Changes in cell surface glycoproteins during Dictyostelium development analysed using monoclonal antibodies

One of the challenges that developmental biologists face is the understanding of how individual cells can differentiate and form an organized mass. In the developmental pathway of the cellular slime mould Dictyostelium discoideum, chemoattractants and morphogens are known to play a part in the formation of cell masses and the generation of cell type patterns within them (Williams et al. 1986). Cell surface components are also thought to play a role in the aggregation and patterning processes, but less is known about the nature and the mechanisms by which these materials affect the processes of development.

During our attempts to create monoclonal antibodies that recognize cell-type-specific plasma membrane ‘markers’, it was observed that a large number of the antibodies recognized oligosaccharide epitopes (as revealed by multiple bands on Western blots and by periodate sensitivity of the epitopes). The amounts of the carbohydrate epitopes present were observed to vary as development progressed. This observation is consistent with findings in other systems, such as cancer cell lines (Hakomori, 1985), sea urchin embryos (Lennarz, 1985), and mouse embryos (Draber, 1987), that cell surface oligosaccharides may play a role in development and differentiation.

The oligosaccharides found in Dictyostelium are predominantly N-linked high mannose species, similar to those found in other eukaryotes (Henderson, 1984; Freeze and Wolgast, 1986). The oligosaccharide precursor is built on a separate molecule (phosphoryl dolichol) and then transferred to the protein. The charge on the oligosaccharides is due to the presence of phosphate and sulphate residues (Gilkes et al. 1979).

Ivatt et al. (1981) suggested that the period when the cell pattern is established (12 h to 16 h) corresponds to a period when the cell surface oligosaccharides in Dictyostelium start to be processed in a different manner. It was suggested that the degree of sulphation and phosphorylation of the oligosaccharides was greatly reduced and the high mannose structure present during the early part of development was processed in such a way that ‘trimmed’ versions of it appeared. Other reports have established that different versions of oligosaccharide-processing enzymes appear in the lysosomes at 8 h of development (such as mannosidase (Moore et al. 1987), acid phosphatase (Loomis and Kuspa, 1984), N-acetylglucosaminidase (Knecht et al. 1985) and beta-galactosidase-2 (Moore et al. 1985)) and that a neutral membrane-bound alpha-mannosidase is induced after lOh of development. The observation that the oligosaccharide content of the cell is changing as development proceeds indicates that the oligosaccharide portion of the cell surface is worthy of further investigation, and an appreciation of the changes that are occurring in the cell surface oligosaccharides may lead to a better understanding of the role of oligosaccharides in the intercellular communication that is present during development.

D. discoideum strain JC4 was grown in association with Klebsiella aerogenes (strain OXF1) on SM nutrient agar (Sussman, 1966). Amoebae were prepared by growth as ‘lawns’ on SM agar under conditions permitting uniform clearing of the bacteria by the feeding amoebae. Amoebae were harvested from the bacterial plates in P buffer (17 mM-KH ₂/Na ₂H phosphate, pH6 ·1) and washed free of the bacteria by centrifugation at 190g for 2 min. After three such washes the cells were resuspended at 2 ×10 ⁸cellsmr ¹ in Lower Pad Solution and allowed to develop on Millipore filters according to the method of Sussman & Lovgren, 1965) until the desired state of development was reached.

Mouse myeloma line NSl/l-Ag4-l was grown in RPMI 1640 supplemented with 2mM-glutamine, 10% foetal calf serum, 50i.u. ml ⁻¹ penicillin and 50 μg ml ⁻¹ streptomycin.

Cell ghosts from cells at the slug (16 h) stage of development were prepared using the method of Sussman & Boschwitz (1975). This method was selected because it rapidly produced cell surface membranes that most resembled whole cell surfaces without the membranes being broken into small fragments. 6 to 8-week-old female Balb/c mice were injected subcutaneously with approximately 100 μg of the cell ghosts emulsified in Freunds complete adjuvant, then boosted subcutaneously 3 weeks later with 100 μg of the immunogen in Freunds incomplete adjuvant. After a further 10 days, the mice were injected with a further 100 μg of immunogen suspended in PBS (10mM-KH ₂/Na ₂HPO ₄, 30mM-KCl, 150mM-NaCl). Three days later the spleen cells were fused with the mouse myeloma line. Fusion, selection and growth of hybrids were carried out as described by Galfre et al. (1977) except that the fused suspension was subdivided into five 96-well plates to ensure the presence of only one hybridoma clone per well.

