A Proteomic Approach Identifies Proteins in Hepatocytes That Bind Nascent Apolipoprotein B*

The biogenesis of apolipoprotein B is quite complex in view of its huge size, hydrophobicity, obligate association with lipids such as cholesterol and triglycerides prior to secretion, and intracellular degradation of a substantial proportion of newly synthesized molecules. Multiple proteins likely serve roles as molecular chap-erones to assist in folding, assembly with lipids, and regulation of the secretion of apolipoprotein B. In these studies, we developed a strategy to isolate proteins associated with apolipoprotein B in rat livers. The purification consisted of two stages: first, microsomes were prepared from rat liver and treated with chemical cross-linkers, and second, the solubilized proteins were co-immunoprecipitated with antibody against apolipoprotein B. We found that several proteins were cross-linked to apolipoprotein B. The proteins were digested with trypsin, and the released peptides were sequenced by tandem mass spectrometry. The sequences precisely matched 377 peptides in 99 unique proteins. We show that at least two of the identified proteins, ferritin heavy and light chains, can directly bind apolipoprotein B. These and possibly other proteins identified by this proteomic approach are novel candidates for proteins that affect apolipoprotein B during its biogenesis. Apolipoprotein B 1 is with lipids including cholesterol esters, phospholipids, cholesterol, and triglycerides, as very low density temperature for 30 min then stopped by adding Tris-HCl to 50 m M and incubating another 15 min. The cross-linked microsomes were collected by centrifugation at 90,000 rpm for 15 min at 4 °C in a Beckman TL-100 tabletop ultracentrifuge. For some experiments, antibodies were conjugated to either cyano- gen bromide-activated beads or cross-linked to protein G-agarose. Cyanogen bromide-activated beads were washed with cold 1 m M HCl for 30 min, followed by brief washes with water and then buffer A (0.1 M NaHCO 3 /0.5 M NaCl (pH 8.5)). Antibody was incubated with the beads for 2 h atroom temperature. Unreacted groups were blocked with 1 M ethanolamine (pH 8.0) fo r 2 h atroom temperature. The beads were washed extensively with buffer A followed by five washes with 0.1 M acetate (pH 4.0)/0.5 M NaCl. Finally, the beads were washed with TXSWB (1% Triton X-100, 100 m M Tris-HCl (pH 8.0), 100 m M NaCl, 10 m M EDTA, 1 m M phenylmethylsulfonyl fluoride) and used for immuno- precipitation. Protein G-agarose was incubated with antibodies in 50 m M sodium borate (pH 8.2) overnight at 4 °C on a rotator. The beads were washed first with sodium borate buffer and then with 0.2 M triethanolamine buffer (pH 8.2). The non-cleavable cross-linker DSS (10 m M ) was added, and the reaction was incubated at room temperature for 1 h. After washing with triethanolamine buffer, the beads were incubated in 0.1 M ethanolamine (pH 8.2) for 10 min at room temperature. The beads were washed in 0.1 M glycine-HCl (pH 2.8) and then in sodium borate buffer. Finally, the washed beads were equilibrated in TXSWB and used for immunoprecipitation.

of several known chaperone proteins including some that are particular to the specialized physiologic role of apoB. Evidence suggests that calnexin, calreticulin, BiP, Erp72, GRP94, and protein disulfide isomerase (PDI) all interact with apoB during its translocation and further biogenesis once it has entered the endoplasmic reticulum (ER) (9 -11). These proteins also function in the proper folding and quality control of other secretory proteins. However, the assembly of apoB with lipids into a lipoprotein particle necessitates additional proteins that may be particular to apoB. The lack of solubility of apoB in an aqueous environment, such as the lumen of the ER, necessitates its co-translational association with lipids (12). This process is facilitated by microsomal triglyceride transport protein (MTP), which plays a crucial role in the initial assembly and regulation of secretion of apoB (13)(14)(15). In a second step that is sensitive to inhibition by brefeldin A, apoB is assembled with a full complement of lipids into a mature lipoprotein particle (16,17). The protein that mediates this addition of bulk lipids is not known.
Molecules of apoB that do not complete the assembly process due to insufficient lipids being available, misfolding, or inhibition of necessary chaperone proteins are targeted for intracellular degradation. Several groups have shown that apoB can be marked for degradation by the ubiquitin-proteasome pathway even during its translocation into the ER (18 -20). ApoB can also enter other routes of intracellular degradation in hepatic cells (21,22). Remarkably, some molecules of ubiquitin-conjugated apoB can be rescued from degradation and secreted when hepatocytes are treated with lipids (23). The variety of pathways of degradation and possibility of rescue implies that several types of proteins might play roles in determining the fate of molecules of apoB.
