Mutagenesis of apobec-1 complementation factor reveals distinct domains that modulate RNA binding, protein-protein interaction with apobec-1, and complementation of C to U RNA-editing activity.

C to U editing of apolipoprotein B (apoB) RNA requires a multicomponent holoenzyme complex in which minimal constituents include apobec-1 and apobec-1 complementation factor (ACF). We have examined the predicted functional domains in ACF in binding apoB RNA, interaction with apobec-1, and complementation of RNA editing. We demonstrate that apoB RNA binding and apobec-1-interacting domains are defined by two partially overlapping regions containing the NH(2)-terminal RNA recognition motifs of ACF. Both apoB RNA binding and apobec-1 interaction are required for editing complementation activity. ACF is a nuclear protein that upon cotransfection with apobec-1 results in nuclear colocalization and redistribution of apobec-1 from the cytoplasm. ACF constructs with deletions or mutations in the putative nuclear localization signal (NLS) still localize in the nucleus of transfected cells but do not colocalize with apobec-1, the latter remaining predominantly cytoplasmic. These observations suggest that the putative NLS motif in ACF is not responsible for its nucleo-cytoplasmic trafficking. By contrast, protein-protein interaction is important for the nuclear import of apobec-1. Taken together, these data suggest that functional complementation of C to U RNA editing by apobec-1 involves the NH(2)-terminal 380 residues of ACF.

other example is C to U editing, the prototype for which is mammalian apolipoproteinB (apoB) 1 RNA (5,6). The editing of apoB RNA is mediated by a multicomponent enzyme complex, in which the nuclear apoB transcript undergoes a site-specific C to U deamination that introduces a UAA stop codon into the edited mRNA. This process occurs in enterocytes of the mammalian small intestine and leads to the production of a truncated protein (apoB48) that is required for the transport of triglyceride and cholesterol into the circulation (reviewed in Ref. 7).
C to U editing of apoB RNA requires a defined sequence context, and considerable information exists concerning the cis-acting elements involved (8 -11). These elements together facilitate the interaction of the apoB RNA template with a multicomponent holoenzyme or editosome to allow optimal presentation of the targeted cytidine to the active site of the catalytic deaminase, apobec-1 (12). Apobec-1 is a 27-kDa RNAspecific cytidine deaminase that functions as a homodimer but in addition requires the presence of other proteins for deamination of the targeted cytidine in apoB RNA (13)(14)(15). Apobec-1 alone, although competent to mediate deamination of cytidines in the mononucleoside form, is inactive on an RNA substrate (16,17). The recent simultaneous identification of a novel gene, apobec-1 complementation factor (ACF), by Driscoll and colleagues (18) and Greeve and colleagues (19) has facilitated the understanding of the protein components involved and their functional role in the apoB RNA-editing holoenzyme. A comparison of the modular structure of ACF has been further facilitated by the recent identification of a related gene, GRY-RBP, which while sharing several of the domains present in ACF and described below does not complement apobec-1 in mediating C to U editing of apoB RNA (20,21).
ACF is a novel 586-residue protein that contains several features of potential importance (Fig. 1). ACF cDNA predicts the presence of three RNA recognition motifs (RRMs) that is of interest in view of the observations confirmed in several studies in which ACF binds to apoB RNA (18,19,22). Indeed, this characteristic was crucial to its original isolation by Driscoll and colleagues (18,22,23). One of the unresolved issues concerning the biochemical mechanisms of C to U editing of apoB RNA, however, is the importance of substrate binding in constraining cytidine deamination to a single nucleotide. This is a particularly perplexing issue, because apobec-1 itself has RNA binding activity albeit low affinity and yet exhibits an absolute cofactor requirement (24,25). Accordingly, a central objective of our investigation was to resolve the requirement for RNA binding in the function of ACF and to identify the domains responsible. For example, in addition to three RRMs, ACF contains a region with six RG residues and a COOH-terminal double-stranded RNA binding domain ( Fig. 1) (18,19), each of which has potential significance in relation to apoB RNA binding activity (26,20).