The hybridomas were screened for activity against whole-vegetative (0h)- and whole-slug (16h)-stage cells which had been immobilized onto the bottom of a Nunc immunoplate at a density of 10 ⁵ cells per well. For assay, 50 μl of culture supernatant was added to both the Oh and 16h containing wells and incubated for 2h. Bound antibody was detected by the unlabelled antibody enzyme method of Stemberger et al. (1970) using rabbit anti-mouse immunoglobulins and mouse horseradish peroxidase-anti-horseradish peroxidase complexes (Polysciences Inc.). Hybridomas that showed a positive reaction with the 16h cells and/or the Oh cells were selected for further analysis. Confirmation of reactivity to the cell surface was obtained using immunofluorescent staining and visualization by a Leitz Dialux 20 UV microscope. Hybridoma supernatants were added to whole D. discoideum cells fixed on slides by methanol for 1 min 15 s at 4°C, followed by rabbit anti-mouse Ig (DAKO) and finally rhodamine-conjugated swine anti-rabbit Ig (DAKO).

Hybrids were cloned twice by limiting dilution and then ascites fluid was prepared in pristane primed mice by intraperitoneal injection of 10 ⁵ hybrid cells. The ascites fluid was collected and EDTA and sodium azide were added before the fluid was stored at –20°C.

Samples (50 μg) of cells or cell ghosts at various times of development were immobilized onto ELISA plates. For the periodic acid degradation, half the samples were treated with 10 mM-periodic acid in 100 mM-acetate buffer (pH 4 ·5) for lh in the dark at 22 °C, while the other half were treated with the acetate buffer alone (Davis & Wheldrake, 1986). For the Endoglycosidase F (Endo F) treatment, the membrane ghosts from 16 h (slug stage) were exposed overnight to the enzyme Endoglycosidase F from Flavobacterium meningosepticum, (Boehringer Mannheim) (6 units diluted in 5 ml PBS with m10mM-EDTA, 0 ·2% Triton X-100 and 1% mercaptoethanol added). An equal amount of cell ghosts were exposed to the buffer solution alone for the same period. The amount of the epitope present was determined by allowing the samples to adhere to a Nunc immunoplate, after which an ELISA test was carried out, using the appropriate mouse anti-Dictyostelium antibodies followed by exposure of the samples to alkaline-phosphatase-conjugated second antibodies. The effects of endogenous alkaline phosphatase activity (normally found to be low under these conditions) was eliminated by subtracting the value of a blank lacking primary antibody.

Cell samples (50 μ g) in 100 μ l of PBS were applied to the wells of a Biorad Dot Blot apparatus and the fluid allowed to soak through the membrane for 30 min before a vacuum was applied to complete the application. After rinsing the well twice with PBS the membrane was removed from the apparatus and soaked for at least an hour in blocking solution (136 g low fat milk powder, 9g NaCl, 1 · 21 g Tris base and deionized water to 1 litre. The NaCl and Tris base is dissolved in 300 ml of water and adjusted to pH 7 · 5 with concentrated HC1 prior to addition of the milk powder and readjusted after milk powder addition with HC1 or NaOH as necessary). Prior to exposure to the primary (anti-Dictyostelium) antibody, the blot was rinsed once in blot wash (50mM-Tris base, 250 mM-NaCl, 0-05 % Tween 20, pH 7 · 5). Strips of dots were cut from the 96-well pattern (8 × 12) and incubated in a primary antibody solution (determined as in excess by titration) for at least 2h, followed by six 5 min rinses in blot wash. (All antibody incubations were done in PBS mixed with blocking solution in the ratio 9:1). The strips were then exposed to the second antibody (rabbit anti-mouse Ig, at a dilution of 1:1000) for 90 min and washed six times for 5 min with blot wash. Exposure to the iodinated third antibody (donkey anti-rabbit at a dilution of 1:1000) was for at least 90min and was followed by six 5 min rinses in blot wash. The membrane was allowed to dry, cut apart and individual dots placed in scintillation vials with 5 ml scintillation fluid for counting of radioactivity. Scintillation fluid contained 0 · 5 litres 2-methoxyethanol, 1 · 5 litres scintillation grade toluene and 10 g butyl-PBD.

