Apobec-1 interacts with a 65-kDa complementing protein to edit apolipoprotein-B mRNA in vitro.

The editing of apolipoprotein-B (apoB) mRNA involves the deamination of cytidine at nucleotide 6666 to uridine. The catalytic subunit of the editing enzyme, apobec-1, is a cytidine deaminase that requires other unidentified proteins to edit apoB mRNA in vitro. We partially purified an activity from baboon kidney that functionally complements apobec-1. The complementing activity was protease-sensitive and micrococcal nuclease-resistant, had a native molecular mass of 65 ± 10 kDa on size exclusion chromatography, and sedimented at 4.5 S in glycerol gradients. Purified recombinant His6-tagged apobec-1 immobilized on beads depleted >90% of the complementing activity from partially purified extracts. These beads edited apoB mRNA in vitro in the absence of exogenous apobec-1 or complementing activity. A functional holoenzyme containing apobec-1 and the complementing activity was eluted from the apobec-1-affinity resin using 0.5 M imidazole, whereas buffer containing 0.4 M KCl eluted only the complementing activity. The carboxyl-terminal 59 amino acids of apobec-1 were not required for interaction with the complementing activity in vitro. Our results demonstrate that the complementing protein interacts directly with apobec-1 in the absence of apoB mRNA.

Apolipoprotein B (apoB) 1 circulates in the plasma in two forms which have different functions in lipid metabolism (reviewed in Ref. 1). apoB100, a 512-kDa protein secreted by the liver, is the structural component of very low density and low density lipoproteins. Enterocytes in the small intestine secrete apoB48, a 242-kDa protein which is required for the absorption of dietary lipid and the assembly of chylomicrons (1). apoB48 represents the amino-terminal 2152 amino acids of apoB100, and both forms of apoB are encoded by a single gene. The transcript that encodes apoB48 is generated by a posttranscriptional RNA editing mechanism that converts cytidine at nucleotide 6666 to uridine, and alters codon 2153 from CAA, which encodes glutamine to UAA, a premature stop codon (2, 3) (reviewed in Ref. 4). The editing of apoB mRNA involves a sitespecific deamination reaction (5), which is catalyzed by cellular proteins that recognize defined sequence elements flanking the edited base. Mutagenesis studies have shown that editing in vitro requires a minimal cassette of 22-26 nucleotides of apoB mRNA (6 -8). The specificity of editing is mediated by an 11nucleotide "mooring" sequence at nucleotides 6671-6681 downstream of the editing site (8). A model has been proposed in which the editing enzyme recognizes the mooring sequence and edits an upstream cytidine. This hypothesis is supported by the finding that the insertion of the mooring sequence into a heterologous mRNA induced editing of a cytidine 5 nucleotides upstream (9,10).
Although the sequence requirements for the editing of apoB mRNA have been well defined, little is known about the enzyme that catalyzes this reaction. The editing enzyme has been partially purified from baboon (11), rat (12), and rabbit (13) enterocytes, but the molecular composition of this activity remains controversial. Estimates of the molecular size of the enzyme have ranged from 120 -125 kDa (11,14) to a 27 S macromolecular editosome (15). The hypothesis that editing is mediated by a multiprotein complex is supported by the cDNA expression cloning of the catalytic subunit of the enzyme, a 27-kDa protein termed apobec-1 (16). Apobec-1 has limited homology to the active sites of other known cytidine deaminases and dCMP deaminases (17). Although recombinant apobec-1 expressed in Xenopus oocytes (17) and bacteria (18) had cytidine deaminase activity, apobec-1 alone did not catalyze the editing of apoB mRNA in vitro unless proteins from other cells or tissues were added to the reaction (16).
The factors which complement apobec-1 were initially detected in cells that synthesize apoB but lack editing activity (16,19). More recently, we (20) and Yamanaka et al. (21) showed that the complementing activity is widely expressed in baboon and rabbit tissues. Although the protein that complements apobec-1 has not been characterized, this activity may represent the RNA-binding subunit of the editing enzyme which recognizes the mooring sequence and docks apobec-1 to the editing site. However, there is currently no evidence to support this model.
