Glycoproteins released into the culture medium of differentiating murine neuroblastoma cells.

C-1300 murine neuroblastoma cells release glycoproteins into the culture medium. The process was studied by prelabeling spinner cultures for 12 to 60 hours with [3H]glucosamine. Then, the medium was removed and replaced with fresh medium lacking radioactive isotope. Soluble material released into the medium during the subsequent 2-hour incubation was collected by trichloroacetic acid precipitation. The released proteins were then separated by discontinuous polyacrylamide gel electrophoresis in buffers containing sodium dodecyl sulfate. The electrophoretograms of glycoproteins obtained from cultures labeled for different lengths of time were very similar; three major radioactive regions centered about molecular weights 87,000, 66,000, and 55,000 were present. When spinner cells were transferred to monolayer culture in the presence of N6,O2' dibutyryl adenosine 3':5'-monophosphate (Bt2cAMP), differentiation (extension of neurites twice the diameter of the perikaryon) was observed. Monolayer cultures grown in the presence of Bt2cAMP and [3H]glucosamine for 12 hours released glycoproteins which gave a gel electrophoresis pattern similar to that obtained using spinner cultures. However, after 60 hours in the presence of Bt2cAMP and [3H]glucosamine, the released radioactive material consisted almost exclusively of glycoproteins of the 66,000 molecular weight class. Similar results were obtained if [3H]fucose was substituted for [3H]glucosamine, or if bromodeoxyuridine (which also induced differentiation) was substituted for Bt2cAMP. Similar experiments using radioactive amino acids were conducted with both spinner and monolayer cultures. Much of the released radioactive material was contained in the same three molecular weight classes as the glycoproteins released by spinner cells prelabeled with [3H]glucosamine, and this pattern did not vary with length of labeling period or type of culture. These results may imply that the glycosylation of released proteins is influenced by agents which can induce differentiation. The origin of this released material is discussed. [3H]Glucosamine-labeled glycoproteins of the molecular weight class centered about 55,000 (discussed above) were isolated by preparative gel electrophoresis. They co-migrated with authentic mouse brain microtubular protein as two closely spaced bands on a number of different electrophoretic systems. This protein fraction was also characterized as complexing with a monospecific antitubulin antibody.

Glycoproteins Released into the Culture Medium of Differentiating Murine Neuroblastoma Cells* (Received for publication, August 19, 1975) ROBERT TRUDING,$ MICHAEL L. SHELANSKI C-1300 murine neuroblastoma cells release glycoproteins into the culture medium. The process was studied by prelabeling spinner cultures for 12 to 60 hours with [BH]glucosamine.
Then, the medium was removed and replaced with fresh medium lacking radioactive isotope. Soluble material released into the medium during the subsequent 2-hour incubation was collected by trichloroacetic acid precipitation. The released proteins were then separated by discontinuous polyacrylamide gel electrophoresis in buffers containing sodium dodecyl sulfate. The electrophoretograms of glycoproteins obtained from cultures labeled for different lengths of time were very similar; three major radioactive regions centered about molecular weights 87,000, 66,000, and 55,000 were present.
When spinner cells were transferred to monolayer culture in the presence of iVe,OZ' dibutyryl adenosine 3':5'-monophosphate (Bt,cAMP), differentiation (extension of neurites twice the diameter of the perikaryon) was observed. Monolayer cultures grown in the presence of Bt,cAMP and [*H]glucosamine for 12 hours released glycoproteins which gave a gel electrophoresis pattern similar to that obtained using spinner cultures. However, after 60 hours in the presence of Bt ,cAMP and [BH]glucosamine, the released radioactive material consisted almost exclusively of glycoproteins of the 66,000 molecular weight class. Similar results were obtained if [aH]fucose was substituted for ['Hlglucosamine, or if bromodeoxyuridine (which also induced differentiation) was substituted for Bt*cAMP. Similar experiments using radioactive amino acids were conducted with both spinner and monolayer cultures. Much of the released radioactive material was contained in the same three molecular weight classes as the glycoproteins released by spinner cells prelabeled with [3H]glucosamine, and this pattern did not vary with length of labeling period or type of culture. These results may imply that the glycosylation of released proteins is influenced by agents which can induce differentiation.
The origin of this released material is discussed.
