatoma cell line, HepG2 cells, was obtained from ATCC (Rockville, MD) and maintained in MEM containing 10% fetal bovine serum and non-essential amino acids (Gibco). HepG2 cells were transfected as above and clonal stable cell lines selected under 850 µg/ml G418 (Gibco). Protein analyses Whole cell protein was prepared for western blotting from selected lines using reporter lysis buffer (Promega Corp) according to the manufacturer s recommendations. HA epitope tagged APOBEC-1 proteins were detected using an anti-ha mouse monoclonal antibody (BabCo, CA) and an HRP-conjugated rabbit anti-mouse secondary antibody. Reactivity was visualized using enhanced chemiluminescence (ECL, Amersham). RNA isolations Total cellular RNA was isolated from transient transfections and stable clonal McAPOBEC and HepG2-APOBEC cell lines using Tri-Reagent (Molecular Research Center, MRC, OH) according to the manufacturer s recommendation. After isopropanol precipitation, RNAs were digested with RQ-DNaseI and with a restriction enzyme(s) having a recognition site between the PCR primer annealing sites of target substrates as described previously (27). These were EcoRI or HindIII and RsaI for human and rat apob RNA substrates respectively. Editing assays The proportion of editing upon apob RNAs synthesized in transfected cells was determined by the reverse transcriptase polymerase chain reaction, (RT PCR) methodology as described previously (27). Transfected human apob RNAs were amplified using SP6 and T7 amplimers; endogenous rat and human apob RNAs were amplified using ND1/ND2 and PCR5/PCR12 amplimer pairs respectively. Editing efficiencies at specific cytidines in the amplified cdnas were evaluated by the poisoned primer-extension assay using [γ- 32 P]ATP (6000 mci/mmol; NEN) end-labeled apob specific primers. Primer-extension products were resolved on a 10% denaturing polyacrylamide gel and quantified by laser densitometric scanning (Phosphorimager Model 425E, Molecular Dynamics). In vitro editing assays containing 20 fmol of in vitro transcribed apob or NAT-1 RNA substrate and protein as McArdle cell extracts or rat liver extracts were performed as described (17,27). Purified RNA was subjected to poisoned primer extension analysis as described above. Previously, direct sequence analysis of apob RT PCR products, as individual clones, provided an assessment of the number of editing sites within a given apob mrna molecule. We evaluated the feasibility of using Taq DNA polymerase based dye -terminator sequencing of purified apob RT PCR products prior to their cloning, to simultaneously assess the editing efficiencies at C6666

1646 Nucleic Acids Research, 1998, Vol. 26, No. 7 and multiple 5 - and 3 -located cytidines. Oligo(dT)-primed cdnas were PCR amplified using SP6 and T7 primers and the purified products sequenced. Editing efficiencies >20% were observed as superimposed cytidine and thymidine on the electropherogram in a ratio approximating the efficiency determined by poisoned primer extension analysis. However, efficiencies of <20% that can be accurately determined by poisoned primer extension analysis, could not reproducibly be observed by sequencing of PCR products and had to be verified by dideoxy sequencing of individual clones (29). Therefore, to accurately quantify editing efficiencies at multiple cytidines in apob RNAs, poisoned primer extension analyses were performed using primers appropriate for each region. Deoxyoligonucleotides Deoxyoligonucleotides used in this study were: ApoB specific primers: MS 1 5 -CAAGACTTTTTAATTTTTCAATGATTTCATC; MS 2A 5 -GCTATTTTCAAATCATGTAAATCATAAC; MS 4 5 -ATGATTTCATCAATAAATATTAGCAATAGC; MS 8 5 -TAGCTTGCTGTGGGAGTTTTCCCAGGGCTG; MS 9 5 -ACTTGTCTCTCCCAATTGAATGAATTCAG; MS 10 5 -TTAAGTCCTGTGCATCATAATTATCTC; JB 96a 5 -GGGCTGCTCTGTATTTTCTTATATACTG. ND1, ND2, PCR5, PCR12 and DD3 (33), and JB 85, JB 132, JB 143, JB 154, JB 155 and JB 160 have been described previously (34). NAT-1 specific primers (31): NAT-1 5 5 5 -CTCGAATTCGA(A/C)TATCT(A/G)AA(T/C)AGTGGAAATGC; 3 end at 1985. NAT-1 5 3. 5 -CCCTCAAGCTTCT(A/G)GTCATTAAGATGTTCAC; 3 end at 2607 The mixed base positions allowed for isolation of mouse, rat, human and/or rabbit sequences. NAT-1 3 5 5 -CTCGAATTCGCCGTAATAGTATATTGCCTG; 3 end at 3499. NAT-1 3 3 5 -CCCTCAAGCTTAAGTATATAAAATCAGGGCATG; 3 end at 3736. NAT-1 PP 5 -GGGCATTAATGATCTCTCAAACTGATCAG; 3 end at 3605. APOBEC-1 primers: APOBEC-1 5 5 -CTCGATATCATG(A/G)CTTC(T/C)GAGAAAGGTCC; APOBEC-1 3 5 -CTCTCTAGATCATCTCCAAG(C/G)CACAGAAGG. The mixed base positions allowed for isolation of human and rabbit sequences. SP6 and T7 primers have been described previously (27). RESULTS Promiscuous RNA editing in rat hepatoma cells over-expressing human, rabbit or rat APOBEC-1 To investigate whether hyper-editing [observed when rabbit APOBEC-1 was over-expressed in transgenic mice and rabbits (30)] or promiscuous editing [resulting from the over-expression of rat APOBEC-1 in McArdle cells (29)] is dependent upon the species