A Single Cell Density-sensing Factor Stimulates Distinct Signal Transduction Pathways through Two Different Receptors*

In Dictyostelium discoideum, cell density is monitored by levels of a secreted protein, conditioned medium factor (CMF). CMFR1 is a putative CMF receptor necessary for CMF-induced G protein-independent accumulation of the SP70 prespore protein but not for CMF-induced G protein-dependent inositol 1,4,5-trisphosphate production. Using recombinant fragments of CMF, we find that stimulation of the inositol 1,4,5-trisphosphate pathway requires amino acids 170–180, whereas SP70 accumulation does not, corroborating a two-receptor model. Cells lacking CMFR1 do not aggregate, due to the lack of expression of several important early developmentally regulated genes, including gp80. Although many aspects of early developmental cAMP-stimulated signal transduction are mediated by CMF, CMFR1 is not essential for cAMP-stimulated cAMP and cGMP production or Ca2+ uptake, suggesting the involvement of a second CMF receptor. Exogenous application of antibodies against either the region between a first and second or a second and third possible transmembrane domain of CMFR1 induces SP70 accumulation. Antibody- and CMF-induced gene expression can be inhibited by recombinant CMFR1 corresponding to the region between the first and third potential transmembrane domains, indicating that this region is extracellular and probably contains the CMF binding site. These observations support a model where a one- or two-transmembrane CMFR1 regulates gene expression and a G protein-coupled CMF receptor mediates cAR1 signal transduction.

Little is known about how specific cell types within a tissue population are able to sense their cell density (see Refs. 1-4 for review). The initiation of development in the simple eukaryote Dictyostelium discoideum provides a model system for this type of cell density sensing. Dictyostelium normally live as individual amoeboid cells that eat bacteria on soil surfaces (5,6). The cells proliferate by fission and starve after overgrowing their food source. Starving cells secrete a diffusible 80-kDa glycoprotein called conditioned medium factor (CMF) 1 (7)(8)(9)(10)(11)(12), and as the number of starving cells in the population increases, the extracellular concentration of CMF increases. When there is a high percentage of starved cells, as indicated to the cells by a high concentration of CMF, they aggregate using relayed pulses of cAMP as a chemoattractant. The aggregated cells then form a fruiting body consisting of a mass of spore cells supported by a thin stalk. Because CMF can diffuse into the surrounding environment, the concentration of CMF is directly related to the density of the cells secreting it (11). Considering starving cells as a subpopulation of the cells, CMF is thus a model for cell density sensing (13).
Starved cells either at low cell densities or lacking CMF do not aggregate, but they can be rescued by the addition of recombinant CMF, suggesting that the function of CMF is to permit cAMP-mediated aggregation only when there is a high density of starving cells (9,14). Pulses of cAMP are sensed by cAR1 receptors (15,16), which activate a heterotrimeric G protein complex, causing G␣ 2 to release GDP and bind GTP (17)(18)(19). Disassociated G␤␥ then activates adenylyl cyclase, whereas G␣ 2 -GTP induces an activation of guanylyl cyclase (15, 20 -33). A cAMP-stimulated Ca 2ϩ influx is mediated partly by a G protein-dependent pathway and partly by a G proteinindependent pathway (34 -36). Exposure of cells to CMF is required for cAMP activation of both cyclases and Ca 2ϩ influx (12).
CMF modulates cAMP signal transduction by regulating the lifetime of the G␣ 2 -GTP conformation (12,14,37). GTP␥S partially inhibits the binding of CMF to membranes, suggesting that some of the CMF signal transduction pathway involves a G protein-coupled receptor. Cells lacking G␣1 do not exhibit either GTP␥S inhibition of CMF binding or CMF regulation of cAMP signal transduction, suggesting that a putative CMF receptor interacts with G␣ 1 (14). We have found that CMFinduced G␣ 1 /␤␥ dissociation activates phospholipase C, which in turn inhibits a G␣ 2 GTPase, thereby prolonging the cAMPactivated G␣ 2 -GTP configuration and promoting the cAMP signal transduction process (14,37).
In addition to regulating cAMP signal transduction, CMF is important for the expression of early developmentally regulated genes including prestalk and prespore genes (7-9, 38, 39). Expression of these genes requires cells to be exposed to both CMF and cAMP and can occur in cells lacking G␣ 1 , G␣ 2 , G␤, phospholipase C, or the cytosolic regulator of adenylyl cyclase (14). This suggested that, unlike the regulation of phospholipase C by CMF, the induction of prestalk and prespore gene expression by the combination of CMF and cAMP is G protein-independent.