[BH]Glucosamine-labeled glycoproteins of the molecular weight class centered about 55,000 (discussed above) were isolated by preparative gel electrophoresis. They co-migrated with authentic mouse brain microtubular protein as two closely spaced bands on a number of different electrophoretic systems. This protein fraction was also characterized as complexing with a monospecific antitubulin antibody.  (4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15). This released or "shed" material could be a product of the normal metabolic turnover of surface membrane glycoproteins, or possibly be a specifically released cytoplasmic material involved in the regulation of cell to cell interactions.

Glycoproteins
As an extension of our interest in membranes of the nervous system, we utilized such an approach to study the differentiation of cultured neuroblastoma cells.
The were isolated from spinner and monolayer cells as described previously (21). Analysis of sugars in the glycoprotein fraction involved limited acid hydrolysis followed by chromatography on an amino acid analyzer (26). Authentic microtubular protein, isolated from mouse brain 1 hour after intracerebral injection of ['Clleucine (27). was used as a standard in some experiments.
In the immunoprecipitation experiments, carrier tuhulin was prepared by the method of Shelanski et al. (28) and monospecific antibody against bovine brain tubulin was prepared by the procedure of Fuller et al. (29).

RESULTS
Cell Growth-Details regarding the growth of this cell line under our conditions have been published (21). In the present studies, cells in the logarithmic phase of growth had a doubling time of 18 hours. Monolayer cells in the presence of Bt,cAMP grew to a confluent monolayer with a doubling time of 30 hours during the logarithmic growth phase. Within a few hours of transfer, most of the cells had flattened, and many of the cells had started to extend neurites. By 24 hours, a majority of the cells had extended neurites; and by 60 hours, about 75% of the cells were differentiated (Le. had extended neurites at least twice as long as the perikaryon diameter). Monolayer cultures treated with bromodeoxyuridine were similarly differentiated.
Treatment with sodium butyrate caused the cells to adhere to the substratrum and slowed the growth rate but did not cause morphological differentiation. Spinner cultures, regardless of drug treatment, and untreated monolayer cultures retained an undifferentiated morphology.

Sugar
Label-Monolayer and spinner cells were incubated for 12, 36, or 60 hours in the presence of radioactive glucosamine. The material released during the subsequent 2-hour incubation, in fresh serum and label-free medium, was isolated and subjected to polyacrylamide gel electrophoresis.
In a typical experiment where the monolayer cells were prelabeled with 10 &i of ['Hlglucosamine for 36 hours, the distribution of trichloroacetic acid-insoluble radioactivity was: surface membrane, 45,000 cpm; soluble cytoplasmic proteins, 36,000 cpm; and released materials, 22,000 cpm. Correcting for yield (assuming 65% for surface membrane (21), 95% for cytoplasmic proteins, the extent of cell breakage W), and 100% for released protein, the distribution of trichloroacetic acid-insoluble radioactivity was: surface membranes, 38% soluble cytoplasmic protein, 20%; released material, 12% with the remaining radioactivity being in the nuclear and debris fractions. Fig. 1 shows the electrophoretic analyses of glucosamine-labeled proteins released into the culture medium of Bt,cAMP-treated cells during differentiation. At the 12-and 36-hour time points, the released proteins were grouped into three major molecular weight regions centered about apparent molecular weight of about a, 87,000; b, 66,000; and c, 55,000; respectively (Fig. 1A). Each of these classes of proteins was defined by its position on the gel: Region a (gel slices 10 to 15 + one slice); Region b (gel slices 18 to 23 += one slice); or Region c (gel slices 24 to 31 + one slice). Each class was probably composed of more than one glycoprotein; Region c was usually resolved into two bands. There was some labeled material at the origin (gel slice l), as well as a small peak of radioactivity (gel slices 54 to 56) just before the dye front (gel slice 57 f two slices). The gel pattern was not affected by extraction of the protein with chloroform/methanol, 2/l, v/v prior to electrophoresis.
The relative distribution of radioactivity between the major glycoprotein classes changed with increasing time of culture in the presence of Bt,cAMP.
In a series of eight time course experiments, the time of onset of this preferential appearance of Region b glycoprotein was variable. However, in each case, by 60 hours, primarily Region b glycoprotein was being released into the medium (Fig. 1C). There were no further marked changes in the label pattern in the subsequent 24-hour period (Fig. 24). If fucose was used to label the glycoproteins, the same development pattern was observed, resulting in glycosylation of primarily one molecular weight class (Region b) of released glycoproteins after 60 hours in Bt,cAMP (Fig. 20). Another agent which induced morphological differentiation, bromodeoxyuridine, also induced a time-dependent increase in the proportion of released label present as Region b glycoprotein (Fig. 2B). If a monolayer culture was treated with Bt,cAMP for 48 hours, and then radioactive glucosamine was added for 12 hours, the gel pattern of the protein released in a subsequent 2-hour incubation (in serum-free medium) was the same as that of Fig.  1c.