source of APOBEC-1 separate clonal McArdle cell lines, expressing equivalent levels of human, rabbit and rat HA epitope-tagged APOBEC-1 proteins were selected by western blot analyses (Fig. 1A). C6666 in human apob RNA (humapob wt), expressed from transiently transfected human apob cdna was edited with high efficiency in each of the APOBEC-1-expressing cell lines (Fig. 1B). High levels of promiscuous editing at C6655 were also detected in each cell line. Editing of C6666 was also stimulated on the endogenous McArdle cell, rat apob mrna (ratapob wt), by all three sources of APOBEC-1. Promiscuous editing of C6661 on the endogenous transcript was apparent, albeit at a reduced efficiency compared with that observed at the equivalent cytidine, C6655, on humapob wt RNA. Only trace amounts of promiscuous editing at cytidines 3 of C6666 on humapob wt (C6675, C6702 and C6724) and ratapob wt (C6675) were detectable (Fig. 1C and D). In previous transgenic animal experiments, levels of hyperediting >25% were measured at C6743, C6762, C6806 and the alternative editing site at C6802 on human apob RNA (10,30). DNA sequencing analyses of humapob wt RNA specific RT PCR products from rat, human and rabbit APOBEC-1-overexpressing McArdle cell lines did not detect comparable levels of hyper-editing at these or any other 3 cytidines (data not shown). These results indicate that the ability of APOBEC-1 to stimulate editing of C6666 in apob mrna and to promiscuously edit other cytidines is not unique to the rat enzyme over-expressed in the homologous rat cell line. Promiscuous editing in human liver cells over-expressing rat APOBEC-1 The analysis to this point has evaluated APOBEC-1 over-expression in the context of auxiliary proteins from McArdle cells. Given the absolute requirement of auxiliary factors for APOBEC-1 editing activity (4,15 20) the characteristics of promiscuous editing were evaluated in the context of auxiliary factors from human HepG2 cells. HepG2 cells transcribed apob RNA but were incapable of significant RNA editing due to their impaired ability to express apobec-1 mrna (Fig. 2, lane 1 and refs 35,36). These cells expressed auxiliary factors, however, as evident by the 70-fold induction of C6666 editing upon expression of exogenously transfected apobec-1 cdna (lane 2 and refs 35,36). Over-expression of APOBEC-1 also induced promiscuous editing of 5 cytidines C6655, C6651 and C6648 (Fig. 1C and Fig. 2, lane 2). Barely quantifiable editing of cytidines 3 of C6666 (at C6675, C6702 and C6724) was detected in wild type HepG2 cells (Fig. 1C and Fig. 2, lanes 3, 5 and 7) which became pronounced upon APOBEC-1 over-expression (Fig. 2, lanes 4, 6 and 8). The editing efficiency at these 3 cytidines is up to 4-fold higher than that determined at the same cytidines on humapob wt RNA expressed in the McAPOBEC cells (compare percent editing in Figure 1D with that in Figure 2, lanes 4, 6 and 8). Of the 3 hyper-editing sites, C6702 had the highest editing efficiency. A comparison of the editing activities in HepG2-APO- BEC and McAPOBEC cells demonstrated that the former cells support proportionately less 5 promiscuous editing and more 3 hyper-editing. Analysis of RNA sequences determining promiscuous editing To evaluate the RNA sequence requirements for promiscuous editing, a 55 nt deletion construct of humapob wt RNA, 17E37 (nt 6649 6703) was expressed in control McArdle cells and in a stable McArdle cell line over-expressing rat APOBEC-1 (McA-

1647 Nucleic Acids Research, 1994, 1998, Vol. Vol. 22, 26, No. No. 17 1647 A B C D Figure 1. Promiscuous editing is independent of the species source of APOBEC-1. (A) Western blot analysis of total cell lysates from clonal McArdle cell lines over-expressing equivalent amounts of the indicated species source of APOBEC-1. HA epitope tagged proteins were detected as described in Materials and Methods and visualized by autoradiography. Sub-cloning strategies employed to generate hemagglutinin tagged APOBEC-1 proteins resulted in the rat APOBEC-1 fusion protein being only four amino acids shorter than the human and rabbit protein. (B) Poisoned primer extension analyses were performed upon RT PCR-amplified transiently transfected human apob (humapob wt) or the endogenous rat apob (ratapob wt) templates from each clonal cell line in (A). Primer extension stops are indicated to the right of the lanes. Note that unedited RNA in the assayed population would generate a stop at C6666. Editing at C6655 would yield a stop at C6655 and promiscuous editing would yield a stop at C6651. The efficiency of editing at a given 5 -located cytidine is determined as an average of four separate assays from duplicate transfections and is indicated below. (C) The RNA sequence surrounding C6666 within the 448 ribonucleotide humapob wt and ratapob wt RNA substrates is shown. Cytidines assayed for editing are indicated in bold and the primers used for poisoned primer extension analyses are shown beneath each sequence and