One-dimensional polyacrylamide gel electrophoresis was carried out according to the method of Laemmli (1970). For immunostaining, proteins were electrophoretically transferred at 25 V for 2 h from the gels onto nitrocellulose sheets by the method of Towbin et al. (1979), except that 0 · 1 % SDS was included in the transfer buffer. Immunostaining was carried out using rabbit anti-mouse immunoglobulins (DAKO) and ¹²⁵I-labelled donkey anti-rabbit immunoglobulins (Amersham, PLC). Bound antibody was visualized by autoradiography using intensifying screens (Laskey & Mills, 1977) or was quantified using an LKB model 1211 scintillation counter.

The method used was as described above for the dot blot technique except that after the cell samples were applied to the membrane and subsequent blocking, one half of the samples were treated with alkaline phosphatase (Sigma) for 1 h at 22 °C and the other half was subjected only to the buffer solution (10mM-Tris – HCl, ImM-MgCl, pH9 · 6). The amount of antigen binding to specific antibodies was then determined as normal. The activity of the alkaline phosphatase enzyme was checked by serial dilution of the enzyme and assay of activity using nitrotetrazolium blue and 5-bromo-4-chloroindoxyl phosphate as chromogenic substrates. The enzyme solution showed activity down to 1/5000 dilution of the concentration used to treat the samples.

10 μl of ascites fluid mixed with 15 μ l of PBS was added to an ELISA plate well and allowed to attach to the plate for 2h. After this time, the wells were filled with 0 · 2 % gelatin in PBS and the plates were stored at 4°C overnight. The wells were rinsed three times with PBS and then 501 of a 25|Ugml ^_1 solution of horse radish peroxidase (Boehringer Mannheim) was added and allowed to react for 2 h at 20°C. The wells were then rinsed three times with PBS and one time with PBS/ 0 · 2% tween 20. 100/d of the substrate solution (Img 3,3’,5,5’-tetramethyl-benzidine and 151 30% H ₂O ₂ in 10ml PO ₄/citrate buffer pH 6 · 0) were added per well and, after 10 min at 20°C, 50 fd of 2N-H ₂SC>4 was added to stop the reaction and convert the product from its blue form to its yellow form. The absorbance at 450 nm of each well was then determined using a Titretek Multiscan ELISA plate reader.

After injecting mice with cell surface membranes from 16 h developing slugs of D. discoideum strain JC4, a series of hybridomas were isolated that produced supernatants reactive in ELISA tests to whole cells immobilized on plastic surfaces. After cloning, the monoclonal antibodies produced by these hybridomas were assayed for their binding to 16h and Oh cells and were confirmed to bind to the cell surface by microscopic examination after indirect immunofluorescence staining with rhodamine-conjugated second antibody. After further recloning and checking for stability, 20 hybridomas (designated CAB1-CAB20) were retained for further study.