To gain a more complete view of the assembly, degradation, and regulation of the secretion of apoB, we wished to identify proteins that contact apoB throughout its biogenesis in the hepatocyte. We employed a proteomic approach to isolate and identify all proteins that could be co-immunoprecipitated with apoB after treatment with a chemical cross-linker. We demonstrate that some of the novel proteins bind to apoB and therefore are candidates for roles in its biogenesis.

EXPERIMENTAL PROCEDURES
Reagents-Janet Sparks kindly provided us with a gift of polyclonal antibody against rat apoB; goat polyclonal antibody against human apoB was purchased from Chemicon. The monoclonal antibodies 1D1, Bsol1, and Bsol7 were kindly given by Yves Marcel. Protein G-agarose was purchased from Kirkegaard and Perry Laboratories. Silver staining kits were obtained from BioRad. The chemical cross-linkers dithiobis(succinimidyl propionate) (DSP), disuccinimidyl suberate (DSS), and dimethyl 3,3Ј-dithiobispropionimidate-2 HCl (DTBP) were obtained from Pierce. Colloidal Blue staining kits were purchased from Novex. The plasmid TyB2, ER2566 bacteria, and chitin beads were purchased from New England Biolabs. A kit for generating capped mRNA using SP6 polymerase was purchased from Ambion. Rabbit reticulocyte lysate was from Promega, and [ 35 S]methionine Tran 35 S-label (specific activity Ͼ1000 Ci/mmol) was purchased from ICN. Ultrafree MC 0.22 m filter units were purchased from Millipore.
Preparation of Microsomes-Livers averaging 4 g each from adult Sprague-Dawley rats of various ages were excised shortly after euthanasia and transported in ice-cold homogenization buffer (HB: 50 mM sodium phosphate, 50 mM potassium acetate, 1 mM EDTA, 0.25 M sucrose, 1 mM phenylmethylsulfonyl fluoride, pH 7.5). The livers were rinsed four times in HB, cut into very small pieces, and passed several times through a parsley grinder. The liver was homogenized in HB using a motorized Teflon-coated Dounce homogenizer at high speed. Tissue debris was removed by centrifugation at 10,000 rpm at 4°C for 15 min. The supernatant was transferred to a new tube and centrifuged again at 12,000 rpm at 4°C for 15 min. The supernatant then was layered onto 4 ml of HB adjusted to 1.3 M sucrose and centrifuged at 28,000 rpm at 4°C for 3 h using an AH-629 rotor. After removing the supernatant, the jelly-like microsomal pellet was collected and gently dispersed in a small volume of HB using a hand-held Teflon-coated Dounce homogenizer.
Cross-linking-We used DSP and DTBP for cross-linking proteins in our samples. Both cross-linking reagents are cleavable, homobifunctional, have similarly sized spacer arms, and permeate membranes. However, unlike DSP, DTBP is readily soluble in aqueous solutions. DSP was dissolved in dimethyl sulfoxide (Me 2 SO) and added to the sample at a final concentration of 25 mM. DTBP was added directly to the sample at a concentration of 10 mM. Chemical cross-linking was carried out in homogenization buffer. The reaction was incubated at room temperature for 30 min then stopped by adding Tris-HCl to 50 mM and incubating another 15 min. The cross-linked microsomes were collected by centrifugation at 90,000 rpm for 15 min at 4°C in a Beckman TL-100 tabletop ultracentrifuge.
For some experiments, antibodies were conjugated to either cyanogen bromide-activated beads or cross-linked to protein G-agarose. Cyanogen bromide-activated beads were washed with cold 1 mM HCl for 30 min, followed by brief washes with water and then buffer A (0.1 M NaHCO 3 /0.5 M NaCl (pH 8.5)). Antibody was incubated with the beads for 2 h at room temperature. Unreacted groups were blocked with 1 M ethanolamine (pH 8.0) for 2 h at room temperature. The beads were washed extensively with buffer A followed by five washes with 0.1 M acetate (pH 4.0)/0.5 M NaCl. Finally, the beads were washed with TXSWB (1% Triton X-100, 100 mM Tris-HCl (pH 8.0), 100 mM NaCl, 10 mM EDTA, 1 mM phenylmethylsulfonyl fluoride) and used for immunoprecipitation. Protein G-agarose was incubated with antibodies in 50 mM sodium borate (pH 8.2) overnight at 4°C on a rotator. The beads were washed first with sodium borate buffer and then with 0.2 M triethanolamine buffer (pH 8.2). The non-cleavable cross-linker DSS (10 mM) was added, and the reaction was incubated at room temperature for 1 h. After washing with triethanolamine buffer, the beads were incubated in 0.1 M ethanolamine (pH 8.2) for 10 min at room temperature. The beads were washed in 0.1 M glycine-HCl (pH 2.8) and then in sodium borate buffer. Finally, the washed beads were equilibrated in TXSWB and used for immunoprecipitation.