Scatchard plots of CMF binding to whole cells yielded a straight line, indicating a single class of binding kinetics (10). To determine whether there was one type of CMF receptor activating both the G-dependent and G-independent CMF signal transduction pathways (40), we used affinity chromatography to isolate membrane proteins that bind CMF. We identified a 50-kDa protein, CMFR1, which is sensitive to trypsin treatment of whole cells. We obtained partial amino acid sequence of CMFR1 and isolated the cDNA encoding it. The derived amino acid sequence has no significant similarity to any known receptor, although there is some similarity to thiamine biosynthesis enzymes and monooxygenases. CMFR1 has two or possibly three predicted transmembrane domains. Expression of CMFR1 in insect cells caused an increase in CMF binding, whereas disruption of cmfr1 in Dictyostelium by homologous recombination resulted in the loss of ϳ50% of CMF binding and all of its associated G-independent signal transduction. Although the cmfr1 Ϫ cells do not aggregate, the G proteindependent CMF signal transduction pathway for IP 3 was functional in cmfr1 Ϫ cells, suggesting that cells sense the densitysensing factor CMF using two or more different receptors and that CMFR1 regulates some unknown mechanism necessary for aggregation (40). In this report, we identify and characterize distinct properties of CMFR1 with respect to signal transduction and receptor membrane topography and show that a likely reason cmfr1 Ϫ cells do not aggregate is that CMFR1 regulates the expression of adhesion proteins required for aggregation.

EXPERIMENTAL PROCEDURES
Cell Culture-Vegetative Dictyostelium Ax2 wild-type parental and cmfr1 Ϫ cells were grown in suspension shaking culture with HL-5, antibiotics, and vitamins as described previously (41). To obtain starved cells, midlog phase cells (2-5 ϫ 10 6 cells/ml) were harvested by centrifugation at 500 ϫ g for 6 min, resuspended in either PB (3 mM Na 2 HPO 4 , 7 mM KH 2 PO 4 , pH 6.5) or PBM (20 mM KH 2 PO 4 , 0.01 mM CaCl 2 , 1 mM MgCl 2 , pH 6.1, with KOH), collected by centrifugation again, and resuspended at 1-10 ϫ 10 6 cells/ml at 21°C. Cells were then shaken as for growth conditions for 3-6 h. Conditioned starvation medium was made following Brock and Gomer (41). Low cell density assays for CMF and cAMP-induced expression of the SP70 marker were carried out using immunofluorescence of 2500 cells/well seeded in eight-well slides as described previously (42).
CMF and cAMP Binding-CMF binding was measured as described previously (10,40). For cAMP binding, starved cells were washed three times in cold PB and resuspended in this buffer to 2 ϫ 10 7 cells/ml. The binding of [ 3 H]cAMP was then assayed in ammonium sulfate as described previously (43).
Expression of Recombinant CMF Active Site Fragments-A PCR was performed using the primers 5Ј-GCCATATGCGTCCACTCAATTG-GAAA-3Ј and 5Ј-CGGGATCCTCAAGAAGTTATTGGAATGAG-3Ј using the Advantage 2 PCR kit (CLONTECH, Palo Alto, CA) and Dictyostelium genomic DNA as the template. A product from 519 to 780 bp of the CMF gene encoding the region from amino acid (aa) 101 to 188 (fragment 1, F1) was obtained (Fig. 1). Similar reactions were performed to obtain fragments encoding aa 101-180 (F2), 120 -188 (F3), 110 -188 (F6), and 110 -180 (F7). All of these DNA fragments contained a NdeI site at the 5Ј-end and a BamHI site at the 3Ј-end. The DNA fragments were digested and ligated into the NdeI and BamHI sites of pET15b (Novagen, Madison, WI). After confirming insert sequences (Lone Star Laboratories, Inc., Houston, TX), the expression constructs were transformed into Escherichia coli BL21(DE3)pLysS (Novagen), and recombinant protein was expressed following the Novagen prokaryotic expression instructions. The expressed CMF fragments were purified with a B-PER His 6 Spin Purification Kit (Pierce). To generate the shorter fragments F4 (aa 120 -180), F5 (aa 120 -170), F8 (aa 130 -180), F9 (aa 130 -170), F10 (aa 120 -165), F11 (aa 120 -159), and F12 (aa 120 -153), similar PCRs were done except that 5Ј NdeI and 3Ј XhoI cloning sites were added, and PCR products were digested and ligated into the expression/secretion vector pET-22b (Novagen). CMF peptides were expressed in E. coli BL21(DE3)pLysS cells following the same protocol as above. The growth medium containing the secreted proteins was clarified by centrifugation, and 49.42 g of ammonium sulfate was added per 100 ml of clarified growth medium at 0°C. Precipitated CMF protein fragments were collected by centrifugation at 20,000 ϫ g for 15 min, resuspended in 20 mM Tris-HCl, pH 7.5, and then purified using the B-PER His 6 spin purification kit. Potential O-linked glycosylation sites were searched for using the DictyOGlyc 1.1 server (available on the World Wide Web at www.cbs.dtu.dk/services/DictyOGlyc/) (44).

