Peptide signaling during terminal differentiation of Dictyostelium

Abstract

A wide variety of mechanisms have evolved for intercellular communication in metazoans, but some of the signaling molecules were already used in their predecessors. The social amoeba, Dictyostelium discoideum, is known to use peptides to trigger sporulation within fruiting bodies, but their sequences have not been defined. We found that the peptide signal spore differentiation factor 2 (SDF-2) is processed from acyl-CoA binding protein, AcbA. The mammalian homolog of AcbA is processed to diazepam binding inhibitor that binds to the GABA_A receptor in the brain and to peripheral 1,4 benzodiazepine receptors. Although Dictyostelium has neither GABA_A nor peripheral-type benzodiazepine receptors, we find that both a diazepam binding inhibitor peptide and diazepam (Valium) can mimic SDF-2 in a Dictyostelium bioassay. Mutants lacking AcbA sporulate well only when developed in chimeras with WT cells. Using a yeast system we show that ligand binding to the SDF-2 receptor histidine kinase, DhkA, inhibits phosphorelay, which can account for its ability to induce rapid sporulation.

Terminal differentiation of prespore cells in the fruiting bodies of Dictyostelium discoideum has to be carefully regulated so as to allow them to climb the extending stalk before encapsulating into nonmotile spores. Premature encapsulation results in spores at the base of the stalk where they cannot benefit from the advantages of height for dispersal. Expression of the spore coat gene, spiA, starts in cells next to the prestalk cells at the top of fruiting bodies and passes as a wave through the underlying mass of prespore cells, suggesting that a signal emanates from the prestalk cells (1). By use of an in vivo assay where isolated cells expressing partially constitutive PKA activity are allowed to develop in buffer containing cAMP, rapid sporulation was found to depend on addition of peptides isolated from developed Dictyostelium cells (2, 3). The mixture of peptides could be separated into cationic and anionic fractions, spore differentiation factor (SDF) 1 and SDF-2, respectively. Biochemical characterization of SDF-2 activity showed it to be a pepsin-sensitive, trypsin-insensitive acidic peptide of 15–35 aa; however, there was insufficient material for sequence determination (3).

SDF-2 acts through a two-component system that relays phosphate to response regulators including the cAMP phosphodiesterase RegA (4–6). The receptor is the histidine kinase DhkA, which accumulates during development on the surface of both prespore and prestalk cells. Strains carrying null mutations in dhkA do not respond to SDF-2. Moreover, when these strains were transformed with a construct in which the extracellular domain of DhkA is interrupted by a myc epitope, the affinity for SDF-2 was found to be 100-fold less than that of WT DhkA (5).

Production of SDF-2 depends on the product of tagC, a transmembrane protein of the ABC family fused to a serine protease (3, 7). tagC is expressed during the aggregation stage of development but only in prestalk cells. Because signal generation depends on a prestalk process, this interaction may explain why induction starts in the prespore cells nearest the top where the prestalk cells are localized (1).

When cells of the PKA partially constitutive strain (KP) are developed in monolayers, they do not release SDF-2 until primed with low levels of SDF-2 (3). They then rapidly release a large amount of SDF-2. KP cells lacking either TagC or DhkA do not respond to priming and fail to generate any SDF-2 (3).

We found that acbA encodes an 84-aa protein that can be cleaved at trypsin sites to generate SDF-2. We have confirmed our previous observations on the effects of SDF-2 isolated from cells with synthesized peptides corresponding to the tryptic products of AcbA and directly showed that ligand binding to DhkA inhibits its ability to phosphorelay. The sequence of SDF-2 is 44% identical to the human diazepam binding inhibitor neuropeptides, which bind to the GABA_A receptor. Surprisingly, both the human neuropeptide and its agonist diazepam (Valium) were found to induce sporulation in Dictyostelium.