described in Materials and Methods. UGAU motifs are shown as boxed areas. (D) Editing efficiencies at 3 -located cytidines on the amplified transiently transfected humapob wt or ratapob wt templates as determined by poisoned primer extension assays are stated beneath as an average as detailed in (B). POBEC cells). Relative to the 448 nt humapob wt RNA (Fig. 3A, lane 1 and Fig. 3B), 17E37 RNA was edited at C6666 with low efficiency (lane 3 and refs 12,37). Previous findings suggested that the absence of 3 apob mrna in 17E37 was responsible for its reduced editing efficiency (12). Editing of C6666 in the 17E37 RNA construct increased 30-fold in McAPOBEC cells (lane 4). Despite this large stimulation in C6666 editing, only a relatively modest amount of 5 promiscuous editing (at C6651) could be demonstrated compared with that seen on the 448 nt humapob wt RNA (lane 2). Promiscuous editing of apob specific cytidines 3 of C6666 could not be detected on 17E37 RNA. The data suggested that additional 5 and 3 flanking sequence (found in humapob wt) may be necessary for efficient promiscuous editing activity. To examine this possibility, RNA sequences 5 of C6666 were evaluated using the translocation construct, 3 TL described previously (27,37). Translocation of the mooring sequence, along with the entire 3 end of humapob wt to a position just downstream of an otherwise unedited cytidine (C6434), directed low level editing activity to this cytidine and thereby demonstrated that the mooring sequence was required for editing (37). A 15-fold increase in C6434 editing was observed on transiently expressed 3 TL RNA in McAPOBEC cells (Fig. 3A, lanes 5 and 6 and ref. 29) compared with the 7- and 32-fold stimulation of editing activity seen on humapob wt and 17E37 RNAs respectively. Despite the high level of induction at C6434, the relative efficiency of promiscuous editing at C6431 on 3 TL RNA was greatly reduced when compared with promiscuous editing of a comparably positioned 5 cytidine on humapob wt RNA (at C6655) but was significantly higher than that observed at C6655 on 17E37. In contrast, promiscuous editing on cytidines 3 of C6434 in 3 TL RNA demonstrated a similar trace level or

1648 Nucleic Acids Research, 1998, Vol. 26, No. 7 Figure 2. Promiscuous editing in the context of human hepatoma cell line auxiliary factors. Poisoned primer extension analyses were performed upon RT PCR-amplified endogenous human apob mrna from parental HepG2 cells and a clonal cell line that over-expressed rat APOBEC-1 (HepG2-APOBEC). Primers used are given above each pair of lanes. Odd numbered lanes represent the results from wild type HepG2 cells; the even numbered lanes represent the results from the HepG2-APOBEC cell line. The efficiency of editing at a given cytidine is stated beneath. The primer extension stops are labeled as described in Figure 1. absence of activity to that observed in humapob wt RNA (Fig. 3, note that sites C6675, C6702 and C6724 in humapob wt have been renumbered as C6444, C6471 and C6493 in 3 TL). Taken together, the data demonstrated that for a given RNA construct, over-expression of APOBEC-1 did not affect all cytidines proportionally, and therefore suggested that sequence context must be important. Moreover, the editing efficiency of a cytidine within the context of the tripartite motif (wild type editing) appeared to be under different control to that at cytidines of promiscuous sites. Construct 3 TL has less 5 flanking sequence than exists in the humapob wt construct and this may have given rise to the reduced amount of promiscuous editing. Construct US-6434 6 has a comparable amount of 5 and 3 flanking sequence as humapob wt but has a wild type tripartite motif encompassing C6434 as a result of site-directed mutagenesis. The construct therefore contains entirely different 5 and 3 flanking sequences than humapob wt but contains C6434 with more of the 5 sequence found in the 3 TL construct (Fig. 3B). US-6434 6 therefore represents an ideal substrate for a further evaluation of the relative contributions of bulk 5 and 3 RNA and tripartite motif to promiscuous editing. Editing of US-6434 6 at C6434 was stimulated 7.5-fold in McAPOBEC cells relative to control McArdle cells (Fig. 3A, lanes 7 and 8 and Fig. 3B), comparable with that observed at C6666 in humapob wt but less than that observed at C6434 in 3 TL RNAs. Interestingly, no promiscuous editing could be detected 5 of C6434. Only trace or no promiscuous editing activity could be detected 3 of C6434 (Fig. 3C). The data suggested that despite ample 5 and 3 flanking sequence for C6434 editing, promiscuous editing was impaired in US-6434 6. Given that promiscuous editing occurred in the context of 5 flanking sequences of humapob wt, 17E37 and 3 TL (all with common 3 flanking sequence but different 5 flanking sequences), failure of US-6434 6 to promiscuously edit must be related to its unique 3 flanking sequence (Fig. 3B). In this regard, the most striking difference