Intrahepatic Assembly of Very Low Density Lipoproteins PHOSPHORYLATION OF SMALL MOLECULAR WEIGHT APOLIPOPROTEIN B*

The possibility that apo-B is phosphorylated was examined using cultured rat hepatocytes. Rabbit antiserum prepared against rat apo-B was found to specif- ically react with both large and small molecular weight apo-B (by electroblotting assay and by immunoprecip- itation of [s6S]methionine-labeled proteins synthesized and secreted by hepatocytes). Following a 4-h incuba- tion with [S2P]orthophosphate, immunoprecipitation, and sodium dodecyl sulfate electrophoresis, an auto- radiographic band corresponding to small molecular weight apo-B was obtained from cells and medium. Compared to the relative abundance of ‘‘P which was associated with secreted small molecular weight apo- B, there was little (if any) detected in large molecular weight apo-B. Addition of excess unlabeled apo-B (ob- tained from rat serum) totally competed with the specific antiserum for this radioactive protein, indicating it was antigenically related to apo-B. Moreover, isolation of the ”P-labeled apo-B electrophoretic band, fol- lowed by acid hydrolysis and phosphoamino acid analysis, showed that at least 20% of the s2P originally associated with small molecular weight apo-B was in the form of phosphoserine.


B, there was little (if any) detected in large molecular
weight apo-B. Addition of excess unlabeled apo-B (obtained from rat serum) totally competed with the specific antiserum for this radioactive protein, indicating it was antigenically related to apo-B. Moreover, isolation of the "P-labeled apo-B electrophoretic band, followed by acid hydrolysis and phosphoamino acid analysis, showed that at least 20% of the s2P originally associated with small molecular weight apo-B was in the form of phosphoserine. Control experiments ruled out the possible contamination of apo-B with phospholipid as well as the possibility that the phosphoserine produced by acid hydrolysis could have been derived from phosphatidylserine. To examine the relevance of these data to the in vivo state, rats were injected with [s2P]orthophosphate. Immunoprecipitation of their livers followed by autoradiographic analysis showed the presence of in small molecular weight apo-B. These data show for the first time that small molecular weight apo-B is synthesized as a phosphoserine containing protein.
VLDL' secretion requires a complex sequential series of intracellular events. These include apolipoprotein and lipid *This was supported by National Institutes of Health Grant HL30560 from the National Heart, Lung and Blood Institute, Public Health Service, Department of Health, Educational and Welfare and by National Cancer Institute Grant CA-34517. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
5 Established Investigator of the American Heart Association. To whom correspondence should be addressed.
' The abbreviations used are: VLDL; very low density lipoprotein; 33 biosynthesis, association of the lipid and protein moieties into lipid-protein aggregates, condensation into secretory vesicles, and targeting to the sinusoidal surface membrane for secretion into plasma (1-3). The molecular interactions through which the nascent VLDL particle is assembled and subsequently vectorially shuttled through this multiorganelle pathway are unknown.
Although there are numerous discrete apolipoproteins associated with VLDL, evidence suggests that apo-B in particular is essential for secreting triacylglycerol-rich lipoproteins. Apo-B exists in at least two molecular weight forms (4-7), both of which are initially secreted by rat liver as a component of VLDL (6-8). In humans, genetic deletion of plasma apo-B abolishes VLDL secretion (9). These results suggest that one or both molecular weight forms of apo-B is required for VLDL assembly. Because of technical problems in working with apo-B, such as self-association and the inability to maintain its solubility in aqueous solutions, little is known about its structure, sequence, and chemical composition. Moreover, as a result of insufficient chemical data, the unique molecular feature which affords apo-B the ability to direct VLDL assembly and secretion remains unknown.
In this report, we show that the small molecular weight form of apo-B is secreted by cultured rat hepatocytes in a form which contains phosphorylated serine residues, as evidenced by specific immunochemical isolation and phosphoamino acid analysis. In marked contrast, there is little or no phosphorylation of the large molecular weight form of apo-B. These novel findings may provide a clue to the molecular interactions through which apo-B containing lipoproteins are assembled, secreted, and metabolized.

MATERIALS AND METHODS
All culture reagents, rats, and chemicals were obtained from sources previously described (10). Radioactive chemicals were obtained from ICN Biochemicals, Irvine, Ca.