The change in pattern of released glycoprotein in Bt,cAMPtreated cells was not primarily a function of cell density. Monolayer cultures were plated as several different concentrations of cells, so that following a go-hour prelabeling period with ['Hlglucosamine in the presence of Bt,cAMP, different subconfluent densities were reached on the flasks. The gel patterns in each case were similar to those of Fig. 1C.
Monolayer cultures not treated with any drug, or treated with sodium butyrate, did not differentiate, and the electrophoretogram of released glycoproteins did not show an increase in the proportion of radioactivity in Region b. The lack of response to sodium butyrate (which slows growth) demonstrated that the specific change in the pattern of released glycoproteins induced by Bt,cAMP or bromodeoxyuridine was not due only to the slower growth rate observed in the presence of these drugs. Neither the presence of serum during the 2hour release period nor extensive extra washes with serumfree Minimum Essential Medium, serum-containing medium, or 0.001 M EDTA in Ca*+-and Mg+-free medium affected the electrophoretic pattern or subcellular distribution of radioactivity.
The results with the differentiating monolayer cultures were in contrast to the result obtained in spinner culture in the absence of Bt,cAMP (Fig. 3). In spinner cultures labeled for 12, 36, or 60 hours, the same three major molecular weight groupings in Regions a, b, and c were recognizable in the polyacrylamide gel pattern. The ratio of label incorporated in the three major glycoprotein regions did not change markedly with time, although the relative increase in background labeling with time caused these groupings to appear less prominent.
The observed specificity in release of glycoproteins ( Fig.  1C) was not dependent upon morphological differentiation. If spinner cells were treated with 1 x 10ms M Bt,cAMP ( Fig.  2C) or 1 x 10m5 M bromodeoxyuridine, the pattern of glycoproteins released in a 2-hour period after 60 hours of prelabeling with [3H]glucosamine resembled the results obtained using differentiated monolayer cells (Fig. 1C) much more than the results obtained using untreated spinner culture cells (Fig. 3C).
The differences between spinner and monolayer cultures in molecular weight distribution of radioactivity incorporated into released proteins was observable only when a sugar label was utilized.
When [3H]leucine was used to prelabel spinner culture cells and Bt,cAMP-treated monolayer cells from 12 to 60 hours, no marked differences were observed in the electrophoretograms of glycoproteins released into the culture medium in the subsequent 2-hour period. Fig. 4 illustrates the results from the 60-hour time points (the difference in position of the leading peak at slices 53 and 55 was not reproducible and was due to gel variation).
Similar resuits were obtained when a mixture of SH-amino-acid was used instead of leucine alone. The electrophoretic pattern of the leucine-labeled released proteins consisted largely of the same three molecular weight classes of proteins obtained from glucosamine-labeled cells (Fig. 5). One observable difference was the presence of another peak, gel slice 36 of Fig. 5, in the leucine-labeled proteins.
The experiments represented in Figs. 1 to 4 were repeated with 2.5% calf serum and 2.5% fetal calf serum present during the 2-hour release period. The resulting electrophoretic patterns were very similar to those obtained when no serum was present during the release period.
The release of material into the culture medium during the 2-hour release period was not due to nonspecific leakage of cell-soluble proteins through the surface membrane. Fig. 6 illustrates directly the more complex pattern of soluble cytoplasmic glycoproteins relative to released glycoproteins. The difference is much more dramatic after 60 hours of differentiation.
Evidence that the radioactive sugar was on a glycoprotein molecule was obtained by demonstrating that following di- gestion with the proteolytic enzyme Pronase (0.7 mg/ml for 24 hours at 37"), material originally migrating in Regions a, b, and c formed a smear in the low molecular weight regions of the polyacrylamide gels. Following limited acid hydrolysis of the [3H]glucosamine-labeled released glycoprotein, more than 95% of the radioactivity was eluted with glucosamine or galactosamine from an amino acid analyzer. Also, substituting ["Clglucosamine for [BH]glucosamine did not change the gel electrophoretic patterns. The label was not on a mucopolysa& charide, since, when cells were grown in the presence of B"SO, and a delipidated membrane protein fraction prepared and subjected to electrophoresis, no radioactivity migrated with Regions a, b, or c.

