A secreted 80 × 10³M_r protein mediates sensing of cell density and the onset of development in Dictyostelium

In the presence of a food source, Dictyostelium discoideum cells grow vegetatively as unicellular amoebae. Upon starvation, up to 10⁵ amoebae aggregate together using relayed pulses of cAMP as the chemoattractant. The aggregate then elongates and forms into a migrating slug or pseudoplasmodium, which then differentiates into a fruiting body consisting of a ∼ 2 mm high stalk with a mass of spore cells on top of it. Approximately 20 % of the cells are stalk cells and the remainder spore cells (Devreotes, 1989; Firtel et al. 1989).

Precursors to spores and stalk cells (prespore and prestalk cells, respectively) can first be identified in the aggregate stages using monoclonal antibodies against extracellular proteins, antibodies made against bacterially expressed fusion proteins, or promoter/reporter gene fusions (Krefft et al. 1984; Datta et al. 1986; Gomer et al. 1986b; Jermyn et al. 1989; Williams et al. 1989; Haberstroh and Firtel, 1990; Esch and Firtel,1991). Prestalk cells are ‘randomly’ scattered within the forming aggregate and constitute ∼ 15 % of the cells while prespore cells can first be seen as an annulus in the middle of the aggregate 1-2 h later as the aggregate is forming a tip. The prestalk cells then sort to the anterior region, thus forming the tip (Williams et al. 1989; Esch and Firtel, 1991 or unpublished data). In addition, some of the prestalk-specific gene products are also expressed in basal cells. These include prestalk genes inducible by cAMP and those inducible by DIF, a factor necessary for stalk cell differentiation (see Morris et al. 1988). The pattern of expression of the prespore genes and both classes of prestalk genes has been characterized at the level of transcription (see Hjorth et al. 1989, 1990; Williams et al. 1989; Haberstroh and Firtel, 1990).

The induction of cAMP-inducible prestalk and prespore gene expression in wild-type NC-4 cells and axenic strains such as KAx-3, derived from NC4, requires distinct developmental events regulated by at least two extracellular factors. One is cAMP, which 270 R. H. Gomer, I. S. Yuen and R. A. Firtel induces the expression of both classes of genes via signal transduction pathways acting through cell surface cAMP receptors (Schaap and van Driel, 1985; Gomer et al. 1986a; Oyama and Blumberg, 1986; Haribabu and Dottin, 1986). A continuous level of at least 30 UM cAMP after 4 to 6 h of starvation is necessary for maximal gene expression (Mehdy and Firtel, 1985; Gomer et al. 1986a), while cAMP added earlier blocks this expression, presumably by inhibiting the expression of genes that are expressed earlier and are required for establishing the necessary signal transduction systems (Mann and Firtel, 1987; Mann et al. 1987, 1988). This 4 to 6 h of starvation required for NC-4 derived strains is not a requirement for V12M2, an independent wildtype isolate (Gross et al. 1981).

The second extracellular factor necessary for prespore and prestalk gene expression is conditioned medium factor (CMF). Mehdy and Firtel (1985) found that, in submerged monolayer culture, prespore and prestalk gene expression can be induced in NC4 or KAx-3 cells at high cell densities (∼ 10⁵ cells cm^-2) by extracellular cAMP, but not in cells at low cell densities (< (∼ 5 × 10³ cells cm^-2). However, when cells at low densities were plated in buffer previously conditioned by starved cells developing at a high density [conditioned medium (CM)], the low density cells could then express prestalk and prespore genes in response to extracellular cAMP. CMF also appeared to be required for the expression of genes normally induced in aggregation and during the interphase period between starvation and the onset of aggregation. For prestalk and prespore gene expression, CMF needed only to be present for 2–6 h and could be removed prior to the addition of cAMP. This suggested that the factor functioned during the earliest stages of development and was no longer essential once cells were competent to be induced by cAMP (Mehdy and Firtel, 1985; Gomer et al. 1985; Mann and Firtel, 1989). These results also suggested that Dictyostelium cells use a factor present in conditioned medium to monitor cell density and that the entry into development and the expression of selected genes is dependent upon the spatial density of cells. In this report, we fractionate and characterize the active components of CMF. We show that there are two substances secreted by Dictyostelium cells, both of which have CMF activity, and that one is an 80 × 10³M_r protein, which we have purified.

