Biosynthesis of the Vitellogenins IDENTIFICATION AND CHARACTERIZATION OF NONPHOSPHORYLATED PRECURSORS TO AND

Avian vitellogenin consists of two major species des- ignated VTG I and VTG 11. Rooster hepatocytes were employed to identify intracellular forms of the vitello- genins and to characterize biosynthetic intermediates of VTG I and VTG 11. After labeling with [3H]serine, intracellular vitellogenin radioactivity was seen in mature VTG I and VTG I1 but was primarily found in two species, pVTG I and pVTG 11, which showed greater mobilities in sodium dodecyl sulfate-polyacrylamide gels. The pVTG species were identified as vitellogenins by reaction with antibodies against plasma VTG I1 and against the mixture of VTG I and VTG 11. Immunological and peptide mapping procedures were used to re- late pVTG I and pVTG I1 to secreted VTG I and VTG 11, respectively. Pulse-labeling and pulse-chase experi- ments showed that the pVTG species are precursors to the secreted vitellogenins and are thus discrete inter- mediates in the biosynthesis of the vitellogenins. Additional labeling experiments showed that the pVTG species are glycosylated but not phosphorylated. The stages of vitellogenin biosynthesis may be ordered as follows: polypeptide synthesis + glycosylation + phosphorylation -+ secretion. The presence of only small quantities of the phosphorylated vitellogenins intracellularly indicates that when

Avian vitellogenin consists of two major species designated VTG I and VTG 11. Rooster hepatocytes were employed to identify intracellular forms of the vitellogenins and to characterize biosynthetic intermediates of VTG I and VTG 11. After labeling with [3H]serine, intracellular vitellogenin radioactivity was seen in mature VTG I and VTG I1 but was primarily found in two species, pVTG I and pVTG 11, which showed greater mobilities in sodium dodecyl sulfate-polyacrylamide gels. The pVTG species were identified as vitellogenins by reaction with antibodies against plasma VTG I1 and against the mixture of VTG I and VTG 11. Immunological and peptide mapping procedures were used to relate pVTG I and pVTG I1 to secreted VTG I and VTG 11, respectively. Pulse-labeling and pulse-chase experiments showed that the pVTG species are precursors to the secreted vitellogenins and are thus discrete intermediates in the biosynthesis of the vitellogenins. Additional labeling experiments showed that the pVTG species are glycosylated but not phosphorylated. The stages of vitellogenin biosynthesis may be ordered as follows: polypeptide synthesis + glycosylation + phosphorylation -+ secretion. The presence of only small quantities of the phosphorylated vitellogenins intracellularly indicates that when phosphorylation is completed, the vitellogenins are rapidly secreted from the hepatocyte.
The differences in the electrophoretic mobilities of the pVTG and VTG species suggested that sodium dodecyl sulfate-polyacrylamide gel electrophoresis does not accurately estimate the molecular weights of the heavily phosphorylated vitellogenins. This was confirmed directly by showing that the mobility of plasma vitellogenin increased upon dephosphorylation. An independent estimate of vitellogenin molecular weight was made by gel chromatography in 7 M guanidine-HC1. With this method, the molecular weights of the pVTG and VTG species were indistinguishable and in agreement with the molecular weight of the pVTG species as judged by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. These analyses indicate that the vitellogenin polypeptide has M, = 180,000. This value is 60,000-70,000 less than commonly estimated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. The implications of this lower molecular weight are discussed in relation to vitellogenin strut-* This research was supported by Grant AM 18171 from the National Institutes of Health. 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. ture and the egg yolk polypeptides which may derive from each vitellogenin.
Vitellogenin is the major yolk precursor protein synthesized by and secreted from the livers of oviparous vertebrates in response to estrogenic stimulation (1,2). After uptake into the developing oocyte, vitellogenin is cleaved into a complex group of yolk proteins which act as nutrient sources for embryonic development. The enormous increases in hepatic vitellogenin synthesis and vitellogenin mRNA following hormone treatment have focused considerable attention on avian and amphibian vitellogenins as models for estrogen-regulated gene expression (1-3). Recent studies indicate that vitellogenin is not a unique protein, but appears to be a small family of similar proteins encoded by several genes (4,5). At the protein level, heterogeneity has been noted in amphibian vitellogenin (6) and two avian vitellogenins, VTG' I and VTG 11, have been purified from plasma of estrogen-stimulated roosters (7). Comparisons of amino acid compositions, peptide maps, and immunological properties indicate substantial differences in the structures of the two avian vitellogenins. Both vitellogenins are unusually large proteins with apparent M, = 260,000 and 246,000 for VTG I and VTG 11, respectively, as judged by SDS-polyacrylamide gel electrophoresis. Each vitellogenin also contains 2% phosphorus (7) and is glycosylated.2 The vitellogenins are thus large complex proteins which must undergo extensive post-translational modification in the hepatocyte. Surprisingly, little is known about vitellogenin biosynthesis or the influence of the post-translational modifications upon the physical and biological properties of these proteins. Such studies have been difficult because of the complexity of these proteins as well as the inability to analyze specific vitellogenins as discrete molecular species (1,6,8).