In this study, we report the first characterization of the complementing activity in a mammalian tissue. This activity was partially purified from baboon kidney, a tissue which does not express apoB mRNA or apobec-1. Our studies suggest that the complementing activity is a protein or protein complex of 65 kDa that interacts with apobec-1 in vitro in the absence of apoB mRNA. We also show that this interaction does not require the leucine-rich region at the carboxyl terminus of apobec-1.

MATERIALS AND METHODS
Northern Blotting-Total RNAs (20 g) were isolated from baboon tissues and analyzed by Northern blotting (22). Filters were hybridized overnight with a 32 P-labeled baboon apoB100 cDNA insert (nucleotides 6504 -6784) (11), and washed at high stringency (22). The levels of apoB mRNA were quantified by PhosphorImager analysis (Molecular Dynamics).
Analysis of Recombinant Apobec-1-Approximately 4 mg of protein were obtained from 1 liter of bacterial culture, of which 80% was apobec-1, based on SDS-polyacrylamide gel electrophoresis, Coomassie Blue staining, and Western blot analysis. In addition to the 28-kDa apobec-1, a 56-kDa protein was also detected, which varied in amounts between preparations and was absent in extracts of bacteria transformed with the expression vector. In Western blot analysis, the purified 28-kDa apobec-1, but not the 56-kDa protein, reacted with the rabbit anti-apobec-1 peptide antibody. Based on V8 protease mapping (23), the peptide maps of the 56-and 28-kDa proteins were identical when a range of Staphylococcus aureus V8 protease concentrations, and incubation times were tested. This suggested that the high molecular weight form represents a covalently linked SDS-resistant apobec-1 dimer. A few bacterial proteins also copurified with apobec-1. These minor contaminants were obtained when proteins were purified from IPTG-induced M15(pREP4) bacteria that were transformed with the vector only. The proteins from vector-transformed bacteria served as a negative control in the protein interaction experiments.
Western Blot Analysis-Antibodies were generated against a synthetic peptide corresponding to amino acids 10 -21 of rat apobec-1 (16). The peptide was synthesized using the multiple antigen peptide method (24), and injected into rabbits using standard procedures (25). For Western blot analysis, proteins were resolved on SDS-polyacrylamide gel electrophoresis, 12% gels, and transferred to polyvinylidene difluoride membranes (25). Membranes were incubated with the primary antibody (1:2000) and an affinity-purified goat anti-rabbit peroxidaseconjugated antibody (Boehringer Mannheim). Proteins were detected using ECL reagents (Amersham Corp.).
Purification of Complementing Activity-Whole cell extracts were prepared from baboon kidney as described previously (11). After centrifugation at 100,000 ϫ g, the supernatant was fractionated by adding solid ammonium sulfate, and the precipitate from the 15-30% fraction was dialyzed against buffer D (20 mM Hepes, pH 7.9, 100 mM KCl, 20% glycerol, 0.2 mM EDTA, 0.5 mM dithiothreitol). This fraction (80 mg) was chromatographed on a Sephacryl S300 column (1.5 ϫ 50 cm) in buffer D containing 200 mM KCl at a flow rate of 30 ml/h. The column was calibrated with gel filtration standards (Bio-Rad) containing thyroglobulin (670 kDa), bovine ␥-globulin (158 kDa), chicken ovalbumin (44 kDa), equine myoglobin (17 kDa), and vitamin B 12 (1.35 kDa). The void volume was determined by loading blue dextran. The standard curve was plotted as the ratio of (elution volume minus void volume) to (column volume minus void volume), and was used to determine the molecular mass of the fractions with peak complementing activity. For ion-exchange chromatography, fractions were dialyzed against buffer D containing 25 mM KCl, bound to Q Sepharose (Pharmacia Biotech Inc.), and eluted with a linear gradient of KCl (50 -500 mM) in buffer D. Protein concentrations were determined according to Bradford (26).
Editing Assays-The synthesis of RNA substrates and in vitro editing assays were performed as described previously (9). The apoB48 and apoB100 primer extension products were quantified using a Phosphor-Imager (Molecular Dynamics) or a Scan Maker III densitometer (Microtek).