The Dictyostelium axenic strain KAx-3, grown in HL-5 axemc growth medium, was used for all experiments Cells were grown in axenic shaking culture in HL-5 (Firtel and Bonner, 1972). Trypsin-TPCK, bovine serum albumin (BSA), activated beef heart phosphodiesterase, soybean trypsin inhibitor, sodium dodecyl sulphate (SDS), dithiothreitol (DTT), and cAMP were purchased from Sigma, St Louis, MO.

Conditioned medium was made from starving, developing cells similarly to previously described methods (Mehdy and Firtel, 1985) Log phase KAx-3 cells were collected by centrifugation at 200g for 5 min, and resuspended and washed in PBM (0.02M potassium phosphate, 10⁻² mM CaCl₂, and 1mM MgCl₂, pH 6 1) twice. The cells were then resuspended in PBM at a density of 9×10⁶ml^-1 and shaken at 110revs min^-1 for 20 h at room temperature. CM was also occasionally made with cells at 5×10⁶ml^-1. The conditioned medium was then clarified by centrifugation as described above followed by centrifugation at 8000 g for 15 min at 4 °C.

KAx-3 cells were washed as described above and resuspended in PBM at a density of 2 × 10⁶ cells ml^-1. Two of cell suspension was mixed with 200/d of 1.10 dilution of CM in PBM (positive control), PBM (negative control), or fractions to be tested for CM activity. This was placed in a well of Lab-Tek 8-well glass slide (Miles Scientific, Naperville, IL) and incubated at room temperature Six hours after plating the cells, cAMP was added to each well to 300 μ M final concentration. Eighteen hours after plating, the supernatants in the wells were gently aspirated and 200 μ l of room temperature 70 % ethanol was added to each well for 15 min. The slides were then air dried for 1 h. Immunofluorescence was done as previously described using the anti-prespore or anti-prestalk antibodies against spore coat protein SP70 and the prestalk protein pst-cathepsin (CP2) that were used for previous studies (Gomer and Firtel, 1987; Gomer et al. 1986i>). The ratio of positively stained cells to total cells was determined by counting as described (Gomer et al 1986b). For the discoidin CM assay, log phase cells at a density of 2 × 10^s cells ml^-1 were used; no cAMP was added and the cells were fixed at 7h after starvation. Rabbit polyclonal anti-discoidm I antiserum (a gift of Dr Wayne Springer, UCSD) was fractionated by ammonium sulfate precipitation and used at a 1 · 100 dilution for immunofluorescence.

To develop a quantitative assay for CMF and to determine whether there was a heterogeneity in the response of starved cells to conditioned medium, the percentage of cells differentiating into prespore cells was determined At least 3000 cells were counted per assay point Initially, individual column fractions were assayed with at least two dilutions. Fractions containing CM activity were then pooled. Serial factor of 2 dilutions of the pooled fractions from each purification step were then assayed. We define maximum response to occur at the dilution where highest percentage of cells differentiated in that assay. Following Gherardi et al. (1989), the units of activity are then defined as the inverse of the highest dilution of the pooled material that showed 50 % of maximum response. The only variability that we observed in the assay is the absolute number of cells that differentiate. The maximum number of isolated cells in low density cultures that differentiate into prespore cells is approximately 38 % [in migrating slugs, approximately 47% of the cells stain with anti-SP70 antibody (prespore cells)] (Gomer and Firtel, 1987), we observe a maximum in separate experiments of anywhere from 8 to 38 %, similar to what we have observed previously. The reason why more cells differentiate into prespore cells in migrating slugs compared to low density culture is unknown The variation of the maximal number of cells that differentiate in low density culture appears to be due to variations in culture conditions that we and others (M. Mehdy and R. Firtel, unpublished data) have not been able to control It is important to note that the dilution of CMF at which this maximum or half-maximum number of cells differentiates is not affected by this variability Interestingly, comparisons of assays in which a maximum of 8 % of the cells differentiated with assays in which 38% of the cells differentiated shows that there is no difference in the strong fluorescence of the positive cells in either assay, or of the low fluorescence of negative cells.