We report here experiments which identify and characterize two discrete intracellular vitellogenins which are precursors to avian VTG I and VTG 11. The intracellular vitellogenins are found almost exclusively as nonphosphorylated polypeptides, but only the fully phosphorylated vitellogenins are secreted. These and other results permit us to order the stages of vitellogenin biosynthesis as follows: polypeptide synthesis + glycosylation + phosphorylation + secretion. The presence of only small quantities of the phosphorylated vitellogenins intracellularly suggests that, when phosphorylation is completed, the vitellogenins are rapidly secreted from the hepa-tocyte. Additionally, we report the surprising result that the avian vitellogenin polypeptides are 60,000-70,000 daltons smaller than generally estimated by SDS-polyacrylamide gel electrophoresis. This result has important implications for vitellogenin structure as well as for the elucidation of which egg yolk polypeptides are derived from each vitellogenin.

EXPERIMENTAL PROCEDURES
Hormone Treatment, Hepatocyte Preparation, and Radiolabeling-White Leghorn roosters (0.6-1 kg, SPAFAS, Norwich, CN) received an intramuscular injection of diethylstilbestrol (50 mg/kg) in propylene glycol 3 days prior to killing. Laying hens received no treatment. Hepatocytes were prepared by a previously described (9) modification of the two-step collagenase perfusion method developed by Seglen (10). Hepatocytes exhibit viabilities greater than 95% as judged by trypan blue exclusion and decline in viability by no more than 10% during 4 h of suspension culture (9). Radiolabeled amino acid incorporation into protein is linear for at least 4 h (9). Hepatocytes (5 X 10'/ml) were incubated at 40 "C in modified minimal essential medium (9) containing 5% chicken serum and one of the following: ['Hlleucine (100 p~, 39.6 Ci/mmol,New England Nuclear). Cells were removed by centrifugation for 3 min at 60 X g and washed once in iced balanced salt solution (9). Medium was centrifuged again for 1 min at 10, OOO x g to ensure complete removal of cells. Sample buffer of electrophoretic system A (11) was added to the medium; the sample was boiled for 3 min and analyzed by electrophoresis as indicated below or stored at -70 "C until analysis. The washed hepatocytes were solubilized by boiling in electrophoresis sample buffer and analyzed by electrophoresis. For immunological analyses, cells were homogenized at 0-4 "C in 0.02 M sodium phosphate, pH 7.4, 0.15 M NaCl, 0.005 M ethylenediaminetetraacetic acid, 1% Triton X-100, 100 pg/ml PMSF, and debris was removed by centrifugation for 5 min at 10, OOO X g. In several experiments, the influence of proteolysis was tested by including the following inhibitors singly or in combination in homogenization buffer or electrophoresis sample buffer in addition to PMSF: aprotinin (10 units/ml), pepstatin A (1 pg/ml), diisopropylfluorophosphate (5 mM), benzamidine (5 mM), and 2,3-dimercaptopropanol (10 mM).
For in vitro tissue incubation, a liver slice (IO mg) was chopped into four pieces and incubated for 1 h at 40 "C in 0.05 ml of Krebs-Ringer bicarbonate buffer (12) containing 50 or 250 pCi of ['H]leucine or ['Hlserine under an atmosphere of 95% 0%/5% COZ. The tissue was washed with 2 ml of iced Krebs-Ringer buffer, homogenized in 0.15 ml of electrophoresis sample buffer containing 100 pg/ml PMSF, boiled for 3 min, and analyzed by electrophoresis. In vivo labeling was carried out by injecting ['H]leucine (1.5 mCi/kg) intraperitoneally 15 min prior to killing. Liver slices were homogenized directly in electrophoresis sample buffer, boiled, and analyzed by electrophoresis.
Immunological Analysis-Rabbit antibody against VTG I1 (anti-VTG 11) has been described (7). Antibody against the unresolved mixture of VTG I and VTG I1 (anti-VTG M) was prepared in the same fashion (7). Direct immunoprecipitations of hepatocyte extracts or incubation medium were carried out with unlabeled antigen as carrier at 50-75% of equivalence (7). Washed immunoprecipitates were dissolved in electrophoresis sample buffer, boiled, and analyzed by SDS-polyacrylamide gel electrophoresis.
Electrophoresis and Partial Proteolysis-Electrophoresis was carried out as described (System A Ref. 11) with SDS-polyacrylamide slab gels using the buffer system of Laemmli (13). The running gel contained either 5, 7.5, or 10% acrylamide as indicated. Radioactive proteins were visualized by fluorography (14). In several experiments, ["Hlleucine incorporation into cellular and secreted vitellogenin was determined by cutting the appropriate bands from the dried gel. The gel secretions were combusted, and radioactivity was recovered for scintillation counting as described (9). For partial proteolysis mapping (15), liver slices were labeled with pH]leucine or ['Hlserine as noted above and run on a SDS-5% polyacrylamide gel. Dansylated vitellogenins were added to the sample to facilitate identification of the vitellogenin region of the gel, which was excised and lyophilized. The dried gel was rehydrated in the minimum necessary volume of protease digestion buffer (7) containing 0.2 mg/ml V8 protease from Staphylococcus aureus (Miles). After incubation for 2 h at 37 "C, the gel was boiled for 3 min, loaded onto a second slab gel at 90" to the first dimension, and run into an SDS-10% polyacrylamide gel. Lyophilization of the gel segment after f i i t dimension electrophoresis markedly improved the reproducibility of the digestion kinetics in the gel, presumably because the protease enters the gel quickly during rehydration.