Protein Interaction Studies-Proteins were purified from IPTG-induced bacteria as described above. Ni-NTA resin (Qiagen) was equili-brated in buffer E (20 mM Hepes, pH 7.9, 100 mM KCl, 10% glycerol, 15 mM imidazole, 5 mM ␤-mercaptoethanol). Aliquots of the resin (100 l) were incubated for 1 h with 2 mg of purified recombinant protein, which was in excess of the total binding capacity of the resin. The unbound protein was removed, and the beads were washed with 100 volumes of buffer E. The beads were incubated for 2 h at 4°C with 100 g of partially purified complementing activity (Sephacryl S300 fraction). The supernatant was combined with the first wash and used as the unbound fraction. After washing with 200 volumes of buffer E, the beads were resuspended in buffer E and an aliquot was assayed for editing activity. This process was scaled up for elution experiments (2.5 mg Sephacryl S300 fraction/ml apobec-1-Ni-NTA resin). After batch binding, the affinity resin was washed until an A 280 ϳ 0, and packed into a column. Proteins were eluted with 4 bed volumes of buffer D containing 0.5 M imidazole or 0.4 M KCl.

Expression of ApoB mRNA in Baboon Tissues-
We previously demonstrated that the complementing activity is widely expressed in baboon tissues, whereas the expression of the editing enzyme is restricted to the small intestine (20). In humans, apoB mRNA is expressed in a variety of human tissues, including kidney (27). Northern blot analysis of baboon tissues detected a 14-kilobase pair apoB mRNA (Fig. 1). Relative to the levels of apoB mRNA in the liver (100%), the levels of apoB mRNA were 50% in the small intestine, 5% in the spleen, and 1.3% in the stomach. No hybridizable apoB mRNA was detected in other baboon tissues, including kidney ( Fig. 1). We were also unable to detect apoB mRNA in baboon kidney using the more sensitive approach of reverse transcriptionpolymerase chain reaction (data not shown). Based on these results, we chose to characterize the complementing activity in kidney, a tissue that lacks detectable editing activity and apoB mRNA.
Baboon Kidney Extracts Complement Recombinant Apobec-1-Rat apobec-1 was expressed in bacteria as a His 6 -tagged protein, and the protein was purified to 80% homogeneity on Ni-NTA resin under native conditions. Recombinant apobec-1 lacked intrinsic editing activity in the standard in vitro editing assay and when tested over a wide range of concentrations and incubation times (data not shown). Editing activity was restored when apobec-1 was complemented with the 15-30% ammonium sulfate fraction of baboon kidney extract (Fig. 2, second lane).
Partial Purification of the Complementing Activity, a 65-kDa Protein-The complementing activity in kidney extracts was sensitive to pretreatment with proteinase K, but not to micrococcal nuclease (data not shown). These results suggest that the activity is a protein which does not require a free RNA component to complement apobec-1. To partially purify this activity, whole cell extracts of baboon kidney were fractionated with 15-30% ammonium sulfate, which precipitated 80% of the complementing activity but only 6% of the total protein. This fraction contained a large number of high molecular weight proteins, and an additional 6-fold purification was achieved by chromatography on Sephacryl S300. The complementing activity eluted as a broad peak in a volume corresponding to an average size of 65 Ϯ 10 kDa (Fig. 2). In glycerol density gradient centrifugation, the activity sedimented as a single peak around 4.5 S (data not shown). This corresponds to a molecular mass of 65 kDa, assuming that the activity is a globular protein. When the fraction from the gel filtration column was chromatographed on Q Sepharose, the complementing activity eluted in a single peak between 150 and 200 mM KCl.
Characterization of the Complementing Activity-The 15-30% ammonium sulfate fraction was used to characterize the complementing activity in vitro. Maximum editing was achieved with 20 g of partially purified kidney extract and 1 g of apobec-1, and an excess of either complementing activity (Fig. 3A) or apobec-1 (data not shown) inhibited editing. After the addition of complementing activity to apobec-1 in the in vitro editing assay, editing was detected within 5 min and increased for 3 h (Fig. 3B). Preincubation of apobec-1 or the complementing activity with each other or with the synthetic apoB mRNA did not change the kinetics of editing (data not shown). The optimal KCl concentration for the editing activity of apobec-1 and the complementing protein was between 90 and 125 mM, with a pH optimum of 8.0. The complementing activity was heat-sensitive, with 50% of the activity lost after incubation at 40°C for 10 min prior to addition to the in vitro editing assay.