Another issue concerns the topology of C to U editing in relation to the subunits of the holoenzyme. Available evidence suggests that apoB RNA editing is predominantly a nuclear event, although recent work suggests that it may also occur in the cytoplasmic compartment of transfected cells (27)(28)(29). Confocal immunolocalization studies suggest that apobec-1 is distributed both in the nucleus and cytoplasm of transfected cells (20,30,31). However, following cotransfection with ACF, the distribution of apobec-1 becomes predominantly nuclear, suggesting that the interaction of these two proteins leads to translocation of a complex into the nucleus (20). The functional relevance of a putative bipartite NLS in the NH 2 terminus of apobec-1 has been called into question as a result of studies from Smith and colleagues (30), suggesting that two distinct domains are required for nuclear distribution beyond the putative NLS region per se. By contrast, ACF contains a sevenresidue stretch ( 144 PKTKKRE 150 ) that resembles an SV40-type NLS (18,19). This result raised the possibility that the determinants of nuclear distribution of the holoenzyme reside not in apobec-1 but perhaps in ACF.
We have undertaken studies to localize specific domains in ACF responsible for its RNA binding activity and for mediating protein-protein interaction with apobec-1. Our findings indicate that these functional activities are distributed in distinct but partially overlapping regions of the protein, and both activities are required for efficient complementation of C to U editing. Epitope-tagged ACF bearing deletions or mutations in the putative NLS revealed a nuclear distribution, suggesting that this domain is not required for nuclear localization. However, the interaction of the NLS mutants with apobec-1 was impaired, and the distribution of apobec-1 in cotransfected cells remained predominantly cytoplasmic compared with the nuclear colocalization found with wild-type ACF. The collective results suggest that the functional domains of ACF involved in C to U editing of apoB RNA are localized in the regions spanning residues 1-380, particularly the second and third RRM.

MATERIALS AND METHODS
Cloning and Expression of Recombinant ACF Mutants-All mutants were constructed by a two-step PCR method (32). The full-length PCR products were sequenced and subcloned into pTYB1 (New England Biolabs) using the NdeI-SapI sites, which generate a COOH-terminal intein fusion protein as described previously (20). For the expression of GRY-RBP/ACF chimeras, chimera 1 was cloned into pTYB1 using the NheI-SapI sites, whereas chimeras 2 and 3 were cloned into pTYB2 using the NheI-SmaI sites. Each mutant was also subcloned into pCMV-Tag 2B (Stratagene). Using two internal restriction sites for ACF, EcoNI, and BstEI, the mutated sequences were digested from the pTYB1-ACF vector and cloned into pCMV-Tag 2B-ACF. This cloning step generated an NH 2 -terminal FLAG-tagged fusion protein. Recombinant proteins were expressed as intein fusion products and purified to homogeneity using chitin affinity chromatography as described previously (20), using ER2566 cells as recommended by the manufacturer. The cell pellet from a 1-liter culture was resuspended in 50 ml of lysis buffer (20 mM HEPES, pH 8.0, 500 mM NaCl, 1 mM EDTA, 0.5% Triton X-100, 0.5 mM phenylmethylsulfonylfluoride. The lysate was loaded on a 10-ml chitin column, and the recombinant protein eluted in the presence of 1 M NaCl and 50 mM dithiothreitol. Fractions containing the protein of interest were pooled and dialyzed against 20 mM HEPES, pH 8.0, 100 mM KCl, 0.2 mM EDTA, 0.5 mM EDTA, 0.5 mM dithiothreitol, 20% glycerol, 0.5 mM phenylmethylsulfonylfluoride. Purity was assessed by denaturing SDS-PAGE and determined to be Ͼ95% homogeneous in all cases. The secondary structure predictions were obtained using the GORIV method provided by the Network Protein Sequence Analysis web server (npsa-pbil.ibcp.fr).