Immunoprecipitation-Aliquots of microsomes were completely dissolved in TXSWB; no unsolubilized material remained. Antibody against apoB was added in the form of serum or conjugated to beads (as described above). After incubating the sample with antibody for 1 h at 4°C, protein G-agarose beads were added to samples containing free antibody in solution, and the tubes were rotated overnight at 4°C. Subsequently, the beads were washed three times with TXSWB and twice with Tris-NaCl (100 mM Tris-HCl (pH 8.0), 100 mM NaCl). The immunoprecipitated proteins were released by boiling for 10 min in SDS loading buffer containing 0.5 M dithiothreitol (DTT) and separated by SDS-PAGE.
Cloning of Ferritin Heavy and Light Chain Coding Regions and Plasmid Constructions-Total RNA was prepared from a confluent T75 flask of HepG2 cells using 2 ml of solution I (4 M guanidine thiocyanate, 20 mM sodium acetate, 0.5% Sarkosyl, and 0.1 mM DTT). The lysate was mixed by inversion with 0.2 ml of 2 M sodium acetate (pH 4), then 2 ml of water-saturated phenol, and finally 0.4 ml of chloroform. After mixing, the solution was kept on ice for 15 min and then centrifuged at 10,000 ϫ g for 20 min at 4°C. The aqueous phase was transferred to a new tube, and the RNA was precipitated with 2 ml of isopropanol by storing the solution at Ϫ20°C for 30 min before centrifuging at 10,000 ϫ g for 15 min at 4°C. The RNA pellet was dissolved in 0.3 ml of solution I and precipitated again with 0.3 ml of isopropanol. The RNA pellet was resuspended in 75% ethanol at room temperature and spun at 10,000 ϫ g for 15 min at 4°C. The pellet was air-dried and resuspended in 100 l of diethyl pyrocarbonate-treated water.
Poly(A)-containing mRNA was purified from this total RNA. Oligo dT cellulose was washed in 0.1 M sodium hydroxide, rinsed several times in diethyl pyrocarbonate-treated water until the pH value of the wash was neutral, and then resuspended in 0.5 M lithium chloride, 10 mM Tris acetate (pH 7.5), 1 mM EDTA, and 0.1% SDS. The RNA solution was heated to 70°C for 10 min, adjusted to 0.5 M lithium chloride, and incubated with the oligo dT cellulose on a rotator at room temperature for 1 h. The oligo dT cellulose was recovered by centrifugation in a microcentrifuge at maximum speed for 30 s. The oligo dT cellulose was washed twice with 0.15 M lithium chloride, 10 mM Tris acetate (pH 7.5), 1 mM EDTA, and 0.1% SDS. After the second wash, the resuspended cellulose was transferred to an Ultrafree MC 0.22 m filter unit. The cellulose was washed three more times in the filter unit followed by centrifugation for 30 s in a microcentrifuge at maximum speed. After the final wash, the filter cup was transferred to a new tube, and the poly(A)-containing RNA was eluted using 2 mM EDTA/0.1% SDS.
The poly(A)-containing RNA was used to synthesize cDNAs for the ferritin chains. First strand cDNA was synthesized using Superscript II reverse transcriptase (Invitrogen) according to the manufacturer's instructions. The cDNAs were used as templates to generate PCR products encoding ferritin heavy or light chains using primers corresponding to the amino-and carboxyl-terminal ends (R1Nde-FerritinH, GGA-ATTCCATATGACGACCGCGTCCACCTCG; KpnInt-FerritinH, GGG-GTACCCTTGGCAAAGCAGCTTTCATTATCACTGTCTCCC; R1Nde-FerritinL, GGAATTCCATATGAGCTCCCAGATTCGTCAG; KpnInt-FerritinL, GGGGTACCCTTGGCAAAGCAGTCGTGCTTGAGAGTGA). The PCR products were inserted into the plasmid TyB2 at NdeI and KpnI recreating the amino acid residues in the intein so that no additional amino acid residues were encoded between ferritin and the intein. Individual plasmids were verified by sequencing.