in the US-6434 6 3 flanking sequence is the absence of a tandem pair of UGAU sequences found in humapob wt at nt 6695 6698 and 6703 6706. To evaluate the relative importance of the mooring sequence and tandem UGAU repeats as cis-acting elements controlling wild type and promiscuous editing site utilization, two additional constructs were evaluated. In the construct 13-MI, the mooring sequence was eliminated from humapob wt by mutagenesis (Fig. 3B). Transient expression of this RNA revealed no editing of C6666 in control McArdle cells (Fig. 3A, lane 9). A very low level of editing at C6666 could be demonstrated on 13-MI in McAPOBEC cells and only a trace amount of promiscuous editing at C6655 was observed (lane 10). Only trace levels, or no editing activity, could be demonstrated on cytidines 3 at C6677, C6702 and C6724 in this construct. The data corroborate those from previous reports that demonstrated the importance of the mooring sequence for editing at both the wild type (14) and promiscuous (29) sites. In this regard, the absence of the mooring sequence produced a similar-fold reduction in editing efficiency at C6666 and C6655. The construct 13 87C( )6688 6710 contains a wild type tripartite editing motif but lacks a portion of 3 flanking sequence containing the tandem UGAU motifs. In control McArdle cells, the editing efficiency of C6666 in 13 87C( )6688 6710 was comparable with that observed in humapob wt RNA (Fig. 3A, compare lanes 11 and 1). Over-expression of APOBEC-1 stimulated editing of C6666 in 13 87C( )6688 6710 to a similar extent to that measured on humapob wt (Fig. 3A, compare lanes 12 and 2). In contrast, promiscuous editing of the 5 cytidine C6655 was reduced to 13% in 13 87C( )6688 6710 compared with 42% in humapob wt. Promiscuous editing of the 3 cytidine C6724 [renumbered as C6701 in 13 87C( )6688 6710] remained below detection limits (Fig. 3C). Taken together with the editing data from US-6434 6, these results strongly suggest the importance of the tandem UGAU motifs in the promiscuous editing of cytidines 5 of the mooring sequence. The data also suggest a minimal requirement for the tandem UGAU motifs for the editing of C6666 or of cytidines 3 of the mooring sequence. RNAs lacking tandem UGAU motifs 3 of the mooring sequence do not support promiscuous editing of 5 cytidines in McArdle cells Utilizing a modified differential display technique, an mrna encoding NAT-1 was determined to be edited at multiple cytidines (31). Hyper editing of this RNA was dispersed at varying distances 5 and 3 of five mooring sequence like motifs contained within the transcript. To investigate whether NAT-1 mrna was edited in response to over-expression of APOBEC-1 in cultured cells, NAT-1-specific RT PCR products were amplified from control McArdle and HepG2 cells and their apobec-1 cdna stable transfected cell line counterparts. Two separate regions of NAT-1 RNA were amplified (nt 1963 2626 and 3479 3757) which had >99% identity with mouse NAT-1 and contained three and two mooring sequence like motifs respectively (31). The amplicons included 130 and 42 total cytidines respectively, of which 25 and 13 respectively, would be

1649 Nucleic Acids Research, 1994, 1998, Vol. Vol. 22, 26, No. No. 17 1649 Figure 3. RNA sequence requirements defining 5 promiscuous editing. (A) Analysis of promiscuous editing efficiencies in McAPOBEC cells at cytidines 5 of the mooring sequence. Poisoned primer extension analyses were performed upon RT PCR-amplified transiently expressed apob RNA templates as described in Materials and Methods. The construct name and specific primer used for poisoned primer extension is given above each pair of lanes. The odd numbered lanes represent the results from wild type McArdle cells; the even numbered lanes represent the results from the McAPOBEC cell line. The editing efficiency at a given cytidine is stated beneath. The strategy for labeling poisoned primer extension stops is as described in Figure 1. (B) The sequences of the wild type and mutant apob RNA substrates are shown. Ribonucleotides shown in capitals are vector derived. Cytidines assayed for editing are shown in bold and the primers used for poisoned primer extension analyses shown beneath each sequence, described in Materials and Methods and stated in (A). UGAU motifs are shown as boxed areas. (C) Editing of cytidines 3 of C6666 was assayed by poisoned primer extension upon the same templates utilized in (B) with the specific primers shown in (A). The efficiency of editing in McArdle cells (odd numbered columns) and the McAPOBEC cell line (even numbered columns) is stated for each cytidine. expected to be hyper-edited. Sequencing of eight independent clones of each region from each parental and APOBEC-1-overexpressing cell line did not reveal C to U editing (sequence data not shown). The mooring sequence motifs centered at nt 2431 and at nt 3612 were 91 and 82% identical with that found adjacent to C6666 in apob RNA, and each had a cytidine positioned 6 nt upstream, i.e. only one more than that at the wild type apob RNA editing site. Neither cytidine was edited in control or APO- BEC-1-over-expressing cells lines (sequence data not shown). This contrasts significantly with the 3 and 47% editing efficiency of 3 TL RNA in McArdle and McAPOBEC cells respectively, which also contained a longer spacer element. The remaining three mooring sequence motifs also did not support editing of 5 located cytidines as might have been