UV Cross-linking to ApoB RNA-A 32 P-labeled rat apoB RNA template (50,000 cpm at 4 ϫ 10 8 cpm/g) was incubated with 100 ng of wild-type or mutant ACF for 15 min at room temperature in a binding buffer containing 20 mM HEPES, pH 8.0, 100 mM KCl, 1 mM EDTA, 0.25 mM dithiothreitol, and 2.5% glycerol. The RNA was then treated sequentially with RNase T1 (2 units/l final concentration) and heparin (2 mg/ml final concentration). The mixture was UV-irradiated on ice in a Stratalinker (Stratagene) at an energy of 250 millijoules/cm 2 and analyzed by SDS-10% PAGE under reducing conditions. C to U RNA-editing Activity-20 femtomoles of a 470-nucleotide rat apoB RNA was incubated with 250 ng of glutathione S-transferaseapobec-1 and 2 ng of wild-type or mutant ACF in buffer containing 20 mM HEPES, pH 8.0, 100 mM KCl, 1 mM EDTA, 0.25 mM dithiothreitol for 3 h at 30°C. The RNA was reverse-transcribed and cDNA-amplified prior to primer extension, the latter using dideoxy-GTP as described previously (17). The extended products were fractionated by electrophoresis in an 8 M urea-8% polyacrylamide gel and quantitated by phosphorimaging (Molecular Dynamics, Sunnydale, CA).
Protein-Protein Interaction Revealed by Coimmunoprecipitation-Rat-apobec-1 cDNA was cloned into pCMV-Tag 3B and expressed as an NH 2 -terminal c-Myc-tagged fusion protein. In vivo protein-protein interaction studies were performed using Myc-apobec-1 and FLAG-ACF (wild-type or mutant) expressed in COS-7 cells. Cells were grown to ϳ70% confluence and transiently transfected with the indicated plasmid as detailed previously (20). 48 h after transfection, cell lysates were prepared under native conditions, and 5 l of rabbit anti-FLAG IgG (1 g/l) (PA1-984 Affinity BioReagent, Golden) was added. After incubation at 4°C for 1 h, the immune complexes were collected on protein A-agarose beads, which were extensively washed, and the complexes eluted by boiling in SDS-loading buffer for 5 min. The immunoprecipitated proteins were separated by SDS-10% PAGE, transferred to polyvinylidene difluoride membranes, and analyzed by Western blotting using mouse anti-Myc IgG (SC-40 Santa Cruz Biotechnology). Aliquots of cell lysates were also examined directly for the expression of Mycapobec-1 and FLAG-ACF by Western blotting using mouse monoclonal anti-FLAG (F3165 Sigma) or anti-c-Myc.
Immunofluorescence and Confocal Microscopy-COS-7 and McArdle rat hepatoma cells (ATCC, Manassas, VA) were grown to 50 -70% confluence on coverslips. Using 6 l of FUGENE 6 (Roche Molecular Biochemicals) or 10 l of Superfect reagent (Qiagen), the cells were transiently transfected with 2 g of pCMV-Tag 2B expressing an NH 2terminal FLAG-tagged ACF (wild-type, ⌬RRM2, or NLS/A mutant) and 2 g of pCMV-Tag 3B expressing an NH 2 -terminal c-Myc-tagged apobec-1. 48 h after transfection, the cells were fixed with 3.7% formaldehyde, permeabilized with 0.5% Triton X-100, and probed with a rabbit anti-FLAG IgG and mouse monoclonal anti-Myc IgG, followed by Cy3or fluorescein isothiocyanate-conjugated secondary IgG (Jackson Im-munoResearch). Nuclei were visualized using TO-PRO-3 iodide (Molecular Probes, Inc.). Preparations were imaged using a 63X Zeiss plan apochromatic objective and a Bio-Rad MRC 1024 confocal adaptor. A krypton-argon laser was used with epifluorescence filter sets designed for Texas Red (Cy3), fluorescein isothiocyanate, and cyanine (Cy5). The confocal aperture was set at 1.8. Usually 5-12 images at planes separated by 0.5 m were obtained. This increment allows sectioning of the FIG. 1. Schematic representation of functional domains of ACF. ACF has three RRMs represented as gray boxes. The COOHterminal region of ACF contains six RG clusters and a putative doublestranded RNA binding domain (dsRBD) indicated by black bold lines. A putative nuclear localization sequence (nls) is also present in the RRM2 as indicated by a black dot. The numbers below the scheme indicate the position of amino acid residues in ACF. ⌬84 and ⌬55 refer to deletions corresponding to alternatively spliced ACF mRNAs. entire image giving a range of a signal covering every plane of the cell in that image. Images were processed using Adobe Photoshop 4.0 software. For standard immunofluorescence microscopy, stained cells were mounted with Vectashield, and nuclei were visualized with 4,6-diamidino-2-phenylindole (Vector). Images were obtained with a Zeiss Axiashop 2 MOT microscope equipped with a ϫ 40 plan neofluar objective and a 3-CCR camera (DAGE-MTI, Inc.). A Zeiss Attoarc variable intensity lamp was used with filter sets designed for Cy3, fluorescein isothiocyanate, and 4,6-diamidino-2-phenylindole. Pictures were processed using Adobe Photoshop 4.0 software.