Phosphodiesterase activity was measured following Rutherford and Brown (1983) 

SDS-polyacrylamide gel electrophoresis and Coomassie staining were done as previously described (Gomer et al 1986a) Gels were 1.5 mm thick and contained 15% acrylamide. Morrissey’s procedure (Morrissey, 1981) for silver staining was done with the following modification: Gels were fixed for 1h in 50% methanol/10% acetic acid, washed in several changes of water for 1h, and then soaked in DTT, omitting the glutaraldehyde step.

Whole CM was concentrated with a minitan pressure dialyzer and 10x10³M_r cutoff polysulfone membrane (Cat. no PTGC, Millipore, Bedford, MA). The concentrated material (M> _r10× 10³) was sealed in a 12 –14 ×10³M_T cutoff Spectrapor dialysis bag (Spectrum Medical Industrial, Los Angeles, CA) and dialyzed against 2 liters of PBM for 12 h at 4°C. Preswollen DEAE-Sephacel (Pharmacia, Uppsala, Sweden) was prepared by washing, equilibrating, and removing fines in PBM The concentrated and dialyzed CM was applied to a 2.8 × 12.5 cm DEAE column and then eluted from the column stepwise with one bed volume (80 ml) each of 0, 0 1, 0 2, 0 3, 0.4, 0.5 M NaCl in PBM at 4°C. Bio-gel HI P hydroxylapatite (Biorad Laboratories, Richmond, CA) was equilibrated and fines were removed in PBM. Columns (3 ml bed volume, 2.8 × 0.5 cm) were run at room temperature. DEAE fractions containing CM activity were pooled and loaded on the hydroxylapatite column. The column was eluted stepwise with 5 ml each of 20, 50, 100, 150, 200, 250, 300, 400 m_M KPO₄ pH 6.1 containing 1mM MgCl₂ and 10⁻²mM CaC12. The fractions were tested for CM activity The CM-containmg hydroxylapatite fractions were pooled and sealed in a Spectrapor dialysis bag (cutoff 12 –14 ×10³M_r) which was placed in a bed of and coated with Aquacide II (Calbiochem, La Jolla, CA) at 4°C The Aquacide-concentrated materials were mixed without boiling with modified Laemmh sample buffer (10 μM DTT instead of 10 mM) prior to electrophoresis. After protein separation, the gel was evenly divided into 10 slices from top to bottom, crushed, and protein eluted passively into PBM at 4°C with constant mixing for four hours CM activity was tested for each slice. In order to determine further the molecular weight of CMF-H, a similar slice containing CM activity was further divided into 1 mm thick pieces and passively eluted into PBM. All protein concentrations were measured with the Biorad protein assay (Bradford, 1976)

Size fractionation and molecular weight determination was initially done using sieving gel chromatography A 25 cm high × 0.7cm diameter column of Pharmacia superfine Sephadex G-50 was used. One ml of CM was loaded on the column, which was then eluted with PBM at room temperature Fractions eluting from the column were assayed for CM activity. Molecular weight standards (blue dextran, 2000 × 10³M_r, BSA, 66× 10³M_r, lysozyme, 14.3 × 10³M_r,; bromophenol blue, 670 M_r) were loaded on the column before and after sizing the CMF. HPLC sieving gel chromatography was done with a Biorad Biosil TSK-125 300 × 7.5 mm column. The column was prewashed with degassed water and then PBM at 1 ml min . The column was loaded and run in PBM at 0.4 ml min^-1 at room temperature. The column outflow was monitored at 210 nm and 0.4 ml fractions were collected Biorex-70 (Biorad) was eluted as above in PBM with varying concentrations of NaCl. Mono Q, Mono S, and Superóse 12 columns (Pharmacia) were run on a Pharmacia FPLC at 4°C and eluted with NaCl gradients in PBM. CMF was loaded on Phenyl-sepharose (Sigma) in 2 M NaCl in PBM and did not bind to the column. Zinc-chelate (Boehringer-Mannheim, Indianapolis, IN) was activated with ZnCl₂ before use. Heparin-agarose was from Sigma and Affi-gel blue was from Biorad.