1650 Nucleic Acids Research, 1998, Vol. 26, No. 7 Figure 4. Promiscuous editing site utilization and RNA substrate specificity in vitro. DNA sequencing of individual clones of RT PCR-derived NAT-1 amplimers from McArdle and the McAPOBEC cell line revealed no C to U editing events (see text). 40 µg of cell extract prepared from McArdle (odd numbered lanes) and McAPOBEC (even numbered lanes) cells were incubated with in vitro transcribed NAT-1 (3479 3757), wild type apob (6411 6860) and mutant apob ( 3 TL, see Fig. 3B) RNAs. Poisoned primer extension analyses of 5 and 3 cytidines were performed using the primers stated above each pair of lanes. Editing efficiencies at the indicated cytidine are stated beneath. predicted from the extent to which they diverged further from the consensus mooring sequence motif. The data suggested that the intracellular complement of auxiliary factors that permit hyperediting of NAT-1 in mouse liver when APOBEC-1 is over-expressed do not exist in McArdle and HepG2 cells. To rule out the possibility that cellular trafficking or compartmentalization of RNA imposed constraints which prevented NAT-1 mrna from being promiscuously edited, we evaluated the question in the context of editing competent cell extracts prepared from McAPOBEC cells. HumapoB wt RNA supported editing of C6666 when added to McArdle cell extracts and demonstrated enhanced editing in McAPOBEC cell extracts (Fig. 4, lanes 1 and 2). Promiscuous editing of the 5 cytidine C6655 was readily apparent (lane 2) but only trace levels or no promiscuous editing of the 3 cytidines C6675, C6702 and 6724 could be demonstrated (lanes 3 8). These data demonstrated that the in vitro system recapitulates site specific and promiscuous apob RNA editing in cells under conditions of both normal and over-expressed APOBEC-1. Addition of a 278 nt NAT-1 RNA substrate (nt 3479 3757) to control McArdle and McAPOBEC cell extracts under standard editing assay conditions resulted in no editing in either extract (Fig. 4, lanes 9 and 10). This is in contrast with the similarly sized 230 nt translocation mutant of apob, 3 TL, which was edited with 2 and 12% efficiency in McArdle and McAPOBEC cell extracts respectively (lanes 11 and 12). In this regard promiscuous editing of the 5 cytidine C6431 was also observed in McAPO- BEC cell extracts (lane 12). Taken together, the data suggested that NAT-1 was an inappropriate RNA substrate for C to U editing in McArdle and HepG2 cells under either normal or promiscuous editing conditions. DISCUSSION In this study we have evaluated some of the cellular and molecular properties that contribute to editing at C6666, the wild type site of apob mrna, and those that contribute to the aberrant editing of other cytidines (promiscuous editing) when APOBEC-1 is over-expressed. The data corroborate and extend previous findings by demonstrating that promiscuous editing in McArdle cells is apob mrna specific, occurs 5 of the mooring sequence and is dependent on the mooring sequence. Examination of several cytidines downstream of C6666 demonstrated that editing of these sites (referred to as hyper-editing) occurred at levels close to or below the detection limits of the primer extension analysis. This contrasts with studies in APOBEC-1 transgenic mice and rabbits that identified hyper-editing as C to U specific editing that occurred independent of the mooring sequence in apob as well as alternative RNA substrates. Furthermore, mutations of the mooring sequence that reduced site specific editing at C6666 also reduced promiscuous editing efficiencies. These data demonstrated that over-expression of APOBEC-1 in McArdle cells and in transgenic mice or rabbits, while resulting in similar levels of C6666 editing, does not induce aberrant editing of the same cytidines. This finding suggested that the biology of the two experimental systems must determine promiscuous or hyper-editing of apob mrna. Our studies indicated that the characteristics of aberrant site editing were independent of the species source of APO- BEC-1. This suggested that the potential issue of homologous versus heterologous components in the experimental systems may be a minimal consideration in understanding the observed differences in editing site utilization. In contrast, auxiliary factors may have played a major role in determining both the distribution of edited cytidines and the proportion of apob RNAs within the cellular population that were affected. Auxiliary protein factors are just beginning to be characterized from the point of view of the number of proteins involved and their function (18 20,24,30,39,40). It is currently clear that APOBEC-1 cannot edit apob mrna without these factors and that the occurrence of auxiliary factors in nature is not restricted to cells that express apob mrna or APOBEC-1 (5,21,25). The data demonstrated that C6666 editing was stimulated in McArdle and HepG2 cells by over-expressing APOBEC-1 and