Relationships between Released Glycoproteins and Surface Membranes
Glycoproteins-The appearance of glycoproteins in the culture medium may be related to metabolic turnover of the surface membrane. Fig. 7 (-) and ["Clleucine (---). The released material was isolated separately, solubilized, aliquots of each preparation were combined, and then subjected to electrophoresis. One of three similar experiments with similar results.  from Bt,cAMP-treated monolayer cultures labeled for 36 hours. The released material and surface membranes were isolated separately, solubilized, aliquots of each preparation were combined, and then subjected to electrophoresis. One of two similar experiments with similar results.
cultures were prelabeled with [SH]glucosamine for 36 hours and the released glycoprotein was then subjected to electrophoresis with ["Clleucine-labeled mouse brain tubulin, the authentic tubulin co-migrated with the two peaks of Region c. The Region c glycoprotein was further characterized by isolation on a small preparative gel (System b) and tested for co-migration with authentic tublin by three different gel systems (see "Experimental Procedure"). The released matkrial and authentic tubulin were identical by these criteria (Figs. 8 and 9 illustrate the results from two such gels). Similar results were obtained with released glycoproteins collected from cultures incubated in the presence of 5% serum (2.5% calf serum and 2.5% fetal calf serum) during the 2-hour release period.
To test further for the possibility that this double peak in Region c was composed of tubulin, carrier tubulin (28) was added to the preparation and then precipitated with monospecific antitubulin (29). This gave a quantitative precipitation of the carrier and precipitated between 8 and 10% of the total counts in [BH]glucosamine-labeled preparations. Electrophoresis of the precipitate, after dissociation in sodium dodecyl sulfate and urea (System c), showed over 90% of the radioactivity co-migrating with tubulin. Controls done with anticytofilament antiserum (cytofilament = 9 nm of filament) resulted in the precipitation of less than 0.5% of the counts, and these were widely dispersed on gel electrophoresis.

DISCUSSION
It is well known that many glycoproteins are located on the exterior surface of the surface membrane, and that limited proteolytic digestion may release glycopeptides derived from these glycoproteins (30)(31)(32)(33). These glycopeptides may be related to the glycoproteins released during normal growth of cells (12). Viral transformation of normal cells results in changes in the chromatographic properties of the surface peptides released by treatment with proteolytic enzymes (34). It has been suggested that transformed cells differ from normal cells with respect to proteolytic activity at the cell surface (35). It is also relevant to note that antibodies prepared against cell surface material released by HeLa cells inhibited division of these cells (36).
The radioactive glycoproteins found in the culture medium of neuroblastoma cells grown in the presence of radioactive glycoprotein precursors might arise from (a) labeling of serum  (--) from spinner cultures (isolated on a preparative gel) and "C-labeled tubulin (---). The discontinuous polyacrylamide gel contained both sodium dodecyl sulfate and urea (System c). One of three experiments with similar results. from spinner cultures (isolated on a preparative gel) and "C-labeled tubulin (--). The isoelectric focusing gel contained a pH 3 to 10 gradient (System d).
proteins by cell surface enzymes; (b) leakage of intracellular material due to cell damage or death; (c) products of surface membrane metabolism; or (d) specific secretion of certain intracellular proteins. The first possibility is unlikely, since all of the peaks were labeled with leucine, and the addition of amino acids at the cell surface has not previously been reported. Also, if serum or bovine serum albumin was added during the 2-hour release period, neither the amount, nor gel electrophoretic pattern of the released glucosamine-labeled material was changed. The second possibility is unlikely because of the simplicity of the gel electrophoretic pattern of released glycoproteins (Fig. 6). Also, note the marked effect of Bt,cAMP which should not influence a nonspecific mechanism (Fig. 2C).