Expression of Recombinant CMFR1 Outer Loop Protein-A YES
vegetative Dictyostelium cDNA library was obtained from Dr. Eugeno De Hostos, and PCR was performed using Pfu polymerase (Stratagene, La Jolla, CA) with the primers 5Ј-GGAATTCCATATGAGAGAAGGTA-GAAAAGTTGC and 5Ј-CCGCTCGAGTTCGTGACCGAATTTAGCC containing the 5Ј NdeI and 3Ј XhoI cloning sites. A 999-bp DNA fragment was gel-purified using GeneClean III (Qbiogene, Carlsbad, CA), ligated into pET22b (Novagen), and cloned using the One Shot E. coli transformation system (Invitrogen). The resulting C-terminal His tag expression vector encoding aa 101-429 of CMFR1 was then used to transform BL21-DE3 cells (Novagen), and the transformants were induced in Terrific Broth containing 1 mM isopropyl-1-thio-␤-D-galactopyranoside for 3 h at 37°C (45). Cells were extracted with B-PER, and the 37.5-kDa detergent-insoluble protein was solubilized in 8 M urea and purified on a nickel column following the manufacturer's protocol (Novagen). Urea was removed, and the protein was renatured by dialyzing 3 ml of protein in 500 ml of PBM with four changes over 48 h at 4°C. IP 3 , cAMP, and cGMP Production and Ca 2ϩ Uptake-The production of intracellular IP 3 in response to a 30-s pulse of CMF was determined following the procedure of Van Haastert (46) using an IP 3 3 H assay system (Amersham Biosciences) to quantitate IP 3 with the exception that 100-l aliquots of neutralized cell supernatants were used. The production of cAMP in response to a pulse of the functional analog 2Ј-deoxy-cAMP was determined following Van Haastert (46). Cells starved at 10 7 cells/ml for 6 h were collected by centrifugation and resuspended at 10 7 cells/ml. These were stimulated with 10 M 2Јdeoxy-cAMP in the presence of 10 mM dithiothreitol. At 0, 3, and 5 min after stimulation, the cells were lysed, and cAMP was quantitated using an isotope dilution assay kit (Amersham Biosciences). The production of cGMP in response to a 10-s pulse of 0.1 M cAMP was determined in a similar manner following the procedure of Kesbeke et al. (47), using a cGMP assay kit (Amersham Biosciences). Cellular Ca 2ϩ uptake in response to a pulse of cAMP was determined as described by Milne and Devreotes (48) with the following modifications. Cells were starved for 6 h at a density of 5 ϫ 10 6 cells/ml and were resuspended to a concentration of 10 7 cells/ml in HK buffer (20 mM Hepes, 5 mM KCl, pH 7.0). The cell suspension was incubated in uptake buffer for 30 s, stimulated with a 10 M pulse of cAMP, and then incubated for an additional 40 s. The cells were then collected by centrifugation at 1400 ϫ g for 30 s and washed with 1 ml of ice-cold HK buffer containing 10 mM Ca 2ϩ , and radioactivity was measured by liquid scintillation counting.
Antibody Purification and Western Blots-The generation of antibodies 672 (Ab1) and 673 (Ab2) against the CMFR1 epitopes used for Western blots and immunofluorescence has been described previously (40). IgG was enriched from antiserum using ammonium sulfate (49) and used for Western blots and immunofluorescence. The concentration of total IgG in antibody preparations Ab1 (2 mg/ml) and Ab2 (3.5 mg/ml) was determined by SDS-polyacrylamide gel electrophoresis comparing the antibody preparation to a series of known concentrations of purified rabbit IgG (R & D Systems, Inc., Minneapolis, MN). Fab fragments of CMFR1 antibodies were generated and isolated from ammonium sulfate-fractionated IgG using an ImmunoPure Fab Preparation Kit (Pierce) following the manufacturer's directions. The concentration of Fab fragments (0.4 mg/ml for both Ab1 and Ab2) was determined using the Bio-Rad protein assay and purified rabbit IgG as a standard.
Immunofluorescence Microscopy-Approximately 8 ϫ 10 4 cells in 200 l were placed in individual wells of an eight-well glass slide and were allowed to attach for 1-6 h in HL5 or PBM and then were fixed with 2% formalin, 0.2% glutaraldehyde, 0.02% Triton X-100 in PBM for 5 min. Alternately, cells were fixed with 50% ethanol, 10% formalin, 6.5% acetic acid, 0.4% picric acid for 20 min. Slides were washed in PBM followed by 15-min washes in 1 mg/ml sodium borohydride at 4°C and then PBS containing 0.1% SDS and 0.5% Nonidet P-40. Cells were then incubated for 1 h with Ab1 or Ab2 at a 1:75 dilution in the PBS/SDS/ Nonidet P-40 buffer, washed with the same buffer for 30 min, and washed in PBS containing 0.05% Nonidet P-40 for 10 min. Slides were then incubated with Alexa Fluor 488 goat anti-rabbit antibody (Molecular Probes, Inc., Eugene, OR) at a 1:300 dilution in PBS-N for 45 min. Following two 30-min washes, the slides were mounted as previously described (42). Cells were examined with a Nikon Microphot Fx with a 1.4 NA 60ϫ lens, a Deltavision (Applied Precision, Issaquah, WA) deconvolution microscope with a Zeiss 1.4 NA 100ϫ lens, or a Zeiss LSM-410 confocal microscope.