1651 Nucleic Acids Research, 1994, 1998, Vol. Vol. 22, 26, No. No. 17 1651 that promiscuous editing occurred predominantly on cytidines 5 of C6666. In contrast with McArdle cells, significantly higher (4-fold) levels of editing of cytidines 3 of C6666 occurred in HepG2 cells. The relative amount of 5 promiscuous editing in HepG2 cells at C6655 compared with C6666 (5.5 28%; Fig. 2) was less than that observed at these sites in McAPOBEC cells (42 84%; Fig. 3). This is an interesting difference in that McArdle and HepG2 cells could be viewed as similar in that they are both hepatoma cell lines, express apob mrna and express sufficient auxiliary factors to support enhanced C6666 editing when APOBEC-1 is over-expressed. However, these cells differ in species origin and, importantly, in the fact that McArdle cells express their own APOBEC-1 and edit apob mrna whereas HepG2 cells have low or no expression of APOBEC-1 and do not significantly edit apob mrna without apobec-1 cdna transfection. These two cell lines must be viewed, therefore, as related but not identical auxiliary factor contexts. In addition, we cannot rule out the possibility at this time that the multiple proteins which make up the auxiliary factor requirement may not be expressed at equal levels in the two cell lines or (due to endogenous APOBEC-1 expression) may not be equally accessible to the over-expressed enzyme. A reasonable proposal from these data is that promiscuous editing in McArdle cells and hyper-editing in transgenic mouse and rabbit liver is in part due to different contexts of auxiliary protein factors. We cannot formally rule out the possibility that the location of additional editing sites is determined by differences in the levels of active APOBEC-1 between the two cell lines. However, our previous data indicated that increasing levels of over-expression of APOBEC-1 resulted in higher levels of editing both at C6666 and additional editing sites, but did not alter the location of these sites within apob mrna (29). Additionally, there are no differences in the relative subcellular localization of the over-expressed enzyme between the two cell lines that could account for the altered utilization of 5 and 3 editing sites (41). It is further unlikely that differences in the relative expression levels of apob mrna between the two hepatoma cell lines caused the shift from 5 to 3 editing site utilization (or vice versa), as we have shown previously that the editing efficiency of a given apob mrna is determined by RNA sequences surrounding the editing site, the proximity of introns and the levels of APOBEC-1 (27,29), but not by the level of the apob mrna substrate (27,42). Previous studies have demonstrated that when APOBEC-1 was over-expressed, cytidines as far as 50 nt 5 of the mooring sequence were edited, thus distinguishing promiscuous editing from editing at C6666 in lacking a spacer or enhancer element requirement (29). The data in this report corroborate these findings and extend them to promiscuous editing in HepG2 cells over-expressing APOBEC-1. An important extension of the current study was the identification of important RNA sequence 3 of the mooring sequence which selectively affected the efficiency of promiscuous editing of cytidines 5 of C6666. Deletion of a pair of tandem UGAU motifs [RNA constructs 13 87C( )6688 6710 and 17E37] selectively impaired promiscuous editing 5 of C6666 but did not impair the enhanced editing of C6666 resulting from APOBEC-1 over-expression. Similarly, there were no UGAU motifs immediately downstream of the mooring sequence within the apob RNA construct US-6434 6 and this RNA supported enhanced editing of C6666 by APOBEC-1 over-expression but did not demonstrate detectable levels of promiscuous editing. For the same reasons cited above, different levels of expression of the various apob RNA substrates would not alter the editing site preference of the over-expressed APOBEC-1 (27,42). The UGAU motif is a predominant feature of the RNA sequence surrounding the editing site in apob RNA. Expected to occur at random once every 256 (4 4 ) nucleotides, it occurs eight times in the 240 nt between 6497 and 6737. The RNA construct 13 87( )6688 6710 retained one UGAU motif 3 of the mooring sequence after deletion of the tandem pair of UGAU motifs. Promiscuous editing was substantially reduced on this construct relative to that seen on humapob wt. The complete ablation of promiscuous editing in US-6434 6 was unique to this construct. US-6434 6 also