The activity of the reconstituted enzyme was completely inhibited by the addition of basic and acidic proteins, including lysozyme and bovine serum albumin, but not by the addition of total cellular protein from COS cells or reticulocyte lysate (Table I). Moderate concentrations of EDTA, EGTA, monovalent cations, or divalent cations had no effect on apobec-1 and the complementing activity (Table I). In contrast to the general characteristics exhibited by most nucleic acid binding proteins, the editing activity of apobec-1 and the complementing protein was not affected by the addition of heparin, tRNA, singlestranded DNA, double-stranded RNA, or ribohomopolymers (Table I). Although it has been reported that the activity of apobec-1 is inhibited by poly(U) (28), we found that even a 10,000-fold molar excess of poly(U) did not inhibit editing in our system (data not shown).
Complementing Activity Interacts with Apobec-1-To test whether the complementing activity and apobec-1 can interact in vitro, the Sephacryl S300 fraction from baboon kidney was incubated with purified His 6 -tagged apobec-1 immobilized on Ni-NTA resin. Since the recombinant apobec-1 was only 80% pure, we also purified proteins from bacteria that were transformed with vector to use as a negative control (vector-Ni-NTA). Aliquots of the unbound fractions (15 g) were analyzed for complementing activity in the in vitro editing assay under conditions where this activity was the limiting component. The complementing activity did not significantly bind to vector-Ni-NTA resin (Fig. 4A, lane 3) or to Ni-NTA resin (data not A, editing assays (60 l) were performed with 1 g of purified His 6 -tagged apobec-1 and increasing amounts of complementing activity (15-30% ammonium sulfate fraction of a baboon kidney extract) as indicated. The samples were analyzed as described in Fig. 2. B, in vitro editing assays (20 l) containing partially purified complementing activity (20 g) and purified His 6 -tagged apobec-1 (1 g) were incubated for 5 min to 8 h as indicated.
shown). Near quantitative depletion of the complementing activity was obtained when 100 g of protein were incubated with 1 mg of resin-bound apobec-1 (Fig. 4A, lane 4). Binding to the apobec-1-Ni-NTA resin was saturable at higher concentrations of complementing activity (data not shown). After extensive washing, the beads were assayed for editing activity in the absence of exogenous apobec-1 or complementing activity. Editing activity was detected in the apobec-1-Ni-NTA beads which had been incubated with complementing activity (Fig. 4B, lane  3), suggesting that a functional holoenzyme had been generated on the resin. No editing activity was detected in the vector-Ni-NTA beads that had been incubated with complementing activity or in apobec-1-Ni-NTA beads alone (data not shown).
In Fig. 4C, the in vitro binding assays were scaled up to perform affinity chromatography using apobec-1 as a ligand. Complementing activity was bound to apobec-1-Ni-NTA, and proteins were step-eluted with buffer containing either 0.5 M imidazole or 0.4 M KCl. The imidazole-eluted fractions contained a functional editing enzyme that contained both apobec-1 and the complementing activity. However, only the complementing activity was eluted with KCl, as these fractions did not edit apoB mRNA unless recombinant apobec-1 was added to the reaction (Fig. 4C). Western blot analysis also confirmed the absence of apobec-1 in the salt-eluted fractions, which demonstrates that the lack of editing activity was not due to stoichiometric inhibition by excess apobec-1 (data not shown). The apobec-1 affinity chromatography step resulted in a 300fold increase in the specific activity of the complementing activity. No complementing activity was detected when similar elution experiments were performed with Ni-NTA resin.