Materials and Methods

Strains, Bioassay, and Chemicals. Cells of WT strain AX4 were grown in HL5 medium and developed on buffer-saturated nitrocellulose filters (8). Prespore and prestalk cells were separated from dissociated slugs on Percoll gradients (7). Western analyses of AcbA were carried out on 5 × 10⁷ cells collected from the filters, washed, and lysed in 0.1% SDS. Samples were centrifuged to remove particulate matter, and protein concentration was determined by Bradford assay. Samples containing 10 μg of protein were electrophoretically separated on gradient (10–20%) polyacrylamide gels in Tris·Tricine (pH 8.6), 0.1% SDS. Proteins were transferred to poly(vinylidene difluoride) membranes that were stained with purified anti-AcbA (see below). SDF-2 bioassays were carried out with the KP strain, which overexpresses the catalytic subunit of PKA (2). After incubation for 24 h at low cell density in buffer containing 5 mM cAMP, the increase in number of spores was determined microscopically 1 h after the addition of test samples. Human octadecaneuropeptide (ODN) peptide was purchased from Calbiochem. Diazepam, flumazenil, and the protease inhibitor tosyl phenylalanyl chloromethylketone were purchased from Sigma.

Recombinant AcbA and Antibody Production. A full-length cDNA (SSG115) encoding AcbA was kindly provided by the Japanese cDNA Project (9). It was PCR-amplified by using primers that introduced a BspHI restriction site at the 5′ end and a XhoI site at the 3′ end before cloning into NcoI–XhoI sites of the bacterial expression vector pET32a (Novagen). The plasmid was transformed into Escherichia coli strain BL21 DE3, and fusion protein was obtained following the Novagen protocol for His tag purification with Talon resin (Clontech). After stringent washes of the resin, the protein was eluted and digested with 30 μg of enterokinase (Roche Diagnostics) for 24 h at 37°C to release the His-tagged thioredoxin, which was absorbed on Talon resin. The purified recombinant AcbA showed a single band of the expected size (10 kDa) after gel electrophoresis. One milligram of the recombinant protein was used to raise antibodies in rabbits. Polyclonal antibodies from 5 ml of the serum from the third bleed were purified on immobilized recombinant AcbA. Staining of Western blots with extracts from WT and an acbA^– null strain showed that the antibodies were specific to AcbA. The acbA gene was transferred to pDNeo2, where it is under the control of the actin15 regulatory region for expression in Dictyostelium (10).

Acyl-CoA Binding Assay. Binding of ¹⁴C-labeled palmitoyl-CoA (Sigma) to purified recombinant synthesized AcbA was measured after incubation for 10 min at 20°C (11). Chilled samples containing 0.2 μM protein and various concentrations of palmitoyl CoA were mixed with lipidex 1000 (Sigma H-6258) and centrifuged at 14,000 × g for 5 min at 4°C, and aliquots of the supernatant were counted in a scintillation counter. At least three assays were performed for each concentration of palmitoyl-CoA.

Yeast Transformation and Assays. DhkA clone p299.1 (5) was PCR-amplified so as to add BamHI sites on both ends of the full-length gene. After cloning in pGEMT-EASY, the DhkA coding sequence was transferred to the yeast expression vector p425TEF (12). The sln1^– strain TM182 and its parental strain SW100 (SLN1⁺) were kindly provided by Tatsuo Kakimoto (Osaka University, Osaka). Strain TM182 was transformed with either empty vector (control) or the dhkA⁺ vector, and transformants were selected on synthetic complete (SC)-galactose plates lacking leucine (13). The strains were grown overnight in SC-galactose liquid medium at 30°C and diluted into media lacking galactose (SC-glucose) with or without 10 nM SDF-2 to an OD₆₀₀ of 0.01. The OD₆₀₀ was measured after 4 days of incubation at 30°C with shaking.

Results

Disruption of acbA Affects Sporulation and SDF-2 Production. We inspected a set of developmentally regulated genes that we recently found in microarray experiments (14) and noticed that a prespore-specific gene, acbA (GenBank accession no. AY878075), encodes a protein that could be processed by trypsin to generate peptides with the physical properties of SDF-2. We used homologous recombination to disrupt acbA and showed that transformants failed to accumulate either acbA mRNA or AcbA protein by using Northern and Western blots, respectively (data not shown). Cells of the acbA^– null strains developed synchronously for the first 20 h but formed fruiting bodies with ≈10% as many viable spores as the WT (Table 1). If we developed these strains together with 10% WT cells, the number of viable spores increased almost to the WT level (Table 1). Most of these spores gave rise to cells with the acbA^– phenotype, showing that this phenotype is noncell autonomous as expected for one resulting from the lack of an intercellular signal. Moreover, no SDF-2 activity could be extracted from the fruiting bodies of acbA^– strains (Table 1). The ability to make SDF-2 and viable spores was restored by transformation with a construct in which acbA is driven by the act15 regulatory region, demonstrating that the defects in the mutant strain are all derived from the disruption of acbA (Table 1).