Wheat germ agglutin (WGA)-conjugated agarose (2 mg WGA ml¹ resin) was from Sigma A 0.2ml bed volume column was poured in a 1 ml plastic syringe plugged with glass wool. Columns were washed with PBS, PBS/1M NaCl, PBS/0.25M N-acetyl glucosamine (Sigma), and finally PBS Samples were loaded on the column and reloaded twice. The column was then eluted as washed, omitting the last PBS step; the column was additionally eluted with PBS/0 25 M N-acetyl glucosamine (Sigma) after sitting overnight in that solution All operations were at 4°C. Aliquots of eluates were passed over G-50 spin columns before being assayed for CMF activity. One ml of Concanavalin-A (ConA)-conjugated Sepharose (12 mg ConA ml^-1 resin, Sigma) in a 0.6 mm diameter column was washed as above, using 20 % methyl a- D-mannopyranoside (Sigma) as the sugar, and similarly loaded and eluted.

To assay for discoidin I gene induction, early log phase cells (5 × 10⁵ml^-1) growing in shaking culture in HL5 were supplemented with CMF or PBM alone After 3h, cells were harvested, and RNA was isolated and sized on denaturing gels, and analyzed for discoidin I RNA levels by hybridizing blots with a nick-translated discoidin I cDNA probe as previously described (Mehdy and Firtel, 1985)

Results

As previously described, Dictyostelium cells in low density submerged monolayer culture require a factor (CMF) secreted by high-density starved cells in order to initiate differentiation (see Fig. 1) (Mehdy and Firtel, 1985; Gomer and Firtel, 1987). As shown in this assay, dilute, starved cells are not capable of inducing prespore and prestalk gene expression (as determined by immunofluorescence using polyclonal antibodies against a prestalk and a prespore protein) in response to cAMP, unless cells are first cultured in the presence of conditioned medium from developing cells (see Materials and methods).

To determine the approximate size of CMF, whole conditioned medium was fractionated on a Sephadex G-50 column. As shown in Fig. 2, multiple peaks of CMF activity are observed. The high M_r peak contains two subpeaks. The second subpeak (fractions 11 and 12) is variable in appearance from experiment to experiment and thus may be a degradation product of the first peak (see below). The large peak has an approximate M_T of >66 × 10³. A second set of peaks (fractions 27 – 40) are composed of molecules<14 × 10³M_T. The CM activity in each of the peaks is sufficient to allow the induction of both prespore and prestalk antigen expression (data not shown). Similar results were obtained using Sephadex G-100 (data not shown).

The high M_r CMF activity (G-50 fractions 8 – 12) has been designated CMF-H while the lower M_T material has been categorized together as CMF-L.

To verify that the low M_r material was not a retention artifact, 20 ml of unfractionated CM were dialyzed against 1 liter of water at 4°C overnight, using Spectrapor dialysis bag (12 – 14 × 10³M_r cutoff). The liter of water, now containing material that passed through the dialysis membrane, was lyophilized and resuspended to a final volume of 20 ml with water. This material had CMF activity at dilutions of up to 1:5. Conversely, the material remaining in the dialysis bag after 5 changes of PBM over 3 days also had CMF activity.

To test whether CMF-H is a protein, the activity was examined for trypsin sensitivity. Conditioned medium contains 16 μ g ml^-1 of protein, as determined by the Biorad protein assay. When CM is sized by electrophoresis on a SDS-polyacrylamide gel, a large number of protein bands can be seen by Coomassie or silver staining of SDS-polyacrylamide gels (see Fig. 3, lane 2). In order to have a marker for protein proteolysis, we added BSA to CM. To our surprise, 50 μ I ml^-1 BSA was not degraded by 1mg ml^-1 trypsin at 37°C for 2h, as determined by SDS-PAGE (data not shown). This suggested the presence of an inhibitor of trypsin in crude conditioned medium. To separate the inhibitor from CMF-H, CMF-H was purified by ultrafiltration, DEAE chromatography, and hydroxylapatite chromatography (see below). When BSA and trypsin were added to the hydroxylapatite-purified CMF-H, we found that the BSA could now be proteolysed. As shown in Table 1, we found that trypsin addition also eliminated the activity of CMF-H. The high M_r CMF activity was not eliminated if the trypsin was boiled or if trypsin inhibitor was added to the trypsin before incubating with CMF-H.