Dephosphorylation of Plasma Vitellogenins-Vitellogenin was dephosphorylated with bacterial alkaline phosphatase (17 units/mg vitellogenin) (BAPC, 27.7 units/mg, Worthington) at 37 "C in 0.025 M 2-(N-morpholino)ethanesulfonic acid, pH 5.5, containing 100 pg/ml PMSF.3 The reaction was stopped by boiling or by the addition of trichloroacetic acid to a concentration of 5%. In the latter case, samples were centrifuged for 5 min at 10, OOO X g, and phosphorus was measured (16) in the supernatant to assess the extent of dephosphorylation. The initial phosphorus content was determined as phosphorus released by alkaline hydrolysis from the trichloroacetic acidinsoluble pellet of samples taken at time 0. Dephosphorylation experiments were carried out with vitellogenin preparations enriched in VTG I1 and preparations enriched in VTG I. When the dephosphorylated vitellogenins were run on SDS-5% polyacrylamide gels, ['HI serine-labeled hepatocyte extracts were run in adjacent gel lanes to determine the mobilities of each VTG and the corresponding nonphosphorylated precursors.
Molecular Weight Estimation by Gel Chromatography-Protein samples were S-carboxymethylated (17) and chromatographed on a Sepharose CL-4B column (55 X 0.9 cm) with 7 M guanidine-HC1 as eluting solvent. Purified VTG 11, a mixture of VTG I and VTG 11, and protein standards were dissolved in or dialyzed into 7 M guanidine-HC1, 0.35 M tris(hydroxymethyl)aminomethane, pH 8.6, 0.05 M ethylenediaminetetraacetic acid, 0.19 M P-mercaptoethanol and held at room temperature for 4 h. Iodoacetic acid was subsequently added, and after 30 min, the sample was applied to the column and eluted at 1.6 ml/h with 7 M guanidine-HC1. Column fractions were monitored at 280 nm, and elution positions were measured in terms of the weight of eluting solvent (18). Excluded and included elution weights were determined with dextran blue and ["Hlleucine, respectively. A calibration curve was constructed as recommended by Ackers (19) and Fish et al. (20) with the following protein standards: human apolipoprotein B (Mr = 250,000), myosin heavy chain (M, = 200,OOO), /3galactosidase (Mr = 116,200), bovine serum albumin (Mr = 68,000), ovalbumin (Mr = 43,OOO), deoxyribonuclease (Mr = 31,000), and ribonuclease A (M, = 14,000). Human low density lipoprotein (1.006-1.06 g/cm') was prepared from fresh plasma by standard procedures (21) except that 100 pg/ml PMSF was included in all solutions.
Apolipoprotein B was isolated and delipidated (22) by two cycles of chromatography on Sepharose 6B in 0.02 M sodium phosphate, pH 7, 1% SDS, 100 pg/ml PMSF (11) and extensively dialyzed against 7 M guanidine-HC1 prior to S-carboxymethylation and chromatography. For the chromatographic analysis of cellular VTG and pVTG species, ['Hlserine-labeled liver tissue was homogenized in iced 10% trichloroacetic acid; the insoluble material was collected by centrifugation, washed three times with iced sterile water, dissolved in 7 M guanidine-HCl and S-carboxymethylated as above. After chromatography, individual fractions were dialyzed extensively against 0.02 M tris(hydroxymethyl)aminomethane, pH 6.8,100 pg/ml PMSF at 4 "C, lyophilyzed, and analyzed by SDS-5% polyacrylamide gel electrophoresis.

RESULTS
Occurrence of Putative VTG Precursors-Vitellogenin accounts for as much as 10% of liver protein synthesis in estrogen-stimulated roosters as judged by ['Hlleucine incorporation (23-26). This feature as well as the high serine content of plasma vitellogenin (7,8) suggested that [3H]serine should selectively label newly synthesized VTG I and VTG 11. The results of Fig. 1 indicate that this is the case. This figure shows the electrophoretic profile of newly synthesized proteins after estrogen-stimulated hepatocytes were labeled with ['Hlserine for 1 h. The SDS-5% polyacrylamide gel was run for twice the time necessary for the tracking dye to reach the gel bottom in order to resolve proteins in the M, = 100,000-400,000 range. VTG I1 and are present in approximately the same proportions as the two plasma vitellogenins (7). The secreted vitellogenins were seen only with hepatocytes from estrogen-stimulated roosters and laying hens but not with hepatocytes from untreated roosters. In addition, these proteins were further identified as VTG I and VTG I1 by reaction with anti-VTG M, an antiserum raised against plasma VTG I and VTG I1 (data not shown).