Oligonucleotides-The oligonucleotides used in this study are listed below. The underlined nucleotides represent restriction sites introduced for subsequent cloning into bacterial (pPCR-Script) and pTYB vectors and mammalian expression vectors (pCMV2B). The boldface nucleotides represent the mutated bases. For each mutant, the first set of PCR was performed using pTYB1-ACF plasmid as template, two external primers, 5Ј primer T7 5Ј-TAATACGACTCACTATAGGG-3Ј and 3Ј primer ACF-R 5Ј-GGTGGTTGCTCTTCCGCAGAAGGTGCCAT-ATCCATC-3Ј (SapI site), and two internal primers surrounding the region to be deleted. The full-length product was obtained using the primers ACF-F 5Ј-GGTGGTCATATGGAATCAAATCACAAATCCG-3Ј (NdeI site) and ACF-R. The internal primers used in the first set of are  2-4). The RNA was extracted, and C to U editing was analyzed by primer extension assay. The data from three independent assays were averaged (mean Ϯ S.D.) and are represented as a percent U. A representative experiment is shown. The relative mobility of the primer (P), edited U, and unedited C products are indicated. D, RNA binding activity of ACF mutants. UV cross-linking was performed by incubating 100 ng of recombinant protein with a 32 P-labeled 105-nucleotide apoB RNA (RB105), spanning the edited base. After treatment with RNase T1 and UV irradiation, the products were analyzed by SDS-10% PAGE. The molecular weight markers are shown at the left of the gel. Lane 1, wild-type ACF; lanes 2-4, ACF mutants. This is representative of three independent assays. E, coimmunoprecipitation of wild-type and mutant ACF with apobec-1 in transfected COS-7 cells. COS-7 cells were transfected simultaneously with DNA encoding c-Myc-tagged apobec-1 and FLAG-tagged ACF (lane 2) or FLAG-tagged ACF mutants (lanes 3-5). After 48 h, cell lysates were prepared, and extracts were analyzed by Western blot with anti-FLAG (upper panel) and anti-Myc IgG (middle panel). Lower panel, the extracts from transfected COS cells were immunoprecipitated with anti-FLAG antibody. The immunoprecipitates were separated by SDS-10% PAGE and analyzed by Western blotting using anti-Myc IgG. As a control, an extract from nontransfected COS cells was analyzed by Western blot and immunoprecipitation (lane 1). The data from panels C-E indicate that the NH 2 -terminal half of ACF is required for the apobec-1 complementing activity. This is representative of three independent assays.

RESULTS AND DISCUSSION
Deletional Analysis of ACF: Functional Implications from Natural Mutants of ACF-We constructed two deletion mutants (⌬N84 and ⌬55) and a full-length ACF clone with an 8-amino acid insertion (⌬ ϩ 8aa) both based upon earlier (19,20) and more recent findings (44) of alternatively spliced ACF transcripts in human liver and small intestine (Fig. 1). The dominant transcript produced in vivo, however, encodes the wild-type ACF, analogous to that used in the current construction. Each splice variant of ACF was expressed as recombinant protein (Fig. 2B) and used in an in vitro RNA-editing assay with apobec-1, and the extent of C to U editing was determined. As shown in Fig. 2C, lane 1, wild-type ACF demonstrates Ͼ80% C to U editing activity, similar to the value observed with the FIG. 3. Functional analysis of predicted domains in ACF reveals that apobec-1 complementation activity involves both the binding of ACF to apoB RNA and interaction with apobec-1. A, a representation of ACF and GRY-RBP proteins. The predicted functional domains identified in these proteins are indicated as follows: RRM motifs (gray boxes) and RG and RGG clusters (black bold lines). Sequence alignment between ACF and GRY-RBP reveals a region unique to ACF (331-385), the missing region in GRY-RBP indicated by a dashed rectangle. A putative bipartite nuclear localization sequence has been identified at the COOH terminus of GRY-RBP indicated by two black dots. The ACF mutants were constructed by deleting entire functional domains indicated by interrupted lines. The deletion mutants are identified by the missing region. B, 2 g of purified recombinant protein was analyzed by SDS-PAGE and stained with Coomassie Blue. C, 2 ng of wild-type ACF (lane 1) or mutants ACF (lanes 2-6) were added to 250 ng of glutathione S-transferase-apobec-1 in an in vitro C to U editing assay. The data represent the mean Ϯ S.D. from three independent assays. A representative assay is shown. D, RNA binding activity of mutants. 