We also examined the heat sensitivity of unfractionated CM (containing CMF-L and CMF-H) and the material retained after extensive dialysis (CMF-H). Both fractions were heat stable and lost <10% of their CM activity after incubation at 80°C for 25min.

From preliminary studies, it was found that an approximately 1:10 dilution of this whole conditioned medium induced the highest percentage of cells to differentiate (Fig. 4A). In order to separate the CMF-H from CMF-L, minitan ultrafiltration was used to retain and concentrate materials larger than 10 × 10³. Dialysis against PBM was performed using Spectrapor (cutoff 12 – 14 × 10³M_r) to further ensure no retention of CMF-L. The concentrating and dialysis step did not detectably affect the protein composition (Fig. 3, lane 3). This material was then loaded on a DEAE-Sephacel column and most of the CMF activity was eluted by 200mM NaCl. The DEAE ion exchange chromatography purified CMF-H 12-fold and probably removes an inhibitor of the CMF since the total units of activity were increased almost 2-fold. Fractions containing CMF activity were pooled and then chromatographed on a hydroxylapatite column. Assays were done on several fractions eluted from the hydroxylapatite by each phosphate concentration. The amount of potassium phosphate required to elute CMF-H varied from 150mM to 250mM. The hydroxylapatite column effeclively purified the CMF an additional 15-fold as judged by the increase in specific activity (Table 2). Subsequently, the fractions containing peak activity were pooled, concentrated with Aquacide, and electrophor-esed on SDS-polyacrylamide gels. This method has been successfully used to purify several growth factors (Matsushima et al. 1985; Singh and Bonin, 1988; Parnell et al. 1990). Initially, the gel was evenly divided into 10 slices, which were ground and the proteins contained ineach slice eluted into PBM at 4°C overnight. The polyacrylamide fragments were removed by centrifugation at 15 000 g for 10 min and the supernatants were then assayed for CMF activity using dilutions of 10³ to 10⁴ (see Table 2). There were two peaks of activity, one having a AL between 70 and 97 × 10³ and the other between 50 × 10³ and 70 × 10³. A second gel was then run and the region between the 45 and 97 × 10³M_r protein marker bands was divided into 1mm thick slices and similarly eluted and the eluates then tested for CMF activity. The two CMF activities were located to protein bands of 80 × 10³M_r and of 65 × 10³M_r and resulted in a total recovery of 32 % of the activity (Table 2). When the supernatant eluted overnight or the acrylamide gels fragments from the 80 × 10³M_r band prepared as described above were sized by SDS-PAGE, we observed a 65 × 10³M_r band but no 80 × 10³M_r band (data not shown). Elution of the proteins from crushed gel slices for 15 or 30 min instead of overnight (see Fig. 3, lanes 6 and 7) shows some 80 × 10³M_r protein, especially with the shorter elution time. Re-electrophoresis of the crushed acrylamide immediately after excising the 80x10³M_r band shows even a greater percentage of 80 × 10³M_r material and less 65 × 10³protein (Fig. 3, lane 8). Thus, under the conditions of elution, the 80 × 10³M_r CMF protein appears to be degraded to a 65 × 10³M_r protein that still retains CM activity. The 65 × 10³M_r band was present in all purification steps. Presumably, other proteins present in cruder fractions inhibit the degradation of the 80 × 10³M_r protein. These results suggest that the 80 × 10³M_r band isolated from the SDS-PAGE gel slice is the native CMF-H activity.

To determine the lowest amount of CMF required to effectively induce cell differentiation, serial dilutions of purified CMF were assayed. As shown in Fig. 4B, the range of concentration in which the highest percentage of differentiated cells was induced is narrow. Too much or too little CMF present results in submaximal response. Dose-response curves for each purification step look similar to Fig. 4 (summarized in Table 2).