In contrast to the profile of secreted proteins (Fig. 1, lane 2), newly synthesized cellular proteins (Fig. 1, lane I ) show two prominent bands designated pVTG I and pVTG I1 with apparent M , = 200,000 and 190,000 respectively, in comparison to 260,000 and 246,000 for secreted VTG I and VTG I1 (7). As seen here and in experiments described below, pVTG I and pVTG I1 are in similar proportions to secreted VTG I and VTG 11. The cellular pVTG bands were observed only with hepatocytes from estrogen-stimulated roosters and laying hens, but not with hepatocytes from untreated roosters. In addition, the pVTG bands were seen when estrogen-stimulated liver slices were labeled in vitro with either ["Hlserine or ['Hlleucine and when liver proteins were labeled in vivo with ["Hlleucine (data not shown). The occurrence of the pVTG species, therefore, is dependent upon estrogen but is independent of the in vitro labeling conditions and the preparation of hepatocytes. These results suggest that the pVTG species were intracellular vitellogenins. Note that the electrophoretic mobility of apolipoprotein B of very low density lipoprotein (11, 27-29) is the same in cellular (Fig. 1, lane I ) and secreted (Fig. 1, lane 2) proteins. This result indicates the absence of significant proteolytic activity in the cell extract and suggests that the greater electrophoretic mobilities of pVTG I and pVTG I1 as compared to the secreted vitellogenins did not result from proteolysis. Additional experiments to eliminate the role of proteolysis are described below.
Identification ofpVTG as Nonphosphorylated Vitellogenins-The pVTG were further identified as vitellogenins by virtue of their reactivity with anti-VTG M, an antibody raised against the mixture of VTG I and VTG I1 purified from plasma (7). As shown in Fig. 2 (lane 3), pVTG I and pVTG I1 were selectively precipitated by anti-VTG M while apolipoprotein B was not (compare the immunoprecipitate in lane 3 with the profile of total ["Hlserine-labeled hepatocyte proteins in lane I ) . Note that a faint VTG I1 band can be seen in the immunoprecipitate (lane 3) and a VTG I band can also be detected with a longer fluorographic exposure (data not shown). This result indicates that small quantities of cellular VTG I and VTG I1 are present, but the great majority of the newly synthesized vitellogenin occurs as the pVTG species. Densitometric tracings of either ["Hlserine or [:'H]leucinelabeled hepatocyte proteins show that pVTG I and pVTG I1 account for 94-98s of the cellular vitellogenin after labeling periods of 1,2, or 3 h (data not shown).
Post-translational modifications of the cellular vitellogenins were examined by labeling hepatocytes with [:'H]glucosamine or "Pi. After labeling with [:'H]glucosamine, cellular proteins show significant incorporation into pVTG I, pVTG 11, and VTG I1 (Fig. 3, lane I ) , each of which is precipitated by anti-VTG M (Fig. 3, lane 2). Longer exposure of the fluorograph shows labeling of cellular VTG I as well (data not shown). The ['H]glucosamine-labeled vitellogenins (Fig. 3, lane 2) have the same or very similar electrophoretic mobilities compared to the corresponding pVTG and VTG species labeled with ['Hlserine ( Fig. 3, lune 3). This result indicates that both the pVTG and VTG species are at least partially glycosylated. In studies to be reported elsewhere,' we have observed that intracellular vitellogenins labeled with [:1H]serine in the pres- ence of the glycosylation inhibitor, tunicamycin (30,31), have slightly greater, but clearly distinguishable, electrophoretic mobilities than pVTG I and pVTG 11. This result suggests that each vitellogenin molecule within the pVTG I and pVTG I1 bands contains carbohydrate. In contrast to this result, after labeling hepatocytes with "Pi, cellular proteins (Fig. 2,  lane 2; Fig. 4, lane I ) show extensive labeling of the VTG bands but only very minor labeling of the pVTG bands. It is clear that pVTG I1 (Fig. 2, lane 2) shows virtually no labeling with 32Pi while pVTG I shows minor labeling. Comparison of the distribution of ["Hlserine and radiophosphate among the pVTG and VTG species (Fig. 2, lanes 1 and 2), however, indicates that pVTG I as well as pVTG I1 contains very little phosphorus. Secreted proteins (Fig. 2, lane 4; Fig. 4, lane 4) show radiophosphate incorporation only into VTG I and VTG 11. When cellular (Fig. 4, lane 1 ) and secreted (Fig. 4, lane 4 ) "lP-proteins were reacted with anti-VTG M, both cellular (Fig. 4, lane 2) and secreted (Fig. 4  Specific Identification of p VTG Z and p VTG ZZ-The similarity in the proportions of newly synthesized pVTG I and pVTG I1 (Figs. 1 and 2) as compared to VTG I and VTG I1 in secreted proteins (Figs. 1 and 2) or in plasma (7) suggested that pVTG I and pVTG I1 might be specific precursors to VTG I and VTG 11, respectively. These relationships were evaluated via immunological and peptide mapping procedures. First, cellular proteins labeled with ['Hlserine were reacted with either anti-VTG M or antiserum raised against plasma VTG I1 (anti-VTG 11) (7) to determine whether anti-VTG I1 showed selectivity for pVTG 11. The immunoprecipitates were electrophoresed on an SDSd% polyacrylamide gel and the pVTG II/pVTG I ratio was determined by densitometry of the fluorograph. The ratio with anti-VTG I1 was 7 while the ratio with anti-VTG M was 3. pVTG I1 is thus selectively precipitated by antiserum specific to plasma VTG I1 (7). The fact that some pVTG I is precipitated by anti-VTG I1 may indicate a minor degree of cross-reactivity not previously detected (7) or may reflect co-precipitation of pVTG I that is present in heterodimers with pVTG 11. The latter possibility is likely in view of the recent finding that dimers of VTG I and VTG I1 are present in rooster plasma: A similar selectivity was also shown for cellular and secreted "'P-VTG species. As shown in Fig. 4 VTG I1 while anti-VTG I1 (lunes 3 and 6) selectively precipitated [32P]VTG 11. The selective reactivity of cellular pVTG I1 and VTG I1 with anti-VTG I1 relates these species specifically to plasma VTG 11. The poor reactivity of pVTG I with anti-VTG I1 as compared to anti-VTG M indicates immunological dissimilarity between pVTG I and pVTG 11, suggesting indirectly that pVTG I is related to VTG I.