Interaction with the Complementing Activity Does Not Require the Leucine-rich Region in Aqpobec-1-The carboxyl terminus of apobec-1 contains a leucine-rich region (amino acids 182-210) which is required for editing activity (16,18,21). To test whether this sequence is required for interaction with complementing activity, a deletion mutant that lacked amino acids 171-229 (apobec-1/Kpn) was expressed in bacteria (Fig.  5A, lane 3). Purified apobec-1/Kpn edited apoB mRNA in vitro with low efficiency, possessing only 0.2% of the activity of the wild type protein (data not shown). However, apobec-1/Kpn immobilized on nickel-NTA resin was able to deplete approximately 90% of the complementing activity in the in vitro binding assay (Fig. 5B, lane 4). To confirm that the complementing activity had bound to the truncated apobec-1 protein, the apobec-1/Kpn-Ni-NTA resin was eluted with 0.4 M KCl (Fig. 5C). The recovery of complementing activity from the apobec-1/Kpn  (20 g) was preincubated with various test reagents for 10 min prior to addition to the in vitro editing reaction. Editing assays were performed for 3 h at 30°C with 11 fmol of synthetic baboon apo-B100 RNA, and 1 g of purified 6xHistagged apobec-1. The reaction products were detected by primer extension, as described under "Materials and Methods."  4. Complementing activity interacts with His 6 -tagged apobec-1 in vitro. A, partially purified complementing activity (100 g) was incubated with apobec-1-Ni-NTA or vector-Ni-NTA beads (100 l) for 2 h. Equal amounts of the starting material and the unbound fractions (15 g) were assayed for complementing activity in the presence of purified apobec-1. Lane 1, buffer; lane 2, starting material; lane 3, unbound to vector-Ni-NTA; lane 4, unbound to apobec-1-Ni-NTA. B, complementing activity was incubated with apobec-1-Ni-NTA beads as described above. After extensive washing, an aliquot of the beads (20 l) was assayed directly for editing activity. Lane 1, buffer; lane 2, 15 g of partially purified complementing activity (starting material) and 1 g of apobec-1; lane 3, apobec-1-Ni-NTA beads after incubation with complementing activity. C, partially purified complementing activity was bound to apobec-1-Ni-NTA beads (0.5 ml), and proteins were eluted with 2 ml of buffer D containing either 0.5 M imidazole or 0.4 M KCl (0.5 ml/fraction). The imidazole-eluted fractions were pooled, and an aliquot (10 l) was assayed alone for editing activity. Aliquots of the KCl-eluted fractions (10 l) were assayed for editing activity in the presence or absence of apobec-1 (1 g) as indicated.
affinity column was comparable (within 2-fold) to the recovery from the wild type apobec-1 column in three separate experiments.

DISCUSSION
This study is the first characterization of the complementing activity in a mammalian system. We show that this activity in baboon kidney is a 65-kDa protein or protein complex that interacts with apobec-1 in vitro in the absence of apoB mRNA to generate a catalytically active enzyme. Although it has been difficult to purify the native editing enzyme to homogeneity in the past, this was primarily due to the lack of a high affinity purification step. Our results demonstrate that an apobec-1 affinity column can be used to isolate either a holoenzyme that contains both apobec-1 and the complementing activity, or the complementing activity alone. Apobec-1 is a 27-kDa protein that is known to dimerize, although the functional significance of this dimerization is not known (29). It is intriguing that the sum of the sizes of an apobec-1 dimer and the 65-kDa complementing activity is consistent with the minimal molecular size of 125 kDa that was previously reported for the editing complex in baboon (11) and rat (14) enterocytes. However, our studies have not definitively established that the complementing activity is a subunit of the native editing enzyme.
Our in vitro results suggest that the proteins involved in editing have a high affinity for each other, and that a significant lag period is not required for the interaction of apobec-1 and the complementing activity with each other, or with apoB mRNA. This is in contrast to studies by Smith and colleagues who suggested that a long lag period was required for the assembly of higher order complexes, or editosomes, on apoB mRNA (15,30). The hypothesis that the editing enzyme may be associated with other proteins involved in RNA processing is attractive, since several lines of indirect evidence suggest that editing may occur during the splicing and polyadenylation of apoB mRNA in vivo (2,19,31). However, several laboratories were unable to demonstrate a lag period for the in vitro editing reaction with the native enzyme (11)(12)(13). In addition, it has been reported that the enzyme has a buoyant density of pure protein, without a nucleic acid component (12). This discrepancy may reflect differences in extract preparation, since nuclear, cytosolic S100, and whole cell extracts have all been used as sources of the editing enzyme by various laboratories.