Fusion Protein Preparation-The plasmids encoding the ferritin fusion proteins and TyB2 itself were transfected into the ER2566 strain of Escherichia coli, and overnight cultures were prepared from single colonies. The next day, the culture was diluted 1:10, and after 1 h 0.2 mM isopropyl thiogalactopyranoside was added. After 5 h, the cells were pelleted by centrifugation at 5000 ϫ g for 10 min at 4°C. The pellet was resuspended in phosphate-buffered saline and disrupted over the horn of a sonicator (Heat Systems-Ultrasonics, Inc.) for a total of 5 min using pulses of 1 min interrupted by cooling on ice. After sonication, the solution was brought to 1% Triton X-100 and mixed gently by inversion, and aliquots were spun in a microcentrifuge at maximum speed for 5 min at 4°C. The supernatants were pooled in a new tube and incubated with chitin beads on a rotator at room temperature for 7 min. The beads were collected by centrifugation at 500 ϫ g for 5 min at 4°C and washed three times with phosphate-buffered saline. Aliquots of the fusion proteins bound to chitin beads were displayed by SDS-PAGE and quantitated by Coomassie staining.
Binding Assay-The plasmid pSPB29 encoding apoB29 (the aminoterminal 29% of apoB) downstream of the SP6 promoter was created by inserting a linker encoding a termination codon at the first AccI site in the plasmid pSPB48 (pSP64 with the 5Ј-untranslated region of Xenopus globin and apoB48 coding region). This plasmid was used as a template for transcription by SP6 polymerase using a kit. Subsequently, the mRNA was translated into [ 35 S]methionine-labeled protein using rabbit reticulocyte lysate. Two microliters of lysate containing the translation product was diluted in 250 l of TXSWB and incubated with 6 g of the fusion proteins bound to beads on a rotator at room temperature for 4 h. The beads were collected by centrifugation at maximum speed in a microcentrifuge and washed three times with 1 ml of TXSWB and two times with Tris-NaCl buffer. The washed beads were then incubated in SDS loading buffer with 0.5 M DTT for 30 min and then boiled for 10 min. The proteins were resolved by SDS-PAGE and visualized by Coomassie staining followed by autoradiography. Coomassie staining was performed to verify that equivalent amounts of fusion proteins were used in the binding assays.

RESULTS
Strategy to Identify Proteins That Bind to ApoB-Our approach to isolate proteins bound to apoB involved two forms of purification (Fig. 1A). First, we separated intact microsomes containing nascent apoB and associated proteins from cytosolic, nuclear, plasma membrane, and other organellar compartments. Once microsomes were prepared, the proteins were cross-linked using cleavable, homobifunctional, chemical crosslinkers reactive toward amino groups. In the second purification step, apoB and cross-linked proteins were co-immunoprecipitated using polyclonal antibody against apoB. During preparation of the isolated proteins for SDS-PAGE, the sam-ples were treated with 0.5 M DTT, which cleaves the disulfide bond in the chemical cross-linking agents. The co-immunoprecipitated proteins were then separated by SDS-PAGE and visualized by staining (Fig. 1A).
We anticipated needing a large quantity of antibodies for immunoprecipitating rat apoB. However, antibodies against rat apoB are not commercially available. Because apoB is such a large protein, it is likely that several epitopes are conserved in both rat and human apoB. Therefore, we screened commercially available antibodies raised against human apoB for reactivity against rat apoB. We specifically sought an antibody that would immunoprecipitate rat apoB as efficiently as an aliquot of antiserum raised against rat apoB that was kindly provided by Janet Sparks. Aliquots of rat liver homogenate were subjected to immunoprecipitation using some polyclonal and some monoclonal antibodies against human apoB. For comparison, the polyclonal antibody against rat apoB was used for immunoprecipitation of one aliquot. The immunoprecipitated apoB was visualized by Western blotting using the antibody against rat apoB (Fig. 1B). The monoclonal antibodies appear specific to epitopes in human apoB that are not present in rat apoB (Fig. 1B, lanes 2-4). The three commercial polyclonal antibodies against human apoB had varying efficacy in immunoprecipitating rat apoB (Fig. 1B, lanes 5-7). We found that polyclonal antibody against human apoB obtained from Chemicon International immunoprecipitated rat apoB48 and apoB100 as well as antibody raised against rat apoB (Fig. 1B,  compare lanes 1 and 7). We used antibody from the same lot from Chemicon for all subsequent experiments.