100 ng of wild-type ACF (lane 1) or ACF mutants (lanes 2-6) were incubated with radiolabeled RB105 and were UV-irradiated. The cross-linked products were analyzed by SDS-10% PAGE. The numbers on the left indicate the molecular weight markers. This is representative of three independent assays. E, coimmunoprecipitation with apobec-1 in transfected COS-7 cells. Plasmids encoding apobec-1 and wild-type or ACF mutants were transfected into COS-7 cells. The expression of each protein was analyzed by Western blot using the indicated antibodies (upper and middle panels). The extracts from transfected cells were incubated with anti-FLAG IgG and precipitated with protein A-agarose. The precipitated complexes were resolved by SDS-10% PAGE and analyzed for the presence of Myctagged-apobec-1 by Western blot using anti-Myc IgG (lower panel). This is representative of three independent assays. ⌬ ϩ 8aa alternatively spliced form. However, two other splice mutant forms of ACF, ⌬55 and ⌬N84, failed to demonstrate complementation activity (Fig. 2C, lanes 2 and 3). ApoB RNA binding was virtually eliminated with the ⌬55 and ⌬N84 forms (Fig. 2D, lanes 2 and 3), and none of the isoforms demonstrated coimmunoprecipitation of apobec-1 (Fig. 2E, lanes 3 and 4). These results together point to the NH 2 -terminal half of ACF as containing crucial functional domains.
Deletion Analysis of ACF: Role of RRMs and Other Potential RNA Binding Domains-The amino acid coordinates for the next set of deletions are shown in Fig. 3A. We expressed mutants lacking either RRM1 or RRM2 but were unable to express an RRM3 deletion in isolation. However, as noted above (Fig.  2B), the ⌬55 mutant eliminates a portion of both RRM2 and RRM3. Deletion ⌬(331-385) removes a region of 54 residues that is unique to ACF, no comparable region being expressed in GRY-RBP (Fig. 3A). As alluded to findings noted earlier, GRY-RBP lacks complementation activity in C to U RNA-editing assays, and we speculated that this particular region in ACF may be of relevance in this regard (20). Deletion ⌬(331-385) and a finer deletion ⌬(380 -402) were also used to examine the role of the RG domain in ACF, because RG domains have been demonstrated in other proteins to exhibit RNA binding activity (33,34). Finally, ⌬(438 -515) was constructed to eliminate the putative double-stranded RNA binding domain from the COOH terminus of ACF (33). Three of these mutants retained detectable C to U editing complementation activity, suggesting that the loss of the corresponding domain is tolerated at least partially. These mutants include ⌬RRM1, the ⌬(380 -402), and ⌬(438 -515) mutants (Fig. 3C, lanes 2, 5, and 6). The ⌬(438 -515) mutant in particular demonstrated close to wild-type levels of C to U editing activity (Fig. 3C, lane 6). The results suggest that the double-stranded RNA binding domain is unlikely to play a major role in the complementation function of ACF, despite predictions from folding algorithms that apoB RNA in the region of the edited base adopts a conformation that resembles double-stranded RNA (35)(36)(37). This said, apoB RNA binding activity was reduced with both the ⌬(380 -402) and ⌬(438 -515) mutants (Fig. 3D).
Two mutants revealed no C to U editing complementation activity. These include the ⌬RRM2 and the ⌬(331-385) mutants (Fig. 3C, lanes 3 and 4). These findings were replicated in assays using up to 100 ng of recombinant protein (data not shown). Neither protein exhibited RNA binding activity as inferred from the UV cross-linking analysis (Fig. 3D). However, ⌬(331-385) revealed wild-type levels of apobec-1 interaction (Fig. 3E, lane 6), whereas ⌬RRM2 failed to coimmunoprecipitate with apobec-1, suggesting that protein-protein interaction involves a region NH 2 -terminal to residue 330.