Antibody Preparation-An ultracentrifugation fraction ( d < 1.21 g/ml) obtained from rat serum was delipidated according to the method of Cardin et al. (11). Individual molecular weight forms of apo-B were purified by gel filtration on Sepharose CL-GB according to the method of Sparks and Marsh (12). Analysis of the fractions on SDS-PAGE showed that some contained only one band (silver stained) which corresponded in molecular weight to either large molecular weight apo-B or small molecular weight apo-B. Antibodies against the pure individual apo-B forms were raised in rabbits using procedures previously described (10).
Antibody specificity was assessed using the electroblotting technique of Towbin et at. (13) (see Fig. 1). Although antibodies raised against either apo-B form showed relatively the same titers and specificity, antibodies raised against the large molecular weight apo-B form were used for these studies.
Preparation of Hepatocytes-Hepatocytes obtained from male Sprague-Dawley sucrose-fed rats (to induce VLDL secretion (10)) were plated after collagenase digestion using methods which have been described in detail (14). After incubating for 4 h with DME containing 20% calf serum, the medium was changed to serum-free DME. The serum-free medium was phosphate-and methionine-free.
Protein 3zPand 36S-labeling Studies-After incubation with [3'P] orthophosphate ( al. (15). Boiling buffer B was immediately added to the plastic culture dish containing the cells. Visualization via phase contrast microscopy showed that all of the cells were rapidly lysed and came off the dish. The solution containing the cells was diluted with buffer B without SDS in order that the final concentration became 0.15%. Separate aliquots of medium and cells were placed in plastic 1.5-ml tubes and an appropriate amount of rabbit anti-apo-B antisera (previously shown to precipitate all of the apo-B contained in each sample) was added. The solutions were incubated a t room temperature for 18 h after which protein A bound to Sepharose (Pharmacia) was added a t a concentration previously shown to bind all of the rabbit immunoglobulin in each sample. After an additional 4-h incubation, the Sepharose beads were isolated by centrifugation in a microfuge (Fisher Scientific) for 5 min. The Sepharose pellet was washed six times with buffer D (as described by Faust et al. (15)). A sample containing 100 pg of rat serum protein ( d < 1.21 g/ml) was then added as a cold carrier. The samples were then boiled for 15 min in sample buffer containing 8 M urea, 2% SDS, 10% glycerol, 5 mM mercaptoethanol, and 10 mM Tris/glycine (pH 8.3). The samples were tnen subjected to SDS-PAGE using the system of Laemmli (16) except we used a 3-10% linear polyacrylamide gradient. Molecular weight standards run on the same gel permitted an accurate molecular weight determination. Gels containing %labeled proteins were treated with Enlightening (New England Nuclear). Gels were dried onto filter paper and then fluorographed ('"S-labeled) or autoradiographed ("P-labeled) using Kodak-XRP-5 x-ray film and Dupont Cronex Lightening Plus intensifying screens a t -76 "C for the period of time indicated in legends. After developing the x-ray film, individual proteins in the dried SDS-PAGE gel were identified and cut out and their radioactivity was assayed as described (10).
In Viuo Labeling Studies-Male Sprague-Dawley rats were fed chow + 30% sucrose in their drinking water to increase the synthesis of VLDL (10). After three days, rats were injected intraperitoneally with [32P]orthophosphate (10 mCi/2OO g body weight). Three h later, rats were bled and their livers were removed. A total lipoprotein fraction ( d < 1.21 g/ml) was obtained by ultracentrifugation of serum (17). The livers were homogenized in 40 ml of buffer D (15). Both the fraction ( d < 1.21 g/ml) and the liver homogenate were immunoprecipitated and subjected to SDS-PAGE. Phosphoamino Acid Annlysis of Apo-B-After autoradiographic identification of the apo-B band, it was cut out of the dried polyacrylamide gel and the radioactivity was assayed by scintillation counting. The dried gels were then placed in a 5-ml plastic pipette and were subjected to electrophoresis after loading 100 pg of bovine serum albumin dissolved in 0.1 ml of the sample buffer (see above). After 18 h of electrophoresis, there was no Coomassie blue dye left on the gel and all of the dye was located in the dialysis bag which was affixed to the bottom of the pipette. Analysis of the radioactivity isolated in the bag showed an 80% yield. The contents of the bag were made to 20% trichloroacetic acid and centrifuged a t 2000 rpm a t 4 "C for 20 min. The recovery of radioactivity in the precipitate averaged 60% of the original. The precipitate was extracted two times with acetone, and once with ethanol. There was no loss of radioactivity by either wash. The precipitate was brought up in 6 N HCI, placed in a capillary tube which was then flame sealed, and finally heated a t 110 "C for 2 h. The free amino acids were then subjected to paper electrophoresis using procedures previously reported (18).