contains one UGAU motif 3 of the mooring sequence but this motif is 18 nt further 3 relative to the mooring sequence than that found in 13 87( )6688 6710. We currently do not know whether the distance of the UGAU motif from the mooring sequence, or other yet to be identified sequences, account for the extremely low levels of promiscuous editing upon the US-6434 6 RNA substrate. A more detailed comparative analysis of the differences between humapob wt and US-6434 6 3 in their 5 flanking sequence or RNA secondary structures will be required to completely resolve this issue. The naturally occurring mrna NAT-1 lacked tandem UGAU motifs proximal to mooring sequence homologs but contained four mooring sequence like motifs ranging from 73 to 91% identical to the canonical apob motif. This mrna was hyperedited extensively in transgenic mouse liver but was not edited in McAPOBEC or HepG2-APOBEC cells nor under in vitro editing conditions. In terms of the sequence preferences described here for promiscuous editing, only one of the hyper-edited cytidines, at nt 2069, in NAT-1 RNA was flanked by a UGAU motif. Moreover, UGAU motifs only occurred 18 times in 3803 nt of mouse NAT-1 RNA, i.e. once every 211 nt, close to its expected random frequency of occurrence. At one level, the data support our overall proposal that auxiliary factor context determines promiscuous editing site utilization. The complete lack of editing activity on NAT-1 mrna was, however, surprising and suggested that the RNA sequence requirements determining apob mrna editing at C6666 may not be sufficient in all RNA contexts. This possibility has been suggested through the analysis of apob-albumin chimeric RNA constructs in which apob sequence 3 and distal to the mooring sequence was implicated as affecting the efficiency of editing at C6666 (12). In contrast, the mrna encoding human neurofibromin (humnf1) was predicted, by comparison with apob, to contain a tripartite motif and in fact did support C to U editing (43). Resolution of these conflicting results will require that additional edited mrnas be identified and their sequence requirements compared. The apob RNA sequences from rat, human, mouse, rabbit (44) and several other species (45) are largely identical in the vicinity of the editing site. The tandem pair of UGAU motifs is found at identical positions in the apob RNAs from humans and rabbits, but not from mice or rats. This might explain the reduced hyper-editing upon rabbit and human apob RNAs compared with mouse apob mrna in transgenic mice, whilst there was no significant difference in the extent of promiscuous editing upon each RNA substrate (30). Furthermore, the very low levels of promiscuous editing upon the endogenous McArdle cell apob RNA, compared with the transfected human apob substrate, is now easily explained, as rat apob lacks this pair of tandem UGAU motifs. Clearly, however, the presence or absence of the tandem

1652 Nucleic Acids Research, 1998, Vol. 26, No. 7 UGAU motifs is not the decisive mechanistic switch between 5 localized promiscuous editing and 3 hyper-editing. Promiscuous editing has been reported in only two species under normal levels of APOBEC-1 expression (45). In apob RNA of liver and intestine from dogs and the intestine from cows, C to U editing was observed at C6655. It is noteworthy that apob RNA from these species, like that from humans, includes the tandem UGAU motif. Other apob RNAs that possess this tandem UGAU motif include horse, cat and sheep, although no promiscuous editing was detected on their liver or intestine apob RNA. The significance of this is unclear but it is interesting in that, like humans, sheep and cats have no detectable hepatic editing activity. Conversely, horses and sheep exhibit lower (72 and 40% respectively) intestinal editing activity compared with other species in which >90% editing efficiencies were observed. In conclusion we have demonstrated that promiscuous editing of cytidines 5 of the wild type site at C6666, induced by experimental over-expression of APOBEC-1, is independent of the species origin of the catalytic sub-unit of the editosome. A comparison of rat and human sources of auxiliary factors revealed, however, differences in the relative proportions of 5 and 3 additional editing sites suggesting cell-type-specific differences in auxiliary proteins. Promiscuous editing was dependent upon the mooring sequence but was primarily regulated by a pair of tandem UGAU motifs located 3 of the mooring sequence. ApoB RNA and alternative RNA substrates hyper-edited in the livers of APOBEC-1 transgenic animals that lack these motifs were not promiscuously edited in McArdle or HepG2 cells. We propose that RNA substrate specificity and editing site fidelity require a combination of unique RNA sequence and appropriate stoichiometry of APOBEC-1 and homologous auxiliary factors in the editosome. ACKNOWLEDGEMENTS The authors wish to thank Jenny M.L.Smith for the preparation of the figures. This work was supported in part through a Public Health Service Grant DK43739 and grants from The Council for Tobacco Research and the Rochester Area Pepper Center for Research on Aging awarded to H.C.S. REFERENCES 1 Chen, S.