RNA Analysis-RNA was isolated from 5 ϫ 10 7 cells either at the vegetative stage or at various times during starvation in shaking culture with or without pulses of cAMP, essentially as previously described (40), using the RNeasy mini kit (Qiagen Inc., Valencia, CA). Electrophoresis of RNA (10 g/lane) and transfer to Duralon UV membrane (Stratagene) were done following the manufacturer's directions. Equal loading was verified by ethidium bromide staining of rRNA in the gels. The cDNA probes for cAR1 (15), gp80 (50), discoidin (51), and phosphodiesterase (PDE) (52) were labeled with [ 32 P]dCTP by random hexamerprimed DNA synthesis (Amersham Biosciences) and hybridized with the blots in a 60°C reaction containing 0.25 M Na 2 HPO 4 , 0.25 M NaCl, 0.5% SDS, 1 mM EDTA, 10% polyethylene glycol (M r 8000), pH 7.2. Blots were then washed with 50 mM Na 2 HPO 4 , 0.5% SDS, pH 7.2, for 30 min at 25°C and 15 min at 60°C. Autoradiography on preflashed Kodak X-Omat AR5 film was done at Ϫ70°C.
SP70 Expression in Response to Glutathione S-Transferase-CMF-Sepharose-Approximately 10 4 glutathione-Sepharose 4B beads (Amersham Biosciences) were incubated with 1.6 g of recombinant glutathione S-transferase-CMF, prepared as described (10), for 30 min at 4°C in 0.2 ml of PBS. The beads were then washed five times with 1-ml volumes of PBS, two times with PBM, and finally resuspended in 0.2 ml of PBM. To assay SP70 expression, Ax2 cells were starved in eight-well glass slides at 10, 7.5, and 5 ϫ 10 3 cells per well, and after 6 h, either ϳ1500, ϳ1000, or ϳ500 glutathione S-transferase-CMF-Sepharose beads were added to the wells together with cAMP (300 M). Cells were fixed and stained for SP70 12 h later, and positive cells were scored as described above.

RESULTS
Overlapping but Distinct Regions of CMF Activate the Two CMF Pathways-CMF stimulates both G protein-dependent and -independent signal transduction pathways (14). We previously identified a putative receptor for CMF, CMFR1, which is necessary for CMF-stimulation of G protein-independent prestalk and prespore gene expression but not G protein-dependent IP 3 production (40). This suggested, but did not prove, that there are two receptors for CMF. To test the hypothesis that there are two separate receptors for CMF, we first determined whether a difference could be detected between the two receptors with respect to CMF binding affinities by measuring the K D for 125 I-labeled CMF binding to parental and cmfr1 Ϫ cells. As shown in Table I, there was no detectable difference in the K D between the two cell types. The K D for both cell types was what we previously observed for wild-type cells (10,40), whereas the number of CMF receptors for Ax2 cells was considerably less than that reported for Ax4. This sort of difference also occurs for cAMP receptors, with Ax2 cells having considerably less cAMP binding than Ax4. 2 As previously observed (40), there are about half the number of CMF receptors on cmfr1 Ϫ cells compared with parental cells. The similar K D for CMF binding to parental and cmfr1 Ϫ cells suggests that the G protein-coupled receptor (the only expected CMF receptor in cmfr1 Ϫ cells) has the same K D as the mixture of CMFR1 and this receptor, so by this criterion we cannot demonstrate the existence of two different receptors.
We then determined whether differences in the two receptors could be detected with respect to the region of CMF that activates the receptor. Twelve different recombinant peptide fragments of the CMF active site were generated and examined for their ability to stimulate SP70 (a protein expressed in a subset of prespore cells) accumulation and IP 3 production. As shown in Fig. 1, a fragment encoding amino acids 120 -180 of CMF (fragment 4) had essentially full activity for stimulating both pathways. We previously found that the EC 50 of purified or recombinant CMF is ϳ300 pg/ml (8,9). The EC 50 of fragment 4 is considerably lower (indicating a much higher specific activity). This is in agreement with our observation that small breakdown fragments of CMF have a much higher specific activity than the entire protein (53). Deletion of amino acids 171-180 (creating fragment 5) resulted in almost complete loss of fragment-stimulated IP 3 production with little effect on SP70 accumulation. Further deletion of amino acids 166 -170 (fragment 10) abolished the ability of the CMF fragment to stimulate SP70 accumulation. No significant decrease in either IP 3 production or SP70 accumulation occurred by deleting amino acids 101-120 with the C terminus fixed at 180 or 188 ( Fig. 1). However, deleting only amino acids 101-109 decreased SP70 accumulation, indicating that having just amino acids 110 -129 rather than 101-129 or 120 -129 interferes with the CMF active site. Deletion of amino acids 120 -129 to create fragment 8 resulted in a significant decrease in both IP 3 and SP70 expression (Fig. 1). Thus, the essential stimulatory ability of the CMF resides in amino acids 120 -180, with amino acids 171-180 being more crucial for activating the IP 3 pathway, 166 -170 being necessary for gene expression, and 120 -129 being necessary for both pathways. The differential capacities of the various CMF fragments strongly suggest that distinct regions within the CMF active site operate on two different receptors and pathways.