RESULTS AND DISCUSSION
Specificity of Antiserum and Immunoprecipitation-The specificity of the antiserum was determined by electroblotting and by immunoprecipitation of cells and medium was labeled with [%]methionine (Fig. 1). Both methods showed that the antiserum specifically reacted with both molecular weight forms of apo-B.
We used this specific antiserum to examine the possibility that apo-B is phosphorylated. Using the same preparations of cells and experimental protocol as was used for the [35S] methionine studies, we examined the incorporation of ["PI orthophosphate into immunoprecipitable apo-B. The results show that "P becomes associated with a secreted protein (ie. isolated in the medium) which has similar migration (SDS-PAGE) to the small molecular weight form of apo-B (isolated from rat plasma) (Fig. 1). We were unable to detect a discrete

FIG. 1. Incorporation of [%]methionine and ['*P]orthophosphate into apo-B by cultured rat hepatocytes. Four h after
initially plating hepatocytes in DME + 20% calf serum, the culture medium was changed to serum-free DME, which was methionineand phosphate-free (lanes 3-6). After  "P-labeled band for large molecular weight apo-B. Additional experiments show that the autoradiographic technique used is capable of detecting 5% of the radioactivity found associated with small molecular weight apo-B. Thus, if large molecular weight apo-B is secreted in a phosphorylated form, it contains less than 10% of the 32P which is associated with small molecular weight apo-B.* It is possible that in other metabolic conditions or in other species the relative phosphorylation of the different molecular weight forms of apo-B might be different from that observed in these experiments.
Immunoprecipitation of the cell lysate also showed the presence of radioactivity corresponding to the small molecular weight form of apo-B (Fig. 1). To ascertain whether this band was, in fact, due to apo-B, excess (50 pg) of unlabeled apo-B (obtained from rat plasma) was added to the cell lysate and the medium prior to adding the antisera. The results showed the complete disappearance of the band corresponding to small molecular weight apo-B in the medium and cells (data not shown). In several but not all experiments, 32P-labeled proteins that had molecular weights less than the plasma apo-B standards were isolated from the cell lysate. The autoradiographed bands corresponding to these proteins did not disappear when excess unlabeled apo-B was added to the immunoprecipitation buffer, suggesting they are not antigenically related to apo-B. Additional studies showed that these bands also were precipitated with serum from nonimmunized rabbits and protein A bound to Sepharose suggesting that these proteins have some affinity for these reagents.
Phosphoamino Acid Analysis of Apo-B-To define the covalent linkage of the 32P in apo-B, the radioactive (i.e 32Plabeled) protein corresponding to small molecular weight apo-Based on our experimental finding that large molecular weight apo-B is secreted a t a rate which is 30-5076 of the rate of small molecular weight apo-B secretion (10). B was cut out of the SDS-PAGE gels, eluted by electrophoresis into a dialysis bag, and subjected to acid hydrolysis and paper electrophoresis. The results show that small molecular weight apo-B contains phosphoserine as the only detectable phosphoamino acid (Fig. 2). There also was the presence of ["PI phosphate (free inorganic phosphate). Since unincorporated ["LP]phosphate was removed by precipitation with trichloroacetic acid and washing, it is probable that the free phosphate was not present in the unhydrolyzed sample. It is more likely that the free phosphate was produced during the acid hydrolysis step. Quantitation of the amount of radioactivity in phosphoserine showed that a minimum of 20% of the 32P originally associated with small molecular weight apo-B was in the form of phosphoserine. The 20% recovery of phosphoserine from apo-B after acid hydrolysis is similar to the recoveries found for other phosphoserine-containing proteins (19,20). We can not rule out the possibility that some of the :3zP associated with apo-B and hydrolyzed to free phosphate may have been derived from phosphorus forms in addition to phosphoserine.