Table 1. Comparison of strains.

Strains	Spore viability, %	SDF-2 production*, units
AX4	80-100	104
acbA-	2-20	>0.02
90% acbA- + 10% AX4	60-100	NA
acbA-/acbAOE	80-100	5 × 103

^*SDF-2 was collected from fruiting bodies containing 10⁷ cells, purified on anion exchange resin, and assayed as described (3). NA, not applicable.

Trypsin Treatment of AcbA Generates SDF-2. acbA encodes a protein of 84 aa with a sequence that is 50% identical to the 87-aa human acyl-CoA binding protein (ACBP) (Fig. 1). This small protein binds C₁₄-C₂₂ acyl-CoA esters and is involved in membrane mobilization (15, 16). We expressed a full-length cDNA of acbA in E. coli and purified the recombinant protein. AcbA was able to bind C¹⁴-palmitoyl-CoA in a 1:1 molar ratio with an apparent K_d of 0.35 μM similar to mouse and plant homologs (11, 15). As expected, treatment of AcbA with trypsin destroyed the ability to bind to this acyl-CoA. In the in vivo sporulation assay, purified AcbA had no activity, whereas trypsin-treated AcbA rapidly induced sporulation (Fig. 2). Treatment with pepsin, either alone or after trypsin treatment, destroyed all of the activity in the bioassay. Because the activity generated by trypsin treatment of recombinant AcbA could be purified in a manner similar to that of SDF-2 isolated from cells, it appeared likely that AcbA was the precursor of SDF-2.

Fig. 1.

AcbA from Dictyostelium (Dicty AcbA) aligned to human ACBP. The neuropeptides TTN and ODN generated by proteolysis at trypsin sites of ACBP are indicated as well as the three synthetic peptides derived from Dictyostelium AcbA. Identities are given in bold.

Fig. 2.

Responses to peptides and diazepam. Rapid spore induction was determined in low-density populations of the sporogenous strain (KP) incubated in the presence of cAMP for 18 h (3). Shown are peptide 1 (▴), trypsinized AcbA (▿), peptides 2 and 3 together (▪), peptide 2 (○), peptide 3 (□), diazepam (•), and human ODN (▵). Values are the averages of three to five independent determinations with error bars showing 1 SD.

Synthetic SDF-2, Human ODN, and Diazepam Induce Sporulation. In mammals ACBP is processed to form diazepam binding inhibitor, which modulates the GABA_A receptor in the brain and affects various peripheral organs (17–19). In the brain, ACBP is processed to an anionic peptide of 36 aa (triakontatetraneuropeptide, TTN) and its C-terminal 20-aa fragment (ODN) (Fig. 1). Both of these peptides have been shown to inhibit binding of 1,4 benzodiazepines such as diazepam (Valium) to the GABA_A receptor in the CNS (18). TTN and ODN are also present in many peripheral organs, including the duodenum where they activate voltage-dependent L-type calcium channels via peripheral benzodiazepine receptors in intestinal endocrine cells, resulting in the release of the gastrointestinal hormone cholecystokinin (20, 21). The sequence of AcbA of Dictyostelium shows that trypsin could generate a peptide of 34 aa with 44% sequence identity to human TTN (Fig. 1).

We synthesized the 34-aa peptide predicted to be generated from AcbA as well as two internal peptides (Fig. 1). When these peptides were tested in the bioassay, the 34-aa peptide was maximally active at 0.02 pM, the same concentration at which trypsin-treated AcbA was active (Fig. 2). Peptide 2 and the ODN homolog, peptide 3, were maximally active only at higher concentrations, 2 and 100 pM, respectively (Fig. 2). Adding both peptides 2 and 3 together gave maximal activity at 0.1 pM. Human ODN was also active in the bioassay but only if added at 1,000-fold higher concentration than the Dictyostelium homolog. Addition of an equally acidic peptide with an unrelated sequence (EFDGEEYDIPESKGTWSKDDEE) did not induce sporulation even when added at 10 μM (data not shown).