Partial proteolysis mapping (15) was carried out to relate pVTG I and pVTG I1 to cellular and secreted VTG I and VTG 11. After in vitro labeling with ["Hlleucine or ["Hlserine, tissue slices and medium were combined and solubilized directly in SDS sample buffer, and proteins were resolved on an SDS-5% polyacrylamide gel. The gel region including the pVTG and VTG bands was excised and treated with V8 protease (see "Experimental Procedures"), and the resultant peptides were analyzed in a second dimension SDS-10% polyacrylamide gel. As shown in to secreted VTG I and VTG 11, respectively.
When processed in the oocyte, vitellogenin gives rise to two groups of proteins the amino acid compositions of which reflect the nonuniform distribution of particular amino acids within the vitellogenin polypeptide. Vitellogenin regions corresponding to the yolk phosvitins are very rich in phosphoserine and poor in leucine (1,8,32) while regions corresponding to yolk lipovitellins contain both serine and leucine with leucine in slight predominance (8, 33). V8 protease cleavage products of vitellogenins or vitellogenin precursors should also reflect the nonuniform distribution of these amino acids. Cleavage of plasma VTG 11, in fact, yields a peptide distribution such that most of the phosphopeptides run in the M , = 80,000-100,000 range and most of the nonphosphorylated peptides run at lower molecular weights on SDS-10% polyacrylamide gels (7). Region A in Fig. 5 is the molecular weight range which should contain many of the serine-rich, leucinepoor phosphopeptides of VTG 11. Note that the region A peptides of pVTG I1 do not have corresponding peptides of identical mobility in the VTG I1 digest. In contrast, the region A VTG I1 peptides appear to run somewhat slower than the http://www.jbc.org/ Downloaded from region A peptides of pVTG 11. Since the vitellogenin phosphates reduce the electrophoretic mobility of vitellogenin (vide infra), slower mobilities of the VTG I1 region A peptides may be anticipated if these peptides, in fact, arise from the serine-rich vitellogenin region. To determine whether this is the case, V8 protease maps of pVTG I1 and VTG I1 were compared after hepatocytes were labeled with either ["HI serine (Fig. 6A) or [:'H]leucine (Fig. 6B). The labeling intensities of the region A peptides in relation to the peptides outside region A confm that region A contains serine-rich peptides. Furthermore, two additional serine-rich pVTG I1 peptides outside region A (indicated by stars in Fig. 6 A ) appear to contain little or no leucine (compare to Fig. 6B). Note that, in general, the serine-rich peptides of pVTG I1 do not have corresponding peptides of identical mobilities in the VTG I1 digest. In contrast, the leucine-rich peptides of pVTG I1 (Figs. 5 and 6B) outside region A have corresponding peptides of identical mobilities in the VTG I1 digest; three such peptides are indicated by arrows in Fig. 6. The lack of correspondence in the mobilities of the serine-rich peptides of pVTG I1 and VTG I1 may reflect differences in the protease cleavage sites as a result of the phosphates in VTG 11. Alternatively, the phosphates of the VTG I1 peptides may alter the electrophoretic mobilities as occurs with intact vitellogenin. In either case, these data ( Fig. 6) illustrate the nonuniform distribution of serine and leucine in the pVTG I1 and VTG I1 polypeptides. In addition, the serine-rich, but not the leucinerich, peptides of VTG I1 appear to be modified so as to have altered electrophoretic mobilities in comparison to the serinerich pVTG I1 peptides. This feature is consistent with the clustering of phosphates in the serine-rich region of VTG I1 as well as the influence of the phosphates upon the electrophoretic mobility of intact vitellogenin (vide infra). Precursor Character of pVTG I and pVTG 11-The precursor character of the pVTG was confmed with both pulselabeling and pulse-chase procedures. Fig. 7 illustrates the kinetics of ['Hlserine incorporation into cellular vitellogenins when hepatocytes were analyzed after 15,30,60, and 120 min of labeling. With a 3-day film exposure, no labeling is detected at 15 min (Fig. 7, lane I ) ; only pVTG I1 is seen at 30 min (lane 2), pVTG I and pVTG I1 are seen at 60 min (lane 3), and pVTG I, pVTG 11, and VTG I1 are seen at 120 min (lane 4). With an 11-week film exposure, the 15-min sample shows both pVTG I and pVTG I1 but not VTG I1 (lane 5). In contrast, the 120-min sample of similar intensity (lane 4) shows a distinct VTG I1 band as does the 30-min sample after an 11-week exposure (lane 6). In other experiments, we were unable to detect VTG I and VTG I1 among secreted proteins with labeling times of less than 30-35 min (data not shown). It appears, therefore, that ['Hlserine is incorporated into the pVTG species prior to its appearance in either cellular or secreted VTG I and VTG 11.