Co-immunoprecipitation of Proteins Cross-linked to ApoB-
In an effort to detect proteins that bind to apoB, we prepared microsomes, treated some aliquots with chemical cross-linkers, and subjected the solubilized proteins to immunoprecipitation with antibody against apoB. Uncross-linked samples showed a sharp band of apoB near the top of the gel and a few other bands of faster migration (Fig. 2, lane 1). Two broad bands corresponding to the eluted heavy and light chains dominate each lane (Fig. 2). When the intact microsomes were treated with the cross-linker DSP, some new bands are visible (Fig. 2, lane 2; e.g. see the 18 -29-kDa range). When cross-linking was carried out using DTBP, many new bands emerge (Fig. 2, lane  3). This pattern of bands was consistent despite variations in the concentration of microsomal protein during cross-linking (data not shown). Furthermore, we always detected more bands using DTBP compared with DSP (data not shown). We selected DTBP for further studies because a larger amount of proteins are cross-linked to apoB with this reagent (Fig. 2). Because the length of the spacer arm between the two reactive imidoester groups of DTBP is 11.9 Å, these bands correspond to proteins that contact or are within close proximity to apoB.
The prominent bands corresponding to heavy and light chains posed two problems. First, the increased amount of antibody necessary for a larger scale preparation for protein sequencing would overload and distort the gel lane. Second, the broad bands obscure underlying and neighboring bands rendering them undetectable. Attempts to elute apoB from the polyclonal antibody while leaving antibody bound to the solid matrix were unsuccessful. Therefore, we cross-linked the antibodies to protein G-agarose and used the conjugated antibodies for immunoprecipitation. Antibodies were conjugated via coupling to cyanogen bromide-activated beads or by cross-linking antibodies on protein G-agarose using the non-cleavable agent DSS. Once conjugation was complete, the beads were washed several times to remove unbound antibodies. We verified that the conjugated heavy and light chain did not elute visibly from the beads despite boiling in SDS and DTT (data not shown). Using conjugated antibodies, we compared the proteins immunoprecipitated from aliquots of uncross-linked microsomes (Fig. 3, lane 1) FIG. 1. A, strategy for the purification of proteins associated with apoB. See "Experimental Procedures" for details. B, one polyclonal antibody against human apoB can efficiently immunoprecipitate rat apoB. Rat livers were homogenized and solubilized in TXSWB. Equal dilutions of antibody were added to aliquots of the homogenate, and immunoprecipitation of rat apoB was carried out. The washed immunoprecipitates were separated on an 8% polyacrylamide gel and transferred to Hybond C Super membrane overnight. The blot was probed with antibody against rat apoB and visualized via chemiluminescence. Bands for apoB100 and apoB48, both of which are synthesized in rat liver, are seen in some lanes. The antibodies used for immunoprecipitation of the aliquots is indicated in the labels above the blot: rat, polyclonal antibody against rat apoB; 1D1, Bsol1, and Bsol7, monoclonal antibodies against human apoB; X and Y, two commercial polyclonal antibodies against human apoB; C, Chemicon polyclonal antibody against human apoB.

FIG. 2. Many proteins can be cross-linked to apoB.
Microsomes were prepared from rat livers, and some aliquots were treated with either of the chemical cross-linkers DSP or DTBP. After quenching the cross-linking reaction, the microsomes were collected and solubilized in TXSWB. Antibody against apoB and then protein G-agarose beads were added to the samples, and the immunoprecipitated proteins from the washed beads were separated on a 15% gel and visualized by silver staining. In lane 1 (Un), the microsomes were not treated with a cross-linker. In lanes 2 (DSP) and 3 (DTBP), microsomes were treated with the respective cross-linkers prior to immunoprecipitation. The positions of molecular mass markers are indicated on the right.
with microsomes treated with DTBP (Fig. 3, lane 2). The conjugated antibody co-immunoprecipitated several proteins that are not readily evident in the uncross-linked sample. Furthermore, the conjugated antibody co-immunoprecipitated a very similar pattern of proteins from cross-linked microsomes as did unconjugated antibodies (free in solution) that were collected subsequently on protein G-agarose (Fig. 3, compare lanes 2 and 3). The pattern and intensity of the bands is very similar regardless of whether the antibodies were conjugated. The lack of heavy and light chains from the conjugated antibody reveals other bands (Fig. 3, lane 2) and permits increasing the scale of the purification without overloading the gel with protein. Thus, conjugated antibody offers significant advantages without significantly altering the recovery of co-immunoprecipitated proteins.