We also examined other chromatograhic properties of CMF-H. 100 μ l of the hydroxylapatite purified fraction was loaded on a HPLC Biosil TSK-125 column and the column was eluted in PBM. Fractions were collected and aliquots were assayed for CM activity and electrophoresed on SDS-polyacrylamide gels and silver stained. CM activity and the 80×10³M_r protein were both present in the same single fraction. In the elutions from DEAE, hydroxylapatite and other chromatogra-phy resins tested, the 80×10³M_r band always coeluted with the high M_T CM activity; when CM activity was in two or more fractions, the 80x10³M_r protein was similar in these fractions and the amount of the 80× 10³M_T band paralleled the amount of CM activity in the fractions. On the Biosil column, this 80 ×10³M_rband eluted shortly before BSA. Similar results, although with less resolution of the protein peaks, were obtained with a Pharmacia FPLC Superose-12 sieving gel (data not shown).

A variety of other column resins gave no further improvement in purification after DEAE and hydroxylapatite chromatography. These included Biorex-70, heparin-agarose, phenyl sepharose, affigel blue, Mono Q, Mono S, and zinc agarose. The 80 ×10³M_rCMF-H eluted in the flow through of all the resins except affigel blue, Mono Q, and Mono S.

CMF-H partially purified by DEAE chromatography was run over Concanavalin A (ConA) and Wheat Germ Agglutinin (WGA) columns. For ConA, a 1ml bed volume column was loaded with 5.3 mg of DEAE-purified CMF-H. For WGA-agarose, a 0.1ml bed volume column was loaded with 0.12 mg of DEAE-purified CMF-H. The columns were washed with high salt and eluted with the appropriate sugars (see Materials and methods). Many of the proteins found in whole conditioned medium bind to ConA, including CMF-H. Biologically active CMF-H can be eluted from ConA by 2 % methyl-α -D-mannopyranoside; no activity is detectable in the flow-through or other washes. By SDS-gel electrophoresis, silver staining and bioassays, all of the CMF-H in the hydroxylapatite fraction binds to ConA. Similar experiments with WGA-agarose showed that CMF-H does not bind to WGA. One of the proteins from whole CM that binds to WGA has a M_r similar to CMF-H, but it does not have CMF activity and does not fractionate with CMF-H on DEAE.

Discoidin 1 is a developmentally regulated lectin that is required for proper development, whose function in Dictyostelium development is analogous to fibronectin in mammalian cells. It is not expressed in low density log phase cells and is induced at ∼ 2 h of starvation and maximally expressed during aggregation (5–7 h). In addition, discoidin I is induced in growing cell populations as the density increases (Rosen et al. 1973; Rowekamp et al. 1980; Poole et al. 1981; Devine et al. 1982; Springer et al. 1984; Crowley et al. 1985). CMF was also shown to potentiate the developmental expression of discoidin I genes (Mehdy and Firtel, 1985). Aliquots of the HPLC Biosil PSK-125 fractions were diluted 1:30 and used in a CM assay in which discoidin I protein expression was detected by immunofluorescence (Fig. 5). Discoidin I expression was induced only by the material in the fraction containing the 80 ×10³M_rband.

Recent experiments by Clarke et al. (1987) have shown that discoidin I can be induced in low-density vegetative cells that are grown in growth medium that was previously conditioned by high density cells, cells which are expressing discoidin I. This group has also shown that this induction of discoidin I is in response to a secreted, non-dialyzable factor present in the growth medium of cells grown to a high density but not in medium from cells grown to a low density. In order to examine whether our CMF-H can induce discoidin I expression in growing cells, log phase KAx-3 cells in HL-5 axenic growth medium were incubated with or without Biosil-punfied 80 ×10³M_rCMF-H (see Materials and methods) and discoidin I mRNA levels were assayed by RNA blot hybridization Discoidin I mRNA levels were barely detectable in control cells (addition of PBM), were high in cells incubated with CMF, and were also high in mid/late log phase growing cells (3 – 6 × 10⁶ml^-’) (data not shown).

Although CMF-H and the factor described by Clarke et al. (1987) have similar effects on discoidin I expression and are both not dialyzable, they do not appear to be the same molecule. We have used the CMF-H purification protocol to attempt to purify CMF from HL-5 growth medium conditioned by a high density of vegetative cells. No CMF activity or 80 ×10³M_rprotein was recovered (data not shown). In addition, the factor described by Clarke et al. (1987) is heat labile, while under similar conditions, CMF-H is not (see above).