A pVTG I pVTG II VTG I VTG II pVTG I pVTG II
As noted above, only 2-6% of cellular vitellogenin is present as the VTG species with labeling times up to several hours. This result suggests that secreted VTG must arise directly from pVTG or via short lived cellular VTG intermediates. To determine the quantitative relationship between cellular pVTG and the secreted VTG species, hepatocytes were labeled with ['Hlleucine for 40 min and chased with unlabeled leucine for 45 and 90 min. Cellular and secreted proteins were then analyzed by electrophoresis, and the radioactivity in the vitellogenin bands was determined by scintillation spectrometry after combustion of the gel slices to ensure quantitative recovery of radioactivity. No distinction was made between cellular pVTG and VTG species since the cellular VTG are such a small fraction of the total cellular vitellogenin. As shown in Table I, 30% of the cellular vitellogenin radioactivity was secreted during the 90-min chase. The extracellular vitellogenin recovered at each time point closely approximates that chased from the hepatocytes and can only be accounted for by the radioactivity initially present in the pVTG species. These data confm that the secreted vitellogenins arise from cellular pVTG I and pVTG 11. and VTG species suggest that SDS-polyacrylamide gel electrophoresis is an inappropriate method to estimate the molecular weights of the vitellogenin polypeptides. We have, therefore, explored the influence of the phosphates upon the mobility of plasma vitellogenin and estimated the molecular weights of the various vitellogenin species by an independent method. Fig. 8 shows the results of an experiment in which a plasma vitellogenin sample enriched in VTG I1 was progressively dephosphorylated. The electrophoretic profile of the untreated sample (Fig. 8, left, lune 1 ) is shown in comparison to the mobilities of the VTG and pVTG species. As VTG I1 was dephosphorylated (lunes 2-10), the band broadened and its mobility increased until the leading edge had the mobility of pVTG 11. This mobility shift closely parallels dephosphorylation as judged by the release of vitellogenin phosphorus (Fig. 8, right). The dephosphorylation and mobility shift were maximal by 2 h and remained unchanged for a further 3 h (lunes 7-10). In other experiments employing successive treatments with alkaline phosphatase, the mobility of the leading edge of the VTG I1 band did not increase further, but the band became much sharper. Similar results have been obtained with plasma VTG I in which case the dephosphorylated protein had the same mobility as pVTG I (data not shown). In addition, note that two very faint bands (Fig. 8, left) which run slightly ahead of the VTG I1 band also shift in mobility during treatment with alkaline phosphatase. We have determined that these proteins are also heavily phosphorylated as judged by the incorporation of radiophosphate during in vivo labeling experiments (data not shown). These results indicate that the phosphates markedly influence the mobility of vitellogenin in SDS-polyacrylamide gels.

Molecular Weight Estimations of the Vitellogenins-The
Independent estimates of vitellogenin molecular weight were obtained by gel chromatography of reduced and alkylated proteins using 7 M guanidine-HC1 as solvent. Fig. 9 shows a calibration curve which employed P-galactosidase, myosin, and human apolipoprotein B (22) as high molecular weight standards. Plasma VTG I1 and dephosphorylated VTG I1 showed elution volumes corresponding to the M , = 170,000 to 190,000 range. When the mixture of VTG I and VTG I1 was analyzed, both proteins eluted together in the same molecular weight range. Hepatocyte proteins labeled with ['Hlserine were analyzed in a similar fashion. Subsequent to chromatography, column fractions were analyzed by SDSd% polyacryl-

TIME (hours)
Coomassie blue. The gel was then prepared for fluorography to visualize the cellular ["Hlvitellogenin which was run in an adjacent gel lane. The mobilities of the pVTG and VTG species are indicated. The dephosphorylation of vitellogenin was monitored by measuring phosphorus in the trichloroacetic acid-soluble fraction as a function of treatment time with alkaline phosphatase (right). amide gel electrophoresis and fluorography. The results from several experiments showed that pVTG I, pVTG 11, VTG I, and VTG I1 elute at the same volume with the peak fractions corresponding to the M , = 170,000 to 184,OOO range. The chromatographic behaviors of these species as well as the plasma vitellogenins were reproducible and clearly indicative of molecular weights less than myosin (Fig. 9). Znfluence of Proteolysis-Three types of experiments were carried out to eliminate the remote possibility that the observed differences between the pVTG and VTG species were due to proteolysis. First, the pVTG species were the predom-inant cellular vitellogenins seen whether hepatocytes or tissue slices were boiled directly in SDS electrophoresis sample buffer (as in Fig. 1, lane I ) or homogenized to prepare a cell extract for immunochemical procedures (as in Fig. 2, lanes 1  and 3). This was also the case when cells or tissue slices were homogenized directly in trichloroacetic acid. Second, the predominance of the cellular pVTG species was unaltered by a variety of protease inhibitors (see "Experimental Procedures'') whether the inhibitors were included in homogenization buffer for cell extract preparation or in electrophoresis sample buffer used for direct solubilization of hepatocytes. Third, mixing experiments were carried out in which medium containing ["Hlserine-labeled VTG I and VTG I1 was included in buffer used to homogenize unlabeled hepatocytes. No breakdown of the 'H-vitellogenins was seen upon subsequent electrophoresis (data not shown). These data eliminate the remote possibility that the pVTG species arose through proteolytic cleavage of VTG I and VTG 11.