-H., Habib, G., Yang, C.-Y., Gu, Z.-W., Lee, B.R., Weng, S.-A., Silberman, S.R., Cai, S.-J., Deslypere, J.P., Rosseneu, M., Gotto, A.M., Jr, Li, W.-H. and Chan, L. (1987) Science, 328, 363 366. 2 Powell, L.M., Wallis, S.C., Pease, R.J., Edwards, Y.H., Knott, T.J. and Scott, J. (1987) Cell, 50, 831 840. 3 Hadjiagapiou, C., Giannoni, F., Funahashi,T., Skarosi, S.F. and Davidson, N.O. (1994) Nucleic Acids Res., 22, 1874 1879. 4 Teng, B., Burant, C.F. and Davidson, N.O. (1993) Science, 260, 1816 1819. 5 Yamanaka, S., Poksay, K.S., Balestra, M., Zeng, G.-Q. and Innerarity, T.L. (1994) J. Biol. Chem., 269, 21725 21734. 6 Lau, P.P., Zhu, H-J., Baldini, A., Charnsangavej, C. and Chan, L. (1994) Proc. Natl. Acad. Sci. USA, 91, 8522 8526. 7 Oka, K., Kobayashi, K., Sullivan, M., Martinez, J., Teng, B.-B., Ishimura-Oka, K. and Chan, L. (1997) J. Biol. Chem., 272, 1456 1460. 8 Navaratnam, N., Bhattacharya, S., Fujino, T., Patel, D., Jarmuz, A.L. and Scott, J. (1995) Cell, 81, 187 195. 9 Anant, S., MacGinnitie, A.J. and Davidson, N.O. (1995) J. Biol. Chem., 270, 14762 14767. 10 Smith, H.C. (1993) Semin. Cell Biol., 4, 267 278. 11 Smith, H.C., Gott, J.M. and Hanson, M.R. (1997) RNA, 3, 1105 1123. 12 Backus, J.W. and Smith, H.C. (1994) Biochim. Biophys. Acta, 1217, 65 73. 13 Driscoll, D.M, Lakhe-Reddy, S., Oleksa, L.M. and Martinez, D. (1993) Mol. Cell. Biol., 13, 7288 7294. 14 Backus, J.W. and Smith, H.C. (1992) Nucleic Acids Res., 20, 6007 6014. 15 Harris, S.G., Sabio, I., Mayer, E., Steinberg, M.F., Backus, J.W., Sparks, J.D., Sparks, C.E. and Smith, H.C. (1993) J. Biol. Chem., 268, 7382 7392. 16 Navaratnam, N., Shah, R., Patel, D., Fay, V. and Scott, J. (1993) Proc. Natl. Acad. Sci. USA, 90, 222 226. 17 Smith, H.C., Kuo, S.R., Backus, J.W., Harris, S.G., Sparks, C.E. and Sparks, J.D. (1991) Proc. Natl. Acad. Sci. USA, 88, 1489 1493. 18 Yang, Y. and Smith, H.C. (1996) Biochem. Biophys. Res. Commun., 218, 797 801. 19 Mehta, A., Banerjee, S. and Driscoll, D.M. (1996) J. Biol. Chem., 271, 28294 28299. 20 Yang, Y., Yang, Y., Kovalski, K. and Smith, H.C. (1997) J. Biol. Chem., 272, 27700 27706. 21 Driscoll, D.M. and Zhang, Q. (1994) J. Biol. Chem., 269, 19843 19847. 22 Phung, T.L., Sowden, M.P., Sparks, J.D., Sparks, C.E. and Smith, H.C. (1996) Metabolism, 45, 1056 1058. 23 Giannoni, F., Chou, S.C., Skarosi, S.F., Verp, M.S., Field, F.J., Coleman, R.A. and Davidson, N.O. (1995) J. Lipid Res., 36, 1664 1675. 24 Funahashi,T., Giannoni, F., DePaoli, A.M., Skarosi, S.F. and Davidson, N.O. (1995) J. Lipid Res., 36, 414 428. 25 Inui, Y., Giannoni, F., Funahashi, T. and Davidson, N.O. (1994) J. Lipid Res., 35, 1477 1489. 26 Bostrom, K., Garcia, Z., Poksay, K.S., Johnson, D.F., Lusis, A.J. and Innerarity, T.L. (1990) J. Biol. Chem., 265, 22446 22452. 27 Sowden, M., Hamm, J.K., Spinelli, S. and Smith, H.C. (1996) RNA, 2, 274 288. 28 MacGinnitie, A.J., Anant, S. and Davidson, N.O. (1995) J. Biol. Chem., 270, 14768 14775. 29 Sowden, M., Hamm, J.K. and Smith, H.C. (1996) J. Biol. Chem., 271, 3011 3017. 30 Yamanaka, S., Poksay, K.S., Driscoll, D.M. and Innerarity, T.L. (1996) J. Biol. Chem., 271, 11506 11510. 31 Yamanaka, S., Poksay, K.S., Arnold, K.S. and Innerarity, T.L. (1997) Genes Dev., 11, 321 333. 32 Yamanaka, S., Balestra, M.E., Ferrell, L., Fan, J., Arnold, K.S., Taylor, S., Taylor, J.M. and Innerarity, T.L. (1995) Proc. Natl. Acad. Sci. USA, 92, 8483 8487. 33 Driscoll, D.M. Wynne, J.K., Wallis, S.C. and Scott, J. (1989) Cell, 58, 519 525. 34 Backus, J.W., Schock, D. and Smith, H.C. (1994) Biochim. Biophys. Acta, 1219, 1 14. 35 Giannoni, F., Bonen, D.K., Funahashi, T., Hadjiagapiou, C., Burant, C.F. and Davidson, N.O. (1994) J. Biol. Chem., 269, 5932 5936. 36 Hirano, K.-I., Min, J., Funahashi, T., Baunoch, D.A. and Davidson, N.O. (1997) J. Lipid Res., 38, 847 859. 37 Backus, J.W. and Smith, H.C. (1991) Nucleic Acids Res., 19, 6781 6786. 38 Shah, R.R., Knott, T.J., Legros, J.E., Navaratnam, N., Greeve, J.C. and Scott, J. (1991) J. Biol. Chem., 266, 16301 16304. 39 Schock, D., Kuo, S.R., Steinberg, M.F., Bolognino, M., Sparks, J.D., Sparks, C.E. and Smith, H.C. (1996) Proc. Natl. Acad. Sci. USA, 93, 1097 1102. 40 Lau, P.P., Zhu, H.J., Nakamuta, M. and Chan, L. (1997) J. Biol. Chem., 272, 1452 1455. 41 Yang, Y., Yang, Y. and Smith, H.C. (1997) Proc. Natl. Acad. Sci. USA, 94, 13075 13080. 42 Collins, H.L., Sowden, M.P., Smith, H.C., Sparks, J.D. and Sparks, C.E. (1997) Circulation, 96, Suppl S, 3041. 43 Skuse, G.R., Cappione, A.J., Sowden, M.P., Metheny, L.J. and Smith, H.C. (1996) Nucleic Acids Res., 24, 478 486. 44 Teng, B., Black, D.D. and Davidson, N.O. (1990) Biochem. Biophys. Res. Commun., 173, 74 80. 45 Greeve, J., Altkemper, I., Dieterich, J.-H., Greten, H. and Windler, E. (1993) J. Lipid Res., 34 1367 1383.