T o rule out the possibility that [32P]phosphoserine produced by acid hydrolysis of apo-B was derived from phosphotidylserine, three control experiments were performed. First, cells and medium were labeled with [S2P]orthophosphate using the same protocol as described in Fig. 1. The "P-labeled lipids were quantitatively extracted (10). The "P-labeled lipids (following solvent evaporation) were added to unlabeled medium and cells (they were incubated for 4 h in a manner similar to that described in Fig. 1). Immunoprecipitation, SDS-PAGE, autoradiography, and radioactivity quantitation (p-scintillation counting) failed to show any radioactivity in the apo-B SDS-PAGE band. Furthermore, acid hydrolysis of the 32Plabeled lipids also failed to yield [32P]phosphoserine as assayed in Fig. 2. Secondly, [3H]glycerol was added to cells for 4 h to form 'H-phospholipids. Immunoprecipitation, SDS-PAGE, and radioactivity quantitation by p-scintillation counting also failed to show any radioactivity. The sensitivity of this assay was 10 ng of phospholipid. Finally, phosphatidylserine was subjected to the same procedure as was used to isolate ["P]phosphoserine from the "'P-labeled apo-B (as in Fig. 2). Acid hydrolysis of phosphatidylserine (100 pg) failed to yield any detectable phosphoserine. The quantitative acid hydrolysis of phosphatidylserine produced only free serine (data not shown). This assay was capable of detecting 0.4% of the phosphoserine which could have been produced from 100 pg of phosphatidylserine. Since acid hydrolysis of 32Plabeled apo-B yielded 20% of the 32P in the form of phosphoserine (Fig. 2)., these data rule out the possibility that phosphatidylserine was the source of the ["P]phosphoserine.
In Viuo Demonstration of Apo-B Phosphorylation-To determine if apo-B is phosphorylated in uiuo, sucrose-fed rats were injected with 10 mCi of [32P]orthophosphate. After 3 h, rats were bled, the serum was separated into a total lipoprotein fraction ( d < 1.21 g/ml) by ultracentrifugation. Analysis of the fraction ( d < 1.21 g/ml) by SDS-PAGE (silver stained) clearly showed the mass presence of both molecular weight forms of apo-B (data not shown). Immunoprecipitation of the fraction ( d < 1.21 g/ml) showed a faint but definitive autoradiographic band for only the small molecular weight form of apo-B (data not shown). Immunoprecipitation of the livers of these rats showed the presence of 32P in small molecular weight apo-B (Fig. 3). It is unknown if some of the 32P-labeled small molecular weight apo-B obtained in the serum may have also been produced by the intestine.
Additional studies showed that incubation of VLDL (containing 32P-labeled small molecular weight apo-B) with fresh rat serum for 8 h caused a dramatic loss of 32P (258 f 24 cpm before incubation; 78 & 15 cpm after incubation; n = 3 in each group). These results indicate that rat serum contains a process which can, in fact, dephosphorylate small molecular weight apo-B.
Recent studies show that apo-A-I (21-22) and apo-A-II(23) are synthesized by the liver as propeptides. Pro-A-I peptide is normally cleaved in plasma by a protease (24). It is possible that the phosphate group on small molecular weight apo-B may behave similar to the NHz-terminal extention of a propeptide (i.e. after being secreted by the liver, phosphorylated apo-B is dephosphorylated).
Phosphorylation may be a general characteristic of a specific class of amphipathic proteins. Several other amphipathic proteins which associate with lipids to form lipid-protein aggregates in a manner similar to VLDL apo-B are phosphorylated. For example, avian liver secretes lipoproteins which act to transport lipid to the oviduct for egg development (25). Two of these proteins derived from vitellogenin are polyphosphorylated (26). Casein, a major protein in milk, is hydrophobic, excreted complexed with phospholipid and triacylglycerol, and is polyphosphorylated on serine and threonine residues (27). In addition, myelin basic protein, which is also a hydrophobic protein complexed with phospholipid, is also phosphorylated on serine and threonine residues (28).
Staehelin and Arntzen (29) propose that phosphorylation of proteins in chloroplasts provide the molecular forces nec-  fig. 1) was cut out of the SDS-PAGE gel, eluted by electrophoresis, and hydrolyzed with acid, and the free amino acids were separated by paper electrophoresis. The autoradiogram was obtained after a 1-week exposure. The migration of standard phosphoserine is indicated. Standard phosphothreonine and phosphotyrosine are well separated from phosphoserine (data not shown).