Role of CMFR1 in cAMP-mediated Signal Transduction-We previously found that CMF regulates cAMP signal transduction via a pathway involving an unknown G protein-coupled CMF receptor, G␣1, and phospholipase C (14). These studies could not exclude the possibility that a G protein-independent CMF signal transduction pathway also regulates cAMP signal transduction. We therefore examined whether second messenger pathways, which are activated by cAMP and its G proteincoupled receptor cAR1, are altered in cmfr1 Ϫ cells. Cells were starved for 5 h at low cell density (10 6 cells/ml) prior to signal transduction assays to prevent cAMP accumulation, pulsing, and up-regulation of cAR1. This basal starvation condition was also relevant in that assays for cAMP induction of SP70 gene 2 L. Tang and R. H. Gomer, unpublished results. expression are done at low cell density. Table II shows that both wild type Ax2 and cmfr1 Ϫ cells bound similar amounts of cAMP under basal conditions. However, a dramatic difference was observed after 5 h of pulsing with cAMP. A nearly 7-fold increase in binding was seen with wild-type cells, whereas little change was found for cmfr1 Ϫ cells. This indicates that while CMFR1 may not be necessary for basal levels of cAR1, it is involved in the mechanism for cAMP-pulse-induced up-regulation of cAR1.
When wild-type and cmfr1 Ϫ cells were stimulated with 10 M 2Ј-deoxy-cAMP for 3 and 5 min, an ϳ2-fold increase was observed in cAMP production (Table III). Similarly, as shown in Table IV, stimulation of wild-type and mutant cells with 0.1 M cAMP for 10 s resulted in ϳ2-fold increases in cGMP levels. In both cell types, there also appeared to be no significant difference in basal levels of cAMP or cGMP. This suggests that CMFR1 does not regulate cAMP-stimulated cAMP or cGMP accumulation.
A pulse of cAMP also causes a transient Ca 2ϩ influx in starved Dictyostelium cells (48,54). Basal Ca 2ϩ influx was similar in Ax2 and cmfr1 Ϫ , and whereas relatively little cAMPstimulated Ca 2ϩ influx was observed for cmfr1 Ϫ , it was not significantly different compared with the 25% increase seen for wild-type cells (Table V). cAR1 levels have been shown to be directly proportional to the extent of cAMP-induced Ca 2ϩ influx (48,54). When Ax2 cells were pulsed with cAMP, a nearly 3-fold increase in cAMP-stimulated Ca 2ϩ influx was observed, whereas cmfr1 Ϫ cells showed no significant cAMP-stimulated Ca 2ϩ influx (Table V). This lack of response is consistent with the inability of cmfr1 Ϫ to up-regulate cAR1 (Table II) and indicates that, for unknown reasons, pulsing cmfr1 Ϫ cells with cAMP abrogates their cAMP-stimulated Ca 2ϩ influx. Together, the data indicate that CMFR1 is required for some but not all of the cAMP-stimulated Ca 2ϩ influx.
CMFR1 Regulates Gene Expression at the Level of mRNA Accumulation-CMF regulates the expression of early genes such as discoidin I and cAR1 as well as prestalk and prespore genes (7,8,12). We previously found that CMFR1 mediates the CMF induction of prestalk and prespore protein expression (40). To determine whether CMFR1 mediates the CMF regulation of discoidin I and cAR1, we examined their expression in cmfr1 Ϫ cells. After 3 h of starvation, cAR1 mRNA levels in cmfr1 Ϫ were approximately half that observed for wild-type cells, and although there was a slight increase after 6 h of pulses, cAR1 mRNA levels were significantly less in cmfr1 Ϫ versus wild-type (Fig. 2). The role for CMFR1 in the non-pulseinduced discoidin I gene as well as that for the pulse-induced gp80 gene is even more dramatic, since essentially no detectable messages were observed. This impaired expression was confirmed in cmfr1 Ϫ cells by immunofluorescence and Western blots, which also showed significant attenuation of levels of the adhesion protein gp24 (data not shown). The absence of CMFR1 has a more complex effect on phosphodiesterase mRNA, although normal levels of the vegetative species were seen in cmfr1 Ϫ . Whereas normal levels of the developmental species were also observed at 3 h, it was pulse-down-regulated in contrast to wild-type; after 6 h, the levels were significantly reduced unless the cells were pulsed. All levels of the developmental species were significantly less than wild-type except for 3 h without pulsing. It therefore appears that CMFR1 regulates the expression of a variety of genes expressed during early development.