Our results also confirm previous studies which demonstrated that apobec-1 and complementing activity need to interact in a defined ratio (18,19). The decrease in editing efficiency that occurs when apobec-1 or the complementing activity are in excess may be due to the formation of nonfunctional complexes between homologous species, in contrast to a functional heterologous interaction between apobec-1, the complementing activity, and apoB mRNA. The domain in apobec-1 that interacts with the complementing protein appears to be highly conserved, since rat apobec-1 can interact with the complementing activity from baboon (this study), chick (16), and human (16,32), and rabbit apobec-1 can interact with rat complementing activity (21). Although apobec-1 is not a conservative protein (33), the leucine-rich region in the carboxyl terminus of apobec-1 is highly conserved across species, and it has been proposed that this sequence may interact with the complementing activity (16,18,21,29). In this study, we showed that deletion of the leucine-rich region reduced editing efficiency to 0.2% of wild type levels, but that the truncated protein was still able to interact with the complementing activity in in vitro binding assay. These results suggest that the domain for interacting with the complementing activity lies within amino acids 1-170 of apobec-1, and experiments are currently in progress to test this hypothesis. Although the function of the leucine-rich region is not known, it may be involved in RNA-binding (18) or homodimerization (29).
In addition to the 65-kDa protein reported here, several other candidate proteins for the complementing activity have been proposed. An "enhancement factor" of 49 kDa has been partially purified from chick enterocytes, which stimulated editing when added to the native editing enzyme in rat enterocyte extracts in vitro (34). UV cross-linking experiments have identified proteins of 60 -66 kDa (p66) and 40 -44 kDa (p40) in rat liver and enterocyte extracts, which bind to apoB mRNA downstream from the editing site at nucleotides 6671-6674 (15,17,35). However, no direct functional evidence has been presented demonstrating that these binding proteins have complementing activity or that they are involved in editing. were analyzed by Western blotting using a rabbit anti-apobec-1 peptide antibody. Molecular mass markers are indicated on the right. B, in vitro binding experiments were performed with the partially purified complementing activity and apobec-1-Ni-NTA or apobec-1/Kpn-Ni-NTA beads, as described in Fig. 4A. The starting material and unbound fractions (15 g) were assayed for complementing activity in the presence of purified apobec-1. Lane 1, buffer control; lane 2, starting material; lane 3, unbound to apobec-1-Ni-NTA; lane 4, unbound to apobec-1/Kpn-Ni-NTA. C, partially purified complementing activity was bound to apobec-1/Kpn-Ni-NTA (0.35 ml) and bound proteins were eluted with 0.4 M KCl as described in Fig. 4C. Aliquots of each fraction (10 l) were assayed for complementing activity in the presence of recombinant wild type apobec-1. plex in rat liver extracts that may be required for the efficient editing of apoB mRNA in vitro, but the ability of p240 to complement recombinant apobec-1 was not tested. Several candidate auxiliary proteins in extracts of McArdle cells have also been reported to bind to an apobec-1-affinity resin in two recent preliminary studies (37,38).
The simplest model of the editing enzyme is that it is composed of the catalytic subunit, apobec-1, and an RNA-binding subunit that docks apobec-1 to its target cytidine in apoB mRNA. Although this model is consistent with our data that demonstrate that apobec-1 and the complementing activity interact in vitro, it has not been established that the complementing protein represents the RNA-binding subunit of the enzyme. Recombinant apobec-1 expressed in bacteria has an RNA-binding activity, with a preference for AU-rich sequences (28,39). Binding was of low specificity, which suggests that the sequence-specific recognition of apoB mRNA may require the complementing activity or the holoenzyme. The widespread expression of the complementing activity in tissues which do not synthesize apoB mRNA or apobec-1 indicates that this activity may have other cellular functions (16, 19 -21). Although the complementing activity may function as a sequence-specific RNA-binding protein that is also involved in the editing of other mRNAs, this activity may also represent a generic RNA-binding protein which has other roles in RNA metabolism. The complementing activity may serve as an RNA chaperone which facilitates the folding of apoB mRNA into a structure that is a more favorable substrate for the enzyme. The activity may also stimulate the binding of apobec-1 to apoB mRNA, or the dissociation of products after editing has occurred. Similar functions have been reported for the nonspecific RNA-binding proteins, HnRNP A1 and p7 nucleocapsid proteins, which facilitate hammerhead ribozyme catalysis (40,41).