Although anti-protein monoclonal antibodies have previously been produced by injection of Dictyostelium proteins into mice (e.g. JABI, Wallace et al. 1984), it was found that the overwhelming majority of the antibodies that were isolated by injection of cell surface Dictyostelium membranes (including all of CAB1 -20) were directed against oligosaccharides epitopes. This is presumably due to their highly antigenic nature and their exposed position on the cell surface. The anticarbohydrate nature of the antibodies was shown by their binding to many bands on Western blots of cell surface glycoproteins and by their sensitivity to periodate treatment (Fig. 1) and to endoglycosidase F (Fig. 2). Compared to the protein antigen recognized by JABI (ST310), the antigens recognized by the CAB antibodies were very sensitive to periodate and glycosidase treatment. Even under the optimal conditions of treatment used, however, the antigens were not completely destroyed. Such incomplete effects of periodate have been noted previously for other carbohydrate antigens (Feizi & Childs, 1987). It is noteworthy that CAB7 antigen was distinct in being resistant to glycosidase F treatment. Henderson (1984) and Ziska & Henderson (1988) have previously noted the presence in Dictyostelium of glycoproteins that are resistant to such endoglycosidases and found that after removal of fucose from their structures they all became sensitive. It is likely, therefore, that CAB7 recognizes a fucose-containing epitope, although this possibility was not directly confirmed.

When the ability of the CAB antibodies to bind to cells that had developed for various times on filters was assayed using the dot blot method, it was found that different patterns could be identified. These have been classified as types A to E (Table 1). Type A showed a constant increase with development from aggregation through to the formation of the multicellular slug (Fig. 3). Type B showed the reverse of this with a steady decline over the period of aggregation. (Checks with bacterial extracts showed that this was not due to the digestion of residual bacteria). Type C showed a constant amount of antigen throughout the period of aggregation and development. Examples of this type were relatively common but only 4 examples were included in the CAB series studied in detail. They are less interesting than other types but represent useful controls. Type D showed a peak at one particular time (in the case of CAB7 at 10 h). The peaks of the others in this group were at 12 h (CAB3) and 14 h (CAB 16). Type E was potentially the most interesting as members of this group showed dramatic changes during the period (12 to 16 h) when it is known that cell pattern is established and positioning (or sorting out) of the two cell types (prespore and prestalk cells) is observed.

Members of this group showed variable binding at Oh but always a steeply increasing curve from 12 to 14 h. Three examples are shown in Fig. 4 with differing Oh binding. (The binding at 16 h is probably near the peak although the experiment becomes technically unsatisfactory using cells after 16 h of development due to increasing problems of dissociation of the aggregates, particularly as stalk cells start to form).

When the time course study of type E was repeated using Western blot analysis, similar results were obtained as with the dot blot technique but additionally the time course of binding to the multiple bands was revealed. Examples are shown in Figure 5. The dramatic rise at 14 to 16 h is very noticeable and in the case of CAB14 no binding before this time could be discerned even with long exposure times of the blots on the X-ray film. Examination of the 16h lane of the three Western blots (Fig. 5D) indicates a strong similarity for all three antibodies, possibly indicating that at this time the same glycoproteins are recognized by the three antibodies, although CAB4 and CAB 13 can recognize only some of these glycoproteins at earlier times. From this evidence, the pattern of changes could be due to new antigens being introduced into the cell surface membranes or to processing of the glycoproteins already present.

As one of the known processing events that occurs during Dictyostelium development between 12 and 16 h is a reduction in charge associated with the cell surface (Crandall, Amatayakul and Newell, unpublished), it seemed worthwhile treating the cell surface antigens of cells harvested before 16 h (between 4 and 10 h) with alkaline phosphatase to remove some of the phosphate charges and then assessing their binding to the type E antibodies. The results of this analysis (Fig. 6) reveal that while binding to antibody types B, C and D were decreased by the treatment, the binding of the type E antibodies was increased. The data suggest that while removal of phosphate groups may hinder binding of antibody types B,C and D, their removal may expose the epitopes recognized by type E antibodies and may be a normal part of the processing of these antigens between 12 and 16h in vivo.

One approach to determining the composition of the antigenic material being recognized by the type E antibodies is to determine the binding of these antibodies to various oligosaccharides of known structure.