CMFR1 Is a Transmembrane Protein with an Extracellular Domain-We wanted to determine whether antibody binding could mimic CMF-mediated stimulation of prespore gene expression, which could subsequently also help define the membrane topography of CMFR1. A similar strategy using a Myctagged protein has been reported for the Dictyostelium transmembrane histidine kinase, DhkA (55). Thus, antibody Ab1 was raised against an amino acid sequence of CMFR1 that resides between the first potential transmembrane domain and the second potential transmembrane domain, whereas Ab2 was raised against a region between the latter domain and the third potential transmembrane domain. When cells were incubated with either Ab1 or Ab2, SP70 accumulation was induced over a range of dilutions, whereas the respective preimmune preparations were ineffective (Table VI). Whereas the dilution profile of stimulation was similar for both antibodies, Ab1 induced a higher overall degree of expression. Since both antibodies were stimulatory when applied extracellularly, it would appear that both of their binding sites are extracellular, and thus the second possible transmembrane domain is not a true transmembrane domain.
To determine whether antibody activation of CMFR1 occurs through the receptor dimerizing by the divalent IgG binding or direct stimulation of the receptor, Fab fragments (which have a single antigen binding site) were prepared. An even greater degree of SP70 accumulation was observed for both Fab1 and Fab2 compared with the IgG (Table VII). The efficacy of Fabinduced SP70 accumulation indicates that monovalent binding to either CMFR1 antigen site can substitute for CMF-mediated stimulation. We also found that when CMF was bound to beads 5-17-fold larger than Dictyostelium cells, the CMF activated SP70 accumulation (data not shown). Together, the data strongly suggest that CMF does not need to be internalized to activate CMFR1 and that the Ab1-binding and the Ab2-binding domains of CMFR1 are both extracellular.
To further test the hypothesis that the regions of CMFR1 that are recognized by Ab1 and Ab2 are extracellular, a recombinant protein containing the sequence between the first and the third potential transmembrane domains was generated. Since this protein contains the binding sites for Ab1 and Ab2 as well as the possible CMF binding site, it should competitively inhibit both antibody-and CMF-induced gene expression. As shown in Table VII, when cells were incubated with Fab1 or Fab2 that previously was incubated for 40 min with recombinant outer loop protein at a 1:3 molar ratio, a significant reduction in Fab-induced SP70 accumulation was observed. Similarly, CMF-induced SP70 accumulation was negatively affected by the presence of outer loop protein (Table VIII). At a 1:100 molar ratio of CMF to outer loop, SP70 accumulation was not observed, and it was reduced by 50% at a 1:1 ratio; complete CMF induction did occur, however, at a 1:0.1 ratio or less (Table VIII). The ability of the CMFR1 outer loop protein to inhibit both antibody-and CMF-induced SP70 accumulation supports a one-or two-span transmembrane model for CMFR1 and strongly suggests that the extracellular CMF binding site resides between the first and third potential transmembrane domains. Cellular Localization of CMFR1-To further examine the cellular localization of CMFR1, wild-type and cmfr1 Ϫ cells were stained with anti-CMFR1 antibodies using immunofluorescence. As shown in Fig. 3A, there was staining of Ax2 cells, whereas no staining was observed in cmfr1 Ϫ cells (Fig. 3B). When we examined the staining pattern more closely using deconvolution microscopy, the fluorescence was associated with the periphery of cells, with some punctile staining in the interior of cells (Fig. 3C). There was no discernible difference in the staining observed with Ab1 versus Ab2 (data not shown).

DISCUSSION
One Ligand Activates Two Different Receptors-We previously observed that disrupting cmfr1 abolishes some but not all aspects of CMF signal transduction, suggesting that there are two different CMF receptors. In this report, we show that there is a difference in the domains of CMF needed to activate the G␣ 1 -mediated and the CMFR1-mediated CMF signal transduction pathways. This strongly supports the hypothesis that there are indeed two different receptors for CMF. Although there exist nonfunctional decoy receptors that bind ligands such as Fas ligand and interleukin-1 (56,57), there are very few known examples of one ligand activating two different functional receptors on a cell. These include two distinct domains of thrombin activating different pathways and cellular responses through separate receptors on fibroblasts (58) and the CD80 and CD86 ligands both activating CD28 and CTLA-4 receptors on T cells (59).