The likelihood that released proteins represent material "shed" from t.he surface membrane because of normal metabolic turnover has been suggested by several investigators (12,37,38). If this is the case, it is clear that not all surface membrane proteins are being shed intact at the same rate. Only the proteins of Region a migrate with major membrane components (Fig. 7). At least one protein in this region is on the outer surface of the membrane as defined by its availability for lactoperoxidase-catalyzed iodination (21). We also have evidence indicating that peptides originating from this iodinated protein may subsequently be released into the culture medium.l Relative to Region a, proteins in the molecular weight classes of Regions h and c are sparsely represented in surface membranes. It is possible that the released proteins in these molecular weight classes are either rapidly turning over proteins which are minor surface components, or that these lower molecular weight peptides are released from larger proteins by some type of proteolytic action. It is likely that the fourth possibility listed, specific secretion of certain soluble cytoplasmic components, also accounts for some of the released material. Region c-released glycoprotein corresponds in molecular weight to a double peak observed in the soluble cytoplasmic proteins (Fig. 6), and is not present as a major component of the surface membranes.
It has previously been shown that treatment of Chinese hamster ovary cells with Bt,cAMP can cause changes in glycosylation of the released proteins (15). We have shown that Bt,cAMP or bromodeoxyuridine induces changes in the pattern of molecular weight distribution of released glycoproteins of neuroblastoma cells. This may be due to relative differences in glycosylation of different proteins under different culture conditions. Another possible interpretation of the data is that there is a change in the relative rates of release of the three classes of glycoproteins under different conditions. This hypothesis is less likely because the distribution of leucine-labeled released proteins does not change with time; although if glycosylated proteins comprised only a small percentage of the total released proteins, changes specific to the glycoproteins might not be observed with leucine as a label. In monolayer culture, application of Bt,cAMP or bromodeoxyuridine caused a marked morphological differentiation along with induction of characteristic neuronal enzymes (18)(19)(20). Under certain conditions where the cells do not differentiate morphologically, these agents can still cause an increase in certain enzyme activities, i.e. biochemical differentiation (21,39). The increase in the proportion of radioactivity released as Region b glycoprotein by spinner culture cells (Fig. 2C) upon application of Bt,cAMP is, therefore, not incompatible with this glycoprotein being associated with morphological differentiation.
Other relevant investigations of surface proteins in neuroblastoma cells include our previous work (21), showing that preferential appearance of a glycoprotein of molecular weight 105,000 in the soluble cytoplasm and surface membrane of neuroblastoma cells is induced under the same conditions which resulted in an increase in the proportion of radioactivity released as Region b glycoproteins into the release medium.
Brown (40)  9353 centered at molecular weight 50,000, similar to those reported in this investigation.
It was also noted that upon addition of [*H]leucine into the culture medium, the cells incorporated radioactivity into trichloroacetic acid-insoluble material almost immediately, but there was about an hour lag period before there was significant incorporation into extracellular protein.
Whatever the reason for shedding or secretion of glycoproteins in neuroblastoma cells, the data presented here demonstrate that this is a relatively specific phenomenon not due to random damage of cells. Furthermore, release of such glycoproteins, and possibly, their glycosylation, is under a highly specific physiological control.
Our results indicate that Region c glycoproteins released by the spinner culture co-migrate with tubulin upon electrophoresis.
The observed molecular weight and isoelectric point relationship between tubulin and Region c-released glycoproteins might be coincidental, but definitive proof awaits further investigations.
A preliminary experiment (in collaboration with Dr. H. Wisniewski of the Albert, Einstein College of Medicine) where vinblastine was added to the release medium revealed an ordered ultrastructure, similar to that seen when tubulin is precipitated by vinblastine (43). Tubulin has been found in neuroblastoma cells (44) and is involved in neurite formation (45). It has been reported that the concentration (per mg of protein) of intracellular tubulin is the same in differentiated and undifferentiated neuroblastoma cells (46,47). Our observation of a specific release of a presumptive glycosylated tubulin into the culture medium might then imply a difference in metabolism of tubulin in undifferentiated, as compared to differentiated, neuroblastoma cells. The incorporation of glucosamine into this material is noteworthy. Although it has been reported by several laboratories that tubulin is a protein containing about 1.3% carbohydrate (4%50), this assumption has been challenged (51). The study of the metabolism of this material in subcellular and extracellular fractions of neuroblastoma cells under various physiological conditions may provide information relevant to the specialized role of microtubules in neurons.
It is also pertinent to note that actin-like proteins are found in brain cells (52,53), cultured nerve cells (54), on the external surface of fibroblasts (55), and localized near the surface membrane of neuroblastoma cells (56,57). Some of these actin-like proteins co-migrate with muscle actin on sodium dodecyl sulfate-containing gels, and have been reported to have a molecular weight of about 45,000 (54,58), similar to the [9H]leucine-labeled protein centered about gel slice 33 of Fig. 4.