The extracellular phosphodiesterase and its inhibitor are ∼ 50 ×10³M_r proteins and are secreted by Dictyostehum cells during the aggregation stage that are essential for normal development (Toorchen and Henderson, 1979; Orlow et al. 1981; Franke and Kessin, 1986; Gerisch, 1987; Kessin, 1988). Phosphodiesterase (PDE) hydrolyzes cAMP in the vicinity of the cell, allowing cAMP receptors to desensitize. The PDE is developmentally regulated and induced by cAMP. The PDE inhibitor, PDE-I, is also developmentally regulated and rising cAMP levels inhibit PDE-I expression. The PDE-I modulates PDE activity by increasing the K_m of PDE for cAMP. Previous results by Kessin and collaborators (Faure et al. 1988, 1989; Kessin, 1988) have shown that both of these proteins are essential for regulating the early stages of development. To determine if CMF-H is either the phosphodiesterase or its inhibitor, partially purified 80 ×10³M_rCMF-H was sized by SDS-PAGE along with both proteins (a gift of J. Franke and R. Kessin, Columbia University). The 80 ×10³M_r CMF-H did not comigrate with either protein, as expected from the known M_r(s) of the PDE and its inhibitor (see above). In addition, a series of dilutions of both proteins from ∽ 0.3mgml^-1 to 3 × 10⁻⁸ mg ml^-1 or the two together could not substitute for CMF in the CMF assay. A direct assay of 21 ng of SDS-polyacrylamide gel-purified 80 ×10³M_rCMF showed no detectable phosphodiesterase activity. Serial factor of 10 dilutions of activated beef heart phosphodiesterase from 0.8 mg ml^-1 down to 0.08 ng ml^-1 could not substitute for CMF. CMF activity could also not be mimicked by a series of dilutions of the HL-5 growth medium, various concentrations of bovine serum albumin, a wide variety of salts, pH changes, or an increased or decreased partial pressure of oxygen.

Upon starvation, Dictyostelium cells initiate a developmental program that controls the sequential expression of classes of developmentally regulated genes. Previously, we showed many of the genes involved in aggregation are not induced when starved cells are plated at low densities and such low density cells do not become competent to induce prestalk and prespore genes in response to cAMP. Addition of CM to these cells allows these developmental processes to proceed (Mehdy and Firtel, 1985; Gomer et al. 1985), suggesting the presence of a secreted factor required for the entry into the multicellular phase of development.

From the size fractionation of whole conditioned medium, we showed the presence of two distinguishable molecules, both of which have conditioned medium activity (Fig. 2) and allow the differentiation of prestalk and prespore cells in response to cAMP. We do not yet know whether the CMF-L is a breakdown product of CMF-H, possibly an expected ∽ 15 × 10³M_r peptide obtained in the cleavage of 80 × 10³M_r CMF-H to the 65 × 10³M_r protein, or whether there is a physiological reason for the existence of two CMFs. If CMF-L is derived from CMF-H, we might expect that this is a specific precursor/product relationship rather than non-specific proteolysis since trypsin destroys CMF-H biological activity. One possibility is that the two CMFs probably have different diffusion coefficients and thus may act over different distances. In addition to being required for developmental competence to express prestalk and prespore genes, we show that CMF also induces discoidin I expression during development (Fig. 4). The factor identified by Clarke et al. (1987) in growth medium does not appear to be CMF based on purification properties and heat sensitivity.

CMF-H is a protein as shown by its trypsin sensitivity and purification to a single band on silver-stained SDS-polyacrylamide gel. The sensitivity of cells to as low as 0.3 ng ml^-1 of CMF-H suggests that there may be a specific cell surface receptor and signal transduction and amplication system for CMF, such as exists in other systems for growth factors. Sequencing the 80 × 10³M_r CMF will be required to determine whether it is related to any known polypeptide growth factor. The fact that the 80 × 10³M_r CMF does not bind to heparin-agarose suggests that this CMF is not closely related to heparin-binding growth factors (Deuel, 1987). Although most growth factors are smaller than 80 × 10³M_r, some, such as the 130 × 10³M_r Mullerian inhibitory substance (Cate et al. 1986) are larger.