We previously found that the CMF active site lies within an TABLE III Production of cAMP in response to cAMP stimulation in Ax2 and cmfr1 Ϫ cells Cells were starved for 5 h, harvested, extensively washed, and resuspended in PB. For each experiment, cAMP production was measured in duplicate at 0, 3, and 5 min after stimulation with 2Ј -deoxy-cAMP. Values are the means Ϯ S.E. from at least three separate experiments; the ratio of the average of the 3-and 5-min values to the 0-min value was calculated separately for each experiment, and the mean Ϯ S.E. is shown as the -fold increase. The -fold increases are not significantly different at p Ͼ 0.4.    Ca 2ϩ uptake by Ax2 and cmfr1 Ϫ cells in response to cAMP stimulation Cells were starved for 6 h at a density of 5 ϫ 10 6 cells/ml without or with pulses of cAMP every 6 min. The cells were then harvested, extensively washed, and incubated in uptake buffer for 30 s followed by an additional 40-s incubation either without or with 10 M cAMP. The amount of 45 Ca 2ϩ taken up by cells was determined in duplicate, and the values are the means Ϯ S.E. from at least three separate experiments. The ratio of the cAMP-stimulated value to no cAMP value was calculated separately for each experiment, and the mean Ϯ S.E. is shown as -fold increase. A paired t test indicated that the -fold increase in Ca 2ϩ uptake is not significantly different between unpulsed Ax2 cells and unpulsed cmfr1 Ϫ cells at p ϭ 0.25.  2. Early gene expression is decreased in cmfr1 ؊ cells. Total RNA (10 g) prepared from wild-type (WT) Ax2 (left panels) and cmfr1 Ϫ (right panels) cells at the vegetative stage (v), and 3-and 6-h intervals of starvation with or without cAMP pulses (p), was resolved by 1.2% agarose/formaldehyde gel electrophoresis and blotted to nylon membranes. Blots were probed using cAR1, discoidin, gp80, or PDE cDNA fragments. The gel was stained with ethidium bromide to verify that equal amounts of rRNA were present in each lane of samples.

TABLE VI
Antibodies against CMFR1 induce SP70 accumulation Ax2 cells were starved at low cell density in monolayer culture on glass slides in the presence of various dilutions of anti-CMFR1 Ab1, Ab2, or preimmune antibodies. Six hours after starvation, cAMP was added, and SP70 accumulation was assayed 12 h later. Ϫ, less than 1% of the cells expressing the marker; ϩ/Ϫ, 1-25% of the cells expressing the marker; ϩ, 25-45% of the cells expressing the marker. Identical results were obtained in three separate experiments. For a positive reference, the complete recombinant CMF protein was used at 1 ng/ml and scored ϩ; no SP70 expression was seen when cAMP was not added.
88-amino acid region of the 570-amino acid CMF polypeptide backbone (10). We found here that the active site can be narrowed to 60 amino acids (amino acids 120 -180) with a predicted mass of 6532 Da. We previously found that cleavage of sites between 120 and 180 with trypsin (53), chymotrypsin, and cyanogen bromide 3 destroys CMF activity. Furthermore, a series of ϳ30-mer peptides beginning at aa 101, 115, 130, 145, and 160 showed no CMF activity, 4 suggesting that the active site of CMF spans the entire 60 aa between 120 and 180. Our delineation of the CMF active site stimulating CMFR1 and the receptor activating the G␣ 1 pathway indicates that both receptors are activated by essentially the same region of CMF.
Assuming that there are only two types of CMF receptors, the observation that the K D for CMF binding to cmfr1 Ϫ cells is 2.1 nM (Table I) suggests that this is the K D of the G proteincoupled CMF receptor. We observed that wild-type cells have approximately twice the number of binding sites for CMF compared with cmfr1 Ϫ cells, suggesting that the increased amount of binding is due to the presence of CMFR1. Since Scatchard plots showed an apparent single class of CMF binding sites on cells (10), this suggests that CMF binds to CMFR1 with a K D of 2.1 nM. This then implies that at any given CMF concentration both CMF receptors have roughly the same percentage of occupancy. We have observed that for both the CMF-stimulated production of IP 3 and accumulation of SP70 and CP2, the maximal stimulation occurs at roughly 1 ng/ml (13 pM) CMF (10,14). Ax4 cells contain roughly 40,000 CMF receptors (10). However, we find that Ax2 cells appear to have only about 660 CMF receptors per cell. It is unclear why there is such a large difference between the number of receptors per Ax2 and Ax4 cells.
CMFR1 and G Protein-independent Signal Transduction-cAR1-activated G␣2-GTP and G␤␥ stimulate cGMP and cAMP production, respectively (60). CMF regulates cAMP and cGMP production by regulating the lifetime of the cAR1-activated G␣ 2 -GTP via G␣1, G␤, and phospholipase C (10 -12, 14, 37, 40). CMFR1 is not required for cAR1 activation of cAMP and cGMP accumulation under conditions where cAR1 levels are not upregulated by pulses of cAMP (Tables III and IV). Although we find that CMFR1 plays an important role in cAMP-induced pulse up-regulation of cAMP binding (Table II) and gene expression (Fig. 2), CMFR1 does not appear to regulate any pathway downstream of cAR1.