Purification and Sequencing of Proteins-Using this approach to isolate proteins that co-immunoprecipitate with apoB, we purified sufficient masses of each band for protein sequencing by mass spectrometry. In the previous analytical scale experiments, silver staining was used to detect bands that in some cases were at the threshold of detection in the range of 1 ng. To ensure adequate recovery of peptides from the gel, we anticipated needing roughly 50 -100 ng of each band. Microsomes were prepared from the livers of four adult rats, treated with DTBP, solubilized, and subjected to immunoprecipitation with conjugated antibody against apoB. After cleaving the cross-linker, the released proteins were separated in one lane by SDS-PAGE and visualized using a colloidal Coomassie stain. This staining kit has a reported sensitivity of 10 ng. The gel used for sequencing is shown in Fig. 4. In view of the cost of protein sequencing, we excised and pooled bands of similar migration into nine pools.
The excised bands were given to the Harvard Microchemistry Facility for sequencing. After the proteins were digested with trypsin in the gel, the released peptides were sequenced by microcapillary reverse-phase high pressure liquid chromatography nano-electrospray tandem mass spectrometry on a Finnigan LCQ quadrupole ion trap mass spectrometer. The peptide data was correlated with known sequences using the algorithm Sequest (24) and programs developed at the Harvard Microchem-istry Facility (25). Finally, the peptide sequences were reviewed manually for fidelity and consensus with known proteins.
A total of 377 peptides were matched to 120 known data base entries and 99 unique proteins. The proteins identified in each pool of bands is shown in Fig. 5. The peptides ranged in length from seven to 31 amino acid residues. On average, the median lengths were nearly 14 amino acid residues. The chance of an exact match of two peptide sequences over 14 amino acids is about 1:1.6 ϫ 10 18 . This calculation lends considerable credence to the identity of the matched protein, and the presence of longer or multiple peptides corresponding to the same protein strengthens the identification further. Two contaminants did show up in our results. Despite an effort to handle the gel with washed gloves as little as possible, keratin was detected by this highly sensitive procedure in four of the nine pools of bands and accounted for 18 of the matches overall. This contamination also could have occurred during handling of the glass plates prior to pouring the gel. The other contaminant stems from the immunoglobulin chains that eluted from the beads probably in small amounts because they account for 22 of the 377 peptides. Thus, 40 of the 377 peptides could be the result of contamination. The rest of the matches appear to derive from the samples. Nearly all of the peptides correspond to proteins of molecular mass that match their group indicating that significant degradation did not occur. (Note that the range of apparent molecular masses of the matched proteins, estimated from where the bands were excised, is approximate especially for proteins of slower migration.) The principal exceptions are peptides from keratin at scattered places on the gel, uricase (also called urate oxidase), ␣ and ␤ globin, one peptide matching a cytochrome P450 enzyme, and two peptides that match PDI and BiP found in pool 3 (in the 20-kDa range). There were 40 unique ribosomal proteins identified. Presumably, these proteins were cross-linked to chains of apoB that had not finished elongation; the isolation of translation elongation factor ␣ also supports this idea. Most of the remaining proteins are hepatic enzymes and not structural proteins, although two forms of actin did occur. Cytochrome P450 enzymes accounted for 12 matches. Of the 38 remaining proteins, some are known to contact apoB during its biogenesis such as PDI and BiP. We next investigated whether some of these newly identified proteins can bind apoB.
Ferritin Chains Can Bind ApoB Directly-Proteins identified by cross-linking and co-immunoprecipitation could contact or  2 and 3). The solubilized microsomes were subjected to immunoprecipitation with either antibody conjugated to cyanogen bromide-activated beads (lanes 1 and 2) or free antibody against apoB in solution followed by the addition of protein G-agarose (lane 3). The washed immunoprecipitates were displayed on a 15% gel followed by silver staining. The positions of molecular mass markers are indicated on the left .   FIG. 4. Preparative gel used for isolating proteins for sequencing. Microsomes were prepared from livers of four adult rats, crosslinked with DTBP, and solubilized in TXSWB. A total of 14 mg of microsomal protein was subjected to immunoprecipitation using an antibody cross-linked to protein G-agarose with the non-cleavable reagent DSS. The immunoprecipitated proteins were released by incubating the washed beads with SDS loading buffer and 0.5 M DTT. The proteins were electrophoresed on a 15% acrylamide gel and stained using colloidal Coomassie. The bands were excised using sterile scalpels, grouped into 9 pools, and sent for microsequencing. The numbers and lines on the right indicate the pools.