The diazepam binding inhibitor peptides compete with diazepam for a site on the GABA_A receptor (18). Therefore, we tested whether diazepam would function in Dictyostelium. Addition of 1 nM diazepam induced rapid sporulation (Fig. 2). The diazepam antagonist, flumazenil, not only blocks the effects of diazepam when added to 10 nM but also blocks the effects of 0.1 pM SDF-2 (data not shown). The apparent affinities of diazepam and flumazenil for Dictyostelium are similar to those for GABA_A and peripheral benzodiazepine receptor (22–24). Although Dictyostelium has no receptors similar to GABA_A or peripheral benzodiazepine receptor, the recognition site for SDF-2 appears to be surprisingly similar to the 1,4 benzodiazepine receptor binding site of humans.

Ligand Binding to the SDF-2 Receptor DhkA Inhibits Its Kinase Activity. The SDF-2 receptor in Dictyostelium is a membrane-associated histidine kinase, DhkA, that relays phosphate via a histidine in RdeA to response regulator regions on several proteins (4, 5, 25, 26). Phosphorylation of the response regulator region of RegA activates its cAMP phosphodiesterase activity, resulting in a decrease in cAMP and PKA activity (3, 26–28). Because inhibition of PKA blocks the ability of SDF-2 to induce rapid phosphorylation, it is likely that ligand binding to DhkA acts through PKA (3). To directly determine the effect of peptide binding to DhkA we used the yeast system in which phosphorelay from a histidine kinase is necessary for growth in the absence of galactose induction of a suppressor protein tyrosine phosphatase (12, 13). Cells of yeast strain TM182 (sln1Δ; GAL1-PTP2) grow in media containing galactose but not in glucose media because the single-histidine kinase Sln1 of yeast is deleted. We transformed these cells with a construct carrying Dictyostelium dhkA and found that, unlike the parental strain, they would grow in glucose media. However, when we added 10 nM peptide 1, the cells were unable to grow in the absence of galactose (Fig. 3). Addition of 10 nM peptide 1 to parental (SLN1) cells had no effect on their growth in glucose medium (data not shown). It appears that DhkA functions as a constitutive kinase in yeast that is inhibited upon ligand binding.

Fig. 3.

Growth of sln1^– yeast cells. Strain TM182 (sln1^–) transformed with empty vector (open bars) or the dhkA expression vector (stippled bars) were inoculated into synthetic complete-glucose medium in the presence or absence of 10 nM SDF-2 as indicated. After incubation for 4 days at 30°C, growth was measured (OD₆₀₀) (average of at least three independent experiments). Only cells expressing dhkA grew significantly under these conditions, and their growth was inhibited by SDF-2. Results are representative of three independent experiments.

Primed Cells Process AcbA. Microarray analyses showed that acbA mRNA is present in vegetative cells, decreases during early development, and then reaccumulates late in development but only in prespore cells (14). Analyses of Western blots showed that AcbA is present at all stages of development but is highly enriched in prespore cells by the slug stage (Fig. 4). However, SDF-2 activity only appears after 20 h of development when fruiting body formation is underway (29). KP cells developed in monolayers do not release SDF-2 activity but priming with low levels of SDF-2 after 20 h of development leads to rapid release of up to a 1,000 times more SDF-2 (3). Without priming no SDF-2 can be measured even if the cells are lysed with Triton X-100, indicating that SDF-2 is generated after priming (Fig. 5A). Washing the cells before priming such that any extracellular AcbA would be removed did not affect SDF-2 production after priming. As expected, KP cells in which acbA is disrupted produced no added SDF-2 after priming (Fig. 5A). To determine whether primed KP cells can process AcbA when it is added to the extracellular buffer, we washed primed cells to remove SDF-2 and then incubated them with 1 pmol of recombinant AcbA. Unprimed cells produced no measurable SDF-2 activity, whereas primed cells produced >2 × 10⁴ units, 10 times the endogenous level (Fig. 5). In the absence of added AcbA, primed cells did not generate any additional SDF-2 (Fig. 5B). Addition of the serine protease inhibitor tosyl phenylalanyl chloromethylketone to primed cells just before addition of AcbA blocked all production of SDF-2 (Fig. 5B). Primed KP cells carrying a null mutation in acbA generated as much SDF-2 from AcbA as did primed KP cells, further demonstrating that the SDF-2 activity was generated by processing of AcbA. Production of SDF-2 depends on expression of the prestalk gene, tagC, which encodes a member of the ABC transporter family fused to a serine protease (3, 30). We found that primed KP cells carrying a null mutation in tagC were unable to process exogenously added AcbA into SDF-2 activity (Fig. 5B). tagC^– KP cells appear to develop well under monolayer conditions because they respond to added SDF-2 by rapid sporulation (3), but no SDF-2 is generated from AcbA in the absence of this protease. It is likely that priming results in making the TagC activity of prestalk cells accessible to the extracellular AcbA.