Cells lacking CMF also do not exhibit cAMP-stimulated Ca 2ϩ uptake but the response can be rescued by a 10-s exposure of these cells to CMF (12). Although the cAMP-stimulated Ca 2ϩ uptake has been reported to be independent of G␣ subunits 1, 2, 3, 4, 7, or 8 and the G␤ subunit (48) and is directly proportional to cAR1 levels (48,54), Ca 2ϩ influx has also been found to be reduced by 50 -70% in G␣ 2 -and G␤-negative strains (35). Since the G protein-coupled CMF receptor seems to function by activating G␣ 1 and G␤ (14), this receptor could be involved with cAMP-stimulated Ca 2ϩ uptake. There is evidence that CMF could affect Ca 2ϩ uptake via fast and slow pathways. CMF rapidly affects the cAMP binding site kinetics of cAR1 (43) and can also down-regulate cAMP binding both in wild-type and in cmfr1 Ϫ cells (40,43). Thus, simply the binding of CMF to its G protein-coupled receptor could rapidly cause a permissive conformation of cAR1 for Ca 2ϩ uptake. CMF could also affect cAMP-stimulated Ca 2ϩ uptake more slowly via CMFR1, since decreased levels of cAR1 mRNA observed in cmfr1 Ϫ cells (Fig.  2) correlate with the lack of cAMP pulse-induced increases in cAMP binding (Table II)  Fab fragments prepared from Ab1 and Ab2 IgG digests were preincubated with or without recombinant CMFR1 at a 1:3 molar ratio for 40 min at 21°C. Ax2 cells were then starved at low cell density in multiwell slides in the presence of various dilutions of the Fab/rCMFR1 mixture and assayed for SP70 accumulation as described in Table VI. The assay results were identical in three separate experiments. For a positive reference, recombinant CMF at 1 ng/ml was used and scored ϩ.
CMFR1 appears to play an important role in the regulation of several genes involved in the growth differentiation transition, including those for the cell surface adhesion proteins gp24 and gp80. Indeed, the most dramatic characteristic of cells lacking CMFR1 is the aggregation-minus phenotype. Whereas gp24 is involved with the initial formation of filopodia-mediated intercellular contacts and aggregation, gp80 plays a later role in cellular streaming (61). Antibodies against gp24 (62) and gp80 (63) block aggregation or cause abnormal streaming, thereby arresting further development. Thus, the inability of cmfr1 Ϫ cells to aggregate into mounds and streams can be explained by significant impairment in the expression of these genes (Fig. 2). In contrast to almost undetectable gp80 and discoidin messages, expression of both cAR1 and PDE genes is somewhat attenuated (Fig. 2). Aside from pathways such as Ca 2ϩ influx, which is directly proportional to cAR1 levels, there do not appear to be significant effects of reduced cAR1 and PDE levels, since the cAMP-mediated signal transduction pathways examined appear normal, and pulses of cAMP can still down-and up-regulate PDE messages at 3 and 6 h of starvation, respectively. In addition, it is unlikely that CMFR1 affects expression of CMF, since the levels of CMF secreted by cmfr1 Ϫ cells are ϳ75% of those observed for Ax2 wild-type cells. 5 Membrane Association and Topology of CMFR1-Immunofluorescence showing that some CMFR1 is near the periphery of the cell, combined with the observation that when CMF is attached to beads that are much larger than the cell this bead-CMF complex can stimulate CMFR1-activated gene expression, suggests that some CMFR1 is on the plasma membrane. An examination of the predicted amino acid sequence of CMFR1 indicated that there were three potential transmembrane (PTM) domains (40). Our observation that antibodies directed against a domain between PTM1 and PTM2 when added to live cells induce SP70 expression suggests that this region is extracellular. Similarly, the ability of antibodies against a domain between PTM2 and PTM3 to induce SP70 expression indicates that this region is also extracellular. This then implies that the potential TM2 is not a TM region and suggests that TM1 and/or TM3 are true TM domains. The ability of the Fab fragments of the antibodies to stimulate SP70 expression (Table VII) suggests that the antibodies were directly stimulating the receptor rather than artificially causing receptors to dimerize. CMFR1 could be a one-span type I receptor such as the bacterial chemotaxis receptor for aspartate, the human growth hormone receptor, or the human asialoglycoprotein receptor (64,65). However, topology with TM1 as the single TM domain would leave only the ϳ23 N-terminal amino acids of CMFR1 in the cytosol; if a single TM region was TM3 there would be the ϳ60 C-terminal amino acids in the cytosol. On the other hand, if CMFR1 is a two-span receptor with both the N and C termini inside the cell, it would have a topography similar to bacterial and Dictyostelium histidine kinase receptors (55,66,67).
In summary, the antibody-and bead-coupled CMF stimulations as well as immunofluorescence observations suggest that some CMFR1 is located on the extracellular side of the plasma membrane, consistent with it being a receptor. The lack of CMFR1 does not affect CMF-mediated cAMP-stimulated cAMP or cGMP production, but it does affect cAMP-regulated gene expression. Taken together with the observation that different domains of CMF activate the G protein-dependent and G protein-independent CMF pathways, it appears that CMFR1 is one of at least two different receptors for the cell densitysensing factor CMF.