FIG. 5. Table of proteins cross-linked to apoB identified by sequencing. The pooled bands from Fig. 4 were sequenced by mass spectrometry (see "Purification and Sequencing of Proteins" for details). The size range of the pooled bands is only an estimate. Proteins that were identified by exact peptide matches are listed with their accession number in the National Center for Biotechnology Information protein data base. If peptides matched proteins from different species, the first match encountered in the data base is shown. The molecular mass for each full-length protein in the data base was calculated from its amino acid sequence; some of the matched data base entries are for fragments (denoted frag). The number of matching peptides and their median length are also shown. Within each pool, the proteins are listed in descending order based on the number of matching peptides. In pool 3, band 1a (10 kDa) was included that is not readily visible in the scan of the preparative gel.
simply be located within about a 12-Å range of apoB. We investigated whether some of the proteins discovered can bind apoB directly. We chose to investigate the binding of apoB by ferritin heavy and light chains due to their interaction detected in other types of experiments. 2 We found that small amounts of apoB can be co-immunoprecipitated from HepG2 cells in the absence of a cross-linker using antibodies against ferritin; at these levels of co-immunoprecipitation, however, it is difficult to rule out crossreactivity of the antibodies with epitopes on apoB (data not shown). Furthermore, if binding is brief in cells, the association might not be detected easily without using a cross-linker.
To circumvent these issues, we investigated whether these 2 S. Hevi and S. L. Chuck, unpublished observations.

FIG. 5-continued
Identification of Proteins That Bind ApoB 22015 candidates can bind to apoB directly in vitro. We constructed plasmids encoding fusion proteins with ferritin heavy and light chains. First, we cloned the coding sequences for ferritin heavy and light chains by reverse transcription-PCR using mRNA from HepG2 cells. The PCR products were engineered into the plasmid TyB2 and verified by sequencing. The resulting plasmids encoded ferritin heavy or light chain fused to an intein and chitin-binding domain (Fig. 6A). These ferritin fusion proteins and the product of unmodified TyB2 (the multiple cloning site, intein, and chitin-binding domain) were expressed in E. coli and purified by binding to chitin beads. The binding assay was carried out using equal masses of the three proteins (bound to beads) and [ 35 S]methionine-labeled apoB29 (the amino-terminal 29% of apoB) that was synthesized using rabbit reticulocyte lysate. ApoB29 is a large protein (molecular mass, ϳ150 kDa) that still can be translated efficiently in vitro. We found that both of the ferritin fusion proteins bound apoB29, with the fusion protein encoding ferritin heavy chain having the greatest affinity (Fig. 6B, lanes 1 and  2). In contrast, the intein and chitin-binding domain by themselves did not bind apoB29 well (Fig. 6B, lane 3). Furthermore, the lack of binding of globin or other unlabeled proteins in the lysate to the ferritin fusion proteins, as ascertained by Coomassie staining of the gel (data not shown), indicates that the binding of apoB by ferritin is specific. Thus, ferritin heavy and light chains not only can be cross-linked to apoB in microsomes prepared from rat livers but also can bind apoB directly. These proteins are candidates for roles in the biogenesis of apoB. DISCUSSION We developed a biochemical approach to purify proteins associated with apoB. This strategy has its advantages and its limitations. Proteins in microsomes derived from rat livers were treated with a chemical cross-linker. This step enables detection of proteins that might bind apoB too weakly to remain associated throughout the immunoprecipitation procedure. The downside of using a cross-linker is that proteins that are in close proximity (i.e. within about 11.9 Å, the length of the spacer arm) but not actually in contact could become crosslinked to apoB. Thus, the use of a chemical cross-linker results in decreased specificity as the price for increased sensitivity. Relatively few proteins, however, are co-immunoprecipitated with apoB when the microsomes were not treated with a chemical cross-linker (see Fig. 3, lane 1, and the few bands seen in Fig. 2, lane 1). Therefore, we chose to use a cross-linker. In the second stage of purification, the microsomes were solubilized, and the proteins were co-immunoprecipitated with antibody against apoB. Polyclonal antibodies often cross-react with other proteins. However, the relative lack of proteins other than apoB in the uncross-linked sample attests to the specificity of the immunoprecipitation step (Fig. 3, lane 1). The combination of microcapillary high pressure liquid chromatography and tandem mass spectrometry is a powerful and sensitive tool for protein sequencing. This sensitivity has its limitations, too. Even small amounts of contamination with human keratin from handling the gel, glass plates, or container for staining can be detected. Some proteins might have been masked by the large number of ribosomal proteins cross-linked to nascent apoB or diluted below detection by pooling bands. A few anticipated proteins, including MTP, calnexin, calreticulin, and GRP94, did not appear in the list of matches. Apparently, these proteins failed to be cross-linked to apoB or were recovered in inadequate amounts from co-immunoprecipitation, or too few peptides of suitable lengths were released from the gel or reverse phase high pressure liquid chromatography for detection. Furthermore, a peptide from rat MTP may have been unmatched by the algorithms used. The sequence for rat MTP is not currently in the data base. MTP from humans, hamsters, and cows are highly homologous and about 85% identical in sequence, and rat MTP might match MTP of other species to a similar degree. The stringent matching process used, however, requires identity over the entire tryptic peptide and not just homology. All of these limitations of our biochemical approach were accepted in view of the lack of a tractable genetic or in vitro system to study the biogenesis of such a large and complex protein. Despite these shortcomings, this strategy did identify proteins that bind to apoB.