Of biological interest are the parallels between CMF and a 17 × 10³M_r secreted polypeptide designated C-factor encoded by the bacterium Myxoccoccus xanthus (Kim and Kaiser, 1990). Like CMF, C-factor appears to be used by the developing M. xanthus cells to sense cell density at the onset of development and is required for initiation of multicellular development. Moreover, like CMF, C-factor is not active undiluted and is only active over a narrow concentration range (4fold for C-factor and 10-fold for CMF).

Our data indicate that cells do not initiate the multicellular developmental program when plated at a low cell density, as determined by the expression of preaggregation and aggregation phase genes or by the ability to induce prestalk and prespore gene expression. Dictyostelium cells will express early genes such as discoidin I or the serine esterase D2 and become competent to differentiate into prespore and prestalk cells, if, and only if, they are exposed to sufficiently high levels of a factor (CMF). CMF is made and secreted by some or all cells from starvation through late development, although maximal activity is present during the preaggregation and aggregation stages (Mehdy and Firtel, 1985). Previous results and those presented here suggest that there is an insufficient concentration of CMF when cells are developed at low densities for the cells to initiate the multicellular developmental program. When starved cells are at a higher density, the concentration of CMF in the medium surrounding the cells is then sufficient to induce cells to differentiate. Presumably, during development, a cell needs to sense whether the local cell density is sufficient to permit aggregation to occur. We believe that CMF acts as this cell-density sensor and that it induces the onset of development (see Mann and Firtel, 1989). The presence of two activities, one a discrete 80 × 10³M_r protein and the other a group of small molecules, may be important in a further refining of the function of CMF in vivo.

It is at first puzzling why cAMP could not be the factor that allows Dictyostelium cells to sense whether they are near a sufficient concentration of other cells to aggregate. However, the cAMP pulse relay mechanism requires that cAMP be rapidly degraded in the vicinity of the cell to prevent saturating the cAMP receptor; this degradation would presumably make cAMP unsuitable for use as a factor that mediates the mass sensing.

Many of the genes expressed during the preaggregation and aggregation stages encode gene products used by the cells during aggregation. These include the serine esterase D2, contact site A (a cell adhesion molecule), a cAMP receptor, and a G_α protein (G_α2) subunit that couples the cAMP receptor with phospholipase C and other downstream effectors (Gensch et al. 1975; Mann et al. 1987,1988; Klein et al. 1988; Rubino et al. 1989; Kumagai et al. 1989; Firtel et al. 1989). The genes are maximally induced by secreted pulses of cAMP that interact with the cAMP cell surface receptor and activate G protein-mediated signal transduction processes. Recent studies (Mann and Firtel, 1989) have shown that a basal level of expression, which is variable for the four genes described above, is induced upon starvation. From results of these and previous studies (Mehdy and Firtel, 1985; Gomer et al. 1985), we believe that CMF is responsible for this basal level of expression. The receptor and G_α2 are part of the signal transduction system required for chemotaxis and pulse-induced gene expression, and thus, basal expression of these genes is essential for establishing the initial phases of the signaling system. CMF would allow this initial phase to proceed when the cells were at a sufficient density to aggregate and proceed through multicellular development. Once this is achieved, CMF does not appear to be absolutely required for later development, although it is still expressed at the slug stage (Mehdy and Firtel, 1985; Gomer et al. 1985). Subsequent studies with purified CMF should help delineate its mode of action.

We wish to thank Paul Price and Matt Williams for use of and help with the HPLC, Navin Khanna and Annegrethe Hjorth for helpful discussions, and Jo Anne Powell for helpful suggestions on the manuscript The extracellular phosphodiesterase and inhibitor were a gift from Jakob Franke and Richard Kessin and the anti-discoidin I antibodies were a gift from Wayne Springer. R.H.G. was supported by an American Cancer Society (California Chapter) semor postdoctoral fellowship This research was supported by USPHS grants to R A.F. and R.H.G.