Taken together, the findings from this series of studies indicate that diminished RNA binding activity with preservation of apobec-1 interaction is tolerated but diminishes complementation activity. However, mutations that abolish RNA binding, FIG. 4. Mutations in RRM2 and the region-spanning residues 331-380 reduce apoB RNA editing. A, top panel, schematic representation of ACF protein. Two specific pairs of residues were mutated to alanine residues. The resulting mutants are identified as L359A,L368A (L359ϩ368/A) and R371A,R375A (R371ϩ375/A). Middle panel, sequence alignment of the RRM2 domains of hnRNP C1/C2 and ACF. The conserved amino acid residues are in gray boxes. The point-mutated residues of ACF are highlighted in bold. The point mutations were constructed by changing the conserved Gly-141 and the basic KKKR amino acids to alanine residues. The mutants are identified as G141A (G141/A) and NLS/A. The corresponding secondary structure of the RNA binding domain, predicted from the GORIV algorithm, is shown below. The RNA binding domain contains four ␤-sheets interrupted by two ␣-helices. B, an analysis of 2 g of recombinant mutants by SDS-10% PAGE and Coomassie Blue staining. C, 2 ng of wild-type ACF (lane 1) or mutants (lanes 2-5) were added to 250 ng of glutathione Stransferase-apobec-1 in a C to U editing assay. The data represent the mean Ϯ S.D. from three independent assays. A representative assay is shown. D, RNA binding activity of ACF mutants. RB105 was in vitro transcribed with [ 32 P]UTP and incubated with 100 ng of wild-type ACF (lane 1) or mutants (lanes 2-5) for 15 min. After UV irradiation, the samples were analyzed by SDS-10% PAGE. The molecular weight markers are indicated at the left of the gel. This is representative of three independent assays. E, coimmunoprecipitation with apobec-1 in transfected COS-7 cells. COS-7 cells were transiently transfected with plasmids encoding apobec-1 and wild-type ACF or mutant ACF. The expression of the mutants and apobec-1 was analyzed by Western blot using the indicated antibodies (upper and middle panels). The extracts from transfected cells were immunoprecipitated with anti-FLAG IgG, and the immunoprecipitates were analyzed for the presence of Mycapobec-1 by Western blot using anti-Myc IgG (lower panel). This is representative of three independent assays. despite preserving protein-protein interaction, fail to complement C to U editing.
Point Mutagenesis within RRM2 and the Region-spanning Residues 330 -380 -The observation, in which a mutant ACF protein lacking RRM2 failed to bind apoB RNA and also failed to coimmunoprecipitate with apobec-1, pointed to this region as containing crucial functional domains that are in turn required for C to U editing complementation activity. As illustrated in Fig. 4A, RRM2 contains two consensus ribonucleoprotein (RNP) sequences, RNP1 and RNP2 (38,39), separated by conserved hydrophobic residues (40,41). Based on recent studies demonstrating the importance of the conserved glycine residue in RNP2 in mediating the binding of hnRNPC1 to a poly(U) substrate (42), we introduced a single point mutation into ACF at the corresponding position (G141A, Fig. 4A). This RNP2 domain is in immediate proximity to the putative NLS in ACF (18,19), which contains a cluster of basic residues (Fig. 4A). Clusters of basic residues, particularly arginine and lysine, are a feature of some RNA-binding protein domains, notably the Rev peptide (43), suggesting to us that mutagenesis of these residues in RRM2 might disrupt RNA binding. Accordingly, we generated the ACF mutant NLS/A (Fig. 4A). In addition, we also attempted to localize the residues between 331 and 385 that contribute to the loss of function, particularly apoB RNA binding, observed with the deletion mutant ⌬(331-385). We selected two pairs of residues, (Leu-359 ϩ Leu-368) and (Arg-371 ϩ Arg-375), each of which was mutated to alanine (Fig.  4A). These leucine residues were selected following the recent report of Dreyfuss and colleagues (42) stating that specific leucine residues at the COOH terminus of the RNA binding domain of hnRNPC1 influence the RNA binding activity of the protein. Arginine residues were selected following the prediction that these residues exert electrostatic interactions with RNA in the Rev-RRE complex (43).