Many ribosomal proteins were cross-linked to presumably incompletely elongated chains of apoB. When lipid synthesis is limiting in HepG2 cells, apoB proteins that appear to be fulllength can be incompletely synthesized and remain in functional association with the ribosome and translocation machinery in the ER membrane (26). Thus, it is not surprising that such a large number of ribosomal proteins were cross-linked to apoB in rat liver microsomes.
This biochemical approach yielded 50 proteins other than ribosomal proteins and contaminants. Some known chaperone proteins were identified, including members of the PDI family and BiP. Despite a lack of evidence for degradation, peptides matching PDI and BiP were found not only in their appropriate size range but also in pool 3. These single peptides may have FIG. 6. Two of the identified proteins, ferritin heavy and ferritin light chains, bind apoB. A, diagram of fusion proteins. Fusion proteins were expressed that consist of ferritin heavy or light chain followed by an intein and a chitin-binding domain. The control fusion protein was expressed from unmodified TyB2 and encoded the region of multiple cloning sites (MCS), intein, and chitin-binding domain. The protein domains are not represented to scale. B, the three fusion proteins were expressed in E. coli and purified using chitin beads. The bound proteins were then incubated with [ 35 S]methionine-labeled apoB29 that was synthesized using rabbit reticulocyte lysate for 4 h on a rotator at room temperature. After washing the beads, the proteins were eluted in SDS loading buffer, separated on a 15% gel, and visualized by fluorography. The positions of molecular mass markers are indicated on the left. been cleaved from smaller proteins than these known chaperones that serve similar functions. One matched protein, fatty acyl co-A ligase (27), is an enzyme involved in triglyceride biosynthesis, which can affect the secretion of apoB (28 -30). Fatty acyl co-A ligase and apolipoprotein B can be found in the ER, and both are enriched in the mitochondria-associated membrane fraction (31). Our data raise the possibility that apoB is in close contact with this enzyme that synthesizes lipids to be transported. An enzyme that converts cholesterol esters to free cholesterol, cholesterol ester hydrolase (32), was identified. This result may fit in with the growing body of evidence that the pathways for hepatic cholesterol metabolism and lipoprotein secretion intersect (2,33). Another steroid metabolic enzyme, 11-␤ hydroxysteroid dehydrogenase (34), was also found. Five matches are with uricase (urate oxidase), which accounted for 39 peptides overall. However, this enzyme cannot play a role in the biogenesis of apoB in man because there is not a human homologue (35). Similarly, L-gulono-␥lactone oxidase, a key enzyme in ascorbic acid biosynthesis, is missing in humans rendering us susceptible to scurvy (36). Twelve different cytochrome P450 enzymes were isolated. As a family, these microsomal, heme-containing enzymes are involved in the conversion of metabolites for excretion and synthesis of steroid hormones, but they have not yet been implicated in the biogenesis of apoB. Many other proteins also were isolated including some that bind GTP, participate in vesicle trafficking, play a key role in the conversion of vitamin A to retinoic acid, and detoxify substances (e.g. glutathione S-transferase, epoxide hydrolase, and cytochrome P450 enzymes). These proteins might be intracellular neighbors that never interact; alternatively, they might bind apoB and play novel roles in its biogenesis.
Our strategy did identify some novel proteins that can bind to apoB. We show that ferritin heavy and light chains can bind apoB. This result raises several questions. Do ferritin heavy and light chains play a role in the biogenesis of apoB? Where do these cytosolic proteins bind on apoB? Do the iron storage and lipoprotein secretion pathways intersect? What other proteins that were cross-linked to apoB likewise bind apoB? We are currently investigating these questions.