Fig. 4.

AcbA during synchronous development. (A) Axenically growing Dictyostelium cells of strain AX4 were washed and deposited on buffer-saturated filters at time 0. The cells aggregated by 12 h and formed fruiting bodies by 24 h. At the times shown, 5 × 10⁷ cells were collected and prepared for Western analyses (see Materials and Methods). Each sample had 10 μgof protein. After transfer of electrophoretically separated proteins to poly(vinylidene difluoride) membranes, blots were stained with antiserum to recombinant AcbA diluted 1:500. The band at the expected size of AcbA (10 kDa) is indicated by an arrow. (B) Western blot of purified prespore and prestalk extracts. No staining was observed on Western blots of extracts prepared from acbA^– cells (data not shown).

Fig. 5.

Priming and processing. KP cells were incubated in the presence of cAMP for 20 h and either left untreated (open bars) or primed with 0.1 pM of the synthesized SDF-2 peptide (filled bars). (A) The supernatants were collected and assayed for SDF-2 activity. Addition of 0.5% Triton X-100 to unprimed cells did not release SDF-2 into the supernatant. Primed KP cells carrying a null mutation in acbA did not release measurable SDF-2. (B) Five minutes after being primed KP cells were washed to remove SDF-2 and then incubated with or without 1 pmol of recombinant AcbA. The supernatants was assayed for SDF-2 activity after 1 h. The serine protease inhibitor tosyl phenylalanyl chloromethylketone (TPCK) was added to 10 μM just before addition of AcbA. Primed KP cells carrying a null mutation in acbA (diagonal stripes) generated SDF-2 from AcbA, whereas KP cells carrying a null mutation in tagC (vertical stripes) did not generate measurable SDF-2 from AcbA even after being primed with 10 pM SDF-2. Results are representative of three independent experiments.

Although these experiments have shown that cells overexpressing the catalytic subunit of PKA respond to SDF-2 when developed as monolayers, cells isolated from culminating fruiting bodies of WT and acbA^– null strains that do not harbor the KP construct also respond to added SDF-2 by rapid sporulation (Table 2). Cells from WT fruiting bodies also released large amounts of SDF-2 within 10 min of stimulation, but acbA^– cells did not, as expected from the behavior of their KP derivatives (Fig. 5A).

Table 2. Effect of synthetic SDF-2 on developed Dictyostelium cells.

Strains	Fold spore induction by SDF-2	SDF-2 produced upon induction
Ax4	2.9 ± 0.4	1.5 ± 0.5 104 units per 103 cells
acbA-	3 ± 0.5	<2 units per 103 cells

Cells were developed on filter supports for 20 h. Culminating fruiting bodies were collected in buffer, dissociated, and washed twice with buffer containing 1 mM cAMP. Cells (10⁵) were deposited in 2 ml of buffer containing 1 mM cAMP in each well of a six-well plate with or without 10 pM synthetic SDF-2. After 10 min an aliquot of the supernatant was collected for SDF-2 activity determinations. No SDF-2 (<2 units per 10³ cells) was detected in the supernatant of cells incubated in the absence of SDF-2. After 1 h spores were counted in a microscope, and the ratio between the level observed with SDF-2-treated cells and untreated cells was calculated.