Recombinant proteins (Fig. 4B) were examined for C to U complementation activity. The results demonstrate a 90% re-   4C, lanes 1-3). C to U complementation activity was completely abolished in the L359A,L368A mutant and reduced by almost 90% in the R371A,R375A mutant, suggesting that these residues are important determinants of ACF function. These mutant proteins were further examined in relation to apoB RNA binding and apobec-1 interaction. The NLS/A, G141A, and R371A,R375A mutants produced a modest decrease in apoB RNA binding (Fig. 4D, lanes 2, 3, and 5), as inferred from visual inspection of the UV cross-linking assay. The role of specific residues cannot be determined with certainty from this mutagenesis study, and further work will be required to examine whether, for instance, other basic residues would function interchangeably. In addition, confirmation of the UV cross-linking assays will need to be made through quantitative analysis of RNA binding. With these reservations notwithstanding, the L359A,L368A mutant demonstrated no apoB RNA cross-linking activity (Fig. 4D, lane 4), implicating these residues as crucial to the RNA binding function of ACF. Apobec-1 interaction was disrupted in the NLS/A mutant ( Fig. 4E) but preserved in the other mutants, consistent with the prediction from the initial experiments described above suggesting that the protein-protein interaction domain is located in a region proximal to residue 257. An important consideration in our interpretation of the mechanisms underlying the effects of these mutations on ACF function concerns the structural conformation of the protein.
Using the GORIV algorithm, there was no overall change predicted with the G141A, NLS/A, or L359A,L368A mutants. However, the R371A,R375A mutant is predicted to change a coiled structure into an ␣-helix downstream of the mutation. Thus, the observed reduction in C to U editing complementation activity, without obvious disruption in either RNA binding or apobec-1 interaction, may reflect this potential conformational change in the R371A,R375A mutant. However, further analysis will be required to confirm this suggestion.
Subcellular Distribution of ACF: Role of Putative NLS-A further series of experiments was undertaken to examine the role of the putative NLS in directing the expression of epitope-tagged ACF to the nucleus of COS-7 and McArdle hepatoma cells. COS-7 cells were selected, because they barely express detectable levels of endogenous ACF and do not express apobec-1 or apoB RNA (20). McArdle rat hepatoma cells were selected, because they express all the requisite components of the apoB RNA editing machinery in addition to endogenous apoB RNA. Transfected cells revealed that wild-type ACF was localized predominantly in the nucleus of both cells (Fig. 5, A and B). Surprisingly however, both the NLS/A mutant and the RRM2 mutant also localized to the nucleus in both COS-7 cells (Fig. 5A, panels C and E) and McArdle rat hepatoma cells (data not shown). These data suggest that the putative NLS is not required for localization of ACF to the nucleus. Colocalization studies using confocal microscopy demonstrated superimposable images for apobec-1 and wild-type ACF (Fig. 5B, panel C), indicating nuclear colocalization of both proteins as reported previously (20). By contrast, cotransfection of apobec-1 with either the ⌬RRM2 or NLS/A mutant demonstrated a predominantly cytoplasmic localization of apobec-1 (Fig. 5B, panels G and K). Under these circumstances, apobec-1 failed to translocate to the nucleus, raising the possibility that nuclear import requires interaction with ACF. These results further confirm the in vitro observations indicating that the apobec-1-interacting domain involves at least the RRM2 domain of ACF. Further experiments will be required to map the structural elements involved in the trafficking of ACF to the nucleus and the requirements for the nuclear import of apobec-1.
Concluding Comments-The general conclusions concerning domains involved in apoB RNA binding and apobec-1 interaction are summarized in Table I and the accompanying Fig. 6. The cumulative evidence points to overlapping regions in the NH 2 -terminal 380 residues of ACF that mediate apobec-1 interaction and apoB RNA binding. Apobec-1 binding appears to be confined to the region-spanning residues 144 -257, whereas RNA binding involves residues 150 -380. Both functions appear necessary to support C to U complementation activity. With the exception of the RRMs, none of the other predicted functional domains (NLS, RG domain, or double-stranded RNA binding domain) appear to be required for complementation activity. Taken together, these results suggest that ACF has a modular functional domain that is contained within the NH 2terminal 380 residues. Further delineation of the residues responsible for apobec-1 interaction and RNA binding is clearly warranted and may shed light on the biological role for this interesting protein.