Such an approach, if undirected, could be lengthy and unfruitful. Fortunately, as a result of a general analysis of the types of sugar units found in the cell surface oligosaccharides undertaken as a joint venture with the Oxford Glycobiology Unit, the sugar xylose was found to be present in Dictyostelium membranes (S. Amatayakul, I.E. Crandall, R.A. Dwek, P.C. Newell, and T.W. Rademacher, unpublished data). A known xylose-mannose oligosaccharide is horse radish peroxidase (Ashford et al. 1987; McManus et al. 1988), and this is highly antigenic in rodents. We determined, therefore, whether horse radish peroxidase contained the epitope that was recognized by our anti-Dzciyostelium antibodies. The technique used the binding of the horse radish peroxidase by monoclonal antibodies, which were themselves bound to the bottom of plastic wells, and testing for the presence of the peroxidase activity by ELISA. The results (Fig. 7) showed that members of type E antibodies bound the horse radish peroxidase strongly, while the protein-recognizing antibody JABI and other antibody types bound only slightly (e.g. CAB 7) or not at all (e.g. CAB11). Crude competition experiments between horse radish peroxidase and Dictyostelium cells for binding CAB antibodies (type E) indicated that the CAB antibodies had a similar avidity for the horse radish peroxidase as they did for cell surface epitopes (unpublished data).

If the antigens recognized by our antibodies were connected with cell pattern formation, it seemed possible that they would be specific for either prespore or prestalk cells. Immunofluorescence studies using type E antibodies with sectioned 16 h slugs, however, failed to show obvious differences between the prestalk and prespore regions. To test this result more quantitatively, the dot blot assay was used to assay binding of the antibodies to cells that had been mechanically dissected from the front 1/5 of slugs (prestalk cells) or from the rear 3/5 (prespore cells). The results revealed that none of the CAB antibodies was cell-type specific. However, while the ratio of prespore/prestalk binding was in most cases close to unity, most members of type E showed a consistent and significant bias in favour of binding to prestalk cells (e.g. ratios for CAB 1 (1:1-5); CAB 14 (1:1-3); CAB 10 (1:1-4) although some members showed this only very weakly (e.g. CAB 13 (1:1-1) and CAB 4 (1:1-2).

If cell-specific antigens on the cell surface of prestalk or prespore cells were uncommon, we might reasonably have missed them with our 20 monoclonal antibodies. Prespore-specific antibodies have been reported previously such as the polyclonal antibody of Takeuchi which binds to intracellular prespore vesicles (Takeuchi, 1963) and the monoclonal antibody MUD 1, which binds to the prespore cell surface (Gregg et al. 1982). A cell-surface antibody that bound only to prestalk cells would be a useful tool and as there were no reports of its isolation, we attempted to isolate such an antibody using cell ghosts of dissected prestalk cells from 16 h slugs injected into mice.

However, even after 19 fusion experiments in which approximately 5000 hybridoma supernatants were screened by ELISA for the ability to bind specifically to prestalk cells but not to prespore cells, no positive wells were found. Only hybridoma supernatants showing quantitative differences were seen. We have since heard that Dr Gerald Weeks (University of British Columbia, Canada) simultaneously and independently carried out a similar screening programme to ours with the same results. Clearly, if a specific prestalk cell surface antigen exists it is rare or not very antigenic.

The results of our analysis with specific monoclonal antibodies indicate that the cell surface oligosaccharides of Dictyostelium change drastically during aggregation and development, with the most noticeable changes seen during the 12 to 16 h period when cell pattern formation is occurring. Previous reports have indicated that the composition of the protein component of the cell surface undergoes apparently much less radical change as development proceeds (Parish, 1983).

How cell surface glycoproteins might be involved in cell recognition or cell sorting during pattern formation in Dictyostelium is far from clear. Although a number of drastic changes to the cell surface sugars occur at the time of appearance of the two cell types, these do not appear to be cell-type specific. This implies that any role that they may have relies on the relative amounts of the oligosaccharide species present rather than on the presence or absence of particular species.

We wish to thank Ken Johnson, Frank Caddick and Mary Crandall for the figures, Jon Shatwell, Sue Amatayakul, David Ashford, Tom Rademacher and Raymond Dwek for discussions and suggestions, June Smith for technical help, and the SERC for financial assistance to PCN.