Discussion

acbA mRNA is present in vegetative cells but is rapidly turned over during early development before reaccumulating during late development in prespore cells but not in prestalk cells (14). The protein product, AcbA, follows the same time course and is found only in prespore cells at the slug stage. Strains in which acbA is disrupted grow and develop well but form significantly less viable spores. Spore formation can be rescued by developing mutant cells together with a small number of WT cells, indicating that they are receiving a signal they are unable to generate themselves. Spore formation can also be rescued in the mutant cells by introducing a construct in which the actin 15 regulatory region is used to drive acbA expression. Because this regulatory region is active in both cell types, preferential expression in prespore cells does not seem to be critical. Transformation with the act15::acbA construct not only rescued sporulation but also SDF-2 production.

We purified bacterially expressed AcbA and showed that the full-length protein was inactive in the bioassay for SDF-2 but that a trypsin digest of AcbA rapidly induced sporulation at 0.02 pM (Fig. 2). A synthetic 34-aa peptide corresponding to the sequence of AcbA flanked by trypsin sites was found to be equally active. When added together, peptides 2 and 3, which correspond to the two halves of the 34-aa peptide, were active at slightly higher concentrations. When added separately, they were only active when added at levels 10- to 100-fold higher. It appears that the 34-aa peptide corresponds to natural SDF-2 but that ligand binding is not seriously compromised by cleavage at the central trypsin site. Moreover, we found that anti-AcbA antibodies added to the bioassay blocked spore induction by any of the synthetic peptides or SDF-2 purified from fruiting bodies (data not shown).

Ligand binding appears to inhibit the kinase activity of the SDF-2 receptor DhkA because the synthetic peptide blocks growth of a yeast strain in which the single-histidine kinase Sln1 is replaced by DhkA (Fig. 3). In WT yeast Sln1 relays phosphate to the small H2 protein Ypd1 except after a shift to high osmolarity (30). Cells fail to grow in the absence of Sln1 unless they express a galactose-induced suppressor phosphatase. Growth of this strain in glucose absolutely depends on phosphorelay to the small H2 protein Ypd1 (12, 13). Because expression of DhkA in this strain allows growth in glucose, it appears that DhkA can phosphorelay to Ypd1. When DhkA binds its ligand SDF-2, it is no longer able to allow growth in glucose media because its kinase activity is inhibited.

During culmination in Dictyostelium SDF-2 binds to DhkA present on both prespore and prestalk. Reduced phosphorelay to the internal phosphodiesterase RegA via the small H2 protein RdeA results in a decrease in phosphodiesterase activity, leading to accumulation of cAMP and resulting activation of PKA (3, 25–28). The increase in PKA activity leads to the rapid processing of AcbA into the SDF-2 peptide and rapid sporulation. During culmination acbA is expressed in prespore cells, whereas tagC is expressed in prestalk cells (7). It appears that AcbA is released from prespore cells and proteolytically cleaved by prestalk cells to generate SDF-2, which then triggers rapid sporulation of prespore cells (Fig. 6). Initially SDF-2 would be formed at the interface of prespore and prestalk cells in culminating fruiting bodies.

Fig. 6.

Proposed signaling pathway leading to sporulation. AcbA is released from prespore cells and proteolytically cleaved by TagC on prestalk cells to generate the SDF-2 peptide. SDF-2 binds to the histidine kinase receptor DhkA present on the surface of both prespore and prestalk cells and inhibits phosphorelay via RdeA to the cAMP phosphodiesterase RegA. The activity of RegA is reduced when its response regulator region is not phosphorylated, leading to an increased accumulation of cAMP generated by the late adenylyl cyclase ACR. Encapsulation is triggered by the resulting increase in PKA activity (3, 5). A similar pathway in prestalk cells results in extracellular TagC activity (3, 30).

The surprising finding that peptides generated from a conserved precursor protein, which are functional in mammalians, are also functional in Dictyostelium suggests that, even though the peptide receptors are unrelated, other aspects of this signaling system may have been conserved. In the brain the precursor of TTN and ODN (ACBP) is found predominantly in astrocytes (31, 32) yet the neuropeptides act on the GABA_A receptors that are exclusively found on neurons. Likewise, peripheral 1,4 benzodiazepine receptors are present in sensory neurons after injury, whereas diazepam binding inhibitor is found in surrounding satellite cells (33). Furthermore, TTN stimulates migration and chemotaxis of neutrophils by a mechanism that is independent of peripheral benzodiazepine receptor (34). Perhaps these peptides mediate a similar interplay among cell types.