DISCUSSION
The experiments reported here identify and characterize discrete intracellular vitellogenins, pVTG I and pVTG 11, and establish that they give rise to secreted vitellogenins. The pVTG are identified as vitellogenins by specific reactivity with antibodies to the plasma vitellogenins (Figs. 2-4). In addition, pVTG I and pVTG I1 are made by hepatocytes from laying hens and estrogen-stimulated roosters but not by hepatocytes from unstimulated roosters. Furthermore, pVTG I and pVTG I1 are clearly intracellular species which do not appear in proteins secreted by hepatocytes (Fig. 1, lane 2) or in plasma of estrogen-stimulated roosters (7). The precursor character of the pVTG is established by three findings. First, pVTG I and pVTG I1 contain little or no phosphorus (Fig. 2), indicating that they are intermediates in the biosynthesis of the mature phosphorylated vitellogenins. Second, precursor amino acid is incorporated into pVTG I and pVTG I1 prior to its appearance in cellular and secreted vitellogenins when hepatocytes are continuously incubated with ["Hlserine ( Fig.  7). Third, quantitative analysis of vitellogenin secretion during a pulse-chase experiment indicates that only the cellular pVTG can account for the secreted vitellogenins ( Table I).
The identification of pVTG I1 as the specific precursor to secreted VTG I1 is strongly supported by both immunological and peptide map comparisons. Anti-VTG I1 selectively precipitates pVTG I1 as well as cellular and secreted VTG I1 but is far less effective in precipitating pVTG I. This is in agreement with previous work which showed poor reactivity between purified plasma VTG I and anti-VTG I1 (7). Partial proteolysis mapping of [3H]leucine and ["Hlserine-labeled vitellogenins shows extensive similarity between pVTG I1 and VTG I1 (Figs. 5 and 6 ) and illustrates the nonuniform distribution of serine and leucine in the pVTG 11 and VTG I1 polypeptides. Although the data are less extensive, similar structural relatedness is evident in the proteolysis products of pVTG I and VTG I (Fig. 5). The dissimilarity of the pVTG I and pVTG I1 digestion products as well as the poor reactivity of pVTG I with anti-VTG I1 suggests that pVTG I is specifically related to secreted VTG I but not VTG 11.
The large differences in the mobilities of the pVTG and VTG species as well as the increased mobility of dephosphorylated plasma vitellogenin indicate that SDS-polyacrylamide gel electrophoresis does not accurately estimate the molecular weights of these heavily phosphorylated proteins. Estimates of vitellogenin molecular weight by this method in numerous laboratories (1, 7, 8, 34, 35) yield values of 235,000 to 260,000. In contrast, the mobilities of pVTG I and pVTG I1 correspond to M , = 200,000 and 190,000. A more accurate estimate by gel chromatography in 7 M guanidine-HC1 confiims that vitellogenin molecular weight is much less than estimated by SDSpolyacrylamide gel electrophoresis. The various vitellogenin species are not resolved by this chromatographic method, but all species showed M , between 170,000 and 190,000 with a mean of approximately 180,000 (Fig. 9). The molecular weights of the pVTG species as estimated by electrophoresis exceed this value by only about lo%, a difference which might be explained by the carbohydrate content of these proteins. In studies to be reported elsewhere, we have determined that the plasma vitellogenins contain approximately 1% carbohydrate.' It is unlikely that this low carbohydrate content would significantly influence estimations by chromatography in guanidine-HC1 (36), although some retardation in mobility might be expected in SDS-polyacrylamide gels (37). Indeed, we have noted that the pVTG species made in the presence of the glycosylation inhibitor tunicamycin (30, 31) have slightly greater gel mobilities yielding M, = 190,000 and 180,000 for nonglycosylated pVTG I and pVTG 11, respectively.2 These findings further confirm that the vitellogenin polypeptides have M , = 180,000.
Cell-free translations of intact, highly purified vitellogenin mRNA have also shown that the cell-free product has a greater mobility on SDS-polyacrylamide gels than plasma vitellogenin. In the cases of avian (38,39) and amphibian (40) vitellogenins, the greater mobilities of the cell-free products were attributed to the lack of post-translational modifications such as phosphorylation and glycosylation. Our results on the markedly different electrophoretic mobilities of the nonphosphorylated and phosphorylated vitellogenins support these interpretations. The report by Gordon et al. (41) that nonphosphorylated vitellogenin synthesized in a cell-free system has the same electrophoretic mobility as plasma vitellogenin is likely due to the use of an acrylamide concentration which is too high to permit resolution of proteins in this molecular weight range. We find, for example, that the pVTG and VTG bands are not resolved in SDS-7.5% polyacrylamide gels, particularly when the vitellogenin is run into the gel for only a short distance. Even with an SDS-5% polyacrylamide gel, the pVTG and VTG species are well resolved only with extended electrophoresis times such that proteins of M , = 100,000 have migrated approximately 8-10 cm (Fig. 1).
A polypeptide weight of 180,000 has significant implications for the analysis of vitellogenin processing in the oocyte. After secretion from the liver and uptake into the developing oocyte, the avian vitellogenins give rise to yolk proteins designated a-lipovitellin, P-lipovitellin, and phosvitins. Both lipovitellins contain two or more polypeptide chains which arise from vitellogenin regions containing little phosphorus while the heavily phosphorylated vitellogenin regions give rise to at least two phosvitins (1, 8, 32, 42). The organization of the various yolk polypeptides within the vitellogenin polypeptides is not well understood. We have previously presented evidence that VTG I1 is the precursor to a single yolk phosvitin as well as the 125,000-dalton polypeptide of ,8-lipovitellin (7). The plipovitellin monomeric mass of 200,000 includes 160,000 daltons accounted for by protein (43, 44) which can be resolved into equimolar quantities of the 125,000-and 30,000-dalton polypeptides (1, 8, 42). A VTG I1 polypeptide of 180,000, therefore, could include both P-lipovitellin polypeptides with approximately 25,000 daltons remaining to accommodate the phosvitin region of the molecule. After correction for phosphoms (32, 45) and carbohydrate (46), the polypeptides of either the M , = 34,000 or 28,000 phosvitin (32), but not both phosvitins, could be accommodated. Comparison of phosphorus contents also c o n f i i s t h a t only one phosvitin can be present in VTG 11. With a phosphorus content of 2% of the polypeptide weight (7), VTG I1 contains approximately 116 mol of phosphorus. Within the error of these measurements, this is sufficient to account for the 117-127 mol of phosphorus present in P-lipovitellin (47) plus either phosvitin (32) but far too little to account for the 217 mol of phosphorus in Plipovitellin plus both phosvitins. Similar calculations indicate that VTG I also contains only one phosvitin. These data, therefore, support the view (7) that each avian vitellogenin polypeptide contains a phosphoserine-rich region which gives rise to one or the other of the two yolk phosvitins.
The experiments described here are of particular interest with regard to the biosynthesis and post-translational modifications of the vitellogenins. These experiments resolve, identify, and characterize individual intracellular vitellogenins and biosynthetic intermediates which appear to represent discrete stages of post-translational processing. Such analyses are necessary in order to understand the biosynthesis of these complex proteins, but have not been reported previously. The pVTG are glycosylated vitellogenins containing little or no phosphorus, indicating that some glycosylation reactions occur prior to phosphorylation. While it is not clear whether glycosylation is initiated during or after polypeptide synthesis, it is clear that phosphorylation occurs after completion of the vitellogenin polypeptide and after the addition of some carbohydrate. The question of whether carbohydrate plays a role in targeting which serines are phosphorylated can be raised since yolk phosvitin contains a glycopeptide sequence that includes a block of eight phosphoserines (46). With the caveat that glycosylation may be initiated during polypeptide synthesis, the stages of vitellogenin biosynthesis may be ordered as: polypeptide synthesis + glycosylation + phosphorylation -+ secretion.
The phosphorylation of each vitellogenin requires the addition of over 100 phosphates/polypeptide chain. The gradual mobility shift accompanying vitellogenin dephosphorylation (Fig. 8 Figs. 1 and 2) or 32P, (Fig. 4, lane 2). Even after labeling with 32Pi for times as brief as 1 min, significant amounts of such intermediates are not seen.5 The absence of such intermediates suggests that once the phosphorylation of the vitellogenin polypeptide begins, it is rapidly completed. One possibility consistent with this observation is for the vitellogenin to enter or pass through a cell compartment that contains a high kinase concentration and represents only a small fraction of the intracellular transport pathway. Additionally, if vitellogenin is vectorially transported (48), the subcellular site of phosphorylation must be very near the terminal stages of the pathway since the majority of the intracellular vitellogenin is present as nonphosphorylated pVTG I and pVTG 11. It appears that when phosphorylation is completed, the vitellogenins are rapidly secreted from the hepatocyte. Consistent with this interpretation is the observation that [32P]vitellogenin appears in plasma with a lag of only several minutes after "*Pi administration to roosters (49). Schirm et al. (50) have also shown that ["P]vitellogenin accumulates in plasma after inhibition of protein synthesis. These data also suggest that vitellogenin phosphorylation occurs immediately prior to secretion. Although the subcellular site and mechanism of vitellogenin phosphorylation are not known, the ability to monitor the intracellular pVTG and VTG species should greatly facilitate such studies. Furthers. Y. Wang, and D. L. Williams, unpublished observation. more, the pVTG species are an appropriate substrate for the analysis of vitellogenin phosphorylation in cell-free systems. (51) have recently presented evidence that in Xenopus laeuis the multiple yolk lipovitellin polypeptides and the various phosphopeptides are derived from multiple forms of plasma vitellogenin. There is thus agreement that the heterogeneity of yolk polypeptides in both species is due to heterogeneity in the vitellogenin precursors.